Datron 1081 Teardown and repair

[MaxFrister]

I recently acquired a Datron 1081.  The Datron 1081 is a 7½ digit meter and is similar to the Datron 1071 but higher specs. 

Basic 24 hour specs:
                   1071                                  1081
10VDC        +/-2ppm +/- 4 digits         +/-2ppm +/- 1 digit
10VAC        +/-0.04% +/- 40 digits      +/-0.01% +/- 0.005%
10k            +/-5ppm +/- 2 digits         +/-3ppm +/- 1 digit

The 1071 was designed about 1975, the 1081 about 1983.   The 1081 achieves its resolution using a network of 4 selected temperature compensated zener diodes.

My unit was advertised as broken and no display.  Unfortunately, for once, the advertisement was right.  Preliminary checks showed the wrong fuse installed.  After installing the right fuse, the -175V rail was missing.  Further checks showed the transformer damaged.  The previous owner apparently used the transformer as a fuse.  There was also serious corrosion on one of the voltage regulators (a common problem, perhaps related to the wrong flux).

I’m waiting for caps and other parts to arrive to do some basic repairs to the power supply before I continue testing.  In the meantime, I thought I’d post a few pics.

Basic machine



The magic, 4 zeners


Corrosion on TO3 package wires



p.s.,
As I mentioned, I’m probably in the market for a new transformer.  If you happen to have a parts machine and you’re willing to part with it send me a pm (I believe that the 1061,1062, 1065, 1071, 1072, 1081 and 1082 all use the same transformers).

[maginnovision]

I've got 4 datrons and none of them have issues with the 5V reg or wiring, seems like I got lucky. After a bit of cleaning it'll look nice, really sad about the transformer though. I saw that in another unit and passed on it for that alone.

[Johnny10]

Are you in US?

[branadic]

Missing display? I'm working on a replacement solution.
https://www.eevblog.com/forum/repair/datron-4000a-getting-at-test-points/msg2295153/#msg2295153

-branadic-

[DC1MC]

That's not looking like a damaged display (except the broken ones with compromised vacuum, all Datron displays turn on a bit if powered correctly), it looks like a display without power !!!

After you're done with restoring the powers supply fully and getting rid of the 169% dead elcos you may have a nice surprise  .

 Cheers,
 DC1MC

[maginnovision]

That's not looking like a damaged display (except the broken ones with compromised vacuum, all Datron displays turn on a bit if powered correctly), it looks like a display without power !!!

After you're done with restoring the powers supply fully and getting rid of the 169% dead elcos you may have a nice surprise  .

 Cheers,
 DC1MC

Or a reason to sell to me! 

[feedback.loop]

How about rewinding the transformer? It is doable.

[maginnovision]

If it really is the transformer and not the bridge rectifiers I guess you could rewind it but I'd probably just buy another unit for the transformer and keep the rest for parts(relays, adc, display, buttons, etc...)

[MaxFrister]

I spent a long time last night replacing the corroded wires to the LM309K regulator and cleaning up the acid damage.  The design of the power supply does not leave much access for rework.

Thanks for all the good suggestions regarding the transformer.  I may have a lead on a replacement.  Anyway, that problem is in the future.  The current workflow is:

• Wait impatiently for the caps to arrive in the mail
• Install replacement electrolytics (none of the tants are obviously shorted)
• Try to see if basic machine works via GPIB
• Try to see if display is working using an external 200V source

[maginnovision]

Actually once you disconnected all the leads and pull the back away it's pretty clear.

[MaxFrister]

In your picture you can almost see the LM309k leads.  There might be a little more to it....

There are, of course, other ways to approach it.  In my case, replacing the leads for the LM309k required:

• Disconnect and remove the back panel
• Disconnect and remove GPIB board
• Disconnect and remove the Digital board
• Desolder the LM309k leads and transistor leads from the back pcb
• Unscrew the heatsink
• Drill-out the LM309K mounting screw!
• Clean and make new leads
• Attach regulator and transistor to heatsink
• Attach heatsink while trying to allign 6 leads with hidden holes in the back pcb
• Resolder and reassemble

If I had to do it again I would cut the plastic fastener on the transistor and leave the transistor soldered in place, but that will make reattaching it to the heatsink a challenge.

[maginnovision]

Taking the back panel off it took me about 10 minutes to R&R the 309 including replacing the plastic fastener that broke. Add 5 minutes to remove the back panel, gpib board, digital board. Not too bad really. I think they look a lot worse to work on than they are.

Once you get it all working it should be pretty nice.

[DC1MC]

I spent a long time last night replacing the corroded wires to the LM309K regulator and cleaning up the acid damage.  The design of the power supply does not leave much access for rework.

Thanks for all the good suggestions regarding the transformer.  I may have a lead on a replacement.  Anyway, that problem is in the future.  The current workflow is:

• Wait impatiently for the caps to arrive in the mail
• Install replacement electrolytics (none of the tants are obviously shorted)
• Try to see if basic machine works via GPIB
• Try to see if display is working using an external 200V source

The tants are really  tools, it will even look good in a LCR meter and then suddenly, under working voltage, will go directly full-short, no warning signs, nothing 

[MaxFrister]

I have lots of progress to report:

• I replaced the electrolytic caps in the power supply
• The power supply is working except for -175V
• I was able to connect and test the machine via GPIB
• The analog board is mostly working with an occasional Error 5 when cold
• The resistance board fails with Error 6
• The AC board has not yet been tested
• The cal constants are still there

Most importantly, the display worked when driven from an external power supply.  Thanks to a forum member, I also have a new transformer on the way.

I've attached a few more pics of the analog board showing the board over-all, another view of the zener reference, and some precision resistors.

[DC1MC]

Yes, I knew that was not a dead display  , congrats for almost 100% revival !!!

Btw, for the resistance test make sure that the device is switched to 2-wire resistance measurement and nothing is connected on the measuring terminals.

Good luck with the new trafo and keep us informed.

 Cheers,
 DC1MC

[maginnovision]

Only a few more issues to figure out then, pretty good. I actually just bought two more 1062 for the option 12 AC boards but if you've got a transformer coming now I guess I don't need to offer one. Can't wait to see it all better.

[MaxFrister]

The new transformer arrived  .

[DC1MC]

This looks like a perfect display  .

 Cheers,
 DC1MC

[maginnovision]

Not too bad. Pass self test now?

[MaxFrister]

Alas it still fails the resistance selftest. The ac board is still out of the machine and untested.  Tantalums are on order for the ohms board.

I went through the manual last night and prepared an order for the remaining tantalums.  I don't normally do preventative recapping but there seem to be a lot of reports of popping tants in Datron machines.   Unfortunately, replacement will be tedious; there are 7.5-digits of tants (actually ~75).   

I'm curious about the failure mechanism.  At least on the 1081, most of the tants are conservatively rated (35V on 15V).  They did use 6.3V on 5V rails and on the ohms board at least a couple rated 16V for 13V.  I wonder if excessive ripple on the failing electrolytics is damaging them...  Then again the original brand "Union carbide" does not install confidence.

For those considering defering replacement, it appears that "high voltage" leaded tants are becoming rare.  The 68uF/35v has already gone extinct at Mouser, Digikey, and Allied.

[maginnovision]

I replaced all tantalums, 250/400V film caps and electrolytics on all of my datrons(5 * 1062, 1 * 1072, 1 * 4000A) and don't regret it. Only the 1 and .1 are 35 V rated, the higher the capacitance the lower the voltage. I replaced the tantalums with all 35V. I also replaced the bridge rectifiers with 2A rated parts. The film caps I used direct replacements. Cost was about 110$ per meter, the 4000 was about $400. It really doesn't seem like they were appropriately derating as AVX would now recommend so I basically followed their advice.

Once you do that hopefully it's all good. If not you could attempt calibration of whatever boards are failing, at least the hardware Cal even if not fixing the software constants.

[Kleinstein]

Alas it still fails the resistance selftest. The ac board is still out of the machine and untested.  Tantalums are on order for the ohms board.

I went through the manual last night and prepared an order for the remaining tantalums.  I don't normally do preventative recapping but there seem to be a lot of reports of popping tants in Datron machines.   Unfortunately, replacement will be tedious; there are 7.5-digits of tants (actually ~75).   

I'm curious about the failure mechanism.  At least on the 1081, most of the tants are conservatively rated (35V on 15V).  They did use 6.3V on 5V rails and on the ohms board at least a couple rated 16V for 13V.  I wonder if excessive ripple on the failing electrolytics is damaging them...  Then again the original brand "Union carbide" does not install confidence.

For those considering defering replacement, it appears that "high voltage" leaded tants are becoming rare.  The 68uF/35v has already gone extinct at Mouser, Digikey, and Allied.

The failure mechanism with tantalum caps is usually small crack in the dielectric layer that under good conditions would repair with not much degradation. The power on after longer storage can be more critical.  However if too much it can lead to local short and possible a thermal runaway up to fireworks if the power source is too powerful.  So the good failure mode is a short, the bad one is a small fire that might burn through the board.  In many cases one could use a modern low ESR Al based electrolytic cap as well - these were not available in the early 80s.

[Johnny10]

I have replaced all the tantalums on both my Datron 4000 units and 1062's I also use a variac set to 115 Volts.

Have not had any problems since.
Kleinstein is right about units that sit unused being more susceptible to burned out tantalums on turn-on.

Ah! the smell of burning tantalums in the morning.

I know using the variac shouldn't make a difference but it is my experience that certain voltages are excessive at 125V Mains power.

[MaxFrister]

The ohms board is working.  I wish I could claim brilliant trouble-shooting but the reality is that after a couple of hours of proding, probing, cleaning, warming and throwing switches it just started to work.

I installed the ac board and, after sufficient warmup, it too passes self-test.

I'm waiting on two orders of tantalum caps and will install them once they arrive.

It appears the calibration is bad.  This may be as simple as as needing a software adjustment for the reference aging or a serious problem as the equipment required for a full cal is extensive.  I'll assess following the cap replacement.

The reference voltage is slightly out of spec according the the 1071 service manual, but there is no spec that I can find in the 1081 manual.

One disturbing finding is that the service manual has a different version of the digital board.

[maginnovision]

The ohms board is working.  I wish I could claim brilliant trouble-shooting but the reality is that after a couple of hours of proding, probing, cleaning, warming and throwing switches it just started to work.

I installed the ac board and, after sufficient warmup, it too passes self-test.

I'm waiting on two orders of tantalum caps and will install them once they arrive.

It appears the calibration is bad.  This may be as simple as as needing a software adjustment for the reference aging or a serious problem as the equipment required for a full cal is extensive.  I'll assess following the cap replacement.

The reference voltage is slightly out of spec according the the 1071 service manual, but there is no spec that I can find in the 1081 manual.

One disturbing finding is that the service manual has a different version of the digital board.

When you do the tantalums take care not to use an earthed iron on the digital board unless you want to calibrate. Of course when/if you do the battery make sure the unit is on. Whatever the reason it's good that it works.

For the calibration it's not necessarily all that difficult. It's virtually the same as 106x and 107x except you can't coarse adjust the reference you can only manage linearity from the looks of it. The real issue is without a very good calibrator there isn't much you can do with the 1081 that makes sense. The ohms is easy to adjust and the AC is not easy to adjust, I bought a 5200A and am having custom transformers made to manage that bit.

[dacman]

Error 6 is usually caused by high resistance in the input path (or current path) which can be due to a relay contact or the 2-Wire/4-Wire switch.

[MaxFrister]

Because the 1081 has a sealed case, the interior is remarkably clean and free of dust.  Despite that. there is some schmutz on the pcbs.  I’ve been careful not to contaminate them further by wearing surgical gloves while working on it.  (Wearing gloves and using forceps to repair the power supply, I felt like a real doctor.)


Anyway, I’m looking for advice on post-repair cleaning (or not cleaning) of sensitive pcbs. 

The other issue is that the soldermask has become brittle so the cleaning should be as physically gentle as possible.  The service manual mentions using IPA but also warns that even that can contaminate the boards.

[splin]

Is there any chance of getting a copy of the 1081 EPROM images? I have a 1081 that I want to install a GPIB board from a donor 1061 but the GPIB  firmware is different for each model and has to match the rest. Just the GPIB ROM would be a good start as I could dissamble it and adjust any addresses referencing the main ROMS but it would be a lot easier with a matching set.

[MaxFrister]

I will eventually read the eproms for backup, and I can let you have copies at that time.  In the meantime, I have memory dump of 2 other 1081s from another member that I can share with you.  Just pm me an email address.

[doktor pyta]

http://rfscientific.eu/memory-dump-directory

[splin]

http://rfscientific.eu/memory-dump-directory

Thanks - I found those ROM images before but neither contains the GPIB ROM.

[MaxFrister]

I gave the meter exterior some attention.  As usual, it was filthy.  The display got brighter after cleaning both sides of the lens.

The top and bottom covers were scratched down to white fiberglass.  I decided to repaint them.  There are worse color matches than the paint shown in the photo.

[maginnovision]

Have any pictures all painted? Curious how it turned out. (I just bought a 1082 with some case issues)

[MaxFrister]

Here's a pic of the machine temporarily put together.

I'm still working on the tant replacement and trying to track down a problem that only occurs when cold.

[MaxFrister]

The “last” problem with the 1081 was interesting.  When cold, for a 1K resistor, it would read ~10M for the first 10 or 12 minutes.  Then it would read ~700K for a few minutes, and then start to read correctly.  I eventually traced it to a bad Fairchild F40014 inverter on the analog board that was part of the auto-ranging circuit.

While waiting for the obsolete IC to arrive, I finished replacing all the tants.  For the digital board, I used an isolation transformer and lifted ground for the soldering and desoldering irons (I don’t own a battery operated one).   I was able to complete the replacement without loosing the calibration constants.  Hint for the next person: C33 is .68uF, not 68uF.  As a 555 timing cap, the wrong one has the expected result.

For cleaning the analog board, I used MG flux cleaner and a soft brush.  I then blotted the residue from the board with a kim wipe.  Normally I would rinse with distilled H2O but I didn’t want to spread flux residue further.  This seemed to work well and did not damage the fragile solder mask.

The next step is to see if I can improve the calibration without a full recal.

[maginnovision]

Glad to see you've got it all figured out. Hopefully the replacement inverter works well. Luckily the Cal ranges can all be done independently. Do you have a decent calibrator? I was very lucky with mine and the DC/AC calibration seem right on. DC matches my 4000 and both of my ltz1000 references. AC matches my fluke 5200 and Prema 5017sc. I do need to clean mine up still.

[MaxFrister]

...Do you have a decent calibrator? ...

Unfortunately, or maybe fortunately, I’m only at the entrance of the metrology rat hole so I don’t have anything that I can use to calibrate the Datron 1081.  All I can hope  to achieve is to adjust it to correspond to low-grade and uncalibrated standards.

Status update:

1. After leaving the machine unpowered for several days, and the weather turning colder, it booted up in an odd state.  Both DC and ohms were unrealistic (negative ohms?).  After several minutes of this I turned it off and back on and it was fine.  I have not been able to replicate this so far.

2.  Because all the ranges are off in the same direction by roughly the same amount I thought I would attempt the “Standard” adjustment that adjusts for drift of the voltage reference.  I bought what I had read was the correct calibration key, and it was not.  I then tried shorting the key contact and I still can’t get it to shift into “cal” mode.  I’ll have to start troubleshooting that one.

3. I’d like to run some long-term logging runs and see how it behaves with respect to drift.  However, I’m leery of wearing out the display.   On many meters, you can turn the display off via GPIB.  I don’t see a way to do that on the 1081.  I’m thinking of adding a physical switch to shut down the +175V power supply.  There is room at the back of the machine for such a switch.   



This is turning into a bit of a slog.  Not a Tin-style epic slog, but a slog nevertheless. 

[maginnovision]

For cal mode turn all the GPIB switches to on, if it doesn't go into cal mode then.  I believe a switch for long term logging may be good, alternatively just pull the connector. If you hook up to GPIB it really shouldn't matter if you need to go in/out of it when you want the display back.

[MaxFrister]

For cal mode turn all the GPIB switches to on...

Nice one!  I never saw that in the manual.  I spent a while tracing the "cal key" signal through the circuit (sort of fun, 74ls logic and all), but it seemed like it was fine.  I was starting to wonder if it was a timing or software problem.  Apparently it was a software problem with the user.  Thanks for that!

[e61_phil]

For cal mode turn all the GPIB switches to on...

Nice one!  I never saw that in the manual.  I spent a while tracing the "cal key" signal through the circuit (sort of fun, 74ls logic and all), but it seemed like it was fine.  I was starting to wonder if it was a timing or software problem.  Apparently it was a software problem with the user.  Thanks for that!

That tooks also a lot of my time a couple of years ago with my 1071
I thought I reported it here, but I didn't..

[MaxFrister]

Here is the meter in its new home near it's cousin, a nixie version of the HP 5340A:



With GPIB disabled, the machine boots into a slightly different mode (not in hold), and allows manual calibration.  I was able to change the "memory" settings and performed the "standard" adjustment.  It helped get the dc voltage closer but still not very accurate.  It did not appear to help ohms which reads consistently -0.3%.

It is a bit much to expect that a machine with unknown provenance, repairs to the power supply and re-cap'ing all boards would be in calibration.

Work plan includes:

1. Read the 1071 manual and try to figure out how to use the various filter, averaging, and math options.
2. Use it for a few weeks and make sure it is stable.
3. Perform what adjustments that I can using my limited equipment.

Unfortunately, my prom programmer (TL866) apparently does not support the eprom chips so I can't make backups of them.

[maginnovision]

This is the most useful stuff.

[MaxFrister]

The 1081 has continued to operate nominally and I started setting up for some stability logging.  I don’t have access to the 1081 user’s manual so I have to guess what the GPIB commands are based upon the similarity to the 1071.  I found a couple of interesting things:

While I have not found a way to turn off the display for logging runs, issuing the K keyboard command has the side effect of blanking the display until the next trigger.  The result is a brief flash upon read and the “rem” remote annuciator always lit.  Almost good enough.

The manual seems to imply that turning on the low-pass filter should result in 7.5 digits but the only way I’ve found the actually get it to report 7.5 digits is to place it in high-res (averaging) mode.  By the way, that mode is apparently “A1” on the 1081.

Since I don’t actually want it averaging for the logging runs, I’m going to experiment with block averaging and see if I can use a block size of 1 and still get 7.5 digits.

The graphs are from 3 short test runs.  Don’t read much/anything into the data.  Both meters and references were starting cold and the reference has not been aged.

The first run shows the Datron in low-res.  The second and third and the Datron in high-res.  It is interesting to note that the shape and even the noise are similar in these.

I’ve added some temperature logging and I’ll setup for a longer logging run.

[maginnovision]

A rolling average of 16 samples should be the default and seems to give good results for me so I would just use Filter and HiRes without changing anything. Once you enter a number it's a block average, or continuous for 0. I haven't had time to get back to mine to setup logging what are you using for that? I was going to just write my own program but I've had very little time lately.

[splin]

The 1081 low res results show quantization steps of 10ppm which doesn't make much sense for a 6 1/2 digit meter.

[EDIT] What were the 34401A settings - AZ, NPLC?

[MaxFrister]

The data is collected by the worlds dumbest python script.  I've used variations in the past.  I can share it with you after I shake out a few bugs.

The "A1" mode may represent "continuous" averaging.  More GPIB experimentation is necessary.

Indeed, I'm scaling the 10V range Y axis by 10E5 instead of 1E5. 

As for the 34401A settings, it is using the default autozero and NPLC.  A better question is what settings should I use?

After an hour or so of logging, I'm getting failures that I think are related to the calibration constants.  It's a bit hard to tell since I'm turning the display off and the GPIB read only reports "!".  Looks like I have more troubleshooting in my future.

[MaxFrister]

This meter has periodically shown odd behavior related to the digital section.  Yesterday it would run logging for 20-60 minutes and then stop with an error.  With GPIB removed and the display restored it showed “fail” sporadically and the then continuously.  The fail message indicates corrupt calibration constants.

With the calibration constants clearly lost, I took the opportunity to pull and clean all of the ics and sockets on the digital board.  I then gave it a quick adjustment (“calibration”) and started the logging back up.  So far, the logging has run for more than 8 hours without errors.

I’m hesitant to say that I fixed the digital problems.  Every time I think I’m finished  with this meter some new problem crops up.  My plan is to continue testing and if everything looks good, restore the adjustment (“calibration”).

The graph shows stability relative to an HP 34401A measuring a LM399-based reference.  The green dots are temperature.  I changed the Datron logging to “A2” which I believe means hi-res with 16 sample averaging.

[maginnovision]

You never replaced the battery?

[MaxFrister]

The battery has been replaced and the failures first showed up while under power when the battery should be mostly irrelevant.

I've also seen bad initial program loads.  I'm guessing that the bus is (was?) noisy.   Some the ics had visible tarnished pins.  I'm hoping that cleaning fixed it.

I continued the logging run overnight and was just about the declare victory until I looked at the graph.   That is one ugly graph.

[maginnovision]

Did you go through and check the DC board and power supplies using the internal adjustments procedures? That's pretty odd that you lost cal while running. After I repaired the analog board on my 1072, the day I got it, I had an issue with the display showing garbage about 60% of the time when starting up. After doing something similar with the IC's(moving them in sockets to help contacts) it got a little better. I don't think I stopped seeing the issue until after I had gone through and replaced the battery, the caps, and did all of the internal adjustments(I think specifically the reset was mis-adjusted on it). I really hope you can get all of this sorted.

Now that I have even more extra parts(but good) let me know if you find you need anything you couldn't find elsewhere. I've got 2 full spare units now(I took the display from one for my 1082, but even that display is fine it just wasn't really nice). When you were looking for a transformer I'd only had 1 spare and couldn't stand to lose it but I'm pretty well set for spares now.

[MaxFrister]

Good ideas.  I've been focused on getting the meter working and so I have not spent much time calibrating.  I've done most of the digital board adjustments but I skipped over reset signal.  I'll go back and check that one.

I think it is possible that the calibration constants were okay but not being read correctly over the bus.   Of course once I pulled the ics that was no longer true.

I have no idea what to make of the noise on the long logging run.  It correlates with the HP meter so it may be external to the Datron.  There seem to me to be 3 possibilities: source reference noise, AC power problems, or external EMI. 

As for EMI, I am in range of a couple of military installations and they have been known to take out the local cable TV.  As the US is prepping for war with China, Mexico, North Korea, Canada, Iran, and/or Australia who knows what they are doing this week.

I'll setup both meters to read a shorted voltage tonight.  That will take the voltage reference out of the equation and disconnect the 2 machines from each other.

[MaxFrister]

The overnight logging of shorted meter inputs is much prettier.    There is only minor deviation with temperature and no 50ppm excursions.   I still don’t know what that happened in the previous run.  I’ll set it up to run that experiment again.

For the Datron, the settings are “F3R4T2C1A3” with a side of “K1L5”.  That translates to DCV, 10V range, normal triggering, low-pass filter on, and block averaging with the block size set explicitly to 1.

The temperature sensor is programmed to behave like a GPIB instrument so it is trivial to use here.   The sensor is a Dallas 1-wire sensor.  I have a fancier BME280 sensor but I can’t seem to find it.  Somehow I never lose the boat-anchor instruments.

I’ve attached the logging script.  It is about as simple and dumb as can be. 

[MaxFrister]

After years of working nearly flawlessly I pulled it for calibration and some upgrades.

• Every once in a while it fails to boot properly.  It is very intermittent and only happens in cold weather so it has been difficult to find.
• The cal key is missing.  While it is pretty easy to open up and jumper, I'd like to fix it.
• The external trigger connector is missing
• Maybe add some of the missing options

The options on the Datron 1081 were:
Option 10:     True RMS Converter
Option 20:     Four-wire Resistance conveter
Option 40:     Selectable Rear/Reference Input
Option 50:     IEEE 488/GPIB
Option 52:     Remote trigger
Option 70:     Analog Output
Option 80:     115V 60Hz Line Operation
Option 81:     115V 50Hz Line Operation
Option 82:     115V 400Hz Line Operation
[No option listed for 230V Line, presumably no option meant that]
Option 90:     Rack mounting.

My unit had 10, 20, 50, 80.  I'm going to try to add 52 and 40.  I'd add 70 but I can't think of any use for it.

There were also some accessories:

PRT 100    Platinum resistance thermometer Probe(100R)
CS1:          Current Shunt Set
HVP:         High voltage probe
1501:        Deluxe Lead Kit
Someone is selling something that might be CS1 on ebay for crazy money.  If anyone has these, it would great to see pictures.