Datron 4200 problems

[chilternview]

I've recently acquired a Datron 4200 (and also a 4000A, but that's a separate topic). The eBay seller called it 'pulled from a working environment' and showed a picture of it powered up, with frequency 1Kz and voltage 1.000000V displayed.

Well, Datrons do not seem to like being transported. They are big heavy (36Kg) beasts afrer all.

On arrival, it powered up and showed the same display as in the listing. But none of the buttons seemed to work, sigh.
Except one - the Safety Reset one. Then immediately a Fail 5 error was shown, and subsequently powering up has always shown Fail 5, and remains unresponsive. This I gather from the docs is the watchdog timer has timed out / tripped.

So far I have inspected all boards for signs of blown tantalums or other damage (nothing found). I've checked the power supplies:
+5V, +8V unregulated for the digital & analog interface - OK
+15v, -15V, -10V, +36V for the reference divider - OK
+/- 15V, +/-38V for the output control - OK
+/-15V, +/-8V for ths since source and AC boards - OK
+/-15V for the Current board - OK
Not checked any of the high voltage supplies - should I? Which, and where if so?

I've also verified the 6802 is clocking and activity is present on the bus, the SSDA is sending data across the interface, most outputs of M51 are active *except* pin 9 which drives the watchdog signal. Its not clear what is causing the watchdog enable to be inactive, the normally helpful documentation only says "the CPU is informed via SSDA status byte transfer of certain analog malfunctions" - err what might those be?

Any guidance appreciated while I tear my hair out.

[TheDefpom]

I did a bunch of work on my Datron 4700, which is a very similar unit.

Things I found were bad tantalums (it uses 25V tants on the 15V rail, they all NEED changing to 35V), bad power supply electrolytic caps, cracked solder joints, blown thermal fuse on high voltage board, and many other things.

My video series is here in case you want to watch it, it might spark some ideas:
Part 1:


Part 2:


Part 3:


Part 4:


Part 5:

[chilternview]

Thanks Scott, some good tips there!

A bit more reading the schematics and it seems the status byte for SSDA transfer back to the CPU is assembled by the logic in the reference divider board from various status signals, these are as below, with the voltages and logic levels (as the gates have vdd=0v, vss=-15v):

J4-74 400V Fail 0.5v - HI
J4-91 38V Fail -14v - LO
J4-92 15V Fail -14v - LO
J4-37 36/15v Fail -14v - LO
J4-83 _HVST_ 0v - HI
J4-23 OVERTEMP -14v LO
J4-76 _LIM ST_ 0v - HI

So immediately suspect is the 400v fail signal. I'm not sure what HVST is - its the high voltage status detector though so I guess that's a clue - or LIM ST which may be due to some limit being reached. I guess the next thing is to figure out how to probe the 400V supply and see if it's working or not.

[volvo_nut_v70]

I believe there is a toggle switch at the back that allows you to disable the 400V, you can see what happens then.

I have attached my spreadsheet of capacitors and other parts that my help you along.

[chilternview]

I believe the toggle switch for 400V/50V just reduces the 400V supply for testing. I tried it, it has no effect on the fail error alas.

[chilternview]

A bit more testing... switching the 400V supply to 50V does change the logic level on J4-74 to LO (400V Fail). And testing the 400V supply gives normal readings. So the focus is on LIM ST and HVST. Both of these are set logic HI. HVST certainly should not be high, and it's probably the easier one to investigate, also the manual says it can give rise to FAIL5.

[TheSteve]

On one of my Datrons I kept getting an error from an incorrect voltage. Turns out they were all fine and the circuitry monitoring the voltages had failed.

[chilternview]

In fact I've realised that the _LIMST_ and _HVST_ signals, being active low, will normally be at 0v, and go to -15v to signal an error. So it looks like all the supplies are OK - however the watchdog trip disables the 400v enable signal, and so there's no 400V to the PA board (and hence the neons on it don't light). It's all a bit chicken and egg - does a supply problem cause them to be shutdown, or as you say if the protection cirtuitry has a fault then the supply is shut down. I tried disconnecting the 400V supply from the PA board and have verified that the 400V to the board is OK.

So it's really back to square one, why is the watchdog tripping and not resetting. Given that the calibration seals on this unit were intact, it has not been 'got at' so I'm still optimistic it will be a simple (though perhaps no obvious) fault.

[alm]

I assume you read section 2.1.2 from the service manual, volume 1, which talks about how fail 5 is a default error state that happens whenever fail 2 / 3 / 4 occur, and has some hints for spotting the original failure? Otherwise that might be of help.

[wutieru]

I guess this pic might help you.

[chilternview]

Just for info: the unit powers up displaying FAIL 5 and does not show any other failure codes before that. So it's no a limit etc as a possible precursor - the watchdog timer is not getting reset.

[wutieru]

The 4200 has schematics and is not difficult to repair.

[chilternview]

So a little more work on the 4200. I have gone round replacing tantalums and electrolytics, so far only found 2 bad caps, C39/C49 (22uF 350v) on the PA board that measured ~10uF and ESR 20ohms. However the fault remains.

According to the 'Possible Fault Locations' info for FAIL5,

1) Digital PCB - no gated WRT STRB pulses at J2/J3-29 - there are indeed no pulses, the signal _WDOG_ is always HI. And in fact M51 pin9 is always LO. Tried swapping M51 in case it had a stuck at fault but to no avail.

2) Analog IF PCB - no SSDA strobe or watchdog disabled - the SSDA strobe is there and the problem is the watchdog is not being reset due to (1).

3) Ref Divider PCB - again lack of activity on J4-97 means the monostable is not getting reset.

So the puzzle is what is causing M51 pin9 to always be low, instead of generating 8mS pulses to reset the WDOG monostable. It's not clear whether in FAIL5 mode the CPU locks up - the safety reset key seems to be working and generates _FP RST_ on the digital board. And other keys generate _KYBD IRQ_ which should be interrupting the CPU, but don't (the 'B' busy light on the front panel never lights if a key is pressed).

So I'm still baffled, there could be many reasons why M51 pin9 is not pulsing:
- FAIL5 stops the CPU from generating it due to the status byte transfer from SSDA indicating a fault
- a ROM is corrupt
- something else I've not thought about!

[chilternview]

And so I found and fixed the problem.

I had checked the status byte returned to the CPU by the SSDA and all looked OK. The power on reset signals were all OK too. All the power supplies checked out correctly.

But the watchdog signals 'BARK' (someone at Datron had a sense of humour) and BARK DEL stayed constant from power up onwards.

So something was wrong with the monostables and flops M10/M13 on the ref divider board. And checking the reset signal on pins 4 and 10 showed they were staying at -10v. A check on _FP RST_ the reset signal from the cpu board showed it was functioing correctly i.e. going LO about 8mS after po9wer on, and staying low until about 470mS as expected from M53. Now normally the signal goes from +5v to about +3.5v (the difference being the drop when the LED in the optocoupler conducts) but _FP RST_ was switching between 0 and 5V - so the diode might be faulty.

A quick rummage in the spares box found a Vishay MCT6 which was a pin replacement for the original FCD880 optocoupler. And hey presto, on power up the 4200 works!

[volvo_nut_v70]

Excellent work!! Thank you for sharing it! 

[chilternview]

The unit now passes self-test, one non-working front panel switch replaced, all seems to be good except that the 1000V range does not give any output. As the 100V range works correctly, it must be something in the HV transformer / switching circuitry?

[DC1MC]

The unit now passes self-test, one non-working front panel switch replaced, all seems to be good except that the 1000V range does not give any output. As the 100V range works correctly, it must be something in the HV transformer / switching circuitry?

It must be another optocoupler, these are the Kriptonite of Datrons 

[TheDefpom]

all seems to be good except that the 1000V range does not give any output. As the 100V range works correctly, it must be something in the HV transformer / switching circuitry?

There is a thermal fuse on the top edge of the HV PCB, it could be blown, it was on mine when I got it, also make sure you have the internal toggle switch set correctly.

[chilternview]

I checked the switch on the PA board, it is set in the correct position. I can't see where a thermal fuse is though?

Also the LED next to the switch doesn't come on - indicating (I believe) that RL6 is not activated.

Looks like checking some signals is necessary. RL6 is only on if both BARK (J9-66) and _1KV_ (J9-32) are LO. Time to get the probes out again.

[chilternview]

And the problem was with the _1kV_ signal - one of the status registers (M30 on the ref divider board) was faulty. Fortunately I had a bag full of MC14094's to swap it out.

Now the unit seems fully working. Time to calibrate it.

[alm]

Nice job on fixing it! I hope it will now perform within specs after calibration.