Author Topic: Scope Repair : GOULD 1604  (Read 2198 times)

0 Members and 1 Guest are viewing this topic.

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #25 on: July 11, 2020, 01:26:58 am »
Yeah sometimes it's good being isolated on a little island eh !  ;D

You have a talent to motivate me do things I was hesitant to doing... I like that in you, makes me progress ! ;D

I think you are right, I might give these DIP 40 RAM chips a go... it's only a two layer board, so jno ground plane, just two signal layers, and there are no copper fills either, so as far as thermal mass is concerned, it's not that much more challenging than a single layer board.

Thanks to my recent addiction for old tube Tek scopes, I now have that low melting point silver solder, so will help.
Also, I do have somewhere a tiny sample of that outrageously mega super immensely expensive "Chip Quick (tm) " "solder", Daved did a video on that years ago IIRC. You know that special alloy which has an extremely low melting point. The alloy then takes many seconds to ccol down, which gives you more time to wiggle high pin count chips off the board. Of course it's not solder as such, so you need to clean it off afterwards so you can replace it with proper solder.  Anyway, have not used it yet, but I am sure it's in some drawer somewhere... just need to find it !

As for sucker versus braid, I use both, they work well. It's just that sometimes one is more suited than the other, so I used whatever works best for a given job. When I was young and didn't know any better, I hate BOTH the sucker and the braid, because I couldn't get anything done with either of them !
Retrospectively, there was nothing wrong with either of them on a fundamental level... it's just that I of course bought the cheap stuff, because you get whatever your local store has ( no internet back then of course...), and the sucker was crap of course, and the braid was not suited to electronics stuff, was way too big. But now I bought a more expensive sucker, and somehow it works an awful lot better and I love it. It never gets clogged, and has good suction power. The braid I use proper stuff now and it works wonderfully as well.

The braid is excellent, just use the proper size and lots of flux on the joints, and a large chisel tip to presses the braid on the joint and apply heat effectively.
Then you can just drag it quickly over a long series of joints and you are done. The main problem with braid is price ! It's a consumable. If you have lots of big joints, you can go through the stuff real fast, and it costs a fortune. It's plain copper and copper is expensive, so getting it from china or Mars or Neptune is not gonna change that fact : there is no such thing as cheap copper... Chinese copper or French copper is still copper and you gotta pay for it no matter where you get it from !  :-\

On the other hand the sucker you pay for only once, then you can suck as many big joints as you wish, it's free ! However the problem with the sucker is that it's extremely slow because you have to do one joint at a time, and "reload" the spring each and every time, takes time and gets tiring after a while. Also, one you release the trigger so to speak, the sudden release of the spring makes the boards "jump", it's a pain. The brain is perfectly smooth and much faster, just drag it in a smooth, quiet, effortless move... it's a joy.

Other problem with the suck is the size of the tip, it's enormous so not suited for modern tiny joints, as you can't get it to "cover" the joint, joint is too small. The fine braid OTHO covers small joints perfectly.

So that why I tend to use the sucker when I had big joints, and not too many of them. Typically, when I salvage components from an old TV or CRT monitor, where you have beefy power resistors, big transistor, transformers and such. Not many pins, but thick pins with large holes and lots of solder on each joint. This way I don't waste lots of expensive copper, and the big joints is where the big tip of the sucker is not a problem.
However if I have smaller joints, and many of them, like for example a large VFD display, two long rows of very thin pins / pads..... the sucker would take an hour and leave me with an aching wrist, whereas the braid does the job in an eye blink and in all comfort.


 

Online tautech

  • Super Contributor
  • ***
  • Posts: 19614
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Scope Repair : GOULD 1604
« Reply #26 on: July 11, 2020, 01:49:17 am »
Yep all that.  ;D

With IC's as you probably know the pins are sometimes not straight and sit against one side of the hole and trap a little solder and so make extraction difficult. Using a tiny screwdriver with your iron can help hold a pin off the edge of the hole while it cools then you don't have the extraction problems.
I use a magnifying visor for all my soldering work now and it really helps spot these tiny traps.  ;)
Avid Rabid Hobbyist
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #27 on: July 11, 2020, 02:14:15 am »
Yep, I do the same. I first quickly drag the braid to suck all the solder from the rows of pins, then I check them one by one to see if they are free and if not, I reheat them and wiggle them with a pair of tweezers, then hold them in place in the center of the hole as I remove the iron tip.


@shakalnokturn : our posts collided ! You replied to me as I was already composing my reply to Tautech...

Thanks for all the good advice as usual !  :-+ 

Yeah, I really do need to get a better soldering gear... it's very high on my list of priorities for the new lab.
Currently I have my trusty 50 year old Weller iron, 24V AC, only 50Watts but somehow never failed me, more capable than it sounds on paper. Also, the temperature is regulated but not adjustable, you know it's the kind that's regulated by the iron tip itself, so if you want to use a different temperature, you need to buy a different tip that's calibrated for that temperature. One good thing about this iron other than its quality that made it last so many years, is the tips are still available 50 years later, and are real cheap, like a couple Euros a piece !

But I would like a newer iron, with adjustable temperature, ideally a " direct heat " type but I don't think Weller do this type so would need to buy elsewhere. Then I would want 90 Watts minimum to feel more comfortable.... but if I can afford a 130 Watt iron then why not, the better...
Then with that, a collection of tips so that I always have the best tip for the job, to get things done as easily, efficiently and cleanly as possible...

As for flux, after watching hours of videos on Louis Rossmann ' YT channel... I have become a flux addict ! As he is used to say : "There is no such thing as too much flux ! "  ;D  It's very expensive but I get Amtech flux through him in super large 30cc syringes and it really keeps cost down. I now use flux liberally all the time and it makes soldering and desoldering so much easier and cleaner. Why didn't they teach me the virtues of flux at school ? They don't have their priorities straight I am telling you !  ;D

Anyway... two things :

1) One of the RAM chips as you can see is surrounded by two big boys : the battery I just put in, and that big blue Philips axial electrolytic cap. I need to remove those two if I want to be able to grab the chip by its sides to pull on it.

2) HOUSTON, WE HAVE A PROBLEM !!!  Just looked.... the pins on those ram chips.... they are all BENT flat on the PCB on the solder side !!  |O
I guess the god of solder does not like me, is that it ?!   :scared:
« Last Edit: July 11, 2020, 02:17:37 am by Vince »
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 19614
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Scope Repair : GOULD 1604
« Reply #28 on: July 11, 2020, 03:43:24 am »

2) HOUSTON, WE HAVE A PROBLEM !!!  Just looked.... the pins on those ram chips.... they are all BENT flat on the PCB on the solder side !!  |O
I guess the god of solder does not like me, is that it ?!   :scared:
Oh no !  :palm:
Really  :wtf: who really needs to bend the pins on a 40 pin DIP ?  :o

Bugger, my sidecutters are twitching to make it easy for you.
Find some new IC's I guess.  :-//
Avid Rabid Hobbyist
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #29 on: July 11, 2020, 12:50:41 pm »
Yeah that's tough luck !  :--

But I just had an idea that might do it for cheap and be quick to implement, since I am on a budget : once I have cut all the pins to safely remove the chip without damaging the board/pads, instead of doing a PCB adapter to fit it back, which costs money and time to do (as well as get delivered) since I have not done a PCB / CAD in 10+ years... I could maybe simply use a chunk of veroboard filled with pads, the size of the chip, then just solder a couple male to male header strips (I think I have some in stock...), the little bits of the strips that will protrude through the veroboard, the pin-less chip ought to fit snugly into, then I will just have a minimal gap to bridge with solder, to connect the header pins to what's left of the pins of the RAM chip !  Might try that...nothing to lose, if that doesn't do it I will just try my chance with newer incarnations of the 6264 chips, and pray that they are still compatible (enough) electrically !...

But at any rate it will slow me down and I am not going to wait to do that, especially since as I said, I hardly believe that corrosion in the acquisition RAM area could (should) brick the scope.  So for now I will just test the damaged tracks and pins to see if I can spot a blatant problem/cut trace, and if not... will leave it at that for the moment and rather move on to probing around the CPU area to see if it's running or not, and go from there...

« Last Edit: July 11, 2020, 01:54:00 pm by Vince »
 

Online shakalnokturn

  • Super Contributor
  • ***
  • Posts: 1246
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #30 on: July 11, 2020, 02:27:11 pm »
Do you have a universal programmer to check the remains of your 6264 RAM while out of circuit?
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #31 on: July 11, 2020, 02:50:36 pm »
Nope, no programmer here... though it's very high on my list of gear to get to improve the new lab, of course...

What do you have at the back of your mind ? What would you do if we could look at the RAM contents ?

Also.. I did read here and there that dynamic RAM modules in computers could retain their data for a few seconds once out of power, due to the charge being stored in the capacitors, but for static RAM chips I never heard about it, I assumed the data vanished instantly...

 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #32 on: July 12, 2020, 04:14:14 am »
Wow.... made a lot of progress today... 5AM, time flies when you work on this thing, better get some sleep ...

Alright, so I spent many hours inspecting the board up close, in the corroded area and close vicinity, inspectin gvisually and checking for track continuity, and found many problems, some of which are fixed.


Checked the tracks the guy repaired, they tested fine.

Armed with an A3 format print of the Acquisition schematic (attached), and a highlighter to keep track of what I checked, I started checking methodically each and every pin of the 3 RAM chips, up to where they went on the two acquisition ASIC chips. The address and data buses where daunting... especially since the data bus of the three RAM chips are NOT shared ! They are all independent and go to their very own individual port on the acquisition chip !

Here the findings :

1)

Top right corner of the sheet, R78 pull-up, connected to the three Chip Enable inputs of the RAM chips... tried to measure it, reads open circuits, several Mohms... Tried to find it on the board, to no available, couldn't see it. Checked the parts list to see what value it is, to help me locate it on the board... guess what ? According to the parts list their is no such thing as R78 in this scope ! It jumps straight from R77 to R79... with blank line where R78 would be...
Then I though it's probably a red herring again. I think most likely they just decided not to implement it, probably because it was redundant since that input is probably driven by a totem-pole output somewhere, not an open-collector output, so no need for a resistor. What made this plausible is that they seem accustomed to using unnecessary resistors on inputs... see the OE / Output Enable pins, they are all grounded.. via a resistor again. I bet it would work just fine without that resistor eh..
So I decided to declare this a non-issue, and moved on.

2)

 The R/W inputs. They were all connected together, no worries, however the signal didn't make it to U56 nor to the acquisition chip. It was a cut trace but it was hard to spot. On the board nothing looked obvious at all... all that attracted my attention was what looked like  tiny dot of black marker/pen... but that dot was right next to the R/W pin of one the RAM chips, and also right next to U56.... so had to dig further just in case. Bingo...at first you don't see any track connecting the two chips, but that's because ...the track happened to be completely hidden / covered by the white silk screen that was exactly where the trace is ! So I scrapped the silk screen to see what's what, and indeed there was a track connecting my two chips, yeah !!  :D
Then scrapped the black dot and yep, under it, the track was cut open. Area was too confined to repair the the track, so to get me going I just soldered a bit of mod wire straight onto the pins of the chips as it was the only things that was easily accessible with the iron.

Powered up the scope... still no joy, still brain dead. Oh well, it needed fixing anyway, so that's one less problem in that scope, can only be a good thing ! Also, it demonstrates as I hoped, that the seller was not thorough enough and missed some things. So, it means I CAN dream that the problem is fixable and that he just didn't find it because he simply gave up to soon... I like that !  :)


3)

 He removed one the RAM chips... on the bottom side of the board I noticed that one of them did NOT have the pins bents on to the PCB, and that all the solder joints were hand made, and looked like they were done by a cave man should I add ! So I  poured a gallon of flux on that chip and reflowed every pin, just for good measure, and to stop my eyes from bleeding... it really didn't look pretty. Other evidence that he removed that chip, is the biog blue Philips electrolytic cap : it's right next to that RAM chip, you do need to remove the cap so you can grab the RAM chip and pull on it. Well, there is a suspicious solder blob on one of the pins of that cap... sure enough, he didn't bother removing the cap, he simply cut on of the legs so he could lift it,then bent it back into place and soldered the cut pin back together again !  I guess it works.... at least this way he didn't damage the pads !

4)

 Still on the bottom side of the board, in the close vicinity of that RAM chip, I found a big dent in the board ! A chunk of the fiber glass is missing, a little crater when see up close... right on a track ! Bad luck... I fixed the broken trace.... powered up the scope full of hope.... nope, still brain dead.


5)

 After all that, I decided to move on the another part of the Acquisition schematic sheet : the clock circuitry, to get a first shot at probing the CPU clock...

The circuitry is made of a couple oscillator and a couple chips, two custom PAL / programmable chips, 20pin DIP chips. 16V8 IIRC.
There is a 20MHz oscillator that drives the acquisition stuff. I probed the signal on the corresponding pin on the PAL chip, I do get a 20MHz signal in there, all fine. Then I probed for the CPU oscillator, 8MHz.... nowhere to be found !  I probed each and every pin of that PAL, because it's not clear what pin I am supposed to probe since the resolution of the drawing is so poor. Pin 13 maybe, is my best guess. So I flipped the main board so as to be able to probe the oscillator pins directly, no middle man.... and yep, the output of the oscillator is dead as a dodo, a solid flat trace on the scope, it's not even trying ! LOL  It's Kaput, RIP.

Replacements are not exactly plentiful... old through hole  DIL 14 package, 5Volts... Farnell has zillions of oscillators but only one that would match the package, operating voltage and frequency, and it's on back order because well... just in case you hadn't noticed by now, this repair is driving me nuts. Plus, I am skint and don't want to suffer the minimum order of 15 Euros IIRC.
So, luckily a local shop 40 miles away has one, for the same price as Farnell, under 3 Euros,...except I don't doubt that it will be some unbranded crap but well, if it works, it will do... shipping is only 5 Euros IIRC so that will be 7,50 total versus twice that from Farnell. So I think I will go for that..

6)

Still in the signal path of the CPU clock, I found another break in the path : clock comes out of the first PAL (U54) on pin 13 if I read that right, and is supposed to then enter the other PAL (U53) on pin 1. Well, I don't have continuity there ! :-(   Need to investigate....
Those two PAL chips are very, very far away from one another. They are literally on opposing edges of the board, so tracing the... trace, could be "fun", it probably takes a twisty road and switches from one side of the board to the other, multiple times...


So it was quite a fruitful session today, happy camper. The guy really went deep on that one, but butchered it, so it's not quite clear yet if it can be saved. I mean, how can I be sure that I won't miss some of his butchering here and there... I would need a crystal ball to know exactly what he did to that poor scope !


6:15 AM now, time to get some sleep eh ?!  :=\

« Last Edit: July 12, 2020, 04:23:07 am by Vince »
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #33 on: July 12, 2020, 04:16:07 am »
....
« Last Edit: July 12, 2020, 04:25:12 am by Vince »
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 19614
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Scope Repair : GOULD 1604
« Reply #34 on: July 12, 2020, 09:56:01 am »
Nice steady progress Vince.  :-+

The blue resistor ? 2nd from right seems to have a dodgy solder joint:
Avid Rabid Hobbyist
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #35 on: July 12, 2020, 02:22:28 pm »
Yep I noticed that too when I looked at my own pictures on the computer screen. With lots of light and a macro shot, it reveals things that go otherwise hardly noticed.  Will touch up that joint for sure...



 

Online shakalnokturn

  • Super Contributor
  • ***
  • Posts: 1246
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #36 on: July 13, 2020, 03:07:59 pm »
What do you have at the back of your mind ? What would you do if we could look at the RAM contents ?

Not looking for any data there, only a tool to logic-test the I.C. functionality if you must reuse it, sometimes the electrolyte does creep in through pins.
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #37 on: July 13, 2020, 04:06:32 pm »
Oh, stupid me, testing the chips...yes that sounds like a good idea indeed ! Another motivation for getting a programmer....

Just back from my local shop (in my town)... he had ONE suitable oscillator in stock, so no need to pay shipping and wait 3 days to get it from the other less local shop that's 40 miles away in the next town, great !

Can't wait to install it on the board... stay tuned !  ;D
 

Online shakalnokturn

  • Super Contributor
  • ***
  • Posts: 1246
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #38 on: July 13, 2020, 08:04:44 pm »
4)

 Still on the bottom side of the board, in the close vicinity of that RAM chip, I found a big dent in the board ! A chunk of the fiber glass is missing, a little crater when see up close... right on a track ! Bad luck... I fixed the broken trace.... powered up the scope full of hope.... nope, still brain dead.

That looks like crazy bad luck to damage a track accidentally in that location, besides the dent is really clean, did you check what link that was on the schematic?
Just keep in mind that it may have been done willingly by a drill bit on a "factory rework". Are there any added wires glued across this board?
 

Online shakalnokturn

  • Super Contributor
  • ***
  • Posts: 1246
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #39 on: July 13, 2020, 08:10:09 pm »
4)

 Still on the bottom side of the board, in the close vicinity of that RAM chip, I found a big dent in the board ! A chunk of the fiber glass is missing, a little crater when see up close... right on a track ! Bad luck... I fixed the broken trace.... powered up the scope full of hope.... nope, still brain dead.

That looks like crazy bad luck to damage a track accidentally in that location, besides the dent is really clean, did you check what link that was on the schematic?
Just keep in mind that it may have been done willingly by a drill bit on a "factory rework". Are there any added wires glued across this board?
Looking at the solder side photo I'm pretty sure those links have been cut away on purpose as a layout correction so should be left as-is.
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #40 on: July 13, 2020, 09:10:02 pm »
Wow !! Track cut on purpose ?!  I must say I didn't think f that ! But yes if you say it is "common" practice back in the day.... yes indeed I agree it looks like it was done very cleanly by a drill bit, it did strike me at the time.

Also yes, as you saw on the pictures, in this area of the board there is a ton of factory bodge wires and components !

Hmmm... thanks for this suggestion !  :-+


OK, so I installed the new 8 MHz crystal....


1) I do get my 8MHz signal now, and it does get to the CPU, great. However still brain dead. The clock signal looks... absolutely.... A-TRO-CIOUS !  See pics !  It's not much of a digital signal !  Not even sure how a TTL output could produce such a weird waveform...
I thought maybe the two PAL chips that receive this clock, are defective and load the oscillator output in some weird way... so I pulled both PAL chip in an attempt to unload the oscillator, but I still get exactly the same waveform.

So I thought OK it's non-sense, might a probing issue... so checked for the basics : sufficient bandwidth ? Yes... scope is 100MHz,probe too. Probe is set to x10 of course, so as to get the B/W, and I compensated it, it's fine. B/W limit enabled on scope ? Nope, set to sue full B/W.  Channel one defective ? Switched the probe to channel two, still the same...

Need to try a different probe, and a different scope... glad to have 30 scopes in my collection, sure I will find one that will do the job !  :P

The waveform looks like, with some imagination, a mismatched transmission line getting reflections : It looks like a square added to itself with some delay...

2) The output of the other crystal, the 20MHz one, also looks weird, although a different kind of weird ! See for yourself...
It looks "better" in some way... we do have a flat portion at the top, and a decent falling edge. However the leading edge shows massive "dampening" !  :o

3) Talking about bodges, there is a long bodge wire that routes the output of the 8MHz oscillator.... straight to the CPU !!! Yes, it thus bypasses the first PAL chip which according to the schematic is supposed to be the recipient of that clock !


So, I need to suss this out ! Ideally I would want to pull the new oscillator so I can test it outside the scope, with a lab power supply, in a "controlled" environment so to speak, to make sure (or not...) that it works fine. But I fear that would put too much stress on the pads on the board... I mean I have desoldered the old oscillator, put the new one in.. if I pull it again and then put it back in... would be a bit much don't you think... Maybe I can solder a chip socket (oscillator has a standard DIL 14 layout) on the board ?? It's only 8MHz so the extra inductance of the socket won't be that much of a problem I think.

I am hope it's a red herring and just a scope/probe issue, but who knows.

I guess Shaklanokturn might on to something.. I might try removing my repair of that "broken" trace on the bottom, in case it was part of the factory bodge jogs, see if that fixes the waveform....


Also, since the CPU is now getting a clock, though a not very useful /working one, I could not help but start probing the thing, was just too curious !  ;D

I am glad to say that now, it does warm up a little, so it's doing something, whereas before it was dead cold AFAIR.

I probed each and every pin, to see if there was any activity, and the status of all control input pins that maybe cause it to not run. I attached below, the schematic sheet for the CPU, or course. So here it is :

- Adress and Data buses : some lines are high, some low, but there is zero activity on any of them.

- Interrupt input : active low, and all 3 of the m are high, so that's good.

- DMA : it's disabled / hardwired high with a pull-up, but checked anyway just in case... it's high indeed, good.

- R/W output : no activity.

- E and Q : I get a steady 2 MHz square wave, 50% duty cycle, on both these pins. Waveforms looks very clean, though the top of the leading edge / corner, is a bit rounded / dampened, but not the end of the world. Never used a 6809 but according to the datasheet I pulled, these signals are indeed supposed to be identical square waves, though in quadrature, and run at a quarter of the CPU clock ! That's interesting !  8MHz clock, and I get a 2MHz from these pin, steady ? So, that would mean that even though the waveform of the clock is disgusting, somehow the CPU is happy with it, as it manages to derive E and Q with it none the less ? So.... this means that although I do need to figure out the dirty clock signal, it is NOT my main problem, it does not keep the CPU from working.

- I kept the best for dessert, you will forgive me... the RESET line ! It is..... it is.......... make a bet !  :P
Yes, it's active low, and it IS held LOW !  :D   So I have something well defined to investigate / concentrate on now, no shooting in the dark anymore, I like that very much !


So, to summarize :

1) Need to figure out the dirty clock signals : undo the repair I done to that cut trace in case it was done intentionally at the factory. Then try a different probe and scope, in case it's a measurement issue / red herring.

2) Start to investigate why the CPU reset line is held low.


Making progress ! ;D

« Last Edit: July 13, 2020, 09:22:36 pm by Vince »
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 19614
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Scope Repair : GOULD 1604
« Reply #41 on: July 13, 2020, 09:40:07 pm »
The 20 MHz clock looks fit for purpose, not so the 8 MHz.  :-//
Avid Rabid Hobbyist
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #42 on: July 13, 2020, 09:51:13 pm »
That looks like crazy bad luck to damage a track accidentally in that location, besides the dent is really clean, did you check what link that was on the schematic?

OK, I removed my repair of this "broken" track. Now the output of the 8MHz crystal looks much better ! Thanks for your suggestion !  :D

See picture below. I still get a huge one full volt of over-shoot and under-shoot, but at least now it looks much more like  a square wave, phew !
Overshoot probably due to me using the standard 4 inch long ground strap, maybe...

I tried to figure out what that track was on the schematic, but failed miserably : it just goes fro one via to another one. On one en, the track "stops" under a component as well as under the silk screen, so no way to know where the track buggers off... then on the end, the track goes to a via that leads right under an integrated circuit (which is not socketed so can't removed it) so again can't can't see what's happening...

The 20MHz clock however, still looks just as horrible as before, it did not improve or affect it in any way.
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #43 on: July 13, 2020, 09:52:51 pm »
The 20 MHz clock looks fit for purpose, not so the 8 MHz.  :-//

You like the 20MHz clock ? Oh well, I was expecting much better from a TTL output, but well if it's OK to you I will move on to the reset line problem then...
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #44 on: July 14, 2020, 05:13:32 am »
OK some progress again... I worked on the RESET line problem that's stuck low.

Looked at the CPU schematic sheet, attached. Couldn't be simpler, just an RC network connected straight to the Reset pin on the CPU, 5 minutes job... or maybe not.

Somehow they decided to locate that RC network 2 miles away from the CPU, go figure.  Resistor tested OK, diode too, electrolytic cap was not shorted.

I am not too sure about why they put such a big cap in there. 10uF 25V.  25V for a 5V rail... sure you need some derating but 10V would have been enough...
Also, the resistor is only 4,7K ! Couldn't they use a much larger resistor ? That, plus a 10V rating, would have made for a much smaller / cheaper cap, maybe not even a need for a horrible electrolytic cap. Anyway, that's about a 50ms time constant, modulo the 20% tolerance on the cap... I looked at the signal on the scope, single shot at power up, to observe the exponential, just because we can !  ;D  Looks good, and indeed about 50ms. So I knew this was working just fine, problem is not there.

Then since the RC network is so far away from the CPU, I rushed to check if I had continuity between the RC node and the CPU reset pin... NOPE !
There is my problem  ;D

So I started tracing the reset track on the board. Problems started about a quarter inch after the node : the track doesn't go to the CPU.. it immediately stops by a chip right next to it, and connects on pin 4.. of U33. What the ?! It's U33.... which according to the parts list, does NOT exist.. AGAIN an inconsistency between the board and the manual, I am fed up...
So my Reset signal goes to an unknown chip rather than the CPU, and I can't even rely on the schematic anymore now, great, what to do now !  |O

It's a 74HC132 ... a quad NAND chip with... Shmidt Trigger inputs !  Ah OK... reset line, Schmidt trigger... they probably decided to condition the signal before feeding it to the CPU...though its Reset input is ALREADY a Schmidt Trigger input, so that's redundant, strange. Anyway, maybe that chip is dead, so I probed the output of that gate to see if I got continuity.... no. Hmmm... then I thought oh, it's a NAND, it inverts the signal so they need to make it go through a second gate to recover the correct polarity. So  I probed all the pins till I found what other gate it was connected to... then probed the output of that second gate to check continuity with the CPU... no luck, still no continuity. So I started to follow the trace... took an hour, a nightmare... its changes board side so many times that I lost the count of it. Maybe 10 times or so, either using a via, or using a pin on some chip. And it buggers all over the place as well, the Reset signal is not used by just the CPU...
At some point the track hides under a ROM chip.... hopefully socketed so I could remove it to see where the track was going.
Then as I got closer to the CPU, the track is on the component side and goes under one of the Acquisition RAM chips, the one that the guy had removed. No way to know where the track "exits" the chip. I just took a guess... I supposed it was going straight ahead, and noticed that it was going through two pins that looked a bit crusty, not far from the battery that leaked. The same two pins I had already noticed a while back, and why I wanted to removed that chip to check for residual corrosion damage. Well I was right, the break was there ! Right under that chip between those two pins ! The guy went to the trouble of removing that chip... and didn't even fix the broken trace ?! Go figure.
So as a quick fix I soldered a mod wire on the solder side between the CPU and the nearest joint, to bypass that RAM chip, then powered the scope and... the front panel went X-mas tree mode, a real fireworks ! Relays dancing too, sounded good !

So I put the scope together, reconnected the CRT board to the main board, put a few screws here and there to kinda put it back together again, sorta, and powered it up again, and got that :




YES !!! IT WORKS !!!!  :box:

JOY !!!   

Couldn't believe it, that feeling, hard to describe !!! Put so much effort into this repair, so many problems, and just as I was going to give up, it's fixed !!!  :D

First power up, no probes attached but already got waveforms on the screen ?!
They were stored in the module attached at the back of the scope. I also know why there is an RTC in that module now : when I went in the menus, I could see the stored waveforms, they all had an individual time stamp on them.  So, that module still works just fine and its battery is good !

However of course, the RTC on the main board was reset because of the battery replacement. However I reset the time and date and it remembers it, so the Battery replacement went well  8)

I checked all 4 channels in analog mode, all seems fine bar a horrible and varying offset, depending on what the attenuator is set to. I guess there must be some adjustment for that on the board. In my old Tek scopes there is. I think they call it "DC gain balance" or something like that...

So I played with the thing for an hour, checked every menu. I am new to this GOULD stuff, it's my first one !  ;D

A bit weird but I did not see anywhere in the menus, any self test ! Plenty of them in Tek scopes that like to go wrong, but ZERO self test in the GOULD ?! Can't quite believe it but, well... unless there it is hidden and there is a magic button sequence to play to get access to a service menu...

However when I switch to digital / store mode.... I get the readout and cursors and such, because it's handled by the CPU which now runs fine, but I don't get ANY trace on the screen ! Not even a flat line, no, I mean nothing, not a pixel !  So either I don't know how to operate the scope, or... the acquisition section on the main board is kaput !  Oh no......  :(    That's too much for me, the acquisition stuff is too complex, and I am exhausted by this repair.... might come back to it later, once I have recovered some juice / lust, but for now I will just be happy that I managed to bring it back to life despite so many problems and having to deal with the crap that the seller did. Fixing a dead scope is one thing, but fixing a dead scope that has already been visited, taken apart and messed with extensively, is another thing !  >:(

Perseverance paid, it looked hopeless but in the end it was fixed with just a 4+ Euros 8MHz oscillator, not bad !  For a 50 Euros scope, I am not complaining....


« Last Edit: July 14, 2020, 05:16:24 am by Vince »
 
The following users thanked this post: tautech

Online tautech

  • Super Contributor
  • ***
  • Posts: 19614
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Scope Repair : GOULD 1604
« Reply #45 on: July 14, 2020, 05:35:07 am »
 :-//
No YAHOO on the video audio ?  :-DD

Nice fix Vince.  :)

Now it's time to RTFM I guess.  :box:  :-DD
Avid Rabid Hobbyist
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #46 on: July 14, 2020, 05:44:59 am »
Yeah no sound sorry, I noticed that after the fact, had it turned off somehow !  :o

That said you didn't miss much other than a little bit of fan noise...but I do agree it feels weird when it's 100% silent !  :-[



 

Online shakalnokturn

  • Super Contributor
  • ***
  • Posts: 1246
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #47 on: July 14, 2020, 02:18:32 pm »
So for all you know there is no self test implemented at all, (I'd get hold of a ROM dump if available and check it for text string error messages just to be sure) and digital memory is not running due to other cut tracks around the same RAM IC's.
May be worth beeping all their pins.
At least you're getting somewhere.
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #48 on: July 14, 2020, 11:37:59 pm »
Google found me this very interesting 4 page topic on the 1604 in some UK forum devoted to vintage radios :

https://www.vintage-radio.net/forum/showthread.php?t=46501

It's ancient and closed/archived, I couldn't post to it even if I subscribed to that forum.. but still, 4 pages of interesting stuff.

1)

There is a secret key combination at power up to reset the scope to factory default, because apparently there is some non-volatile memory somewhere that cannot be erased otherwise.  Did that. No change... so I guess it was already reset when I got it.


2)

There is a secret Service menu !  Press the '9' key 3 times in a row and the menu shows up ! Sadly it's not much use. Not much in it, and certainly no error logs or the possibility to run self tests.. other than testing the external keypad if you have one.


3)

I gathered the scope does run a self test at power up, but in "silent" mode, and very minimalist/quick, so you don't even notice it. Only way you know that it was performed is that at the end of it, it displays at the bottom of the screen the firmware revision of the scope, and also that of the ROM in the pod you might have installed on the scope. That works for me. I can tell it's very minimalist because I unplugged the large acquisition daughter board (which holds the acquisition stuff for channel 3 and 4. Acquisition for channel 1 and 2 is on the main board) and it powered up just fine, no error message, and works no better or worse than it did before ! So I guess it does a checksum of the ROM on the main board and that's about it, doesn't even bother with testing the acquisition stuff !

4)

Apparently in the UK there are a bunch of these scopes around with leaky batteries...

5)

Didn't think of mentioning it yesterday, as it's only a secondary issue for now, but from a cold start, the trace and readout have some focusing issue, can't get them to be super sharp. Also sometimes the text "wobbles" around and is "noisy" if you look up close.After a few minutes of warming up it gets better and better.  Guys in that thread said it's a typical problem, just a few tired caps n the PSU, on the plus and minus 12V rails. So might recap the PSU if I  manage to fix the acquisition issue.

6)

GOULD is king as far as inconsistent documentation is concerned, as well as countless undocumented firmware or H/W revisions, some being not even retro-compatible. Basically these scopes are a total mess, and GOULD was on drugs when they designed their scopes.

7)

They shed A LOT of light on the interface " pod " (that's how GOULD calls it) that plugs at the back of my scope. The numbers on it now make sense. A lot of mysteries completely solved now : " 1600 / 160A (105) ".  1600 means 1600 series, matches my 1604 model. " 160A" is the name of the waveform processor "package", ie the pod AND the external keyboard. 160 is the model name for the K/B.  "105" refers to the pod itself.
So my 105 pod, is one part of the 160 "waveform processing" package.

How does it work ? When I plug the pod on my scope, I didn't see any measurements functions added in any menu, so I was disappointed and perplex.
But it's there !!! The ROM in the pod holds the code for the measurements functions, but that's it. The scope can then use that code. However the scope has NO provision for adding menu items to let you access these extra features from the front panel. That's where the K/B comes into play !  You plug it to the back of the pod using that RJ11 "telephone" type plug, it's a serial link.   I found pictures of that keypad, it's got 26 keys or so (as we can see from its schematic in the service manual). It allows you to remote control the scope in storage mode : you can use the cursors, run/stop/single functions and.... and also select the measurements functions which are implemented in the Pod's ROM ! 

So if you don't have the K/B you can't use the cool measurement functions. However the K/B is very simple, it just sends a single byte coding the key you pushed, using a serial link. The serial bit stream is not even processed by the pod itself, it only goes THROUGH the pod, without interacting with it... to the main board on the scope. Then the scope goes to the ROM in the pod to fetch the code that it needs to run to perform whatever function you asked for.
Some guy on the forum there, had a keypad and was kind enough to show a picture of it, and make a table of the name of each button, and what byte gets send to the serial port when you push them. Some have just connected a terminal to the pod, to emulate/replace the keypad, and were sending the bytes by hand, and it worked just fine. Just need to add some level shifting if you use a computer (usually outputs 12Volts) to lower it down to TTL levels so as not to fry the main board in the scope...
Others have built there own keypad and programmed some little micro to scan the buttons and send the byte to the pod.

So I if manage to fix the acquisition, I might try to hook up a terminal to the pod to see if I too, can get access to the cool extra stuff... could be fun.




OK so now back to the repair.

Spent some time operating the scope, trying to understand what was going on.

1) In unplugged the pod. Scope lost the date and time ! So that's handled by the RTC in the pod, not on the main board.

2) When I am in storage mode, I can't get a trace but it keeps displaying a stored waveform on all 4 channels. Channel #1 displays a flat line, channel #2 a triangle, channel #3 a square wave, and channel #4 another triangle but at a different frequency. I tried all I could to make these go away... but they just won't. Waveforms are supposed to be stored in the pod, so I removed it. No joy, they still show up. I went into the menus to try to see how to erase them, no joy. Without the pod, all the main board will do is save TWO "reference" traces, two, not 4. Plus, these 2 are empty/deleted, and disabled, set NOT to show up on screen.
Played the magic key combination that resets the scope... still no joy. So I come to the conclusion that these 4 waveforms aren't stored from actual past acquisitions by the user. Instead, they are probably "test patterns", which are hard coded into the Firmware on the main board, which the scope displays when it feels it's right to...possibly precisely because I reset it ! 
So I have stopped pulling my hair about these mystery waveforms, these ghosts... They will probably go away as soon as the acquisition is fixed and actual data comes in, that will wipe these test patterns from the screen.



So what to do next.... as you said yes, most probable is some more corrosion damage hidden under one or more of these 3 acquisition chips, so need to spend some time in that area. I have already tested each and every pin of each and every of these RAM chips, and fixed what was broken, as detailed in my previous posts. However there is still plenty of of tracks on the acquisition schematic that need checking, other the RAM chips. some glue logic, control lines to the acquisition ASICs.  I bet one of these control lines goes under the RAM chips and might be broken... hopefully it's just that and I will stumble on it with enough patience... 

But I will need to dismantle the scope all over a gain... it's depressing, I was so happy to have it all back together and alive... only to put it all in pieces again !  :-\



In any case, I also REALLY need to get some work done in the house or else at X-mas I will still be sleeping in the living room !!!  Need to work on the bedroom and bathroom. So I will slow down on the repair stuff from now on. Want to do both at the same time though, so I guess I shall alternate between the two activities...say work 2 days on repairs, then 2 days working on the house, then back to repairs etc...
 

Offline Vince

  • Frequent Contributor
  • **
  • Posts: 479
  • Country: fr
Re: Scope Repair : GOULD 1604
« Reply #49 on: July 18, 2020, 01:39:44 am »
OK took the scope apart again and worked a bit on it.

Probed a bit more in the acquisition area. Checked that the main acquisition chip (the one that interfaces with the CPU, not the other ACQ chip that handles the RAM data). That chip does get its 20MHz clock, and I see that its R/W line, presumably driven by the CPU, is very active. So I took that as a good sign that the CPU was asking the ACQ ASIC to acquire some data. So it was reasonable to thus expect to see activity on the 3 ACQ RAM chips, no ?
So I probed for that. Zero activity on the address and data buses, same on Chip Select pin. At least it's consistent, no chip select means the ASIC doesn't intend to use the RAM, so why would you have activity on the address bus.

So I looked at what provided that chip select signal for the RAM. As you can see from the schematic posted earlier, there is a bit of glue logic there, in the middle of the ACQ chips and RAM chips : a NOR gate U52b, and U56, a quad two input MUX.

The first ASIC doesn't drive the chip select directly. Instead it feeds it to the input of that NOR gate, whose output then drives the second ASIC, the R/W of the RAM, as well as pin #1 of the MUX / U56. Pin  #1 selects between the two inputs of the MUXes. Two of the 4 outputs of the MUX feed back to the first ACQ ASIC, and the two other drive... yes, the Chip Select of the RAM chips. So it's all very much intertwined...

Now I probed the NOR gate. Nothing coming out of it, so no Chip Select, normal. Maybe the gate is dead ? Nope.. its two inputs are low, and show zero sign of activity. The first ASIC drives one of the inputs, the other input as you can see is connected to ground via a RC network. Not sure what that's supposed to do... maybe a crude method to delay the signal a bit before the NOR gate transfers it to the second ASIC and RAM chips, no idea.

But whatever... you will notice on the diagram that the output of that NOR gate is precisely where I had a broken track ? Remember, that pervert one, hidden under that "block dot"  right next to the output of that gate.  So I switched to "if it looks like a coincidence... it probably isn't ! " mode.

So I wondered : what if that broken trace was AGAIN another factory mod, like the cut trace on the bottom side of the board, that was cleanly drilled ?
The "dot" was indeed very circular/clean too, and concave as well. So I thought maybe again they drilled that track on purpose... and decided to cover the hole with a black marker / pen, to keep it from corroding or god knows what, not really important I guess....

I wasn't sure how that would cure my problem, because for all I see, the only thing driving that line, is the gate ! So if I remove my mod wire that bridged the broken track... there would be nothing left to drive that line, it would be floating !   But I thought.. to hell, it's been already demonstrated several times that this schematic just isn't reliable, so maybe on the actual board there is some chip other than that NOR gate, that's there to drive the line. So I decided to remove my mod wire and give it a go !



Glad I did : ACQUISITION NOW WORKS !!!!  :box:



So for the second time, big thanks to Shakalnokturn for suggesting that the bottom track could have been a factory mod ! The other track on the top side was too !!!


I was so happy, put the scope together AGAIN... and AGAIN I shouldn't not have done so : I AGAIN have more problems, so AGAIN I will have to take it apart ! GRRR !!!!

What's wrong now ? Well, acquisition works but... only on channel one. I still get the hard-coded " test pattern " on the other 3 channels.

ALSO, I have now lost channels 2 / 3 / 4 in ANALOG mode too, were they used to work before !  Oh no !  :-(

So this scope is very depressing indeed... one step forward and one step back ! Not sure I will ever see the end of it... but well, I guess I am making progress none the less, eh ?!.......................

Of course the first suspect is that the mod wire I removed somehow caused this problem, because in this scope the digital and analog stuff are so closely intertwined, it's a complete mess, nothing is very clear, it's all fuzzy. So I thought oh no, it that mod wire cures the acquisition but makes the analog channels disappear, it's a lost cause... I give up.

But before I give up, lemme put the wire back, to see if the 4 channels come back to life in analog mode : NO THEY DON'T ! 

GREAT ! Phew ! So the wire didn't cause the problem, it's a new issue altogether, somehow I prefer that !

It was just a coincidence then, just back luck.

So I guess I made some progress after all : found out that the black dot was a factory bodge, re-activated it and now I have acquisition working, albeit only on Channel 1. But if it works on one channel, then 90% of the chain is working and the rest is bound to be a more minor and less complicated issue, I can hope.

The way I am attacking this is : since I lost channel 2/3/4 on both analog and storage modes, a reasonable assumption is to think that there is one single problem, common to both modes. So at first I would rule out any problem with the acquisition subsystem entirely, and focus more on the analog stuff that's common to both modes.

I attached the overall block diagram for the scope, highlighting the signal paths that I suspect most.

As one of you mentioned, most obvious is to check that the signal traces are not simply way off screen. So I checked carefully, thoroughly for that, in both modes of operation.

In analog mode there is no  " Beam Finder " button like one is used to see in almost every scope ever made. It's a GOULD after all, it's exotic, so I guess they didn't want to do like the rest of the crowd.

So all I could do was to set all channels to GROUND coupling, trigger set to AUTO of course,  then set the attenuators to the least sensitive setting, in this case a whooping 20V (most scopes are 5V usually), then played ad nauseum with the vertical position up/down control switches... no luck. No trace to be seen other than channel #1 that works fine.  Tried running the auto calibration routine that's there to get rid of any offset and gain errors in the front end... no joy. Tried the dreaded " Autoset " button, for once in my life... no luck either.

The only hint of an offset/off-screen trace, I found only in storage mode, and only on channel #2 : if I press the "select trace" button, it turns on the cursors and places one of them, which " follows " the trace, on the selected trace. So using that control, I can see that I am able to select channel #2, because the read out tells me the cursor is using it. Also, if I move the cursor along the trace, well an invisible trace of course, I CAN see the bottom half of the cursor sticking out of the top edge of the screen, suggesting that the trace might indeed be there, off screen at the top. Also, the readout shows a couple arrows that point upward, suggesting that the trace might be indeed up there off screen. So that would indeed point to a vertical offset problem, a huge one... but only for channel 2. Channel 3 and 4, no suck luck.

So I think I may indeed have a vertical problem on channel 2, so will investigate that. But it can't explain why channel 3 and 4 are missing, never mind why in analog mode 2 to 4 are missing.

So I think I have a second issue there. Trying to figure out how I could have channel one working fine in both analog and digital, and no trace at all on all 3 other channels... your guess is as good as mine, well better actually, but my own reasonable guess would be.... how does the scope show 4 traces at the same time, with a single electron gun at its disposal ? Yes, if the " beam switching " mechanism fails, it would explain everything...

As we can see on the block diagram, this mechanism is a mess like the rest : it's done in THREE parts ! ... Apparently the scope was designed as a two channel scope originally, so the circuitry for the first two channels is on the main board, then to implement channel 3 and 4 they had to add that large daughter board that sits on top of the main board. It handles part of the stuff for the last 2 channels. Some of the acquisition, and some of the analog stuff.

So there is a switch for channel 1 and 2, then another for 3 and 4, then a third switch that switches between the output of those first two switches ! Lots of switches !  :-//

So I guess I need to take the scope apart yet again (I am kinda getting the hang of it now... ), and start probing those switches...more generally leave the digital part of the scope, move on to the analog part of the board, and follow the signal path from the BNC inputs down to the output of that final switch. Hopefully I am right and I will find something interesting en route... wish me luck  :popcorn:




« Last Edit: July 18, 2020, 02:03:10 am by Vince »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf