NOS Dana 1992 just went crazy - fixed - broken CPU

[kawal]

I bought a NOS dana 1992 with manual and power cord. Seems like a spare unit  kept for the military. It worked great  for about 10..15 hours and now will not boot.  Trying to boot multiple times would bring it back but now that also is not working.  I read the eprom multiple times out of socket and its always reading back the same, so I don't think it the eprom . Now even multiple reboots are not helping and the unit is stuck.  The screen comes up either blank or with random segments lit up.  First I thought it was the 244 chip that was bad and replaced it . The symptoms are getting progressively worse if that's even possible.

The standby button would work  but nothing expect a reboot would bring it back from standby.  Now its will not react to standby most of the time.  I did check and the gate that is connected to standby is reading the standby button.
I did pull the keyboard and verified each key is working. There are no stuck keys, None of the keys are broken yet.
The reset circuit is  held low at power up and goes high which seems normal.  When the Standby circuit is not reacting I noticed the IRQ pin is being pulled low. 
When IRQ is high the PB port activity seems not normal.  Sometimes there is just spikes on the data pins and other times there is ramping signals  and sometimes some pins show normal square wave activity . Even when activity  looks more normal there is no screen activity  - it remains  at the same condition as when booted up.
The IRQ circuit is pretty complicated  but the more I look at it the more I think the MC146805 is toast.

The  voltage rails all are in spec.  The logic chips get a nice clean 4.995V .

On occasion when the keys are pressed on the front there is a lot more port activity  but nothing is getting updated in terms of LED or displays.

I looked at other threads but not exactly the same issue.
https://www.eevblog.com/forum/repair/more-racal-dana-1992-repair-woes/

Help  appreciated.

[Stray Electron]

  FIRST, don't start replacing parts willy-nilly like the OP in the other thread did! Not only is that expensive, but there is also a butt load of fake parts out there today and there is a very good chance that you're install at least one bad part and cause yourself even more problems. And it's not just the fake parts, but also many 30 year old NOS parts could be failing by now. And add in the risk of damaging circuit runs, solder bridges and the like.  Take your time and actually troubleshoot each suspected circuit and only replace what is necessary.

  I would start by pulling every connector and every socketed IC and make sure that the contacts and the IC pins are clean and aren't tarnished or corroded.  A lot of this old stuff used socketed ICs which is nice from a repair point of view but the problem is that the socket contacts were made of one material and the IC pins another and they slowly develop poor connections due to galvanic corrosion. Especially if keep in any kind of humid environment.  After making sure that they're clean, it wouldn't hurt to apply Deoxit to all of the switch contacts, plugs and sockets, etc.

[TheDefpom]

The front panel buttons are likely the problem, they go bad from age (not from use) and they can then confuse the uP on boot up.

I did a couple of videos replacing them on different units, I had links in some cases too to a source of the buttons I used.

Definitely worth re-capping it too, old is old.

[David Hess]

You measured the DC voltages but did you check for AC ripple?  It sure seems like a worn out aluminum electrolytic capacitor problem.

[coromonadalix]

I have two of them  sleeping somewhere, i did remove all the tact switches on one of them, and never finished  my job  loll

even ordered some tcxo clocks and pcb's  for a reference clock upgrade in them

And one of them  had the mentioned problem,  some tact sw  where stuck "pushed"

[kawal]

  FIRST, don't start replacing parts willy-nilly like the OP in the other thread did! Not only is that expensive, but there is also a butt load of fake parts out there today and there is a very good chance that you're install at least one bad part and cause yourself even more problems. And it's not just the fake parts, but also many 30 year old NOS parts could be failing by now. And add in the risk of damaging circuit runs, solder bridges and the like.  Take your time and actually troubleshoot each suspected circuit and only replace what is necessary.

  I would start by pulling every connector and every socketed IC and make sure that the contacts and the IC pins are clean and aren't tarnished or corroded.  A lot of this old stuff used socketed ICs which is nice from a repair point of view but the problem is that the socket contacts were made of one material and the IC pins another and they slowly develop poor connections due to galvanic corrosion. Especially if keep in any kind of humid environment.  After making sure that they're clean, it wouldn't hurt to apply Deoxit to all of the switch contacts, plugs and sockets, etc.

Did that before anything else.  That is always good advice

[kawal]

The front panel buttons are likely the problem, they go bad from age (not from use) and they can then confuse the uP on boot up.

I did a couple of videos replacing them on different units, I had links in some cases too to a source of the buttons I used.

Definitely worth re-capping it too, old is old.

I tested each switch on the front panel and all are open when not pressed and measure about 250 ohms when pressed.  So i don't think its the keyboard. 

[kawal]

You measured the DC voltages but did you check for AC ripple?  It sure seems like a worn out aluminum electrolytic capacitor problem.

I did measure the ripple and no issues there.  Also checked all the caps with ESR meter and capacity meter.  All test good for ESR and capacity.

[kawal]

I have two of them  sleeping somewhere, i did remove all the tact switches on one of them, and never finished  my job  loll

even ordered some tcxo clocks and pcb's  for a reference clock upgrade in them

And one of them  had the mentioned problem,  some tact sw  where stuck "pushed"

I assumed the keyboard was the issue but i don't see bus activity from the keyboard until keys are pressed. Unless the buttons behave different in the chassis versus out of the chassis I am pretty confident about the buttons being good.  I will check that again . Maybe I can de solder the keyboard board connector and try running the unit without a keyboard.  Maybe easier to pull out the keyboard controller as its in a socket.   

[edpalmer42]

Normally, you can tell by feel if the keyboard is good.  If you slowly press the key and it 'pops' in and then when you slowly release it and it 'pops' out, the key hasn't failed.  IOW a good key gives you good tactile feedback.

Ed

[kawal]

Normally, you can tell by feel if the keyboard is good.  If you slowly press the key and it 'pops' in and then when you slowly release it and it 'pops' out, the key hasn't failed.  IOW a good key gives you good tactile feedback.

Ed

Ed
All keys feel good.  you can feel the pop in both directions.
Maciej

[IanJ]

Given the symptoms outlined in your first post I would suspect bad caps. I understand you measured them and they seem good.......but still!

Also, since the symptoms got worse, and ultimately now fails to boot then that only lends to the same cause.

Try scoping the supply rails and monitor how they come up at power-on......any spikes or noise  or other unusual activity may give you some clues and ideas on how to move forward.

Ian.

[edpalmer42]

I found that my 1992 was oddly sensitive to power glitches.  It worked fine normally, but if there was a power failure, it would reboot even though I had it on a UPS.  I theorized that the power supply couldn't handle the switchover from line power to UPS power.   I didn't record any values in my notes, but I did note that there was very little headroom for the regulators.  The 1992 doesn't have any heatsinks on the back so it can't dissipate a lot of power in the regulators or pass transistors.

Ed

[seedkey]

Check the frequency standard . Make sure it is working

[kawal]

Given the symptoms outlined in your first post I would suspect bad caps. I understand you measured them and they seem good.......but still!

Also, since the symptoms got worse, and ultimately now fails to boot then that only lends to the same cause.

Try scoping the supply rails and monitor how they come up at power-on......any spikes or noise  or other unusual activity may give you some clues and ideas on how to move forward.

Ian.

Ian

I will look into that . In stable state there is no noise on the power rails when  looking at the rails with a scope. I did not check the rails  power on. I did check the reset circuit during power on and that is working.

I also tested removing the interface chip on the keyboard - i thought is was in  a socket but it was not.   Either way  I removed the chip that interfaces the keyboard matrix and still the same. If I don't find anything else I will condemn the MC146805.  Nothing else is left to control the display or explain the behavior. 

BTW  Ian I enjoy watching your channel. Keep up the good work and thanks for the tip.

Maciej

[kawal]

Check the frequency standard . Make sure it is working

Frequency standard is working and the processor  is also getting a nice clock. Feeding and external  10MHz  is not helping either.

[edpalmer42]

Check the frequency standard . Make sure it is working

Frequency standard is working and the processor  is also getting a nice clock. Feeding and external  10MHz  is not helping either.

My 1992 has the OCXO option.  Since I use an external standard, I thought I'd remove the OCXO and reduce the internal heat.  It turns out that the unit always boots up on the internal standard, then recognizes that an external standard is present and switches over.  So an internal standard of some type must be present or the unit won't work at all.

Ed

[coromonadalix]

could be the ocxo  design ?? not enough amplitude ?

dont have the schematic at hands to check if you have some presence  pins etc ...

[kawal]

Check the frequency standard . Make sure it is working

Frequency standard is working and the processor  is also getting a nice clock. Feeding and external  10MHz  is not helping either.

My 1992 has the OCXO option.  Since I use an external standard, I thought I'd remove the OCXO and reduce the internal heat.  It turns out that the unit always boots up on the internal standard, then recognizes that an external standard is present and switches over.  So an internal standard of some type must be present or the unit won't work at all.

Ed

I will log the OCXO bootup in the next few days.

[kawal]

Ok
I got some NOS MC146805E today and replacing the processor did the trick. unit powers up right away.
I did check the 5V rail startup on CPU  and there is a nice quick ramp to 5V no overshoot. 
Picture of the NOS  processor.

[edpalmer42]

Congrats!  It's nice when the fix turns out to be so easy.

1989 Week 48 - that's not NOS, that's NOOOOOOOS! 

Ed

[kawal]

Dana 1992 is alive and well
Video of it starting up connected to a OCXO distribution box (there is an OCXO inside the box)
https://photos.app.goo.gl/BoxZdqDu6QdpAGreA

Also picture of the defective processor.

[David Hess]

That is a great job finding the bad microprocessor.  Big ICs like that are usually the least likely to fail unless abused.

[kawal]

That is a great job finding the bad microprocessor.  Big ICs like that are usually the least likely to fail unless abused.

I agree very strange since this unit had zero hours of usage when i got it. I used it for about 12 hours before it failed.  maybe some manufacturing defect on the CPU that would have been covered by the 1 year warranty that ended in 1990?
Either i am sure happy that its working again.

Maciej

[IanJ]

Ok
I got some NOS MC146805E today and replacing the processor did the trick. unit powers up right away.
I did check the 5V rail startup on CPU  and there is a nice quick ramp to 5V no overshoot. 
Picture of the NOS  processor.

Great news.
I have had knackered processors before, but usually it's because of a power supply issue. However, I have had faulty Z80A processors back in the day for no apparent reason.
Glad you are back up and running, welcome to the test equipment repair family.......😀
Ian.