HP 3582A Repair

[Duce67us]

Hi all - first post to the forum. Might as well dive in the deep end!
I picked up a unresponsive unit locally from a recent widower.  She said it worked in the past.  It looked pretty dirty, but I brought the unit home and downloaded the service manual.   
Initially the unit didn’t do much, except run the cooling fan, and create a single spot on the screen. 
Checking voltages revealed no +5.1vdc in the power supply.  Pulled the card - found and replaced a blown fuse.  Checked for shorted regulator(s) and ran into a number of bad filter caps in the power supply.  Now the power supplies are all happy, and back in spec. 

I have a screen!  Yay! But it is full of garbage - and the front panel seems unresponsive.  Boo!!!
I will attach some pictures 

The Display test is Good! 
However, further testing is pointing to The Processor Group. 
 So I have stripped the unit to the bare bones and have started Signature Analyzing !
Bought a Tek 308 to assist in the process, and Made some extender cards.

I am sure I will have more questions. 

[Duce67us]

Sometime While I was pulling the cards out of the cage, I decided that it might be a good idea to clean all the years of dust out of the motherboard connectors, and inspect the solder joints attaching all the edge connectors.  Pretty sure that I located more than one cracked solder joint. 

I also pulled the front panel, after reading that stuck or broken switches can cause the instrument to hang,  to clean and verify all the switches.  Needed to use deoxit on more than a few to get them working satisfactory.  I neglected to check the RPG before putting the front panel back together, and I will probably regret that later.   

[Dacke]

Seeing those cracked solder joints now makes me want to check my 3582A for the same.  Mine is perfectly functional,  but at least now I know that this is a potential failure point.  If you need me to look at anything on mine or need pictures of an area,  let me know.  If repairing the solder joints doesn't fix the issue,  the service manual states the first step to general trouble shooting is to swap the A1 input boards since they are the same,  and this can isolate a lot of other potential issues.  Not sure if you have tried that.

[Duce67us]

Thanks for the input Dacke!
I figured as long as I am going to try and get this 3582A back in service, I sure need to check everything out during the process.  At one point, I pulled all the cards and swapped both A1 boards, also tried them in one position, A and B by themselves.   No real change. 

Another issue that I had to deal with was with the heat sinks on the A5 card.  The rubber grommets were all flattened and allowed the heat sinks to touch the board.  There is also some foam tape that had collapsed that was supposed to keep the 40 pin ic in their sockets.  You might notice there is a grommet that looks pretty good, not flat like the others. Because the screw on that side wasn’t installed correctly.

Right now I am knee deep in learning to use the Signature Analysis. 
The good news is, the processor is not dead.  My A7 board passes the primitive test, and has the correct signatures on the IDA Bus. 

Tonight I did some SA work on the A9 display board.  Trying to get the display to a readable screen. 
I located some typographical errors in the service manual in Table 8-5-1 Display Controller Signature Analysis
Down in the Signatures for SA Test #1.
U3 pin 1 should be 7AC3, also U5 pin 10 should be HFU6.

[Dacke]

Great.  I hope you keep updating your progress.  I've never done signature analysis or used any data analyzers before so this will be interesting for me to follow.

[Duce67us]

I guess I should say that I also have never done signature analysis before, or used any logic analyzer.  Just the schematic, maybe a pinout/data sheet and a 'scope. 

anyway, the real fun begins, exploring the circuits, and the idea of having a new tool to assist in the process is refreshing. 
So we begin at the Basic Processor test loop - Just the A7, A9 and A10 boards in the card nest. Follow the setup for the SA and probe +5v. and get 9PA2
Great! Now we check the IDA Bus signatures, and they match exactly the table provided in the service manual.
Now move to the Primitive ROM test.   
The manual states that to enter the primitive ROM test, install A6 board (ROM), move jumper A7 (J2) to run and power on, then short out A7(J4) and press RESET on the front panel - Keep A4 shorted.   +5v signature = OH62  should appear in about 10 seconds. 
My board did not ever give a signature.  But that is OK, because there is a pretty extensive ROM SA listing that covers address, chip select, and Data lines. 

[Dacke]

You stated that you replaced some (all?) of the caps in the power supply.  About half of the caps there are axial lead,  and I'm kind of forced to use axial as replacements because the power supply is so cramped for space with the boards stacked so closely together.  Since axials are so much more expensive than radial,  did you use axials and where did you get yours?  Digikey has a very limited selection of mostly general purpose, 85C rated 2000 hour caps so I've been looking around at pricing.

[Duce67us]

Yes,
I priced the axial caps too.  I usually go to Mouser and get a price comparison, then decide if the cost is worth the extra.   In the end, for this unit, I just used what I had on hand, and converted radial over to axial by adding solid wire and heat shrink.  Mostly the 470uf caps were bad through out the boards.  These were the three legged ones, two at the bottom were active.  I just used some gray 3m 4991 VHB double sided tape to help hold the new caps to the boards.   

I hope to poke around a bit more in a few days, and post more results. 

[Duce67us]

The ROM board SA process doesn't have explicit instructions for setup.  It doesn't help matters that the manual has a picture of the A6 board with the SA header incorrectly labeled as J1.  Pay close attention to the setup information given, especially the START and STOP.  My ROM board had the correct signatures for all the address bus, select addresses, chip selects, and STROMS.  When checking the overall data, everything matched except LS Byte #6 line.  The manual gives the signature C1AC, my reading was C1A0.  Manual says once found, simply move the start/stop lines to isolate the misbehaving ROM. 

GREAT! Except, the page is cut off for the signatures on the U3,4, 11 and 12 test. 
 
I need the LS Byte signatures for #6 data line.  There are 4 signatures I would love to have.  The +5 sig when Start/Stop is at TP4( mine was 826P), the +5 sig when start/stop is set at Tp2 and tp4 (7A07)
The LS Byte signature I got was 3306 for (TP4 only) and A7PF for TP2 and TP4.

More to follow
** I revisited this, and have two different ROM boards giving the same signatures.  Looks like the incorrect signature above might be another typo.

[Duce67us]

Holidays and other projects have been getting in the way lately.  Not much progress to report, but here are pictures of the bus extenders I have been using.  I made them quickly, to match the larger 44 pin card edge (22x2) with the correct cutout for the short offset.  The same card can be cut to fit the smaller  36 pin (18 x2) and 30 pin (15x2)  cards too.  I have attached the Gerber files used to manufacture the cards.  Use at your own risk - they match perfectly, but standard disclaimers apply.

[RNA]

Hi, that scrambled text looks like a dodgy ram chip.  You will be able to check that with the signal analyser.  You might also notice the the signal on a faulty ram line will not look the same as the others.

[Duce67us]

Thanks for the insight RNA!  I have been unable to spend any time on the repair, but hope to dive back in real soon.  I will post updates as soon as I can. 

[cncjerry]

Yes, that scrambled text is most likely a ROM problem but they can be replaced and and you can probably get the image on Ko4bb. I would work on that first because there are probably others that are rotted as well.  Once you get clean text and all the voltages are good it should be easy to fix.

[Duce67us]

cncjerry,
I agree that it could be a ROM problem.  However, ko4bb does not have the images.  I have tried to locate a set, asked about in the HP Equipment group(groups.io) and have yet to find anything in the usual places.   That being said, I have verified all the ROM signatures that I have data on * (see above post asking for Specific ROM Signatures) and did finally order a used manual on that auction site.  Waiting on it's arrival.  While on that site I also located and bought the updated RAM board - also waiting on the arrival.  I will likely update the progress as these items arrive.

Thanks for the input.  So much to catch up on around here, I look forward to being able to dive back in...

[rastro]

Hi, that scrambled text looks like a dodgy ram chip.  You will be able to check that with the signal analyser.  You might also notice the the signal on a faulty ram line will not look the same as the others.

I agree.  The RAM chips on the memory card are high failure suspects.  I've replace one chip on the RAM PCB but also found in my replacement supply that another would fail intermittently.  I used a chip programmer to test the RAM(s) out of circuit.  In fact it's going to be much easier to verify RAM and Firmware(ROM) in a programmer than diving in with a LA on the back-plane to start out with. 

The HP 3582A has inherent flaky boot problems - I have 2 units that work 100% - most of the time.  I think there may be some borderline design issues on this model.

[Dacke]

I agree.  The RAM chips on the memory card are high failure suspects.  I've replace one chip on the RAM PCB but also found in my replacement supply that another would fail intermittently.  I used a chip programmer to test the RAM(s) out of circuit.  In fact it's going to be much easier to verify RAM and Firmware(ROM) in a programmer than diving in with a LA on the back-plane to start out with. 

The HP 3582A has inherent flaky boot problems - I have 2 units that work 100% - most of the time.  I think there may be some borderline design issues on this model.

Would that be a programmer like the TL866II?  I was aware that they could test 74 series and CMOS logic but not RAM.

[rastro]

I agree.  The RAM chips on the memory card are high failure suspects.  I've replace one chip on the RAM PCB but also found in my replacement supply that another would fail intermittently.  I used a chip programmer to test the RAM(s) out of circuit.  In fact it's going to be much easier to verify RAM and Firmware(ROM) in a programmer than diving in with a LA on the back-plane to start out with. 

The HP 3582A has inherent flaky boot problems - I have 2 units that work 100% - most of the time.  I think there may be some borderline design issues on this model.

Would that be a programmer like the TL866II?  I was aware that they could test 74 series and CMOS logic but not RAM.

Yes, I use the older model but I believe the new one also supports RAM testing.

http://www.autoelectric.cn/MiniPro/TL866II_Instructions.pdf (see page 2)

[cncjerry]

I go to Artek for manuals but I never checked to see if he has any of the rom images.  You might want to send Art a note to check.  It's hard to tell, and I'm going to look at your pictures again, to see if there is anything recognizable at startup.  I don't remember if that has a battery but maybe removing it if it does and letting the box drain down might change things.  These problems are tough.  I bought a powered solder sucker to fix rom problems in a scope.

[Alex Eisenhut]

Yes, I use the older model but I believe the new one also supports RAM testing.

http://www.autoelectric.cn/MiniPro/TL866II_Instructions.pdf (see page 2)

SRAM only even though the menu says SRAM/DRAM. I didn't find any DRAM parts though.

[rastro]

Yes, I use the older model but I believe the new one also supports RAM testing.

http://www.autoelectric.cn/MiniPro/TL866II_Instructions.pdf (see page 2)

SRAM only even though the menu says SRAM/DRAM. I didn't find any DRAM parts though.

I believe the the HM6116P on the memory card is an SRAM.  Not sure why you are looking for DRAM???
see pictures of actual testing...

[Duce67us]

The first gen RAM boards all used (qty 16) 4kx1 DRAM
There must have been multiple issues with these boards.  Later, HP released the Newer revision RAM module that used (qty 4) 6116 SRAM ICs   

I am waiting on the newer revision SRAM card to arrive.   
But here is a picture of the older DRAM.   

[rastro]

The first gen RAM boards all used (qty 16) 4kx1 DRAM
There must have been multiple issues with these boards.  Later, HP released the Newer revision RAM module that used (qty 4) 6116 SRAM ICs   

I am waiting on the newer revision SRAM card to arrive.   
But here is a picture of the older DRAM.

I think I also remember a 3rd style board referenced in some documentation that has just 2 RAM chips (S/D???).

Other than the 6116 SRAMs I have not explored any other RAM's.  So the MinPro may not support DRAM chips because of the refresh over head.  I don't know...but it would be a nice feature.  I guess I got lucky having the SRAM memories since it made finding faulty memory significantly easier.

[Duce67us]

I wanted to update you all on the status.  After receiving and installing the newer revision RAM board, I had the same symptoms of a non functioning unit with garbage on the screen. 
At this point, in bare bones configuration the A7 (processor) test was passing the SA tests, the A6 ( ROM) mostly passed, and the A8 (RAM) test fell apart with unstable readings. 

The one main clue here is that for the A8 test, A7J2 is in the run position.  So in my case, checking the signals at the A7 board that were present in test mode, and missing or sometimes misshaped in Run mode pointed to faulty control lines from the processor.   

After locating a replacement A7 board and installation,  the unit powers up and is displaying the proper screen.  The data loading indicator is flashing, and moving the input mode switch does change the single / both channel indicators. 

The unit also reads the front panel switches and Both ROM boards pass self test, as does the 12 minute RAM test.   

Next step is to try to step through the operational verification process.   

[Duce67us]

I did recheck the front panel switches, and found that the rpg was dead.  If you search around the web enough, it is possible to retrofit a led in place of the incandescent lamp, but it is NOT a simple swap. 
I just swapped in the same 5v lamp as the original.  I believe it took longer to remove and install the rpg than to make the repair.   

Here is a pic of the inside of the rpg module.