HP 3561A Dynamic Signal Analyzer repair

[Cubdriver]

I snared this little beastie from the 'bay; it arrived today.  It was advertised as 'For parts or not working', with a note that it wouldn't power up and that the fuse was good.  For the price, I figured it was worth a shot:


As noted, it doesn't power up - plug it in, press the power switch and from all external appearances, a whole lot of nothing happens.  No lights, no cooling fan, no nothing.  As advertised, the fuse checked out as good.


Externally, it looks almost brand new - the front panel is in beautiful condition, and other than a missing HP emblem on the carrying handle and a bent fan filter retaining screw it looks great.  The missing emblem makes no difference to me, and the bent fan retainer was fixed in about 30 seconds with some gentle persuasion from a pair of lineman's pliers.  This is as it came out of the box - no cleaning:





Removing the top cover (cake - 4 captive quarter-turn fasteners and off it came, with no loose hardware to lose) reveals the usual solid construction typical of HP instrumentation.  There is beryllium copper fingerstock along the sides and top to bond the cover to the chassis, and the boards are further shielded with sheet metal covers:



This was a surprise - I was expecting a long electrostatically deflected CRT like what you find in a scope; instead it has a short little magnetically deflected one.  It's so cute.  I wonder, if you feed it, will it get bigger?:



There's similar bonding fingerstock for the bottom cover which is unsurprisingly retained by the same quarter-turn fasteners as the top, and the underside of the mainboard is basically a giant plane layer for shielding.  It's protected by a sheet of Plexiglas, not readily evident at a glance in the photo:



Removing a metal cover exposed the power supply section:



Linear regulators on the main board.  I suspect that the blue electrolytics on the removable card immediately above them are the filters:



The initial troubleshooting procedure I found in the manual on Keysight's site began with checking a pair of 12V bias supplies, which were operational and within spec.



The following step was to check the indicator LEDs on another board in the PS section.  The lighted one shown indicates an overcurrent fault. (surprise, surprise!!)



That fault was then isolated to the main part of the instrument by pulling the board out of the edge connector to disconnect it from the array of linear regulators on the mainboard (it feeds them).  When power was reapplied, the overload LED was off.  This came as no great surprise.



Of course THAT troubleshooting sequence is in part II of the manual, which is not readily available online as a PDF that I've found, so further misadventures in troubleshooting will have to wait a bit until I get my hands on a hard copy of the two volume troubleshooting and repair manual.  Off to evilBay...

-Pat

[crispy_tofu]

Wow great score! 
Beautiful condition, hopefully it can get fixed! 

[Vgkid]

Nice pick up.

[Cubdriver]

Thanks, I hope so too!

-Pat

[VK5RC]

Love that attention to detail, keep the photos coming, is this thread better in the "Repair" listing?

[AF6LJ]

Good to see it was complete, nice construction as usual.
Keep us posted on what you find.

If it were me I would go on with troubleshooting for as far as I could get without the second half of the troubleshooting guide.

[SeanB]

Being HP, I would guess a shorted tantalum capacitor, probably on one of the 15V rails. Replace with a 35V part, and for good measure look on the board for all the relatives and do them as well on that rail and other highish voltage rails.

[Cubdriver]

My hope is that it will turn out to be a shorted cap on one of the rails, too, at which point I'll likely shotgun similar ones too to try to head off the next one crapping out and causing grief.

I expect to order a copy of the two volume service manual later today, but in the meantime will perhaps poke about further at it tonight.  Troubleshooting is complicated a bit by the fact that the instrument's schematics are contained in the volume of the repair guide that I don't have access to (and of course the scanned versions would likely be fuzzy and frustrating to read anyway).

In hindsight, yes, this probably should go in the 'repair' section.  I'll see if I can have it moved.

-Pat

[free_electron]

My 2 cents : one of the two TO-3 regulators is dead.  i;ve sene that happen before. They conk out and the circuitry behind gets to eat unregulated power so it overloads.

[Cubdriver]

Did a touch more troubleshooting tonight, and also ordered a copy of the service manual (both parts).  Whatever ails it seems to be in the main body of the instrument - I pulled all but the power supply cards and it still indicates the overload.  I made some resistance measurements in the vicinity of the row of linear regulators, and found one tantalum cap that seemed a bit suspicious (29 ohms where one other was in the neighborhood of 500-600 ohms, and the rest well north of that).  It's the one on the bottom right of this picture, next to the CR201 silkscreen.  Lifting one side of it was no help; it appears to be fine and even with it out of the circuit I still had my overload.  I then noticed that a big electrolytic on the backplane board smack in the middle of the power supply (it lives between the two power supply cards when they're plugged in) looks a bit lumpy near the bottom.  That'll be the next thing to investigate when I get more time.  It's barely visible at the top edge of this image, slightly to the left of center.



They put a seriously physically huge microprocessor chip in this thing.  Apparently they needed a lot of pins, so they just made a huge DIP to encase it.  From what I've learned online, it appears to be a custom labeled Motorola MC68000L; a 32 bit processor with a 16 bit bus.  It looks like a supercarrier sailing among a fleet of frigates.




My misadventures will continue when I get more time and/or when the manuals arrive.

-Pat

[PA2HK]

Nice machine, I also have one. The CPU is indeed a good old Motorola 68K, pretty standard for hardware of this vintage. What makes the HP3561A special is the optional usage of magnetic bubble memory that was offered as an option (option 001 - my unit has that option), with that memory option a 3561A can store waveforms. Bubble memory is a non volatile memory technology of the eighties that was going to be a big thing before the advant of flash memory technologies. Well, the 3561A is the only test equiment model HP used bubble memory in, guess it was not that great.

The fault in your unit is most likely a shorted tantalum, unfortunately there a quit a of these caps that can go bad

[Cubdriver]

After weeks of procrastination, laziness, lack of time due to business travel and generally having too much on my plate, I finally got around to digging further into this now that I have both volumes of the service manual.  The issue causing the power supplies to go into shutdown turned out to be a shorted diode (CR207) in the half wave rectifier feeding the +8V supply rail.  It's an HP 1901-0050 (specs according to the manual are diode, switching, 80V, 200mA, 2nS recovery, DO-35 package); according to what I found online it crosses to a 1N4150 which seems to have slightly different specs, but as it's a rectifier I can't imagine that it's super critical.  Since I don't have any 1N4150s kicking around, I installed a 1N4148 in its place for the time being to check things out.  The replacement rectifier brought it to life immediately.  I plan to get the proper replacement on order shortly, and will then wring it out and see if there are any other issues.

The failed diode and its location in the circuit:


A detail of the cleverly disguised short circuit:


The 'temporary' fix installed (CR208 will likely be changed, too, as it's gotten a bit toasty; its location is not shown in the schematic image, but it's the top right diode of the bridge, directly above CR204, and was likely loaded by CR207 as it was failing):


Amazingly, with the power supply once again doing its thing, the instrument comes to life:


And finally a detail of the display:


I'll dig into it further once I get the proper part installed, and post some more detailed images of the various boards and whatnot as I go through it.  It's an impressive little box, with a lot of stuff packed inside.  I don't recall if I mentioned this earlier, but based on the serial number it appears to be a late 1985 manufactured unit.

-Pat

[BFX]

Congratulation 
Nice unit and repair story

[crispy_tofu]

Fantastic, congrats! 

[ckm]

I have one of these as well - it's throwing an error on boot, but the error is only listed in Volume 2 of the manual, which isn't available online - I'm trying to get my hands on a paper copy, but it's pretty rare....

I should probably open it up & check the rails/status lights even with only Volume 1 of the manual.....  Haven't gotten to that point yet, too many other projects.

[Cubdriver]

What error is it throwing?  Perhaps I can look it up and try to at least point you in the right direction.

-Pat

[ckm]

What error is it throwing?  Perhaps I can look it up and try to at least point you in the right direction.

Thanks.  I brought it home today to try to troubleshoot over the weekend and now it's not powering up.   

I was looking at the same power supply board you found the burnt diode on and discovered that one of the two transformer wires (labeled "1") was touching R203.  The leg of R203 closest to the transformer looks like it got really hot (it's almost black) and the part of wire "1" that was touching R203 is chard - so is about 1mm sq of the case of R203 (not enough to affect the resistor AFAIK, it looks like only the paint). 

Not sure what either wire '1' or R203 do, but it doesn't seem normal that R203 would have gotten so hot.   Maybe an inductive heating of some sort?  R203 seems to be connected to a transform output, while wire '1' seems to be an input.  Without a schematic, it's hard to tell....  FWIW, the diodes where you had the heat discoloration are mint on my unit, no discoloration.

I found a copy of both sections of the service manual and have ordered them.   In the meantime, if you have any thoughts, love to hear....

Thx.

Chris.

[Cubdriver]

R203 and C300 are in series and form a snubber network across the 5V supply windings of T300.  It's a center-tapped full wave bridge setup (shades of an old vacuum tube circuit).  Based on the drawings, I'm assuming that the wire that was shorted to the resistor was connected to the board in the area to the right of R203 where there are three tied in - numbered 1, 2 & 3.  On the drawing, 1 & 3 are labeled 'red' and 2 is labeled 'yel'.  If this is the wire that was shorted against the resistor, then I'm not surprised it overheated - 1 & 3 are the ends of the 5V winding, and 2 is the center tap.  If 2 was shorted to the resistor, it would have been either a dead short to one end of the winding, or, at best, 22 ohms based on R203's value.

Separating them would be a good start.  The resistor should be 22 ohms; the cap you won't be able to measure without lifting one side as it will be in series with the resistor and the transformer secondary winding.

Hope this helps a bit.

-Pat

[Cubdriver]

The full wave rectifier for that 5V supply is CR100; that's the TO-3 packaged device on the heat sink above the transformer.  It's a common cathode rectifier; I assume the pins are the anodes and the case is the cathode, but don't presently have the energy to dig out and open the 3561A to check.  Perhaps tomorrow, if it's necessary.

-Pat

[ckm]

R203 and C300 are in series and form a snubber network across the 5V supply windings of T300.  It's a center-tapped full wave bridge setup (shades of an old vacuum tube circuit).  Based on the drawings, I'm assuming that the wire that was shorted to the resistor was connected to the board in the area to the right of R203 where there are three tied in - numbered 1, 2 & 3.  On the drawing, 1 & 3 are labeled 'red' and 2 is labeled 'yel'.  If this is the wire that was shorted against the resistor, then I'm not surprised it overheated - 1 & 3 are the ends of the 5V winding, and 2 is the center tap.  If 2 was shorted to the resistor, it would have been either a dead short to one end of the winding, or, at best, 22 ohms based on R203's value.

Separating them would be a good start.  The resistor should be 22 ohms; the cap you won't be able to measure without lifting one side as it will be in series with the resistor and the transformer secondary winding.

Hope this helps a bit.

That helps, thanks.  The wire that was shorted was not one of the red-yellow-red wires, it was a white one coming from just to the left of the diode you replaced - part of a pair of two white wires, one with a label on the wire marked '1'.  Goes to the center bottom of the transformer, I assumed it was the input, comes from CR206 & 207.

I haven't tried re-installing the board after separating the wires, that's next.

Edit - corrected the wire descriptions, this is what I get for working from memory

[ckm]

Just want to put a link to http://www.orionsword.net/Electronics/TestEquipment/HP3561A/HP3561A.html in the thread as it has some very helpful tips about repairing these.

[ckm]

Some further tests - I checked C210, which is on the motherboard, with an LCR meter and it read 1.09 ohms, which is a value that is OK for a 25V cap < 200uF.  C210 is a fairly large cap, but the markings are oriented so I can't read them (of course).  It seems like it might be too low for such a large cap.

Also checked the CR100 rectifier, it reads the same resistance forward & reverse (about 0.4 ohms), which also seems wrong.  That might explain why R203 got hot, I'd have to check the schematics (which I don't yet have).

Also, I double checked the two white wires, I don't think wire '1' shorted.  It doesn't seem to have burned through the insulation, just charred the insulation.   Also the wire was only touching the resistor 'jacket' not the bare leg.

I spotted a couple of other diodes which I am suspicious of, namely CR304 & CR302.  Testing them in circuit gives me 18 kohms on both.  I'll have to check where they go when I get the schematics.

But right now, I think the 5v rectifier has gone south....

[ckm]

Forgot to mention, I re-installed the 5v board, same issue.  Strangely, measuring the 5v test point (which doesn't have post on my unit) shows way less than 5v, something like 0.6 ish, but very unstable.

[Cubdriver]

That helps, thanks.  The wire that was shorted was not one of the red-yellow-red wires, it was a white one coming from just to the left of the diode you replaced - part of a pair of two white wires, one with a label on the wire marked '1'.  Goes to the center bottom of the transformer, I assumed it was the input, comes from CR206 & 207.

I haven't tried re-installing the board after separating the wires, that's next.

Edit - corrected the wire descriptions, this is what I get for working from memory

The '1' is what confused me - the terminals the red and yellow wires go to are numbered 1-3.  I didn't realize it was on the wire itself.  I popped the unit open and took a look at the board, and now see more clearly. 

[Cubdriver]

Those two white-sleeved wires are taps on the transformer secondary winding that provides +/-12V to the grounded section of the supply.  The entire winding feeds a full wave bridge that, combined with the grounded center tap of the transformer, forms a pair of two diode full wave rectifiers providing unregulated +/- 14ish volts that is regulated down to +/- 12V on the motherboard.  The sleeved white wires are taps in the winding that are full wave rectified to provide about 12V unregulated that's then knocked down to +8V by the linear regulator (U203) on the motherboard.

C210 is a 3300uF, 25V electrolytic.

C300, R203 and CR100 are in the pink section at the top, as is C210.  C210 is also bypassed by a 22 ohm resistor (I'm not sure I trust the R7 designator shown, as I can't find an A99R7 in the replacement parts list; A99R300 is, however, listed as a 20 ohm 2W metal film, so there might be an error in the diagram...)

The sleeved white wire is in the grey rectangle below the pink one; the pair of them feed CR206 & 207.  On mine, the #1 wire goes to terminal 6 on the board, and feeds CR206's anode.

Hopefully this sheds some further light:



-Pat