[SOLVED] More Racal Dana 1992 repair woes

[Squiddaddy]

I finally got back to doing some tests on my Racal Dana 1992 counter. Trying to get a handle on all of the issues so I know which way to proceed first.
I bought a new batch of switches to replace the ones in it, but I don't have much issue with them yet. So I may do that last.
I am looking at replacing the capacitors, but wanted to know what amount of ripple voltage were you seeing in the bad caps? I am getting around a half to a volt p-p of ripple on the power supply caps.
I was having an issue when I powered it back up after several months off, that the 10MHz output when coupled back to the input was reading 30 MHz. That had me completely stumped. But when I put the scope on the output, it was very low in amplitude. I reseated the IC's in the sockets, and that's when I found the next issue. The IC39, which is one of the unobtainium chips, I'm guessing has a flaky socket, since I can press on it or the board, and get the 10 MHz amplitude back to where is needs to be, or mess it up. Reseated it several times.
Has anyone had any good luck replacing that IC socket? I don't want to damage that chip since it'll scrap my unit.
Also, I still have the issue where I can place the unit in standby, but it will not come back out, and I have to power cycle the unit. According to the schematic, this is also controlled by one of the 'special' chips'. It must not be resetting the IRQ line on the CPU.
The counter is counting the signal from the HP generator well, and not drifting, so I'm not convinced that any of the 'special' chips are dead. But it does show random numbers on the input when first powered up. Applying a signal, or raising the trigger lever fixes that.
Does this unit still seem salvageable, before I go spend the time and money on caps, sockets, etc.?
Thanks!

[coromonadalix]

half a volt is a lot for sensitive equipment like an counter, meter or a signal generator.  Good ic sockets are the machined ones (round pins) they maintain their contacts strenght.

Maybe you may have some cold solders joints,  poor connectors contacts, maybe some humidity

Do you have electronics experience,  i would begin by changing the supply capacitors,  but it should be done professionally without creating pcb stress, component stress  etc ... some desolder pumps give huge kicks on pcb, i always use desoldering pumps.


The reset maybe affected by the capacitors issues ....

Your unit is repairable, dont throw it away,  and once again  i would have helped for free if you where in my country, i have specialized stuff at my job.

If your inputs signals and readings are matched, you know the Dana  is working well,  simply the bad contacts issues may be left to solve.

[Andy Watson]

Inspect the PCB and solder joints carefully. I have read that Racal had problems with their PCB production, problems which they never fully solved. I had similar intermittant problems which I curred (so far) by re-flowing all the joints around the main chip sockets.

[coromonadalix]

Yeah  Maybe the contacts problems should be checked first,  dont change parts yet,  You can assume the counter still works

Keep a log of what you do sections by sections

And please apply anti static security and manupulation  counter measures

[Squiddaddy]

Thanks guys!
I do have electronics experience, and I have the desoldering tools. I have repaired scopes, power supplies, and stuff. The reason I'm being cautious is that this is a sensitive instrument and I don't want to damage any of the unobtainium parts. I couldn't tell if this is a multi layer board, which makes changing sockets riskier.
The counter is now hooked up to a GPSDO and the count is good, so I do believe it functions well. Reading 10.000000015 MHz.
Been powered up for several hours now. I tried standby again, and it got stuck.
It's this intermittent crap that worries me, as I don't want to make it any worse.
My date codes on the chips say 87, so maybe the cap change is in order first off just as a sensible measure. Thanks for the feedback on the ripple values.
Time to stop talking about it and do these one step at a time.
I will start by changing caps and reflowing the sockets.
Appreciate the feedback!

[edpalmer42]

Remember that the 1992 uses a linear power supply so the ripple will be higher than you might expect.  As long as the minimum voltage is comfortably above the minimum voltage for the regulators you should be okay.

Ed

[Squiddaddy]

Well, i managed to get all of the pushbuttons replaced, and i have new capacitors on order. My issue with getting stuck in standby remains, and i started reading the section on the standby and irq circuits. Don't believe the caps are causing the issue, but it won't hurt to replace them.
Not sure why the latch isn't resetting. Haven't wrapped my head around the logic states yet.

[perdrix]

My memories of 1992 repairs involve quite a few SMD parts on the underside of the board that only had solder on one side of the component!!!   Maybe they were trying to wave solder them and the wave wasn't high enough?

Worth checking that anyway.

HtH, Dave

[Squiddaddy]

Will do. I have another intermittent with the rear 10MHz output too. It drops in amplitude occasionally. Guess it's reflow time. Lol
I want to get this going because it's very accurate. Holding it own while monitoring a gosdo.

[rodd]

Hi,
Would you please post the name of the place that has these pushbuttons available?
Thanks


Enviado do meu iPhone usando Tapatalk

[Squiddaddy]

Switches are here.
https://www.aliexpress.com/item/Import-Tact-Switch-6-6-11MM-2-pin-DIP-micro-switch-waterproof-and-dustproof/32725667059.html
Get at least 2 packs as the 1992 has 33 switches to replace. Also, watch TheDefPom's youtube video on changing them.
Scott did a great video on it, and he is a member here.

[Squiddaddy]

I looked over the board under a magnifier, and i do not see any cracked or bad solder joints.
But after doing some simple logic checks, it doesn't appear that the standby latch is getting reset after going into standby.
I don't have a logic analyzer to watch all of the signals, so i'm not sure how to confirm.
I could either be the NAND gates in the latch, or the PortA signal from the 6805.
I don't have a spare 6805, and don't want to remove the 4011 chip if it isn't bad.
Any ideas on how i can confirm the issue?
Thanks.

[Squiddaddy]

I ordered a replacement CPU and a batch of CD4011's. But here's what I'm noticing.
It appears that the NAND latch is getting into a metastable state. When it goes into standby, some of the pins go to about 3.4 volts instead of 5 or 0.
Since this is a simple RS NAND flip flop, both inputs at 0 is an invalid state.
The standby switch goes 0 when pressed, and the CPU is supposed to reset the other side once the IRQ completes.
Can a logic ic become more susceptable to metastability as it ages? Would replacing the CD4011 make sense as a first step?
Or am i chasing the wrong tail?

[tggzzz]

I ordered a replacement CPU and a batch of CD4011's. But here's what I'm noticing.
It appears that the NAND latch is getting into a metastable state. When it goes into standby, some of the pins go to about 3.4 volts instead of 5 or 0.
Since this is a simple RS NAND flip flop, both inputs at 0 is an invalid state.
The standby switch goes 0 when pressed, and the CPU is supposed to reset the other side once the IRQ completes.
Can a logic ic become more susceptable to metastability as it ages? Would replacing the CD4011 make sense as a first step?
Or am i chasing the wrong tail?

It isn't metastability per se, since metastability resolves itself "quickly" - albeit not within one clock cycle.

However, the measurement is interesting. If it is a TTL (not CMOS) output, 3.5V is well within limits. If you are measuring with a voltmeter, then the output could be oscillating and you are measuring the mean output voltage. If the voltage is constant, perhaps the output has lost its "oomph" or is being dragged down externally.

Oscillations could be caused by poor PSU decoupling (check DC and AC voltages!), an invalid input voltage (noise, level), poor solder joints on that IC, or just the IC being faulty.

[Squiddaddy]

Well, things seem to be getting worse. I had let the counter run for several weeks to monitor my gpsdo.
Today, i was going to power it off to work on the standby issue, and i noticed that i have an issue after replacing the buttons.
But i didnt have it when i replaced them. At least i never noticed it before, as i checked things to make sure all of them were working.
Now, if i press certain buttons, the wrong leds light up.
It's no longer responding to the extended keys from the encoder.
I am just so confused.
Anyone have any ideas???

[Squiddaddy]

Well, here is where my testing has led me.
After reading the service manual and tracing out the function of both of my issues, I think I may have found the culprit, and that they are related.

I had the standby issue for a while, but the loss of half the keyboard is recent.
Both of these generate an IRQ flag for the CPU to check to determine the type of IRQ generated.
It doesn't appear that the CPU is seeing these flags, and so does not act on them.
The standby still powers down the DC bus, because that is controlled by a simple flip flop. But by then, the CPU has lost it's way and locked up.
The keyboard is only scanning half of the keys, since it doesn't see the Keyboard Extended flag. So it returns the corresponding key in the other half of the keyboard matrix.
I think the issue is the 74HCT244 buffer that transfers these flags to the CPU.
Does it make sense that the chip could lose 1 or 2 of the buffers, and not the whole chip? It's odd that the issues appeared over time, and the only relation between the 2 issues is the 244 buffer chip.
I have a chip on order, so I hope that I'm on the right track.
What an exercise in logic tracing! 

[coromonadalix]

It can happens, a short, an internal defect, bad manipulation while probing  etc...   nowadays anything can go wrong loll

[MarkL]

My 1992 had cracked traces on the keyboard PCB that disabled some of the keys and caused others to be read as the wrong key, similar to what you're seeing.  If I recall (it was 10 years ago), I did the repair from the back by checking the continuity from key to key for each scan line all the way back to IC1 and IC2 according to the schematic.  I installed 2 or 3 wire jumpers to get around missing continuity.

It doesn't, however, explain the STANDBY issue, since that's not part of the scan matrix.

A '244 could lose a single buffer, but I think somewhat less likely for two of them to go bad.  It's a good theory, but you could still be chasing two separate issues.

[Squiddaddy]

Well, this seems like the more I poke at it, the worse it gets.
I totally feel like I got gut punched now.
I got my parts today, so I decided to replace the 74hct244 buffer in the IRQ section.
I removed the old chip and installed a socket. I figured I'd power it up without the chip to make sure I didn't go backwards.
It powered up, displayed the model and version, and then got stuck. I figured, no big deal, it's missing that chip. Powered off.
So then I put the old 244 in the socket, in case of issue I didn't blow the new chip, and I'd be back to where I was before I started doing anything.
Booted up, got model and version, the got the display. But things weren't right, and when I pressed a key, it went nuts.
I powered it back down and took out the old 244, and powered back up. Now I get random display instead of model and version.
Put in the new 244, still random display on bootup. But it appears to be counting, but the keys are all messed up.
But now standby works. Go figure.
I did get Opr Er on one key, so it is responding to things.
The 74hct244 is not tied to the keyboard and display, so I have no clue what the hell happened, or how to proceed.
Do I now have a piece of scrap, or does anyone have an idea on what I can do next???
Totally sick to my stomach now. It's not hard to replace an ic, but I sure got bit on this one. 

[Andy Watson]

Do you have facility to read the EPROM? IF so, read it several times and check that the results are always the same.
Another thing that migh be worth checking is that the chips have a power supply - check on the actual pins of the chip.

[coromonadalix]

Maybe with all the manipulation,   maybe an cold joint or badly socketed parts   cracked pcb trace as mentioned earlier, maybe working with a variable heatgun or cold freeze  could help ??

Maybe 2 or 3 problems at the same time

[Squiddaddy]

Well guys, I finally got to the bottom of this.
I removed the socket for the 74hct244 and cleaned up the board.
Then carefully installed the socket again. Booted up, all good to the lockup, since the chip was missing.
Installed the bad chip in the socket, and I made it back to my original problems.
So that told me I got the socket installed ok.
I went to install the new chip and i noticed the pins were darker than the original.
These chips came from Digikey, but appear to be NOS TI chips. May have an 83 datecode on them.
Don't they make these in DIP anymore?
I think it's tarnish of some sort. I put the chip in and wiggled it around to scrape the pins a little bit.
Powered up, and everything is back working and both of my issues are gone.
It was definitely the 74hct244 that was bad.
Back to the chip, what's the best way to clean those pins without damaging the chip?

Big thanks to everyone who chimed in to try and help me out!!!
Another Racal Dana 1992 lives on. 

[coromonadalix]

Maybe some deoxit cleaner,  or use a pink eraser to gently scrape the black oxydation ??  the pink eraser will act as a sand paper ...  or maybe source another chip

Oh you have brasso cleaner to polish metal, was effective to clean some oxydation ??

[duak]

TI chips in the 70's were famous for turning black, although I thought they stopped by 1983.  I understand it was because they were silver plated and the sulfur in the air caused sulfides to be formed.  These can be removed either mechanically or with a strong enough acid.  Silver tarnish remover should work.  Pepsi works too because of the phosphoric acid, but I don't recomend it.  Do you have any citrus based cleaner?

That being said, AgS is supposed to be conductive, but I can't say it will work in sockets.

Cheers,