Fool for the 8656A Sig Gen

[jrharley]

Greetings EEVBlog forum - A newbie here, hope everyone is well and gettin' their nerd on. I'm a hobbyist focused on vintage audio and test equipment. Professional background is Architecture, but I've obtained a six year degree from UYT. (Thanks Dave, Paul, Tony, Clive, et al!).

I recently acquired an HP 8656A Signal Generator. Couldn't resist it for the price. Sold as a parts unit, but in pretty good shape overall. Serial number : 2146A02249, which I believe is a later model. Despite it's reputation, I'm going to give the thing a go. All 4 power rails appear to be good to go and the display lights up with basically nonsense. No response from any key. I went straight to the attenuator assembly for a quick look and the fault became obvious. The unit is missing two actuator rods and the actuator cover. The gold contacts for the actuator however, were bagged and stored in the unit. Since these parts are no longer made, my only options are 3D printing or getting lucky with an identical trashed unit that I could pull from. The thing that complicates the latter approach is that this attenuator assembly appears different than others I have seen, which have ball bearings on the forked actuator rods. Not so in my version. The ball bearings have been eliminated from the design.

So, finally getting to my question - Does anyone have a reliable source that could take a couple of the parts I have intact on other sections of the attenuator, scan them and 3D print me some duplicates? I've browsed around and it looks like the capability is out there. Tolerances and material selection are more detailed topics for discussion, but I need to start somewhere. Any suggestions?

Thanks in advance for any help! - JRH

[bdunham7]

Mechanical failure of the attenuator shouldn't cause anything other than incorrect/missing output levels.  You might want to sort all of the other problems out first before you go after that attenuator.  As far as the attenuator parts, I'm only familiar with the grommet/ball-bearing design, could you post good photos of what you have?  Is there a revision shown in a manual anywhere? 

[Stray Electron]

   Where are you located?  I think there is (was?) the same unit in a surplus place about 60 miles from me.

   FWIW I hope that you got it CHEAP!  I hate buying TE that other people have been into. There is almost always parts missing or or it's damaged beyond repair.  When it comes to Ebay or when buying in person, I always check to see if the warranty seals are unbroken on the covers. If they're broken then the value of the items goes WAY down unless they're giving me a guarantee that the TE is working.

[jrharley]

Thanks for weighing in Stray!

Yeah, I hear you. I did get it cheap, and wasn't expecting a working unit. My enjoyment comes from getting these boat anchors up and running again.
Appreciate the feedback. - JRH

[jrharley]

A very good point bd, and food for thought moving forward. I had assumed that since the missing parts left an open circuit in an essential part of the unit, that I ought to start there, but your comment makes a lot of sense. I still have a lot of studying of the intimidating schematic to do, as well as researching revisions ahead of me. I've attached a photo of the plundered attenuator cavity.

Thanks so much for weighing in. -JRH

[jrharley]

Greetings all,

Hope the summer was a good one for everyone.

I'm back to the HP8656A for more punishment, studying the circuits and theory of operation. I'm having difficulty getting any datasheet, pin-out, or other technical info on the microprocessor, U13. Does anyone have a data sheet for the Motorola 1820-2099? I'd like to do a little probing on this guy but don't feel great about doing it blind. Any resource suggestions for such info would also be appreciated.

To briefly recap, the units power supply rails all seem to be solid. Unit does not default to the standard display settings on power up. All keys are unresponsive, but all display segments and LED indicators appear to illuminate. I discovered where the previous bodger had removed a wire from connector J2 and when I jump this back into circuit, the display goes blank except for a single "1" in the amplitude domain window.

At any rate, any assistance in getting a datasheet for the Motorola would be much appreciated!

Best,
JRH

[bdunham7]

I believe that is a Motorola MC6802P plastic PDIP-40.

[bingo600]

I believe that is a Motorola MC6802P plastic PDIP-40.

The X-Ref i have says that too
1820-2099   IC   MC6802

/Bingo

[jrharley]

Thanks guys, you are both spot on from the looks of it.
For some reason, I didn't get notifications of replies to this post, so I've been busy on my own trying to find it and doing some careful examination of the boards in the meantime.

The MC6802 datasheet however, is easy to get and fills the bill. Really appreciate it!

Is there a reason that any of you can think of, why I shouldn't just put a new processor in the socket and see what happens? Yeah, I'm still tracin' and reckonin', but I'm really only a hobbyist in the end, let's face it, this beast is over my head. This is no old Heathkit signal Generator!

Right now though, I don't recognize anything in the circuit that could ruin my day by doing that. Fuse F1 had been jumped with a solid wire when I found it. I tacked a 2A cartridge fuse into that spot for testing, the original is listed as 4A.

Anyway, thanks a lot for the help and the info! I'll post up with any relevant updates moving forward.

JRH

[Tony_G]

Welcome to the world of HP Signal Generators...

Just a couple of suggestions:

• If you haven't already then I recommend that you get the Artek Manuals scan for the unit. They're way better than the scans that appear on the Keysight website - http://artekmanuals.com/manuals/hp-manuals/
• The manuals for this era of units are generally excellent. Section 8 of the manual will take you through the troubleshooting process in pretty decent detail. I'd start working through their steps rather than just trying to address issues as you find them.

Good luck with the unit, look forward to seeing how it turns out.

TonyG

[jrharley]

Thanks Tony G!

Appreciate you weighing in. I'll get the Artek manuals to supplement my file, nice to know of that site.
Onward....


JRH

[jrharley]

Greetings HP Guru's
Update on the 8656A project. Mostly, I've been learning how to read the HP schematics and trying to relate them to what I'm seeing in this beast. Kinda' slow going. Some progress though on a few fronts. I've determined that, even though it looked like someone had removed the high stability timebase option 001 upon first look, the unit never actually had the option installed. Maybe they tried to install it and decided not to. Hard to tell. But the factory label doesn't list the option and I don't see any indication it was ever installed, so I think that is settled. I was investigating this with the idea that maybe if someone removed option 001, they could have left a glitch that was contributing to what I was seeing on power up. That appears to be a dead end.

I swapped out a new microprocessor. It had only a very subtle effect on the display operation. I can go into that subtle difference later, but wanted to post a picture of the display on power up to you guys to see if anyone recognizes the "symptom". Keys are unresponsive - all of them. Obviously the circuitry driving the LED segments is good.

I'll be continuing to try and learn things on this, but check out the power up display and let me know what you think. Much appreciated!

JRH[/img]

[Tony_G]

I don't have an 8656A - Only the 8657B which, as an aside, now has an FM problem that is on the whiteboard to get around to fixing...

Anyway, I'm assuming that you did the normal check power rails thing (including ripple because they specifically call that out) and if they're all good then you have pretty much moved from the troubleshooting steps in BD1 (as they tell you to bail before checking outputs, etc if the unit is frozen) - As an aside, it would really help out if you provided the results of the step checks there as they have very specific results for the internal indicators on things like the Power-on Check (I don't know about anyone else but it'd help me out at least to understand where your unit is failing in the start-up process).

Anyway, I'm working from reading the manual here so forgive me if I'm not adding any value, it looks like you should be exiting BD1 in the "Power-up Check" section. I think that you're failing section 1 there (or somewhere near that) so you're instincts are right, I think, in that it is a digital issue. So in BD4 it seems like you're failing the "Power-on or Reset Sequence" part.

According to the manual from the Keysight website, you should be looking at Service Sheet 14 or 16 - However it appears that the manual they have is missing pages so I can't really provide much help (they don't have the start of the section) and, as is typical of the instruments of the era, they assume that you have a signature analyzer to test the board - I'm at a bit of a loss to provide any more commentary.

My recommendation though is to use the service manual to grab all the test point data you can and then, assuming that the manual from Artek with the missing pages doesn't provide any more insight, drop a post on the HPAK group asking if anyone has suggestions. Link to the group is https://groups.io/g/HP-Agilent-Keysight-equipment

Sorry,

TonyG

[jrharley]

No, this is great Tony G! I really appreciate your feedback.

I'm going to spend a little more time with service sheet BD1. Checking for ripple makes some sense, I missed that. Bottom line though is the power up test is a bust for sure. The unit doesn't step through any of its self diagnostics on power up, it just goes straight to the display you see in the photo.

When following the service instructions, you're right, you quickly run up on the need for a signature analyzer in order to follow the steps. I'm keeping my eye out for one, but for the moment, my test equipment will have to consist of a 250MHz 'scope, a decent DMM and a simple Logic Probe. I also want to check the triacs and diodes in the crowbar protection circuit, but not really expecting that to be the issue since it's tied to the attenuator operation. You never know though.

I will definitely post on the HP site, that's a great resource potentially, so thanks for that. By the way, I think I've watched a video you did on the 8657B. Nice job. There are a lot of similarities in the designs so it was informative and I'll be returning to it for more insight.

So anyway, thanks again for the help and I'll keep you posted with any relevant developments.

Best,

JRH

[Tony_G]

I'd probably start probing the microprocessor lines then - All the standard stuff, clock, address, etc - You're sort of stuck working outwards from there.

One thing that the 8657B has is a hardware-implemented NOP instruction to enable ROM testing - I've never used it but I think there is a switch somewhere on the processor board that switches from "run" to "test" (or words to that effect). You might want to make sure that it is set to run mode. Probably already is but worth checking. Assuming that there is such a switch on the 8656A. In fact, if the implementations are close enough, it might help to get the 8657B manual and see what it recommends in terms of direction for tracking down boot issues.

Good luck - I look forward to hearing what you find.

TonyG

[Tony_G]

Replying to my own post to make sure you see this (not sure if you get re-notified if I modify the post after you've been notified originally).

I created a searchable version of the 8657B service manual for you (the one on the Keysight website isn't searchable) - You can grab it here:

https://1drv.ms/b/s!Amqar8_XQ9Uzk48LDIJuGPej5NWrWA?e=RXh0me

TonyG

[cncjerry]

where are you located?  I have a signature meter.  I don't know the 8656 but here's my quick scan of the thread:

1) does it have option 1, 01, 001 listed on a badge in the back? That is usually the high stability time base, basically an HP ovenized oscillator on a board you plug in. Without it, you usually need the bnc strap between the posts on the rear that would be labeled like 'ref out' to 'ref in".  Sometimes you need to strap them if you have the oscillator.  Many pieces of gear are scavenged for the oscillators.  I pulled them out of selective level generators that were shot.
 
2) if it is halting on post, it could be a rom chip if it has them or something hanging the signal lines. 

3) Your symptoms look like they are related to a power problem though.  I've had a few generators that when your probe the power near the PS board, they look great but you need to check at some of the chips.  sometimes the cables running from the power supply go bad, I have a couple of older generators that need cabling.

4) someone else mentioned about the shafts and I agree, that is the least of you worriers.  I have a 3d printer and CNC gear but do you have the complete parts to copy?  They would have to be drawn as I doubt scanning would give you what you want dimensionally.

I would go back and probe the power pins around some of the ICs.  If the display is getting anything on it then it looks like it is running but hanging on code someplace, possibly from a bad rom chip?  I don't know if I would purchase a signature meter to fix one piece of equipment.  I got mine cheap and I think I've used it twice in 10yrs or so.

Jerry

[jrharley]

Wow guys, thanks for posting up on this!
The effort continues. I'm starting to learn the language for navigating the HP manual schematics and info. Lot's of study going on. Not any soldering.

TonyG- Thanks for the link! Good to have. Not sure if there's a test switch in the system as you describe, but I'll keep an eye out. "Thou shalt test voltages" right? So yeah, I'm probing wherever I can and pouring all over this thing. Some interesting results and on going discovery. I'm starting to learn how to run things down in the info efficiently so, stand by for more updates.

cncjerry - Appreciate you weighing in! I'm in Florida, Tampa Bay area. Also appreciate the thought on loaning the test gear, but I'm realistically a ways away from any signature analysis level troubleshooting. I tend to share your instinct that this is a power related issue and so I'm covering the basics first. I'm currently in the "thorough visual inspection" phase of this project, along with basic probing. The manual has a pretty simple series of tests using a scope to evaluate the signals from the keyboard to the rest of the circuitry. I'm focusing on that thread for the moment.

Thanks again all, I'll keep posting anything relevant.

Best,

JRH

[Tony_G]

Just thoughts I'd check in and see if you've had the chance to get any more data etc

[jrharley]

Hey tony G, thanks for reaching out.
Not much data yet, I've taken a sidetrack to clean keyboard switches and get to know that section. Repaired a missing stand-off assembly on the keyboard stack that was allowing too much flex when some keys were pressed, but I got that stabilized. The more I look, the more evidence of rough handling I discover. It's hard to understand why equipment like this gets abused, but it does.

I've attached a photo of a connector that I found disconnected when I opened it up initially. I've been studying the circuit associated with it and have started doing some simple tests on the keyboard per the manual procedure. The good news is that I do get a 5V DC offset when I press any key. So far though, I'm not getting any evidence of data packets coming out of the keyboard, but I'm still working that, I could have my triggering set incorrectly on the scope. Anyway, comments are always appreciated and I'll keep posting any relevant discoveries.

Thanks for the interest!

JRH

[Tony_G]

I can see how that would be a problem...

It would be interesting to see if you were getting clock into the keyboard shift registers (I think they generate the pulse train but I'd need get the manual again to confirm).

TonyG

[Tony_G]

Long time, no hear (actually I was on the road in MO and UT and then out at the cabin here in WA so I was pretty much out of contact and thus my fault).

Thought I'd just reply to see how you're going? My 8657B is probably my fav of the HP sig gens I have so I'm interested in seeing you get this one going again.

Look forward to hearing from you.

TonyG

[jrharley]

Hey Tony G, no worries, thanks for checking in.
Although I've had to take a few unrelated sidetracks recently, I have discovered something that I didn't notice before. Not sure how I could miss it. At any rate, I've discovered that when I jumper pin #2 into the circuit, (see the photo I posted a few comments ago for the connector), The 5.4 V rail indicator LED goes out! Or at least very dim. Something is pulling that rail down when pin 2 is connected. I simply didn't notice that before. So, I've been walking the dog along the schematic path looking for potential culprits. This means that every surface in my shop is covered in schematic diagrams! I can't use a lack of documentation as an excuse, that's for sure! Anyway, pins 1,2 and 3 on this connector are all originate from the 5.4 V rail, but only #2 causes this. The next stop on the path for pin #2's signal appears to be an IC labeled "DEMUX" on the schematic, which I guess is a de-modulator. There's also a diode straight to circuit ground in the path too, and I'm currently hunting that one down to test it since it just seems more unlikely that it would be an IC problem. Then again, I'm really green when it comes to the digital world.

So, that's where I am. I'll locate and test the low hanging fruit on this signal path and report back with any news. Thanks for your interest! I'll be struttin' around like Richard Pryor and Gene Wilder if I do get this beast working, and I really appreciate everyone weighing in. Best regards,

JRH

[Tony_G]

I decided that I couldn't help anymore without getting the manual so I just bought the Artek copy - Where are you looking at the supplies from J2 (so I don't need to go find it all myself )?

[jrharley]

Thanks TG, I really appreciate your interest in this misadventure!

I'm in the middle of getting my bearings on exactly what you're requesting, so, standby for that info.

At the moment though, I'm replacing the battery on a laptop that was clearly designed to discourage anyone from doing just that, including service techs. Gotta' get this out of the way before diving back in to the HP.

For what it's worth in the meantime, I started with the power supply schematic, sheet 22, page 8-127, 128 I think. J2 is shown in the lower right corner of the diagram and for pin 2 you follow to the "H" in other section schematics for where pin 2's path goes. Yeah, I'm still trying to learn how to follow the bread crumbs......

More info soon,

JRH

[Tony_G]

From the schematic, the 5V line carries over to a bunch of filtering capacitors on 8-86. I would start checking those before I started partying on ICs.

Just as an aside that reference mnemonic works like H 14 - Means the signal passes to point H on service sheet 14 - If you see H 15,16,18 then it means point H of those service sheets.

TonyG

[jrharley]

Thanks TG! 
It's hard to estimate just how much time you just saved me. I was chasin' a bogus Rabbit hole. This is such a great example of how invaluable experience is. I thought that small number 14 was something else, not the link to the next node in the chain. According to page 8-23, table 8, "Schematic diagram notes", that number 14 should be BOLD per the example. Yet, sure enough, when I followed to service sheet 14, there it was....... "Big as Ike", as my Dad used to say. "From the +5V supply". And yes, that is a bunch of caps on that line. I'll get to testing and report back. 

Thanks for nudging me back between the white lines!

JRH

[Tony_G]

NP - Happy to help - You'll find some of those numbers being "handwritten" on the schematics - If I had to guess I would say that this manual was produced at the time of moving from drafted schematics to computer-generated so you'll see these various anachronisms around all the schematics (I don't actually know but I'm sure someone over on the HPAK group would).

TonyG

[jrharley]

Hey TG, 8656A update here, hope it's all good -

I did a one night rundown of the caps on string 22H recently. Most of the caps on this line are .01uF ceramic packages. I have assumed that from the appearance of the caps and the fact that the parts list doesn't identify the type, that they're ceramic. They look to be high quality for sure, still shiny. Since each one is associated with an IC pin, I'm gonna' call 'em "Bi-Pass" caps. Not inclined to suspect a ceramic cap, unless it's baked or cracked. That, in addition to their relative circuit significance, has led me to discount any role they may have in drawing down the 5V rail. (Now that I've said that, this will probably come back to bite my 'arse in 6 months!)

There are also a couple of 'lytics in the chain. C1 is 100uF and C8 is 2.2uF. I measured them in circuit at 1KHZ. C1 was pretty much spot on @ 102, with an ESR of .27ohms. C8 was a different story, but I'm not clear if it would matter at all for the disappearing 5V rail. It measured a whacky 102 uF. ESR was so-so at .65ohms. Again, maybe worth a second look, but I can't see this pulling the 5V rail down since it's not dead shorted.

I'm moving on down the chain to look at service sheets 15,16 and 17. Interestingly, things seem to be converging on the sheets that came up as significant in my "bogus rabbit hole" journey.

So that's where I am. More probing around currently underway.

JRH

[bdunham7]

There are also a couple of 'lytics in the chain. C1 is 100uF and C8 is 2.2uF. I measured them in circuit at 1KHZ. C1 was pretty much spot on @ 102, with an ESR of .27ohms. C8 was a different story, but I'm not clear if it would matter at all for the disappearing 5V rail. It measured a whacky 102 uF. ESR was so-so at .65ohms. Again, maybe worth a second look, but I can't see this pulling the 5V rail down since it's not dead shorted.

They're in parallel in circuit, so you're effectively measuring both each time, along with the PCB trace between them that adds a little ESL.

[Tony_G]

As Bdunham7 said, they're in parallel so I would be inclined to remove them from the circuit and test - If you have enough space, you could just lift a lead and then test.

It certainly could be an IC but I'd want to rule out the possibility of it being a cap on that rail.

From memory, I think you said that the main 5V rail stays fine and it is just that pin 2 line that goes low so I would approach it along the lines of checking each cap and if they're all good then start removing the ICs that are related to each cap while monitoring the voltage.

If you've removed all the ICs and still have a problem then you're looking for a different reason. Sort of a longish process but it would allow you to rule out the caps and ICs. Also if you don't have one I can heartily recommend the Hakko FR-301 as it makes desoldering much easier.

TonyG

[jrharley]

Hey guys, copy that! Thanks for weighing in bd7!

I wasn’t putting much stock in the in circuit measurements either. Lifting a leg on them wouldn’t be too tough to do and it may come to that, we’ll see. Meanwhile, a little more accurate discovery on the caps in question has come to light:

Once I found the right parts list, (DOH!), the cap descriptions made more sense. The bulk of the capacitors are, in fact, ceramic and they’re labeled so in the description. What I didn’t expect to see, since I’ve never seen any before, is that the other type of caps in this chain are axial Tantalum caps that look like regular axial electrolytics. But the parts list designates them with a “TA” in the description and they do exist. So, there you go, learn something every day.

There are about 20 ceramic caps in this part of the circuit, most are .01uF. I decided to just test the resistance across them all, since I could do that from the top side easily. Maybe I could squeeze a little more information out of the board. All of them seemed to have pretty consistent and reasonable readings in the hundreds of ohms, the bulk of them being around 390 ohms. There are also 2 very small ceramics that are identical and specified as being 22pF. One measured about 700ohms and the other one seemed to measure open. Nothing. So, I cobbed a couple 60pF ceramics together in series and tacked them across the suspect cap “Shango style” to see what effect would occur, if any. Not unexpectedly, no change. At the moment, I’m thinking this is a low level power supply issue so I think I will run down a couple of other lines before sucking on IC pins.

The pins 1 and 3 on the J2 connector where this Rabbit Hole originates, are tied together with pin 2 where it connects, on board A10.  I think I’m going to follow 1 and 3 along for a bit, to see what I can learn.

By the way TG, I found the programming switch you referred to a few comments back. It appears on service sheet 16. This one’s set to 5V, which seems to be correct.

So that’s where I am - this is still a very active crime scene. More discovery later, and thanks again for the input guys, it's very much appreciated!

JRH

[Tony_G]

There are about 20 ceramic caps in this part of the circuit, most are .01uF. I decided to just test the resistance across them all, since I could do that from the top side easily. Maybe I could squeeze a little more information out of the board. All of them seemed to have pretty consistent and reasonable readings in the hundreds of ohms, the bulk of them being around 390 ohms. There are also 2 very small ceramics that are identical and specified as being 22pF. One measured about 700ohms and the other one seemed to measure open.

Just thought I'd check-in and see if Santa brought you a working SigGen.

Out of curiosity, are you measuring those caps using a proper LCR meter or just using the resistance mode on a DMM? They should all be "open" from a DMM resistance measurement and with an LCR meter, you'll see the ESR values. Just wondering.

[jrharley]

Hey TonyG,

Thanks for weighing in!, Happy New Year to you and all the HP fan boys on the EEVBlog forum.
Nope, Santa’s delivered nothin’ but coal lumps so far. But learning has still been happening. On the cap testing, I hear you, I’d considered the same things. My resistance readings were with a DMM, but I also have a decent LCR meter that I’ve been using to test caps as well, both in and out, of circuit. What I learned from just slapping a DMM across those caps was that things were pretty consistent across the board. Call it a “qualitative measurement”. The circuit was definitely influencing the readings, even if I don’t really know how, precisely. It was just a data point, even if a marginally informative one.

First impressions sure can fool you sometimes. I’ve realized that my initial conclusions about the power rails on this guy were a little bit “slap-dash”. This has screamed power problem from the very beginning, I just didn’t look close enough. The bottom line at this point, is that when all connections are hooked up as intended, the +5V rail becomes +1.8V, and it only does that when pin 2 of the J2 connector is connected into circuit. I’ve looked at the string on sheet 14 pretty closely at this point. I’ve tested the tantalum caps out of circuit, (they’re bang on), and really doubt any of the ceramics are the problem, so I’m moving onto sheet 15, following the 5V distribution. I’ve determined that I do not have the 5V at chips U29 and U33 per the schematic. Interestingly, with the wire from J2 disconnected, I got essentially no voltage on those chips, but with it connected, I got low voltage, 1.6V and 1.7V respectively. The interesting part is that's really close to the 1.8V faulty rail reading observed previously on board A10.

Following other points on sheet 15 where the 5V should be, has led me to the only resistor network on the board, R9. There are 9 - 3.3Kohm resistors in the package. I’m currently doing some investigating of that package and pondering a way to test it. Not sure if I’ll find a diagram anywhere that shows how all those resistors are hooked up inside the package, but it certainly seems like a possible culprit. Great, I suspect that part will be unobtainable in its current form, but we’ll just have to see if it’s got a bum resistor in it somehow. The package has 10 pins.

So, nothing insightful to report unfortunately. This Bodger is still just poking around in discovery mode. More later with any progress....

JRH

[themadhippy]

There are 9 - 3.3Kohm resistors in the package.

Wot like this?

Great, I suspect that part will be unobtainable

https://www.digikey.co.uk/en/products/detail/bourns-inc/4610X-101-332LF/1089196

[jrharley]

themadhippy!
Thanks for the tip!
I'm just now getting into looking at this component, so this is great! Good to know there's some options.

This guy is a good bit snazzier than the HP part, which is pretty "opaque" as far as construction goes. This one's form factor doesn't allow much room for a lot of arrangement possibilities, so I'm guessing they're all in parallel? I'll run it down, I presume digi-key US can supply too.

Thanks again so much for the link and information, it's a big help!

JRH

[themadhippy]

so I'm guessing they're all in parallel?

no,this is the most common arrangement

[Tony_G]

As Themadhippy said, these aren't in parallel, R9 is being used as a pull-up resistor so I wouldn't expect that it would have a problem dragging your rail low.

If you think it might be then remove R9 and see if the rail stays at 5V and that you get +5V to R9 (Pin 1 I would assume but you should confirm on the schematic) - If it does then you more than likely have a problem with one of the ICs that are using R9 as a pull-up.

If that all works (in that removing R9 gives a good rail) then you might want to try using a 3.3K resistor to just provide +5V to each pull-up location (e.g. R9 pin 1 to R9 pin X one at a time) and see if you can identify which IC is dragging it down.

All of this assumes that removing R9 enables the rail to stay at +5V if it doesn't then the problem is somewhere else.

Sorry, I can't be more specific.

TonyG

[jrharley]

 Thanks guys, much appreciated! I'm afraid I'm obnoxiously long on enthusiasm, but woefully short on knowledge, so this is a big help.

Thanks for the diagram themadhippy, that makes sense now. Now I see why they didn't feel like it was necessary to show the package all in one spot on the schematic, among other things.

tonyG, Bodger Alert - I didn't know that there even was such a thing as a "pull-up resistor". Now that I've looked up the concept, I've got a better idea. Thanks for advancing my understanding!
I think at this juncture, I'm inclined to follow your instinct about R9. If it's unlikely that a fault in the unit will cause the fault I'm chasing, then I think I'll put it on the to do list, but keep moving. After all, a digitally naïve, analog assault on this beast, is all I've really got. But I really like your suggestion on how to test R9 and the chips associated with it, I almost want to do it, just to do it. R9 wouldn't be hard to pull. But "eyes on the prize" right? I'll do some more probing tonight.

Always appreciate the feedback folks,

JRH

[Tony_G]

I think that your process of getting the power rails right is the correct first approach to the unit - I'd keep trying to find out why the +5V rail drops to 1.8 (I think you said that) when connected.

Once the rails are right then you can be more "creative" in trying to find the fault.

TonyG

[jrharley]

Thanks Tony_G,

I will stay the course.

On a side note, it occurred to me that I might want to see if there was any output from the unit when powered up. Just curiosity.
I don't currently have any connectors or adapters that will connect to the RF output on this guy. I know I can rig something up, but it got me looking into the connector a little more closely. What is this thing?

Any clues as to what this connector is, would be appreciated. It's larger than the SMA ham radio stuff I see. The parts list isn't a lot of help as to what this thing is called. I gotta' believe it's at least available, and I'm just clueless. This connector to BNC female would be sweet.

Updates forthcoming.......

JRH

[Swainster]

Hi,

You need a 'N' Male to BNC Female connector to convert the output to BNC. Or better to switch to type N connectors on your cables, and just use N to BNC cables for BNC stuff. If you start working microwave band stuff (300MHz upwards) then N connectors will be pretty much standard for test equipment.

I've been following along with interest as I have a couple of 8656Bs, one of which is working but the other is incomplete. Someone has asked about buying the working one so my interest in getting the non-working one up and running is much higher (it's missing the power transformer and GPIB connector, but according to the previous owner everything else is OK).

[Tony_G]

...but according to the previous owner everything else is OK...

Heard that before Much liked "Pulled from a working environment"...

That said, I think the unit should work without the GPIB board (for clarity I mean A13 and not the main processor board where all the actual GPIB smarts are but you'd need to check the schematic/service sheet) but I'm actually not sure - it certainly won't work without the power supply.

As for the adapters - These probably suck in terms of RF quality but in terms of getting you something that you can look at they should work OKish:

https://smile.amazon.com/Adapter-Connector-Extension-Antennas-Female/dp/B091C5W4HM/

As Swainster said, the output on this unit is a N-Type connector - The ones on HP gear are typically good up to 18GHz (in "as new" condition). The actual connector spec is only good to 12GHz I think from memory but HP did some higher specs up to 26.5GHz but this unit should only have a standard 18GHz one.

There is a whole bunch of information on how to confirm that the connector is in spec but seeing that the 8656A only goes up to 1GHz you're probably fine with whatever the connector is right now.

TonyG

[jrharley]

Thanks guys, BIG time saver! Appreciate it.

I'll source some cheap versions to aid in my testing and order the $30 versions for when I'm testing microwave antennas for a fighter jet. Seriously though guys, I really appreciate this dialogue, it gives me hope that I just might be able to get this piece of equipment up and running again.

Thanks for weighing in swainster! You're in a good place with 8656B's from what I understand, they've got a better rep than the A version. I'm glad you're following the conversation.

Tony-G - Thanks for the info and insight. It seems there's quite a science to RF connectors!
I've had my eye on a suspicious Triac in the power rail that I want to test. (Q3 on the A10 board). Think I'll pull it this weekend and see what I can learn. Continuity testing of it, in circuit. has yielded confusing results.

The Brute Force tech team stumbles on.............

JRH

[jrharley]

8656A update here:

OK, so I pulled the "TRIAC", Q3 from the A10 board, which is more precisely an SCR, and proceeded to beat on it for a bit. Just for grins, I powered the unit up without the device in circuit. No change. Although I don't fully understand how the thing would act in the circuit, it's part of a "Crowbar" arrangement intended to protect the attenuator I think, so it's not entirely surprising that it's condition, in or out of circuit, doesn't change anything here.

As for it's condition, I'm basically getting "open lead" no matter how I measure it. I tried turning it on with a 1.5V AA battery thru a 7ohm resistor, but it didn't seem to stick. No resistance could be measured across the legs after pushing a little current through it. But it's also possible that I've pushed a bit too much current through it. The battery was putting out about 200 mA, if my poorly timed measurement of the battery set up was correct. In a head slapping episode of realization, I decided to check the data sheet and it says the maximum triggering current for the device is 15mA. DOH! I may have screwed the pooch here. Hard for this Bodger to know at this point. The SCR, aka Thyristor, is an RCA 2N3528. The TO-8 package for any substitute for this device seems to be hard to find, based on a brief search, but maybe you folks know of a good substitute. There are other form factors that could sub available, but that would be tacky. Who knows though, the thing could be fine.

In any event, I am pretty certain at this point that it's not a factor in the fault I'm chasing. I put it back in.

Meanwhile, the weekend has revealed that my B&K 1430 3 channel power amp, seems to have lost a channel. Since I've got to have tunes, I need to detour for a while to try and get that guy up and running again.

Appreciate the interest in my efforts folks, and any feedback is welcomed.


JRH

[Tony_G]

I believe that the 2N3528 can be replaced with an NTE5455 but you should compare the datasheets to make sure.

Now that Q3 is part of the 5V crowbar circuit and not specifically related to the attenuator (from my quick reading of the schematic/service sheet) - If it is being activated then it should cause a fuse to blow.

I would recommend that you get a tester - I use one of these https://www.amazon.com/Atlas-ADVANCED-SEMICONDUCTOR-ANALYZER-Tracing/dp/B00O0BMNX2 but you can really use any of these testers (including the ones off eBay).

Let us know how it goes.

[jrharley]

Thanks for the cross reference Tony_G! I'll check out the NTE number and store it for future reference should I need it.

Very good point about the SCR action! I'd already checked the fuses associated with it just for curiosities sake, and they're good. The circuit makes more sense now.

When still hot off the board, I dropped the SCR into a cheap, classic, fish8840 tester just to see what I could see. It seemed to indicate that I had 2 diodes. The tester is supposed to test thyristors, but you never know for sure - it's a cheap tester, operator error, solar flares.......

 Oh, for a suite of those Peak testers.......... Nice kits.

The B&K power amp issue turned out to be just gummed up output level controls. I love it when that happens. So now I'm back to the HP for more terrifying technical explorations.

As always, I really appreciate the feedback.

JRH

[Tony_G]

Good to hear - Good luck with the unit - Look forward to hearing what you next discover.

Just to refresh my memory, you checked the voltages per service sheet 22 and did so by removing all the boards not used in SS22? This would tell you that the rails themselves are good (though it is possible that once a rail is loaded down it slumps in voltage) and let you add a board back in at a time to see if that board causes the rails to droop.

Again, look forward to hearing the progress.

TonyG

[jrharley]

Hey Tony_G,
I see where you’re going with that. Good point.
I haven’t systematically isolated all the other boards, but I did isolate the Audio/Power Supply section on board A10 by disconnecting connectors J2 and J4 entirely. I tested most of the components within the isolated assembly, like the SCR Q3, and I’m reasonably convinced the draw down doesn’t originate in that portion. You can see that there are test points for all three main rails on either end of the rails. I tested the voltage at these points, (running the unit off the 120V mains), once with the #2 wire from connector J2 out, and again, with it in the circuit. First number is with the J2 wire out of circuit.

Testing the pins in test socket J6:
 
 “Left side”
TP 11: 21.8V, 22V
TP 9 :  12V, 13V
TP 10 : -23.4V, -23.3V

“Right side”
TP 13 : 14.9V, 15V
TP 3 : 5.4V, 1.8V
TP 2 : -14.7V, -14.7V

Under each of the voltage call outs for each rail on the “Left side”, there are small “hand written” voltages. These voltages, whatever they are supposed to represent, are much closer to my measured voltages than the rails are designated to be. A different “state” perhaps. No idea at this point, but find it hard to believe it’s coincidence. I need to re-visit the theory of operation info if I can find it on this section.

I’ll study the schematic and see if isolating other boards provides any clues. Thanks!

Onward…….

JRH

[Tony_G]

Sorry bit of a brain fart on my behalf - I was imagining that there was a motherboard but these units didn't have that.

Apologies if this caused lost time.

Did you get a chance to put a scope on those test points? Some of them have very tight ripple tolerances (eg TP13 should only have a 10mV ripple on it with the DC value between 13V-17V)?

Detail is in Table 1 on SS 22.

TonyG

[jrharley]

Nope, I do plenty of time wasting inquiry all on my own!

Busy poking around the thing the last couple of days. I eventually got the hint and tested the test points in table I with a scope. (Pin 2, J2 still out of circuit).
The voltages were as outlined previously, but the ripple was all over the shop. Another good data point, thanks!

Importantly I think, it was the worst on the 5V rail. (Worst delta that is, it was actually more or less the same unstable value for all of the test points, varying from 800mV to 1.2V p-p).

I suppose I'm going to be lifting some cap legs soon. Is it possible that the regulator for the 5V rail is just doing its job poorly? The ripple was 1.2V p-p. Yeah, I think the table is saying it should be 0, so not so good. I guess a voltage ripple like that could cause the logic to fritz out downstream? We'll keep at it.

Thanks Tony_G, you're a Wizard! I feel like I'm still making progress.

JRH

[Tony_G]

Good to hear that I didn't cost you any time.

TPs 9,10 & 11 are unregulated and if they have excessive ripple on them then I would check out C17, C18 & C19 - You tend to find ripple in unregulated rails from the main caps and those tend to be large value caps. You might also want to check out A14 C11, C14 & C15 - At least at the start, if all those capacitors test correctly then you can start on the smaller value ones like A10 C15, etc. Dave did a good video on bypass caps if you want to get more background on why there may be multiple bypass caps on a rail.

I'd work to get those TPs into spec first as the other ones are further down the chain (specifically the other 3 TPs from table 1). When you start looking at the regulated TPs then you should probably check the ripple if not in spec then C23, C24 & C25 - I wouldn't bother with A12 U1, U2 & U3 until you can confirm that the unregulated ripple is correct.

Look forward to hearing the next thing you find.

TonyG

BTW It isn't impossible that the diodes doing the rectification are bad but I'd absolutely rule out the capacitors first. As an aside, getting a good LCR meter, like a DER EE DE-5000 is something to add to the toolkit as well but really you can get quite servicable results with some of those very inexpensive eBay devices.

[jrharley]

Really great points to run down Sir! Much appreciated. I'll look at the schematic with those comments in mind tonight. Meantime, I've attached a photo from the project "journal". It's one of the photos taken randomly through the process. I'm testing a small cap in circuit with the LCR meter. 

Report back soon,

JRH

[jrharley]

Brief 8656A update here.

I guess I'm just going to have to do the work.  With this kind of issue in the supply rails, I've got to get a pretty solid test on the caps, bottom line, and that means lifting a leg on C17,18 and 19. I can quickly flay this board open and test these guys, that's not a problem, but here's where my mind is going -

Let's face it, these caps are 40 years old. Good caps originally for sure, and in circuit, they're not raising too much concern. But, yeah, they're 40 years old and there's ripple out of spec. on the rails they smooth.

I'm trying to get this old beast in a place to be a solid contributor on the bench. It seems inevitable that these guys are going to be replaced. Unfortunately, that "thought thread", runs headlong into the form factor problem.

I've got suitable axial replacements in my stash that could go in here. They are about a 3rd the size though. I'm going to have to come up with an appropriate mounting strategy for modern equivalents. The original zap-strap approach from the factory, won't work in the current arrangement for modern caps.

Not saying it can't be neatly done, but I'm just.............sayin'.

So, I'll test 'em, get good data, and strategize about how to get a replacement solution that's appropriate. I have to admit that I found the mounting solution by HP on these guys to be a bit "how ya' doin'" right out of the factory. I'll ponder for a bit.

Thanks again for the feedback.

JRH

[jrharley]

OK,
So all of the big filter caps have been removed from circuit and tested with the trusty DER LCR meter. They are all out of spec to the high side on capacitance value - not horrible, but significantly. If this was an old all American five, I’d still call ‘em good. ESR on every one was strong, about .03 ohms. I’ll take that value. Still, they’re out of spec. I’m unclear on how higher capacitance values would affect this circuit, and the ripple I’m seeing on it. Replacing them, any way you look at it, is going to be a nosebleed. But it’s probably unavoidable in the long run.

I’m currently thinking that the most “dignified” way to replace these guys, is to re-stuff the originals. I think it could be done by making new cores from radial units rigged in parallel to hit the value. Then I could install the caps back in, just like they came from the factory. I’ll have to do a deep dive on cap dimensions and values, but it’s probably the only cost effective solution. The job will always be obvious to an informed eye, but it just seems to be the least invasive fix.

The million dollar question for me is this: Is this significant to the fault I’m chasing? My gut tells me no. But hey, I’m “Analog Man”, so I may be clueless. Calling all circuit Guru’s ! – should I tack these back in and move on for now? Or should I stop and do the deed to get the big filters into spec?

Any feedback is much appreciated.

JRH

[Tony_G]

I think it comes down to trying to find out why the ripple at the unregulated test points is way out of spec - This really needs to be corrected before you start looking at other things (at least this is how I would approach it YMMV).

I'm assuming here that you have already checked the capacitors on A14 (C11, 14 & 15) as they are in the path from the transformer to the unregulated A10 circuit - I can't remember if you did this so I'm going to go with the fact that you did and they were good.

Now that you've measured the A10 caps and they're all ok (the spec is +75% to -10% so reading high is not necessarily a bad thing) I'd start looking at what else might cause that ripple - For example, those caps (C17,18 & 19) go to a ground point via Note 3 on the schematic so checking that ground is correct would be my next step. If that ground was all good then I'd probably remove J4 to disconnect the circuit further downstream.

If it was still out of spec then I'd probably take a look at C23, 24 & 25 as those appear to be connected via PCB traces to the unregulated test points. If everything was still good but the unregulated voltages still had out-of-spec ripple then I'd look at the various diodes & resistors (with J4 still out) to confirm that everything that is between the unregulated voltage and ground is correct.

Fundamentally, something is causing that ripple and the designers specifically didn't want that there - Whether you address it now or later it does need to be addressed - I'd follow Dave's rule of checking the power rails first.

If all that checks out then it is something way unexpected

TonyG

[jrharley]

I'll take that as a hint to put the big filters back in and keep digging.

I'm embarassed to admit it, but I can't seem to FIND C11,C14, and C15. They're prominent on the schematic for sure, but I'm puzzled as to where those suckers are hiding. The photo of board A14 doesn't show any components on it, and it just looks like a mounting board for the HP IB. I have to be doing something stupid, they're too important. I'll keep looking of course, they have to be here somewhere.....

As always Tony_G, you're interest in my project is extremely helpful, thank you very much!

JRH

[Tony_G]

[Update]
Should have taken a bit more time to think about it - Those caps are in the replaceable parts list with the part number "2425-011-x5u0-102-m" - Looking at DigiKey for a close approximation gives this one https://www.digikey.com/en/products/detail/tusonix-a-subsidiary-of-cts-electronic-components/2425-018-X5U0-102M/4214068 and I would imagine that the ones in your unit will look something like this:



Should have realized that from the schematic where you can see them as feedthrough:



Check A14 or the wiring harness and you should find them.

[Original]
Having not seen the insides of the unit I can't be of much help but there are 2 wiring harnesses connected to A14:

• W19 - Connects to A14 J1
• W16 - Connects to A14 J2 & J3

W19 looks like it connects A14 to both A12 and T1 (and the optional high stability timebase on A16) while W16 seems to be all A14 to A10.

If those harnesses are all plugged in then the caps (assuming that they are in fact discrete components and not some engineering solution using the wires & PCB themselves) have to be on A14.

You might want to try backing into the solution by disconnecting the wiring harnesses for A14 from A10, A12 & T1 then measuring the lines as though they were just caps? They might be doing something custom & tricky.

Look forward to hearing from you - I'm as invested in getting this to work as you are now

TonyG

[jrharley]

A "Feed Through Capacitor"........, who knew? That's a new one on me. Thanks Tony_G!

I think I might have found them.

I'm going to have to re-group, secure board A10 first, and then go to surgery in the back of the unit. See photo attached.

Always appreciate the dialog!

JRH

[Tony_G]

In your PDF - You're pointing at A13 I think - A14 might be the board to the upper right in that image but it isn't the board with the GPIB connector on it.

TonyG

[Tony_G]

BTW Did you check the ground connection for A10? You should probably check for A14 as well when you find those feedthrough capacitors.

TonyG

[jrharley]

Hey Tony_G, you're right. Once I cracked it open again and took a closer look, it became clear where A14 was. I was indeed pointing to A13, my bad.

Now I'm working out the best way to get access to the A14 board assembly. It's mounted through a copper shield that is attached to the back of the compartment. Some disassembly is in order here. This could take some doing, those caps are new to me as a form, so I'm being a bit cautious in the tear down. It almost seems like they are a modern version of a "gimmick capacitor" that one can sometimes encounter in really old gear. Hard to see how they could fail, but hey, leave no stone unturned! I think with hemostats and some long tweezers, I can extract this thing, but we'll see.

I'll look at all ground points for sure. I think A10's ground is via screws to the frame per the schematic notes, and I've periodically checked ground continuity when testing, but I'll keep it top of mind as I go about things. The ground wire under the screw that holds the A14 board and shield assembly looks real solid. Don't think anyone's ever removed it.

Thanks for the feedback!

JRH

[Tony_G]

It might be easier for you to just use the ends of the jumpers/wiring harness rather than try to extract the board/assembly.

TonyG

[jrharley]

Too late! I'm in like Flynn! Couldn't resist.

Since I've got no idea what I'm doing really, I have to learn by examination. I hear what you're saying about the testing points, but I gotta' say, that access to the contacts was not great while assembled, and removing the connectors in place looked risky too. I decided I'm going to air this corner out a bit, clean things up, and educate myself about it. It seems to make sense to take a fair amount of time just seeing what I can see with this part of the circuit and improving things where I can. No other fool should ever be compelled to come back in here!

I may end up pulling board A14 out completely, for a closer examination. I've never seen a set up like this. The caps are "invisible". Very curious to me.

See the attached photo for a view inside the A14/A12 area.

Thanks Tony_ G!

JRH

[Tony_G]

Would be interested in seeing photos from various angles of the A14 board if you do remove it.

TonyG

[jrharley]

Of course! The crime is unfolding.

I'll be looking at what I've uncovered with respect to the schematic tonight, and sticking a meter or two across the pins of the board to see what more I can learn.

Once it was free of its tethers, the A14 board was easy to remove. Photos attached.

JRH

[Tony_G]

Interesting - The caps have to be somewhere This is really turning into a mystery...

The parts list for the A14 board shows the connectors being part of the board:



And the caps themselves are listed as standalone parts:



Coming back and looking at the schematic:



 It shows a 1000pF cap between the ground and J1 P15 & J2 P7 - The note says that the chassis ground is provided by the mechanical connection of the D-Type to the frame - Try measuring between those points to see what is there.

My only other thought is that they are in the wiring harness somewhere so you might want to follow that and see what happens to it between A14 and A10.

TonyG

[Swainster]

Hi guys, I'm still following along. I definitely agree with the idea that the caps are inside the potted D-sub connector.

BTW, I swapped the rear panel from my working 8656b to the non working one and it powered up without issue. Well except for the frequency being pretty far out and 3-6dB down from the front panel setting. So I guess the seller was on the level about it just missing its power transformer.

From your pics, the mechanical arrangement of the rear panel of the 8656a is slightly different to the 8656b, but I seem to think that the overall power supply circuit is pretty similar. Would you like any points measured for comparison?

[Tony_G]

I definitely agree with the idea that the caps are inside the potted D-sub connector.

I think you're absolutely right - I asked on another forum and that was the feedback.

I'm also wondering about the rectifier diodes as a source for the ripple seeing that the caps are all good.

TonyG

[jrharley]

Wow, 'yall! Terrific feedback. Lot's to run down, so it's going to take a bit, but I really appreciate the input. Thanks for weighing in swainster, good to know I can reach out for a cross check on the power supply measurements.

I've managed to test the A14 caps and they all read low. I thought it was interesting that the parts spec's said that they could be low by 20%, but not high at all. A couple are a bit too low, but otherwise this looks good to me. I couldn't help but then look to the hidden board A12. Looking at the schematic, these caps on A12 are all associated directly with the regulators, which I've taken as good so far. All the caps on the board are Tants. So far, they haven't measured well in circuit.

The immediate task at hand here is clear for me though. I've got to modestly restore this area before moving on. No sense in breaking into this thing just to leave it worse than I found it. I'll secure things a little better, clean the area as much as I can with Alcohol, and clean all the connectors. More study is needed in here. This stuff does take time, but it's very peaceful for this old Bodger. Bear with me as I stabilize things.

I've attached a photo of the seldom seen A12 board.

JRH

[Tony_G]

That poor SigGen has had a rough life...

I wonder if it may have been dropped at some point and the chassis has twisted a bit - These things weight at 2/3rds of a hernia so it wouldn't surprise me.

JRH - The GNDs for A12 are all direct chassis connections so while you're cleaning that up you might want to check that it is well grounded - I suspect just putting nuts and washers onto the posts will do that but worth checking in passing I think.

TonyG

[jrharley]

I know right? It's amazing to me that a piece of equipment that cost as much as a new car back in the day, could be abused in so many ways. I don't think the chassis has flexed though. It's a solid sand casting I think, very rigid Aluminum. Which is not to say that it hasn't been dropped. It weighs 37 pounds.
Anyway, the compartment is cleaning up nicely. I've swabbed all the grunge out of it, cleaned all the connectors and pins, and treated them with a light brushing of D-100. I've also discovered how the A12 board is mounted and it makes sense with the schematic diagram now. The "nutless" screws through the A12 board don't need nuts. Closer examination revealed that the holes in the board have a threaded insert in them and the screws are directly coming from the TO-3 package regulators on the other side of the chassis wall. The A12 board is directly coupled to the regulator mounting screws. Changing any caps on A12 would mean pulling those regulators. Right now, I'm thinking I don't need to change any. Once I disconnected the A12 board from the circuit, I got decent capacitance readings on them all. No visual signs of distress either. Grounding all seems solid, a single grounding point at the copper shield mounting bracket binds the ground to the chassis well. The bottom of the shield bracket which covers the A12 board, is uncomfortably close to the top of A12 when installed. I may glue a square fish paper sheet to the bottom of the shield bracket for good measure, we'll see.

I think I'll button this up and move on with some of the testing ideas that have been offered up next. I'll send a photo of the compartment right before button up. Thanks for everything TG!

JRH

[jrharley]

HP Sig Gen fans,
We're moving on to the next dead end here!
Back cavity containing the A14 and A12 boards is back together. See attached photos.

 At this point, I think we've eliminated some variables, so all good there. Still, the original issue remains. The good news is, no harm done, the unit powers up exactly as before. Ready to follow up on the backlog of great testing suggestions!

I'll post up any new developments.

JRH

[jrharley]

whooops, looks like a couple photos didn't attach.......

[Swainster]

Just to check, my understanding is that the current status is:
1) 5V supply is ok without the MCU board connected
2) 5V supply goes down to 1.8V when MCU board is connected
3) It is unknown if the fault is due to excess load from the MCU board or a problem with the supply.

If this is correct then perhaps a few comparisons will help - I made some current measurements of the 5V supply on my 8656B, measured at the 6A fuse:

• Plugged in but 'powered off' -> 64mA
• Powered on in default state -> 3.425A

Measuring just the MCU board current (hopefully I identified the correct pin on J2 as it's a bit different on the 'B' version), powered on -> 761mA

Lastly, when the MCU board in disconnected from the 5V supply and I power up the instrument, then the front panel displays all '0's and the total 5V current is 2.807A

[Tony_G]

The SigGen is now 20% faster due to the lighter weight from dust removal...

Swainster - Did you by chance look at the ripple on your supply? How different is the supply circuit? Here is the relevant unregulated part of the power supply schematic:



Also if you happen to catch an example waveform of the inputs that couldn't hurt.

TonyG

[Swainster]

Hi TonyG,

Although the boards have different part numbers, the unregulated part, and probably 90% of the regulator circuit design, appears to be identical in the 8656B. The only differences that I can spot are that the attenuator in the 8656A uses an unregulated filtered supply with over current crowbar, whereas the 8658B has a proper linear regulator (7815K), and the 8856A has a 110VAC fan, whereas the 8656B has a 24VDC fan running from the rectified 18VAC i.e. before any regulators.

Ripple across the 5V supply filter cap C30 is 1V pk-pk on top of about 10VDC. It looks like a pretty typical post rectifier 100Hz waveform (we are on 50Hz mains here).
Ripple on the 5V supply post regulator is tiny i.e. too small for my scope to pick up. One thing to note is that the 'ground' on the big filter cap is not the same as the regulator ground. There was significant AC ripple between the 2 'grounds' (maybe something like 0.3Vrms) so when measuring the post regulator voltages you cannot leave the scope ground clip or DMM lead on the filter cap - it needs to be moved to the ground test pin provided on the A10 board.

[Tony_G]

Interesting - There is indeed a regulated GND test point that should be chassis GND on pin 1 of J6 but no mention of that in the textual part of the manual - The unregulated side has it's GND (tagged with a 1 on the schematic) and being connected to chassis GND via pin 1 of J5. In fact, I don't think there is a specific GND point mentioned for the TP measures of either side.

JRH - Where did you measure to?

TonyG

[jrharley]

Hey Sig Gen folks, very interesting dialog, enjoying the thoughts!

Interesting comparisons swainster.
I'm generally measuring from the ground points on the respective boards, if I'm following you correctly TG.

I'm beginning to think I need to move on from the basic rails though, things have improved after all the poking, prodding, and re-soldering.

After completing the housekeeping back at the A12/A14 boards, I disconnected connector J4 to isolate test points 9,10 and 11. Voltages were good and ripple was modest, around 8mV. When I connected J4 back in and tested the other side test points, 13, 3 and 2, the ripple and voltages seemed to be in spec more or less, even on the 5.4V rail. Ripple was between .5 and 1.5 mV on that rail. I'm not thinking that's enough to cause a problem. For whatever reason, testing the three rails now, seems to be close enough to spec to eliminate this area from being the culprit. Connecting pin 2 of the J2 connector still causes the 5V rail to drop of course.

Could have been testing inconsistencies on my part, could have been dirty contacts, etc., but it seems to be good to me now. I'm currently taking a closer look at service sheet 14, following the J2 pin 2 wire through the circuit.

All the insights are terrific guys, really appreciate it! I'll report back with any developments after tonights adventure.

JRH

[Tony_G]

Good to hear - The tightest requirement for ripple was 10mV so if they're all below that then you are in spec.

Looking forward to hearing how the rail exploration turns out.

TonyG

[jrharley]

Well, things certainly got a lot more interesting last night! 

I decided that the next step for me, would be to test voltages on the pins connected to each of the by-pass caps along the H string shown on the schematic on service sheet 14. So, I connected the pin2 wire back into circuit and powered the ‘muther up. Much to my surprise, the power up display was different! Obviously, something that was done in the process so far has awakened new elements.

The display read like this: Modulation window was blank, Frequency domain read 100 MHz and the Amplitude domain was blinking “1.8” along with all the adjacent LEDs. This was a big change from power up displays so far! I took a solid mental note of the display, but should have taken a picture. DOH! I was encouraged to see that the display was showing the default boot up value of 100 MHz.

Of course, I couldn’t leave well enough alone and pressed the Reset switch. The display went dark, but it was like it was before when pin 2 was in circuit. When I powered off, I heard an attenuator solenoid “clack” in! (Attenuator cable was in circuit). This was also a new development. Previously, no attenuator action was evident.

This beast is trying to come back to life! despite missing some limbs.

It continues to bother me intuitively that there is currently a big hole in the attenuator assembly. At least some portion of it is “open circuit”. I don’t have a good enough understanding of things to describe what is happening, but having an entire section of the attenuator disconnected just “don’t seem ‘natch’ll.”. At least it confirmed that the 24 V rail is working.

At this point, I haven’t been able to reproduce the different display again, it’s back to acting as it did before, except that now, I can here the attenuator trying to work when powering on and off. It doesn’t sound particularly “happy” though. Incidentally, all the voltages on the H string read around 1.6 to 1.7 V, all 13 of them.

I’ll do some more head scratchin’ and report back.

JRH

[Tony_G]

Interesting - I wonder if there might have been a dry joint/grounding issue on the power supply and you addressed that by cleaning everything up and taking the caps out of circuit to check them? Probably a mystery that will never be solved...

Can you expand on what you mean about this when you get a chance?

It continues to bother me intuitively that there is currently a big hole in the attenuator assembly. At least some portion of it is “open circuit”. I don’t have a good enough understanding of things to describe what is happening, but having an entire section of the attenuator disconnected just “don’t seem ‘natch’ll.”. At least it confirmed that the 24 V rail is working.

TonyG

[jrharley]

Hey Tony_G,
Yeah, I think you're right. It would seem to be the only plausible explanation for the improvements in function. Hard to pinpoint precisely what it could have been, but a lot of stuff was done. Caps and a few other components were partially removed from the board an then re-soldered, switches, keyboard keys, connectors and pins were cleaned, I re-tensioned the pins on a few connectors back there, etc., etc....... I think the lesson I've taken away, is that there's no harm in giving these old kits some TLC as long as you're careful and meticulous in the re-assembly. We've all seen controls and connectors that get dodgy just from sitting around a long time. At any rate, something woke up!

On the text clip - Sure, it's kind of how this all started. You'll have to climb into the "way-back machine" and go look at the initial postings to get the context. When I first opened the 8656A up I found a bag of parts inside and an entire set of actuators missing from the attenuator assembly. All I have from them is the gold contacts.

Depending on the feedback I get, (I'm hoping b_dunham is following along), I'm thinking that I should secure things as they are and get after the task of having some actuator forks for the attenuator fabricated to see if I can repair that assembly.

JRH

[Tony_G]

Thanks - Went back and looked at that, so you're basically missing A9MP3 from here (you have the A9MP2 and A9MP1 parts):



Your best bet there is to recreate the part in a 3D modeling tool from an existing one you have and then see if you can find a place to print it (or do it yourself if you have a printer). The only other realistic approach is to buy a parts mule and then try and recoup the costs by parting it out after you've used all of the pieces you needed - I had to do that with an 8901B to get proper replacement parts for an 8902A that someone had dropped in shipping...

As for the way forward, the thing that is causing me to pause is that you're not getting a booting instrument with an abnormal indicator as described in BD1 - I think it would suck to spend all that effort to fix A9 mechanically to discover that there is a completely unrelated issue with regard to the rest of the instrument. That rail is a problem and I'd hate it to turn out to be some unobtanium HP custom IC that is causing the problem after all that work.

However, I don't have one of these so please take my feedback for what it is worth (clearly a few hundred bytes worth of electricity ) and I defer to BDunham7 (who I think has one of these) as a better expert.

All that said, if you can get me a model, I do have a friend that can print some out for me so I'm happy to send the model to him and ask him to print half a dozen for you.

Let me know.

TonyG

[Swainster]

The attenuator is one of the major differences between the 8656A and 8656B so I can't make any comments on it, however it sounds like your logic/MCU board is not yet functioning properly? (i.e. the board which is supplied by J2 pin 2). I'm with TonyG on this - if it were me I'd concentrate on getting the logic/control side fully working before worrying about the attentuator. I know that you started to tackle this around post #30 but I dont think that you have exhausted this avenue of investigation yet.

In this regard, have you measured the resistance to ground of what you refer to as the "H string"? Ideally you would use a 5 or 6 digit multimeter to see if there was any difference in resistance at the various caps, as the lowest measured resistance may indicate a nearby fault i.e. with the cap itself or the associated logic IC. Another option is the finger test - the faulty part may be dissipating extra power so touching the chip packages may be enough to identify the culprit, or at least narrow down the location.

[Tony_G]

For those following along at home, the A10 board delivers a 5V rail to the other boards in the SigGen:



The CPU board starts to get covered in Service Sheet 14 (the nomenclature being the H in the circle followed by the 14 to refer to the applicable Service Sheet).

JRH has said that the 5V rail only drops when he connects pin 2 of A10J2 up and that is the one that leads to the CPU board:



Now it goes to other places on A14 but there are a bunch of bypass caps on that rail and this is where I would start - Then I would go with Swainster's suggestion - If nothing else leaped out then I would probably begin to identify the relevant ICs touched by the 5V rail and work from there.

A somewhat slow slog if the unit is uncooperative in the fault-finding mission.

TonyG

[jrharley]

Hey HP Sig Gen’rs,

Great feedback and insights as usual, really appreciated! It’s a little hard for this old Bodger to keep up with it all, but that’s not a complaint, it just takes me some time to go through things and take it all in. 

I hear everyone loud and clear on solving the 5V rail, and I concur, don’t get me wrong. I'm afraid that my ignorance of this kind of equipment and circuitry, has my pea brain struggling to comprehend how an electrical gap in the attenuator section couldn’t be contributing to the symptoms I’m seeing. But I’ve been studying the attenuator schematic and I think I’m starting to understand why I shouldn’t be concerned. (Any insights on that front would likely be useful though). I think I’ll just continue to try and sniff out the problem with the attenuator cable out of circuit entirely moving forward, and just get over it.

That said, fabricating the missing attenuator parts is going to be a real task, in and of itself. That’s why I’m going to start the process of finding a shop that can do it. I’m thinking not just any 3-D print set up can do this properly. A precision scan will have to be done, the right plastic will need determining, etc. In any case, I won’t pull the trigger on fabrication until the 5V rail mystery is solved, but there’s some leg work to do on this aspect of the repair.

Tony_G –  Thanks for the printing offer, much appreciated! We’ll see how it goes.
I think I’ve beat the components along the H string to death at this point, I’m convinced there’s no component level issue along there.  All of those IC pins are showing a positive voltage of approximately the same value, 1.6 to 1.7V. By the schematic, I think it should be 5V or thereabouts. I'm thinking the issue lies downstream of this section. I’ve been looking at service sheet 15 for the next place to investigate, giving some “side eye” to chips U1, U22, U24 and U32.

Great test suggestions swainster, thanks! As for the touch test, I’ve done a good bit of that and haven’t found anything finger blistering yet, even after prolonged power up, but it’s a thing I’ll be constantly doing. Also, I measured resistance across all of those caps statically a while back and the caps were consistent in their readings, giving me some confidence that they’re solid.

Anyway, I’m taking a brief sidetrack with a cute little Spectrum Analyzer I acquired recently for a couple of days. It needs some mechanical stabilizing before I can shelve it.

Then I’ll get back to the HP. I’ll report back with any new developments as they occur, and always love the comments folks! Thank you!

JRH

[Tony_G]

I don't think you need to go all "high spec" on those actuators initially - You could just use a micrometer/caliper and simple 3D modeling tools to create a model that would work - That would get you an "operating" attenuator - You can then get the unit all running and see if you need to step up the attenuator components to have something in spec later.

Good luck with the SpecAn - Look forward to hearing the next updates.

TonyG

[jrharley]

Quick 8656A update to ponder here - Maybe someone can explain this observation:

When the unit is powered up now, with the A10J2 pin 2 CONNECTED,  The 5V rail reads 5.4V for a couple seconds, until the attenuator latches in, and then it drops back to 1.8V. (what part of the attenuator is switching in, I'm not sure). If I power up with pin 2 unconnected, it behaves as before.

I can't escape feeling like this is telling me something, these things are dependent on each other in some way guys.

If any one can prove to me why the attenuator can't be affecting the 5V rail, I'm all ears.

I may go dark for awhile following my gut here. Bear with me.

JRH

[Swainster]

Sounds good! I think the point is that the priority should be the problem with the 5V rail rather than the attenuator (which is mostly just a bunch of 24V relays from the control point of view). Of course, if the 5V rail problem can be traced to the attenuator then that is a different matter. Did you try unplugging the attentuator control input (J3 on the power supply board) to see if that makes any difference? Then again, erhaps the reverse power protection circuit is involved somehow - that appears to use a 5V relay

[Tony_G]

Swainster's suggestion is a good idea - The attenuator control section draws from the +24V Unregulated & +5V Regulated rails (as shown on SS7) and there could be a problem with the  ICs (like U6, U11, etc) that control the attenuator - Removing J3 might not solve the problem as the path to GND might be through the IC and not through the output of the IC and into the attenuator segment itself.

You might want to consider removing U6 & U10 as that should break the path from the 5V rail to GND completely - the signal/5V goes into U10 and then into the inputs of U11 & U12 before flowing through to A9.

Interested to see what you find.

BTW It looks like both those ICs are still available from Mouser et al

TonyG

[jrharley]

Back to the HP here -
As always, thanks for weighing in on the project guys!

I've got the little Spectrum Analyzer squared away for later work now and I'm just back from "Hamcation" in Orlando FL. I won't lie, I was scanning for an old junker 8656A, but didn't see one.  A good day though, got some supplies and got some fresh air. Now it's time to get back to the HP.

The J3 connector was actually unplugged when I first opened the unit swainster, and has been in and out of circuit a number of times during testing. You're spot on with it's importance. Having J3 in circuit, or not, does not seem to affect the 5V rail voltage discrepancy in this situation. Now though, if J3 is in circuit, I get some attenuator action. That wasn't happening before.

I'm just getting back into the SS7 schematic Tony_G, printed it out for study. That could take a bit.......

Thanks guys!

JRH

[Tony_G]

Great to hear that you had a good time and a bit of a break from the SigGen follies...

It's a pity that the design of the 8656A has the attenuator control circuits on the A10 board otherwise you could just take it out of the circuit and see what happens...

The ICs that directly control the attenuator are A10U7, U11 & U12 - These take a TTL control signal but will output the +24V to drive the attenuator.

The ICs that handle the control latching are A10U6 & U10 - The CPU will send control signals to these and they'll then work out what they need to send to the driver ICs to toggle the appropriate attenuator segments.

The A10U3 IC just enables the CPU to latch in the data before it is applied to the outputs of U6 & 10.

I would tend to think the issue, if it is in one of these ICs, would be in U6, 10, or 3 but it is certainly possible that U7, 11 & 12 could be drawing too much on the output of U6 & 10 that could be dragging the 5V rail low but I would expect that to cause U6 or 10 to go tango uniform.

• U7, 11 & 12 are SN75468N - Available currently for $1.04 at Digikey - https://www.digikey.com/en/products/detail/texas-instruments/SN75468N/380175
• U3 is an SN74LS123N - Available currently for $0.98 at Digikey - https://www.digikey.com/en/products/detail/texas-instruments/SN74LS123N/277282
• U6 & 10 are SN74LS273N - These seem to be popular/in-demand - Digikey & Mouser have 0 in stock (Mouser has 7K odd on order) and there are only 6 at Newark for $1.20ea - https://www.newark.com/texas-instruments/sn74ls273n/d-type-flip-flop-octal-dip-20/dp/60K6876
  

These are all probably available at Arrow, AvNet, etc though I didn't check as I have accounts with Digikey, Mouser & Newark.

I happen to have a Hakko FR-301 so I'd probably just pull U6 & 10 and see what happens - If the rail is still low then I'd pull U3 and check again - If that didn't fix the problem then I'd look elsewhere. If the problem goes away by pulling U6 & 10 and they pass testing (breadboard it up and try applying power and see what happens) then it could still be because of U7, 11 & 12 so you could rinse & repeat that.

Look forward to hearing what you find.

TonyG

[jrharley]

Thanks Tony_G!

I've been studying the SS7 schematic and your thoughts have helped me see the organization clearer, so, thank you for that.

I'll see what happens pulling U6 and U10, maybe U3. It makes sense to start with those ICs to me too, since removing connector J3 hasn't seemed to have had any effect on the symptoms during testing. I have a decent de-solder gun, so removing those chips shouldn't be too bad. (Gotta' clean that sucker up though!)

In the meantime I noticed something while checking the passives associated with the chips. Capacitors C13 and C14 should be 1uF. They're both associated with U3. I lifted a leg to confirm, but they're both reading at 2.3uF with what seems like pretty bad ESR at around 4ohms. Since they're 10% parts, that doesn't look too good to me. I think I ought to replace those guys no matter what, make sense?

It's interesting to me to finally find a component that doesn't test very good on this beast, since nearly everything tested so far has been acceptable. The resistors in front of each of the caps tested pretty much spot on @ 132.5K Ohms.

 After I pause to reflect on things for a couple days, I'll start pulling some ICs and report back.

JRH

[Tony_G]

They're the external timing caps for the IC - I'm not sure what would happen if you just left the legs lifted but you could certainly do that to see if they're the cause of the 5V Rail problem - You will probably have "undocumented" action from the attenuators but for the purposes of a short run to see if they impact the rail I don't think it would hurt.

TonyG

[jrharley]

HP8656A update on a Friday night!

A10 C13 & C14 replaced with brand spankin' new Vishay Tantalums now, because it had to be done. It could just be me, but the unit seems to "cycle up" faster now. Not sure if that's real, or just my perception.

Anyway, I've removed the A10 U6 chip, no problems. De-soldering gun is doing the deed nicely. The chip just fell out and the pads are clean. Love it when that happens!

After those moves, the power up exhibits a dark display and a 1.8 V .....5 V power rail. I'll remove U10 next and see what happens.

JRH

[Tony_G]

Had a bad week so I'm looking forward to hearing some news from you - Thanks for making my day a little brighter

[jrharley]

DRAT! Sorry to hear it Tony_G.

For what it's worth, I'm sure if you wait a minute, I'll do something so head slappingly stupid with this unit, that it would bring a smile to ANY techs face.

At any rate, I pulled the A10 U10 chip this evening. Same results, which is honestly not unexpected. That's not to imply that I actually know what the heck I'm doing here, just that my gut is telling me that I'm chasing an unlikely component failure. Incidentally, I've restored the J2 connector to original status and I'm thinking I keep it that way, with pin 2 back in place. I'm thinking it doesn't really matter if taking pin 2 out of circuit allows the 5V rail to get up. It's still got the problem. It's an interesting clue, but I think I move forward testing the unit with all wires in place.

The question I'm pondering now is, do I put U6 and U10 back in before moving ahead? Or do I pull the next victim with them still out of circuit. Maybe U3 should be done before I get to that question. Not sure.

I'm Also wondering if there's a safe way to "bridge" the missing attenuator contacts, to at least put the unit into a defined state. Not sure if that makes any sense or not. It just makes my teeth itch that there's no connection across that section.

As always, the feedback I get from you and others, is invaluable, and I really appreciate your interest in this long shot project.

"Chin up" Brother, hope things smooth out for you.

JRH

[Tony_G]

Thanks - Appreciate it.

You're not chasing a "ghost" problem - You know that the 5V rail is being dragged down by something, what you are doing is methodically checking components that can cause the issue - So I wouldn't consider it wasted time.

One quick question - Did you check U2 (located somewhere in the unit, probably towards the rear)? Confirm that the unregulated voltage stays well above 5V and the regulated side drops when Pin 2 is reconnected on J2? It's the 5V regulator and has the part number LAS-1905 (should be 5V +/- 5%) - It's not unheard of for the failure mode to be dropped regulation under larger loads. There should be a vast array of TO-3 regulators that can be used in place - Parametric search time... I'd be inclined to bodge something in place that matches the headline specs to see if the same problem goes on - Maybe measure the current being drawn on the regulated side to see if it is a reasonable amount, that sort of thing. It should be capable of providing up to 5A (probably some more but spec'd only to 5A).

You've tested the caps on the rail per SS14 and the caps on A10 so you should be expanding on what might be causing the problem, again by methodically ruling out components that could contribute to the issue.

It really is "Lather, Rinse & Repeat" now. Sorry

TonyG

[jrharley]

Thanks for the pep talk TG!

It's too funny that you mention U2! I was removing U3 last night, (no change by the way), when I happened to notice an IC labeled U2, down on the far corner of the A10 board. I hadn't noticed it before and I still haven't found it on the schematic, but it was real evident from the back of the board and obviously right in the power circuit. I dug into it a bit and it's an LM393 N, a voltage "comparator" apparently. (But still a regulator of sorts). I've always felt confident that the main regulator U2, located near the transformer, was fine, but I was unaware of this little guy.

So I buzzed out some voltages on it. The 2 outputs read 0 Volts basically. No output. Pin 8 read 23.4V. I'll check the inputs just for grins next, but it seems like the LM393 is knackered. This just might be a good thing, I'll report back when I discover something else.

JRH

[Tony_G]

Quick clarification - I was referring to this U2:



and not A10U2:



Without looking further into the crowbar circuit, I think that the output of the of the comparator should be negative or zero (I think SS22 said that the crowbar throws when the output goes positive but I'd have to go have a look - Yes, that is the case - My quick read is that the current over R75 causes U2 to go positive, which then turns Q2 on and triggers Q1 which blows F1).

It might be worth looking more closely at the TO-3 regulator as I can't see how the attenuator crowbar would affect the 5V rail and exhibit that with Pin 2 being connected/disconnected.

TonyG

[Swainster]

They are here on the 8656A/B, possibly covered by little plastic hats as the case is at the rail voltage on some of them. Actually, I was hoping that you could compare the current drawn on the 5V rail with the measurements I took a few posts back - the idea was to get an idea if the rail was collapsing due to overload or weak supply.
Pic from google (scribbling from me):


On my unit these are the part numbers (cross reference where necessary from various online HP references):


Ideally, U2, U3 and U4 should have plastic covers on them as their case has some voltage with respect to ground.

edit: huh, I guess that I still haven't worked out how to use in-line pics

[Tony_G]

[updated a bit for better clarity]

Great reminder there Swainster - You measured around 3.5A draw from the regulated 5V rail so JRH should have something similar I would expect - The part on the 8656A according to the OSM, is a LAS1905 (1826-0513) whereas your unit (or at least the one in the photo is using this) have an LM396K (1826-0966) - Don't see why that wouldn't work if the output spec for current matched.

JRH - You won't get the same numbers due to voltage etc but if you're drawing "more or less" the same than that I would seriously consider U2 being the issue and bodging in a replacement to test. To clarify what we're talking about here - You would expect that the LAS1905 would drop the voltage to keep itself safe per (or go nova trying to supply it and die):



If you're drawing closer to 10A then the problem is somewhere else I think.

Thanks again Swainster - Appreciate the community here, even though it isn't my SigGen

TonyG

BTW I've never got hosted images to work inline - That is why I use Imgur and the linked BBCode instead - If you ever do get it to work, please let me know what you did.

[jrharley]

Thanks for the U2 clarification Tony_G!

I had mentally skimmed over the A10U2 symbol as being some kind of Op-Amp or something! These schematics are foreign to me  “Ahn I will not lie.”
But that makes things much clearer now. It’s also very interesting to me that it’s involved with the attenuator protection “crow-bar” circuit. A lot of things have pointed to this area since the beginning. Recall that I found F1 bridged with a wire when I opened her up initially. I’ve tacked in a 2A fuse at the F1 position for now, and it hasn’t blown yet. But I think you’re right in that the main regulator U2, might need more scrutiny.

I’m hearin’ your hint swainster, thanks for the insight! I’m studying things and thinking about how I implement your suggestion and trying to keep up with you guys. In the meantime I’m occupied by stabilizing the “damage” I’ve done at this point, getting the removed IC pins and pads ready for re-install, improving solder joints, etc.

Speaking of appreciating the community, it’s really interesting to me how things are moving back towards the main regulators and rectifier in the troubleshooting discussion. The reason being, is that “Buddy”, of the YT channel “The Radio Shop”, did a real solid troubleshooting job on the power section of an 8656A a while back. Link is here:



I think I’m going to look forward to removing, cleaning, testing, and re-seating those TO-3’s. Peaceful work!

Always appreciate the thoughts folks! I’ll report back soon.

JRH

[Tony_G]

JRH - There is a good point he showed in that video - If U2 is delivering too much current then it should also be getting pretty hot (you can look at the lines that I had in the previous image I included).

Looking forward to hearing what you find.

TonyG

[jrharley]

I was thinking the same thing TG.  I realized I've never looked at the thermal condition of those TO-3's, they're under a shroud. After reviewing Buddy's video, it dawned on me as well.
Hey, it's Friday night! so I'll have a little more time to go at it. I'm also interested to test U2 with a power supply to see if it gets wonky with some over voltage.

Almost finished getting things in a state where I can move forward so, lot's to do and uncover. Me take pictures........

JRH

[jrharley]

Quick 8656A update here, work is on going.

Testing U2 with a power supply, current limited to .25A seems to say that she aint regulatin'. Once I climbed past 5 V on the input, the output tracked the input all the way up to 8 or 9 volts, when I shut it down, nothing else to see there. Seems to me it should hold the 5V pretty close.

I checked my stash of regulators for a test bodge, but don't have any 5V units, and definitely no 5 Amp beasts, but I do have a pile of adjustable LM317T's that I suppose I could work up on a piece of perf board to output 5 V. They're 1 Amp components, but maybe good for a quickie, not sure. The original LAS1905 TO-3 packages seem to be available, albeit from varied sources, nothing at the big parts houses that I could find.

I'll ponder it for a bit.........

JRH

[Tony_G]

Not sure that I understand the test setup - Sometimes these regulators, and I've never used the LAS1905 so I don't really know, is that they don't like being used without some form of load and some capacitance between the Vin, Vout and GND to stop the device oscillating.

This may not be the case here but I would go implement the example circuit in the datasheet:



And then change Rload to increase the current draw and measure both the current through and voltage across that load.

If that is what you've done and it isn't regulating then as you said, time to replace it - NTE (which I seem to find has very good availability in local stores) has a part that should work for a bodged in test repair - https://www.ntepartsdirect.com/ENG/PRODUCT/NTE931 - Those shops might even have a proper replacement spec'd to 5A.

TonyG

[Tony_G]

Thinking about it in terms of "bogdability" - You could probably also get away with a TO-220 package like:

https://www.digikey.com/en/products/detail/microchip-technology/MIC29500-5-0WT/771612

Might make it easier to get a test component. Will need to ensure that it has an appropriate heatsink etc.

[jrharley]

Thanks Tony_G! All good thoughts.

I tested U2 just like Buddy did on his video. As I recall, his U2 held things to just over 5V. I'll test it again and loosen up the current limit a little bit to see if that makes a difference, but at this point, I'm thinking he's dead Jim. I won't jump to that conclusion just yet though, I'm going to test the 15V regulators too. All the thermal compound has dried up on these guys, so I'm going to re-seat them anyway, might as well test them while they're out.

Some surplus houses seem to have NOS 1905's for various prices. One even had some 1905B's in stock for about $5 each. From what I can tell it is virtually identical to the 1905, just one small bit of data difference that looks to be better, but not important. Several sources have NOS 1905 units for around $10 each. I would like to stick something in there briefly though, before I order anything to see if the 5V rail perks up.

More to come...


JRH

[Swainster]

Quick 8656A update here, work is on going.

Testing U2 with a power supply, current limited to .25A seems to say that she aint regulatin'. Once I climbed past 5 V on the input, the output tracked the input all the way up to 8 or 9 volts, when I shut it down, nothing else to see there. Seems to me it should hold the 5V pretty close.

I checked my stash of regulators for a test bodge, but don't have any 5V units, and definitely no 5 Amp beasts, but I do have a pile of adjustable LM317T's that I suppose I could work up on a piece of perf board to output 5 V. They're 1 Amp components, but maybe good for a quickie, not sure. The original LAS1905 TO-3 packages seem to be available, albeit from varied sources, nothing at the big parts houses that I could find.

I'll ponder it for a bit.........

JRH

The 5V regulator is an adjustable, so wouldn't regulate if it was disconnected from its sense divider (R78, R79 on the A10 board). As you ran it up to 9V I'm guessing that you tested it out of circuit.

[Tony_G]

Hey Swainster - I don't think that is correct for the 8656A - And I need to defer to the fact that you both have devices and I'm just looking at the schematics & datasheet - The LAS1905 is a fixed regulator device according to the DS and the Schematic shows it being used as a fixed device - The schematic seems to imply that it has a virtual GND instead of an adjustable voltage.

That said, I'm not sure you can test it by simply attaching a voltage source on one side and measuring the voltage on the other but, as I said above I defer to your experience with the actual implementation.

TonyG

[Swainster]

Hi Tony,

I was just going by the schematic in the HP 8658A manual on manualslib.com. That said, even my HP 8656B has at least two different A10 boards, so it is quite possible that there is a fixed regulator version of the 'A'. Although... I thought the 'A' version also has the 5.4V adjustment pot? (funny thing, the 'B' has the "5.4V" pot, but despite the markings on the silkscreen, the manual says to adjust it to 5.25V)

Off topic, I recently acquired a HP 461A from the 'Bay and when I got it open it turned out that the output transistor and... the actuating lever for the attenuator had gone missing. Funny coincidence! I enjoyed watching your video for pointers on what the internals should actually look like (wouldn't mind a few more of those pixels though ). I've not got to the stage of performance tests yet - need to put my metalwork skills to work.

[Tony_G]

Yeah, I don't really understand what they were getting at with the 5.4V as it's labeled the same on the 8656A schematic:



To me what it looks like is a virtual GND for U2 as all rails are measured relative to the chassis GND. Anyway, the LAS1905 is a fixed regulator so JRH should use the same type of component if possible.

On the 461A - I went back and looked and I didn't take any photos on the S1/S3 Cam - Let me know if you want me to grab one for you - The unit is sitting right next to the bench and I'm happy to do that.

TonyG

[jrharley]

This is such great discussion guys! I'm compelled to comment, even as I'm in mid exploration on this segment of the project. But here's a couple developments, and a couple notes:

• Yes, the instrument rails, and pretty much everything else, are referenced to CHASSIS ground in this beast.
  The regulators can be tested in the simple manner described previously, unless I can't believe my "Lyin' Eyes" on the Radio Shop video.
  All these T0-3 packages are being tested out of circuit.

As for developments:

• The +15V regulator, U1, tested rock solid. I ran the voltage up to 18V and it didn't even flinch, holding at 15.00V on the DMM. It's good to go methinks'.
  I'm working on the -15V, U3 now. Right now, things are inconclusive, I'm not currently believing those same "Lyin Eyes"........ stand by as I test it some more.
  Interesting discovery regarding the heat transfer pads under the TO-3's. They're like nothing I've ever seen before, but others are probably experienced with them. Their form is like a traditional mica pad for TO-3's, but they're made differently for sure. They feel and look like they're a silicone coated, double thickness mica, definitely more durable than the classic mica pad.

Anyway, too much going on to download at the moment, but more details later. Awesome conversation guys, I really appreciate it!

JRH

[Tony_G]

Good to hear.

I'm sure that someone who does this all the time could chime in and comment but I've had regulators that will regulate correctly under no/light load but then sag under the expected load. The "proper way" to test them (in my view) is to use an Electronic DC Load (I have one that does up to 300W I think, would have to dig it out as it is in the "don't use often" pile of equipment) - You provide them the correct/designed input voltage and then vary the load up to the designed amount of current and ensure that the voltage doesn't sag.

What I mean in this case is that the unregulated voltage coming in is +9.5V to +13.5V and U2 should kick out +5.19V to +5.61V but that might be the case when U2 is providing 0.1A but when asked to provide 3.5A the voltage might drop to +2V. This would mean that it would "pass" when simply checked with a DMM but "fail" in operation.

Buddy's test was just the DMM version and not a load test - He did test whether or not it would stop regulating with higher voltage and he also pointed out that his U2 was drawing 21mA (from memory) which he, correctly, thought was strange - Where he sort of missed it was that the quiescent current of a LAS1905 expected is 10mA, so his was showing something different from expected but according to the datasheet the max was 25mA - So his U2 could have still be working correctly even though it showed something was beyond what would be expected.

This is why I think you need to actually put the regulator under the expected load to really test whether it is working or not.

All that said, if I didn't have either a DC Load or the desire to bodge up something to test that with the correct power dissipation, then I would just go ahead and replace it.

Looking forward to your next update.

TonyG

[Swainster]

Yes, you are correct, of course - its a 5V regulator. I guess that what I am getting at is that if you disconnect it from its load i.e. board A10, you need to make some other arrangement for the virtual ground (which is basically the same as an adjustable regulator with a 5V reference voltage). In short, to measure the output of A12U2, assuming you have completely disconnected it from A10, then you need to at least ground pin 8 of A12J1, or jumper over a ground to the case of U2. As you say, a suitable dummy load would also be needed for a reliable test. If no electronic load is available then some HP power supplies have a "down programmer" which can sink some current.

[jrharley]

All good points, well taken guys, appreciate the feedback. 
No doubt loading the components is the ultimate test. I've always had an electronic load on the "maybe someday" list, but I don't have one on my bench and acquiring one is unlikely in the near future. It's interesting to browse the Bay for them, they tend to be pricier than I would have expected.
Anyway, to the point - the 5V regulator seems to support exactly the scenario you've described Tony_G. Presumably when pin J2 was connected, it loaded the regulator down. Since it tested poorly with the DMM test also, I'm going to replace that sucker. That's enough evidence for me. Both 15V rails tested good at the test points on the schematic and with the DMM test, so I think I'll just re-seat those guys. (Although I'm still poking around with the -15V regulator, its pin-out is different, so I'm still looking at that for curiosities sake.)

As I've removed the TO-3's from their mounts, it's become clear that things were a little rough under there. It's turned out to be a good thing that I'm re-seating them with fresh thermal compound and cleaning the surfaces. I'll send a pic of the final install when I get a new LAS1905. My overall mindset on this unit is one of "reasonable restoration", so I'm inclined to repair and replace as I go, with the hope that I can bring it back into service, knowing full well that I could ultimately hit a dead end and have to scrap the effort. My fun comes from the journey and the learning. Thanks for helping with that!

More to come I'm sure,

JRH

[Tony_G]

Just dug out my DC load (at least enough to see it) - http://www.maynuo.com/english/pro.asp?tid=97 - These guys I think are the OEM for BK Precision (or at least there was some back and forth about who did what/stole what) - Still not an inexpensive thing to have but useful if you're building your own supplies rather than using off the shelf converters.

To add to the list of projects you have, you might want to look at what GreatScott did to build one:



He does seem to have a newer one in his latest videos but my SearchFu failed me in trying to find anything on that.

TonyG

[Tony_G]

In short, to measure the output of A12U2, assuming you have completely disconnected it from A10, then you need to at least ground pin 8 of A12J1, or jumper over a ground to the case of U2.

Absolutely, my apologies if I wasn't being clear enough in my post - You're completely correct to point out that the "in place" test needs to do those things.

Thanks again Swainster for the comments - Love hearing from you - Again, let me know if you need a more hi-res photo of that cam for your 461.

TonyG

[Swainster]

Hi Tony,

No problem - I am just worried that JRH's latest observations dont seem consistent with the rail drooping under heavy load but being fine at light load. Having the case floating while under test might explain the discrepancy.

Regarding the 461A I am thinking about starting a separate thread, but first waiting for the postman to deliver some RF transistors as possible candidates for the missing output driver.

Hi JRH,

as far as electronics load go, I find that I rarely use them, but on that rare occasion then they are indispensible. I actually have 2 (or maybe 2 and 1/2) electronic loads. The one I'm calling "1/2" is actually a 10k pot cobbled together with an op-amp and a random power mosfet with heatsink from my parts collection - this was improvised to give me a controllable few 10s of mA load for a project, once I had got fed up with playing with combining resistors is various combinations (I also didn't want to risk my decent resistor decade boxes on prototype circuits). The first "proper" -ish load was an aliexpress 80W load like the one discussed on a popular thread here on eevblog. Finally my most recent load is a huge 19" Chroma mainframe with a single 300W module installed (and 3 empty slots for future expansion). I bought this one when I needed a bit more oomph and it was available locally and cheap - I think it was less than USD150. The type that Tony has actually looks really convenient, but like you say, that form factor at least always seems quite expensive on the 'Bay.

I just recalled that I have another home made load which is just a bunch of power resistors mounted on perf board with a fan strapped to the front. Its got a pretty limited input voltage as the fan is driven directly from the load voltage  - it was a another project specific 'special' to stress test a 12V 2A supply for a couple of weeks. Anyway, the point is that you can work around not having a fancy electronic load (but they are nice to have available).

For a another example of a self built load, Kerry Wong has a nice video. I have a feeling that he touched on MOSFET SOA and the difference between linear and switching power MOSFETs

[Tony_G]

Oh yeah, I forgot that Kerry did the video on those - He has some great videos on loads & power supplies - Well worth the sub and watching.

TonyG

[jrharley]

Thanks swainster, you guys know my weak spot for any new project!

I've got a beefy old case from an Astron power supply that I parted out a few years back, that would be a great home for a DIY electronic load. Although they are only the basic guts of a working unit, what do you guys think of those $20 bag of parts kits you can see on the bay and other places, bangood, Aliexpress, Amazon? (Not the USB loads). Just curious since I have enough salvaged parts packed away to probably create a power supply, display, etc. to put something like that together. Plenty of old fans and heat sinks too.

At any rate, new old stock LAS1905 regulators should arrive this weekend and I'll stick one in to the 8656A for the next phase of the journey. Should be interesting no matter what happens. Seem to come from a reputable source in the States, "warranted", and came in at about $12 each, shipped.

JRH

[Tony_G]

JRH - The journey is as important as the destination - So if grabbing an electronic load kit for $20 from a website gets you enough bits to start building your own load then go for it - I just wouldn't put any stock in the numbers they rate and in the actual working of the unit - But, and this is the journey part, getting something that doesn't work, learning why and then correcting it is worth the time as well.

Just make sure that you characterize the device before you depend on it.

TonyG

[Swainster]

Hi JRH,

The other day I tried to list all the projects and repair candidates on my to-do list - I realised that I probably need a spreadsheet to keep them all organised

I'm not sure which kit you mean, but if it's the simple constant current source based on LM324 (i.e. no microcontrollers involved) then that looks like a fun project - for that price (seems to be a low as USD10) then I'd be tempted to add it to my next aliepress buy just for fun (actually you could build it on prototype board but the kit PCB would save a lot of donkey work).

Anyway, if it is that one, then be aware that the kit mosfets (110N8F6 in the listing that I was looking at) are not specified for DC loads i.e. no DC Safe area of operation. You can't expect reliable operation when operating outside the datasheet specs so treat it as a bit of fun and maybe see how far you can push it before the supplied FETs pop. At that point you can investigate SOAs and linear power mosfets etc, and maybe identify some more suitable parts (though proper linear power mosfets are not cheap).

Basically I'm just paraphrasing what Tony says.

[Brumby]

The other day I tried to list all the projects and repair candidates on my to-do list ...

I'm too scared to even consider that.   

[jrharley]

Thanks swainster! I would never have thunk someone would use a device on a DC load, that wasn't rated for DC loads! 
The kits generally do seem to use those devices though.
How do you tell on the spec sheets whether they will have a DC operating zone?

Thanks again for all the comments

JRH

[Swainster]

Hi JRH,

For DC operation I just quickly scan through the datasheet looking for the graph with a 'house' shape - typically titled something like "Safe Operating Area". If it doesn't have a DC line/contour then it pretty much means that the manufacturer has optimised the silicon for switching performance so much so that the DC performance is not worth even worth specifying. Not all switching MOSFETs are this highly optimised so its worthing checking for this graph even if the MOSFET has "switching" in the description.

Note that SOA is not just a MOSFET thing - it is fully applicable to power BJTs as well.

At the opposite end of the scale are so-called linear MOSFETs. As linear operation is not so popular, these tend to be rather pricey. I've attached a quick comparison of the 110N8F6 with the same package similar voltage ixys linear MOSFET to show the difference.

Just to be clear, you can use non DC specified MOSFETs in a DC application, but if you do, then it is up to you to characterise the safe operating conditions (and they may not have much relationship to the datasheet numbers)

[jrharley]

Great info swainster, thanks!

I had a few minutes to study the charts and it's much clearer what you mean now.

Tony_G is right about the philosophy of the journey for sure, I'll have to start planning the next one! Even though like you, brumby, and I suspect everyone else, there's no shortage of projects on the spreadsheet!

Hope to put in a new 5V regulator in the 8656A tomorrow night, so more to come.


JRH

[Tony_G]

Looking forward to what you find when you swap out the regulator.

TonyG

[jrharley]

SA-WEET EVER LOVIN' LIZARD !.....Look what we have....................................


JRH

[Tony_G]

Outstanding - Congrats - Now you can get to work on that output attenuator...

Or you can intercept the feed into A9 and measure there (which might be non-trivial as the schematic seems a little vague on whether there is soldered hardline or connectors).

Either way, great work.

TonyG

[Brumby]

   That's what my 8656B looks like after power up as well!

I love seeing old HP kit up and running!

In fact, I have another (incomplete) 8656B that will likely end up as a parts unit - so if you are in need of anything desperately enough to cover shipping* from Oz, let me know and I will see if I have it.

* If it's only little, I could handle that.

[jrharley]

Thanks Tony_G, brumby! And everyone else, jeez, what a fascinating Rabbit Hole! All the input has just been invaluable!

I've had a chance to re-fit the faceplate and stabilize things enough to flip it over into a proper operating posture, so I can play with it a bit.

It SEEMS as if the keyboard is functional. I'm able to set and adjust frequency, amplitude, and modulation like you would expect. I'm thinking this bodes well for control overall.

I've attached a couple pics of the regulator mount just prior to power up with the new 5V regulator.

More to come I'm sure,


JRH

[Tony_G]

Looks nice - It cleaned up well.

Just a reminder, if you can get me a model I should be able to get some of those actuators 3D printed for you.

Also, those keys should be the classic Bill West design so if now you're pressing them and you find they stick then CuriousMarc did a video on cleaning them on his YT channel.

TonyG

[Swainster]

Great news indeed! Interesting that the screw retains the cover on the 8656a. On my b version, there is just a clip on plastic cover for each package. Also, I don't recall the paint being removed where the regulators are mounted, but I haven't needed to do any work in that area.

Hi Brumby, spares unit you say?... Actually my 'B came with a spares unit which is only missing its power transformer (and a burnt out inlet filter). I understand that the previous owner of the 2 units 'repurposed' the transformer for a DIY project. I'd be interested in buying a replacement transformer to complete the unit. That said, posting a heavy item like that to Singapore is probably pretty painful on the wallet.

[jrharley]

Thanks again swainster, Tony_G, et al, it's a good place to pause I think.

I'll spend a little time cleaning up the scene of the crime, documenting the project file a bit and getting the shop set up for the next challenge on the 8656A. I want to run it through the primary checks for instance, to see what I can learn. Then of course I can start to look at the attenuator. Appreciate the offer Tony_G, we'll see how things go.

In the meantime I'm going to see if I can cal up the little BK Precision spectrum analyzer. I got it for a song on the bay recently. It looks like it had been mishandled by impatient hands, but after some work on things like shafts to potentiometers, it seems to be functioning on a basic level. The "calibration" procedure looks to be pretty straight forward and done from the front panel.

Report back on the 8656A in a bit.......

JRH

[Tony_G]

Fingers crossed for your SigGen workout.

I couldn't tell from the photos or the schematic, do those missing actuators disconnect the N-Type on the front from the rest of the generator or are you just missing the ability to add in attenuation?

TonyG

[jrharley]

Appreciate it TG. Here's a couple pics of the attenuator cavities on first open for reference. We'll be gettin' the sleeves rolled up for this area come this weekend. (I'm not sure I'm understanding your question just yet).

Had to LOL at myself upon diving in to the Spectrum Analyzer calibration. The irony is, that I really need something like the 8656A to do said calibration! We'll get back to that later....

Meanwhile, I'm lettin' off steam by rubbing on the cabinet of a 1948 Delco R-1233, all American five radio. Hoping to get it ready for car wax before weeks end. There's an astonishing amount of tar on it

Later......

JRH

[Tony_G]

I was wondering how the actuators work to switch in and out the various resistors for the attenuator segments - It appears that they slide back and forth rather than lift.

With those missing parts then you won't actually get a signal out of the N-Type connector on the front of the unit (ie while the unit may "work" there will be no signal).

Sorry, I can't help with removing that cover.

TonyG

[Swainster]

Haha, I was thinking that the spectrum analyzer would be useful to cal the sig gen, but seems like you have a chicken and egg situation.

Its always interesting to see how HP and others combined electrical and mechanical engineering, especially in older instruments. My Fluke AC calibrator has a front panel with belts and cogs and so on, all working together like clockwork, and of course there is the famous HP 8640 sig gen (I saw one languishing in a local surplus dealers office, on the random kit shelf).

Regarding your missing plastic parts, from the look of them I'd guess that they are made of injection molded POM (delrin). Sourcing salvaged parts maybe the best option, but if I was forced to tackle their replacement myself, I'd be concerned about using a MSLA resin 3D print as I find that it has poor long term stability (though that could be just be due to my lack of skill and using cheap resin). I'd rather try using an FDM (filament) printer, but I'm not sure if it could manage the small details. The advantage is that, although rare and difficult to print with, it is possible to get POM printer filament. Nylon of any sort may not be suitable for this application (due to being too flexible and also absorbing water may affect the attenuator performance), but it's a relatively popular 3D printing material so probably worth a try. A custom machined POM part would be great, though likely rather costly. It would also be difficult to exactly match the injection molded part as molded parts are "negatives" based on a machined tool i.e. it's difficult (expensive) to get a sharp radiused inside corner on a milled part but easy to get a sharp external corner. The opposite is true for a molded part.

[jrharley]

Always great comments guys, thanks!
Looking forward to diving back into the beast this weekend. It'll be mostly eval and dis-assembly.

Yes Tony_G, those actuator forks slide back and forth. They make a very charming "clack" when they engage. At the same time, it's "a bit how ya' doin'" by modern standards. PS - I can confirm that there is currently NO output signal.

Wow swainster, thanks for sharing the fabrication and materials knowledge! I'm following you 100% on that. These are going to be tough parts to acquire, no matter what the route is. Should be interesting.

I have discussed the problem with one local outfit that seemed to be a good fit for the task..... Shared pics, drawings etc. with them.

They ran away, screaming.  No worries, there are others. Haven't even scoped out the cost scenarios yet.

As always many thanks for the input(s)!

JRH

[Tony_G]

Seeing that they slide back and forth they shouldn't be under that much mechanical stress (famous last words) and as long as you use a material that doesn't impact the RF in the cavity you should be able to make them "fairly easily" - The ends being the most important part I think.

[jrharley]

Yeah, I suppose an "exact replica" isn't necessary TG. Good points, we'll just have to see how this roles.

I believe swainster hit it on the head with the Delrin, injection molded supposition, I think that's exactly what the parts are made of. But to your point TG, Nylon should work.

A model of the parts is going to be the first step on the fabrication route, no matter what, it seems to me. It also seems to me that the cost threshold for fabricating the parts is about $300. Beyond that, I should be able to get a parts unit for in that neighborhood. The trouble with having to go the parts mule route, is that the attenuator design has changed over the production life of the 8656A. Whether or not these parts were kept consistent in form or not, remains to be discovered.

Anyway, I continue to ponder the HP as I keep rubbin' on the Delco.......................................

A Hell of a lot of rubbin'......

JRH

[Brumby]

Hi Brumby, spares unit you say?... Actually my 'B came with a spares unit which is only missing its power transformer (and a burnt out inlet filter). I understand that the previous owner of the 2 units 'repurposed' the transformer for a DIY project. I'd be interested in buying a replacement transformer to complete the unit. That said, posting a heavy item like that to Singapore is probably pretty painful on the wallet.

Partially pulled out the transformer.  Pictures below.  The cut wires were already that way.  Weight is over 2kg which puts it into the range of AUD$62 - AUD$68 for basic shipping.  (Actual cost will depend on packed weight.)  I can include the cover and mounting hardware.  I would be inclined to supply as shown up to the DB15, cutting the 4 smaller wires (long enough to be usable).

If you are interested in proceeding, let me know and I'll do some electrical checks.

[jrharley]

HP 8656A update here, thanks again for all the feedback and support from everyone!

I've been diving in to the attenuator assembly, trying to understand what makes it "tick". I've made some progress, learned how to take the actuator forks out and witnessed first hand the o-ring failure that has been well documented by lazyelectrons. (www.lazyelectrons.com).

All of the remaining actuator forks in my unit have cracking o-rings. We're going to be re-building the entire attenuator assembly, there's just no escaping that. I've been working on how to replace my missing forks, we'll see how that goes. There may be a solution developing as I write this. "I get by with a little help from my friends"....... We'll see!

So, for anyone interested, I'll try to document the re-build here, starting with the attached pics.

JRH

[Swainster]

Hi Brumby, spares unit you say?... Actually my 'B came with a spares unit which is only missing its power transformer (and a burnt out inlet filter). I understand that the previous owner of the 2 units 'repurposed' the transformer for a DIY project. I'd be interested in buying a replacement transformer to complete the unit. That said, posting a heavy item like that to Singapore is probably pretty painful on the wallet.

Partially pulled out the transformer.  Pictures below.  The cut wires were already that way.  Weight is over 2kg which puts it into the range of AUD$62 - AUD$68 for basic shipping.  (Actual cost will depend on packed weight.)  I can include the cover and mounting hardware.  I would be inclined to supply as shown up to the DB15, cutting the 4 smaller wires (long enough to be usable).

If you are interested in proceeding, let me know and I'll do some electrical checks.

Hi Brumby,

Yes, I'm definitely interested but so as not to distract from the restoration of the 'A' I'll drop you a private message. Or at least, I'll give the PM system a try as I've never used it before.

[Swainster]

Hi JRH,

I was just thinking about how I'd have tackled this a few years back, before I picked up CAD. When I was back in the UK I would have just popped out to the shed and grabbed the nearest piece of scrap plastic (maybe a plastic ice-cream tub lid or an old chopping board) and started carving away with whatever tools were laying in the bench. Seems like the fork at least might lend itself to this method of improvisation. The cover might be easier as a 3D print

[Tony_G]

I'm wondering if you could, as a test, not long term, get away with just pushing a cam into some modeling clay and then filling the resulting void with epoxy - Little trimming and sanding like Swainster said and you might have something that would at least give you the ability to do some testing while you look for a full long term replacement.

TonyG

[jrharley]

I hear you guys, and I've been mulling over DIY fabrication options as well. Anything from direct milling out of plastic stock to resin casting are options.

In studying the assembly though, such an approach would be a heavy lift in order to achieve the precision needed I think. Not impossible, but one would spend a fair amount of time doing 2 forks and a cover, (which holds everything in place very precariously).

I've attached a sketch of how this crazy contraption switches. It relies on an actual deformation of the little gold contacts, to make or break the connection. Pretty wild.

As always, I really appreciate the interest and the feedback!

JRH

[Tony_G]

Interesting - I wonder if there was a more complex and tolerance dependent way they could have done it...

TonyG

[jrharley]

I know right? It's a pretty whacky gizmo.
I can't imagine what advantages this approach might have had, but who am I to second guess HP! It's a product of its time I suppose.

Anyway, close ups of an actuator fork are attached for those playin' along at home.........


JRH

[Tony_G]

Do they list the torque value anywhere? I don't see it in the manual.

[jrharley]

I can't remember where I saw that TG. Perhaps on the lazyelectrons site. Upon further reflection, it could also only apply to the later versions. I know I saw it somewhere though.

I suspect that ultimately though, the specification is going to be something like ....."just right". I don't have any way to torque a screw down to precise inch/lb values. Not likely to buy a device that could do that either. I'll tweak it under power till they work!

I've pulled all the covers off and removed all the forks. I'll be spending some meditation time cleaning up the cavities and rebuilding the intact sections while I figure out how to get some replacement parts. May have to park it for awhile, we'll see. O-rings are going to be a short term challenge. I have good dimensions, but need to source some new ones.

Thanks so much for your continued interest in this journey, we may just get it running again yet.

JRH

[Swainster]

I've not seen how they work but guestimating from your pics they look roughly the size of a 3mm cable grommet

[jrharley]

Not far off swainster, thanks!
It seems that the right size for replacement has an outside diameter of 5-6mm and an inside diameter of 2.5-3mm. The tricky part in this case is the cross section of the ring. Based on my measurements of the area at the tip of the fork that retains the ring, the 2 o-rings I'll be replacing the original with, should not be much more than 1mm thick. I've been experimenting with some rings from the stash that fit pretty well, but they're too thick to place two, side by side, at around 1.7mm. The original o-ring was a single ring, with a slot around the circumference. I think the actuator plate from the solenoid rode in this slot on the o-ring, and either pushed against one side, or pulled against the other, to slide the fork.

It'll take a little doing to find the right rings, but we'll get 'em. Could be time for an account at McMaster-Carr...... (The kits all over the interwebs aren't all that appealing so far).

JRH

[Tony_G]

+1 on McMaster-Carr - Have bought from them before and they've always delivered quality parts - This is something that I guess, my economic privilege, allows - Buy from professional providers and you'll get what you order - Buy from eBay or Ali and you'll get what you pay for, maybe...

TonyG

[jrharley]

HP 8656A drama continues...

McMaster - Carr is awesome. I could quickly order 3 different sizes of o-rings, 100 to a pack, for less than $20 including shipping, delivered next day! The "Goldilocks" ring was right where expected in the lot and fit good.

Still too thick though. Getting 2 of them to fit on the fork was not happenin'.

The original part is really best described as a "Grommet". But it's a pretty small one. Haven't seen anything that would be a direct "drop in". No worries though, I think. I found that I could shave the o-rings down to about 1/2 their width and get a pretty good fit. Things are seeming to fit together pretty well in the cavities.

Pics on the way.

JRH

[Tony_G]

I had no idea that there would be a store like this but maybe you can find something here

https://www.theoringstore.com/store/

TonyG

[jrharley]

Thanks Tony_G! I did come across that company as well. Could be an option. Their size classifications and descriptions are a little impenetrable, but they could ultimately be the right source, we'll see.

Meanwhile, I'm just about re-fitted with the existing assembly. I've figured out a way to use the o-rings I have, into at least a shot at getting something to work.

Update pics attached.

Thanks always for the feedback and interest!

JRH

[jrharley]

HP8656A update here ya'll,

I've got the existing components in the attenuator assembly re-installed with new "grommets". Cleaned the traces, and bolted everything back down. We'll see if things are back in place correctly, it was a little surprising how easily the assembly went back together. I still have the gaping maw of the missing section to deal with, but its siblings are at least now; all ready for a power up.

As for what to do about the missing parts, I've reconciled the available options for my situation I think. Tough choices no matter what. Balancing time, money and feasibility seemed to point to a parts unit as the best option overall. It should arrive Monday. As an interesting aside, I naturally did what anyone would do and said to myself, "how much is this thing worth versus just getting a modern one?" Pokin' around the web quickly affirmed the decision to spend just a little bit on this beast. Using bandwidth alone as a comparative criteria, even a modern unit from Asia is north of a $1000. I'll take the 8656's interface any day for that money.

Meanwhile, still rubbin' on the Delco......... I'll post up when the A9 board is re-populated and we're ready to see if we've got any "clackin'" going on.

Later,

JRH

[jrharley]

8656A update -

Well........ in a nutshell, everything has changed with the arrival of the parts unit. Much to my surprise, the parts unit appears to be functional. Since it doesn't make sense to destroy a functioning attenuator to fix a broken one, we are re-thinking everything at this point. It will be a little bit before I can figure out how to proceed, but in any case, the strategy is now to make the best unit I can out of the two units I have. I'll report back with progress when it's informative. Appreciate the interest in this project and all the helpful feedback!

Later,

JRH

[Tony_G]

It's always the way, you go from less than one working unit to more than 1 working unit

I would probably approach this by addressing anything majorly wrong with the working unit - By that I mean, are any chassis rails bent, how is the N type on the front, Fans not working, etc - Swapping good parts from the non-functional unit into the working one.

Then I'd address the cosmetic issues like case panels, front panels, buttons (if there are labels, wear, etc), and so on. The goal is to get the best looking and working unit and then you can either keep the parts mule as a backup or sell it (and clearly state what is missing, damaged, etc).

Personally, I tend to keep the parts mules because this stuff is old, and having it go tango uniform is not unexpected - When I end up with 2+ parts mules then consolidate and sell off or recycle so I have 1 parts mule.

Look forward to hearing what you end up doing - And the follow-on test/calibration cycle...

TonyG

[jrharley]

Of course Tony_G, all of that makes perfect sense in light of the new situation. It’s always very interesting to observe the differences between units, with each one having a different story to tell. Here’s what I’ve observed so far, with “unit A” being the original unit, and “unit B” being the new unit:

The two case designs are different, with unit B more closely corresponding to the ARTEK manual drawings. Unit A has better component cosmetics all around, so combining them is going to improve things.

Unit A’s filter caps are in pretty good shape, unit B’s, not so much. They seem to have seen some heat in the past, though they don’t seem to run hot with a quick test. The listed 13,000 uF cap in B has been replaced with a 24,000 uF unit, which doesn’t test great, …ok, but not great. My current thinking is to just swap cap sets.

The F1 fuse in unit B appears to be original, but is 5A @ 125V, not the listed 4A.

I’m sure that the attenuator grommets in unit B are wearing out, as they do, but I’m going to let that Dog sleep until I can’t.

I will be swapping out the faceplate/keyboard assembly from unit A onto unit B. along with feet and handle hardware.

Quick aside, and maybe I should post this on the HP forum, but does anyone know what the “warm gray” color numbers are for this era of HP gear? They used it a lot back in the day (80’s) and I assume it comes from the Government Standard 595 colors, but don’t know. I’d like to source some touch up paint in that color. The “Beige” color of the same era might also come in handy.

Anyway, that’s just a little “color” on what’s going on. Always appreciate any feedback.

Later,
JRH

[Tony_G]

There are a number of posts on the HPAK group about the grey color - Worth just doing a search.

TonyG

[jrharley]

Thanks for the link Tony_G !, I've joined the group and have started to browse around. Nothing "drop in" as a solution for that paint color yet, but there are some source references that could prove useful.

Anyway, posting up with an 8656A update here on a Sat-dee night!

Feelin' pretty good about where I'm at right now; a functioning sig-gen is nestled into its spot on the bench, awaiting some tests. And it's lookin' pretty snappy while doing so! The "swap about" went well and I've buttoned up the original unit for storage, complete with a status sheet to remind me of how it sits whenever I re-visit it for whatever reason. The swap was basically 3 stages: Filter cap swap, Faceplate/Keyboard assembly swap, cosmetic parts swap. The attached pics are about the first swap step. Next pics will be the keyboard swap.

Later,

JRH

[Tony_G]

Good to hear - Look forward to seeing shots of the finished work and then the results of testing...

They have several posts over at the HPAK group on that color but HP changed the color over time so you have to find the right era to ask. You could always try just taking your panel to a body shop and having them scan the color to give you some matched paint - Might not be the exact new color but would be the right color to blend in on the unit as is.

TonyG

[jrharley]

Yeah, we'll keep pondering the paint thing as we go along. Should be able to come up with something. If I get a good solution, I'll post it up for reference.

I'll be doing a basic start up check on the HP in the waning hours of tonight that I have left. Then I'll start to see what section V of the manual has in store for me for "maintenance adjustments" in the next couple days. So far, the beast seems to be pretty accurate. But then, I don't push it by any stretch. I mostly live with frequencies from the FM band down. A couple pics catches step 2 of the process.

As always, thanks for the feedback!

Later,

JRH

[Tony_G]

Looks good for its age - I have a bunch of spare feet if you need a couple (most of my gear is in a rack so I don't need them).

hoping for some good results from your tests.

TonyG

[ekoloski]

Looking great on the restore!

It makes sense to take the best parts from the two and get one unit back into full working order.

The question then becomes, what to do with the spare? Keep it for parts, or fix it up as a standby. From the sounds of it its just a few caps and the pesky attenuator that need to be repaired? I haven't seen the innards of the attenuator in this model, and I've only got a B to reference, which is entirely different. Could you post a picture of one of the forks?

Looking over the previous posts debating the feasibility of 3D printing a replacement I have to wonder if a FDM print using ABS or HIPS may be an option. They're more difficult to print with but may be better suited. Unfortunately do still absorb moisture to a degree, although less than PLA. Admittedly I am a newb in the 3D printing world, I'm sure someone else can chime in with their thoughts.

[jrharley]

Thanks for weighing in ekoloski!
I got a notification a few days back that showed you had commented, but I couldn't see the post for some reason, didn't seem to show up. Anyway, appreciate the comments and encouragement. While documenting and packing up the original unit for storage, it did occur to me how close the guy was to being up and running its own self! I'm confident it could still serve. Meanwhile, it'll be a good back-up for reference if nothing else. As for the forks, there's some pretty good pics of one right at the upper part of this page, reply #151 I think. Browse to either side of that for additional pics on the attenuator and parts. I'm not a 3-D printing ace either, but the more I examined the forks, the less confident I became that they could be successfully printed.

Thanks Tony_G, appreciate the thoughts and the offer! I never forget! 
 I'm going to take a brief side-track before getting into the maintenance tests on the HP though. I've got that A-300 tube amp to move to the next level. But for what it's worth, several spot checks on the output through my digital scope, showed the 8656A to be bang on with the frequency. I was able to calibrate up the little BK Precision Spectrum Analyzer, no problems. It involved switching back and forth between precise db levels. The HP is awesome for something like that. What a pleasure to use. It feels like my HP 15C calculator, which I still use daily!

Final cosmetic step pic attached.

Later

JRH

[Tony_G]

Looks nice, well done

[Tony_G]

I was taking a look at the performance tests and you're probably going to be a bit SOL on a lot of the tests seeing that you don't have the test gear.

That said, you should be able to do an OK job on the frequency and power output with that BK SA - I've never used one of those but you should be able to do the harmonics with it as well. Not sure how it will go when you start getting down to low-level signals.

The modulation stuff needs some form of modulation analyzer (the manual calls for an 8901A but if you have access to an 8902A then that unit is basically an 8901B with more features, mechanically very similar and I used one to fix up an 8902A I have) - A few tests call for a 11715A AM/FM source and these things are rare and costly to acquire (going for ~$4K on auction sites). I wouldn't bother with those tests.

I think you'll mostly be fine with power & frequency accuracy, the only thing you might need is a true frequency counter to get the 10MHz reference tuned correctly but you can actually work around that by using something like what Alan does here:



Look forward to seeing how you go.

TonyG

[jrharley]

Awesome Tony_G! Very interesting stuff.

A brief glance at the test procedures in the manual, did make my eyes roll back up into my head, but I haven't really had the time to focus on it yet. Thanks for the preview!

I've got to take a break from my passion for a few days to travel up to Georgia to help my Mom. Family stuff. Should be good to clear the mind.

Anyway, I've got a decent counter on the bench, but the video really has piqued my interest, so we'll see how she goes when I get back. I'm not going to be a metrology Nazi on the HP, it seems to be working great for my needs, but I do want to get it as ship shape as possible. I'll post up with anything of potential interest.

Have a great Easter break all......

Later,

JRH

[Tony_G]

Hope you had a good break and things with the family were good too.

If you look at the stats on the SigGen it really is only +/- 1dBm or so which might be within the specs of your SA.

Modulation is really the only thing that you can't specifically measure but you can work around that to get an idea of the output by modulating a signal and then either analyzing the audio output (which is basically what the 8901 does when it demodulates) with a PC audio card - This won't be accurate but really for non-lab work it is close enough.

Look forward to hearing from you with pic's of the cleaned up working unit

TonyG

[jrharley]

Hey Tony_G, thanks for that. They're as good as they can be I suppose. Anyway, appreciate the thoughts on the testing as well.

The BK 2615A Spec An is a really basic unit. I got it for less than a Franklin, shipped. BK Precision seemed to have a bit of a niche back in the day supplying schools and training programs with reliable, rugged equipment that handled the basics well, were easy to use, but dispensed with the bells and whistles. This seems to be one of those. No tracking generator, no start/stop, just span. It's perfect for me to get my feet wet with this kind of analysis. I think the resolution is +/- 2 dB. (on a good day  ) As a side note, my analog scope is a BK precision 30 MHZ, 2-channel, of the same ilk, and I've loved using it. I got it as NOS at a Hamfest for $30 a few years ago. It still had the screen protect film on it. No complaints from this bodger on the BK gear. It should get me in the ballpark no?

As for the 8656A, I'm probably a week away from looking at the performance. Right now, I'm elbow deep into a tube amp chasing down some missing voltages. Why does that sound familiar? 

Thanks for the interest and I'll post up soon on the test set-up.

Later,


JRH

[jrharley]

The Fool for the 8656A  Sig Gen Declares “Victory”–

Four score and seven months ago……..  I started trying to get an old EBay Sig Gen back into service. While I wouldn’t ever claim that it’s ready for a real Lab at this point, I’m pleased to report that for the time being, this thing is doing what I need it to do. That’s victory to me! As is often the case with projects, the journey never goes exactly as one might expect. Such was definitely the case with this one.

Thanks so much to all of you folks who weighed in on this effort! -bdunham7, swainster, et al. It’s great to know that there’s a community of like minded brothers and sisters out there who appreciate the efforts to keep this old stuff alive. Special shout out to Tony_G for his particularly comprehensive and consistent feedback on the project. The journey benefited from his input immeasurably. I encourage all following this thread to check out Tony_G’s work on You Tube. He has a number of thoughtful and knowledgeable videos worth watching. - https://www.youtube.com/@tgsoapbox/videos

The first attached photo shows the 8656A outputting a 90 MHZ, -20dBm signal into my 100 MHZ frequency counter. Second shot shows the counter display. I think I’m good with this for now. I tested the output down into the audio range, which is where I’ll most likely be using the thing honestly. This is, of course, well below its specified bandwidth. I’m very pleased to say that it was very stable and accurate down into the lower ranges as well.

As for the Calibration on it, I could do a few of the maintenance adjustments with the gear I have, but not all of them. At the moment, I’m going to move on to my bloated project queue, before I worry about trying to eek another digit out of the HP. Simple priorities at work there. That said, I welcome any inquiries about my 8656A experience, if I can help anyone out with their own efforts by way of photos or observations. etc.

On to the next boat anchor........

Later,

JRH

[Tony_G]

This is great to see - The journey is worth even more when the destination turns out to be what you wanted - Happy to help out however I could on this - Congrats.

You can now get your full member badge for the TEA thread by acquiring the gear needed to "calibrate" this and then the gear to fix and calibrate that gear and, in short order, you'll have a bunch of other jobs written up on the whiteboard for you to do when you have "free time"

Look forward to seeing your next project post...

TonyG

[Swainster]

Well done JRH, it's always satisfying to finish a long running project (or even a short one, so I'm told )

Just in case anyone is interested in more HP 8656x action, I've started a thread about my own signal generators: https://www.eevblog.com/forum/repair/hp-8656b-repair-attempt-ongoing/. It wont be as interesting as this thread as I don't think there is much mechanical stuff on the 'B, but we will see how it goes.

[VK5RC]

Thanks to jrharley for some personal messages helping me fix my old HP 8656A with a dodgy attenuator. This post and LazyElectrons' YouTube were very helpful. I thought I could add some other issues that came up in my repair.
Disassembly went pretty straight forward until I was trying some different o ring sizes when I broke the attenuator fork! Plastic doesn't glue well so I had to embark on a learning curve of casting and moulding - guided by my close friend Richard - a retired dentist.
Repair was good enough to hold together for a 2 part silicone mould. See picture 1 below.
First efforts were porous (bubbles) and weak so de-gassing in a vacuum pot, and then reinforcing the shaft with some kevlar - pinched out of some ultra light antenna wire. See photo 4. Results seemed ok. See photo 1 of following post for old and new cast fork.

The epoxy was JB Weld, a filled epoxy. The mould was pre-sprayed with a releasing agent. The mould had 3 vents to allow bubbles and over flow to escape.
I found Robert Tolone's youtube channel good for teaching about moulding and casting.

The actuator arms/forks are compressed by a cover (with screws torqued to a very low level) so the height between the top of the ball bearing - that slots in a groove in the top of the fork and the lug underneath the fork that engages the gold plated switch underneath is critical.
Measurement by micrometer of the cast models were pretty close mostly within 0.05mm of the original. (4.43mm). The ball bearings are 2.5mm I am pretty sure - they measure 2.498mm.
More stuff to come...

[VK5RC]

I used a pair of metric o-rings to replace the original hp grommets, I used ID 3mm, 2mm thick o-rings but sanded them down to about 1.2mm using the same technique as jrharley - stuck the o-ring to aluminium tape and used a disk sander. Lost a couple in the shed!
See photo 1 in above post old grommet on LHS new o-ring type on RHS.
Reassembly was a bit tricky - my actuator cover - when it's screws are loosened - doesn't lift up much but by prying it gently up with a screw driver and angling the fork in - see photo 3 below - and with a little encouragement from a long fine flat blade screw driver advancing the grommet part of the fork into the actuator mechanism.
You can confirm that it is lodged by pulling and pushing the fork and you get a reassuring clunk of the actuator solenoid.
Final reassembly was pre-dated by cleaning in IPA and a bit of de-oxit D5 on the contact surfaces.
Testing so far looks good 👍
I have a few spare forks and are happy to forward to fellow HP 8656A restorers.
Again thanks to jrharley for his PMs and this post
Rob
VK5RC

[jrharley]

Great work Robert! Thanks for adding to the information on this notoriously problematic component of the 8656A. The information on the tolerances is good to know.

I think I've come to the conclusion that, even though the attenuator approach seems pretty whacky within the context of more modern equipment, those actuator forks and the solenoid assembly, seem to be pretty tolerant of small dimensional variations mechanically. Nice to see someone go the distance on molding replacement parts as proof of concept.

Well done!

jrh

[Tony_G]

That's outstanding guys - Well done.

When you created the mold, did you 3D print the piece originally? I'm just wondering if you might be able to get away with just printing using something like PLA-CF?

TonyG

[VK5RC]

@TonyG - I used the original glued (epoxied) back together fork  for the original in the moulding process.
I don't think the resolution of a 3D print - unless an exceptional printer - would be accurate enough - and the strength would also be a bit questionable - the 'neck' of the fork - just before the base that holds the grommet is only 2-3mm thick - my repair deliberately made it a bit thicker there - maybe 3-4mm and that is where the kevlar is in the epoxy. Getting that shape into your fav 3D program would be a right PITA as well.

Talking to my dental mate - some dental practices have 3D laser scanners attached to CNC mills that can scan then make your next large filling or section of tooth from ceramic blanks. They have to be super careful as the forces that go through a tooth when you bite down are huge.
Dentists are a fantastic resource about making small things and their materials / bonding knowledge is amazing 👍