Datron 1281 Repair/Maintenance

[chinapp]

Summary: you can use a dual coil latching relay in a 1271/81. You don't have to run down the single-coil version, which appears to have been gone for decades.

p/n S4EB-L2-6V

The 5V version should also work.


The 1281 and 1271 use a Panasonic latching relay that is impossible to find. It is a 6V single coil latching relay. The single-coil design requires that you reverse the coil current direction to switch the relay, which drives circuit complexity.

This relay is unobtainable. However, it turns out that the dual coil version will work just as well. I found a clear case version at our local junk store. Here's a little video

https://photos.app.goo.gl/oHU1koYftAFRfNT49

The reversible drive current in the Datron circuit can effectively flip the dual coil relay.

Essentially, the S4-L-6V and the S4EB-L2-6V are interchangeable. This component is much more accessible than previously thought.

The  S4EB-L2-6V have two coils, and footprint is different, how to interchange ? plz give more details,thx

[Kleinstein]

With many (but not all) 2 coil latiching relays on can wire the 2 coil in series / parallel to use it like a single coil version. If the current in both coils helps the sereis connection could be used with about the same voltage as the 2 coils in parallel.
As not all (though most) relays work this way it would be a good idea to do an upfront test.

[martinr33]

On the S4EB-L2-6V, either coil is capable of switching the relay.

The second coil sits across two pins that are unused in the single coil version (at the other end of the relay). So the pinouts are compatible when using a dual-coil relay in a single coil footprint.

If you think about it, the design has to be symmetrical for a single coil to work. Adding the second coil makes the drive circuits simpler, but does not change the mechanics of the relay.

Also, you cannot use a single coil version in a dual coil footprint, but that is not really an issue as the dual coil devices are available.

[leighcorrigall]

I have been considering including a switch to interrupt the VFDs on my Datron 1281 without cutting into the existing enclosure. This would avoid disassembly problems of the back panel and maintain the original look while preserving the display.

Options for replacing an existing component:
1) Keylock calibration switch (S2)
2) Mains switch (S3)

Both options would require a three-position switch.

On review of the schematics, the simplest approach for interrupting the power to the VFDs is at the fuses (F504, F505) as recommended by Micke. A modified schematic has been attached.

I think I am going to try and replace the mains power switch (S3) because it is already associated with the power supplies. Does anyone know the model number of the switch or any other details?

[Kleinstein]

Another option for a switch without drilling a hole would be a reed switch and a magnet from the outside.

The main switch is a bit tricky for accidentially interrupting power, when only wanting to turn on/off the display. In addition the isolation between different contacts may be a problem.

For the cal switch the this would increase the risk for accidentally enabeling calibration - not sure how important that protection is.

[TheSteve]

How about repurposing some contacts on the DB15? There are for certain unused pins that go from the DtoA IC socket to the DB15 if you don't have option 70 installed. It would be pretty easy to have an external DB15 male with a shell and toggle switch that just plugs in to enable the VFD.

[rigrunner]

How about repurposing some contacts on the DB15? There are for certain unused pins that go from the DtoA IC socket to the DB15 if you don't have option 70 installed. It would be pretty easy to have an external DB15 male with a shell and toggle switch that just plugs in to enable the VFD.

You would need to double check if the pin you are thinking of using is free. SK8 Pin 10 for example, is marked as unused on the schematic, but on the digital boards I have here it is wired via 100K to TL201.

[rk767]

Well, members, I too have become a Datron 1281 owner. My unit comes in fairy decent physical condition. Initially it had the "MEAS, CORRNS, INVALID 2021" error. After clearing the NVRAM I was able to do a quick and dirty ext cal with a 10V source. Power cycling the unit about 10 times I was not able to reproduce the error. This unit originally was supposed to be in working condition, however, the place where it was was not able to repair the "2021 issue", because of that I got a nice discount  Not sure if something causes the NVRAM to corrupt itself or a mishap when the battery was upgraded. Unit passes full self-test, however, fails 2008 and 2058 during ADC calibration. My assumption would be due to not having went through a proper calibration. More details later, first thing is to replace the capacitors as they all seem original.

[leighcorrigall]

... More details later, first thing is to replace the capacitors as they all seem original.

Congratulations.

I have a partial list of electrolytics (including dimensions) posted above in a spreadsheet.

[Le_Bassiste]

I have been considering including a switch to interrupt the VFDs on my Datron 1281 without cutting into the existing enclosure. This would avoid disassembly problems of the back panel and maintain the original look while preserving the display.

Options for replacing an existing component:
1) Keylock calibration switch (S2)
2) Mains switch (S3)

i'd rather "sense" one of the front push buttons, say, the LOCAL key, and then toggle the VFD supply via some MOSFET driven by a photovoltaic optocoupler and some sort of flip-flop (or, low pin-count PIC8-someting for that matter).

[leighcorrigall]

Self Cal password ?

It's user settable.

Move the digital board jumper to the cal password off position and see if that allows access?

This is the off position

Hi Rigrunner,

Do you happen to know the functionality of the other jumpers on the digital board? This would be beneficial to me.

Regards.

[rigrunner]

TL101 Pulls the CPU Bus ERRor pin high in position B, preventing anything signalling bus error.

TL201 Doesn't have any obvious behaviour. Its state is passed through a 100k resistor to pin 10 SK8
TL202 Again, no obvious behaviour change with this jumper. Pins aren't fitted to the PCB in later revisions.
TL203 Option 70 enable/disable. Position A is disable.
TL204 Callibration menu password required. Position A is disable.
TL205 pulls the CPU DTACK low in position B. Debugging only, CPU will trap and crash pretty quickly in position B.

[leighcorrigall]

I experienced a strange, yet subtle problem, while adjusting ranges on the Datron when I had access to a Fluke 5700A 5720A series calibrator last month. The problem was unknown to me until the adjustment/calibration attempt. The instrument would accept the new adjustment then, when I wasn't looking, it would revert back to its previous value or something similar (REVISED SENTENCE). You can imagine how frustrating this was after spending several hours meticulously adjusting each range and documenting the results to discover that the values were completely different after verification. Sometimes the value would even hold after a reboot! At first, I thought it might have been how I exited the calibration menu where it asks for the next calibration date (options being ENTER or QUIT). My hypothesis was that the digital board had some sort of malfunction in storing calibration constants. Although potentially unrelated, I should have taken more care in diagnosing the scrambled screen issue described in the initial post. I thought that the board might have been shorted because the card slot was damaged and the protective sheet under the digital board was a piece of tape and some card stock. Well, that was only part of the problem.

In a retaliatory manner , I decided to replace all of the socketed components on the digital board to address the problem (what a beautiful design feature of the older Datron models). This is because I was unable to repeat the issue with a simple cause-and-effect process and I figured it wasn't that expensive to replace these components anyways. A Hong Kong-based eBay seller sold me nearly all of the obsolete parts required to do this swap (40 USD total with shipping). The rest were purchased new from DigiKey. Attached is a list of the socketed components that were replaced. I compared my older Datron to TheSteve's Wavetek model from what I could piece together from reference pictures.

Part of the replacement process involved programming 27C010, GAL16V8B, and GAL20V8B chips with the TL866II Plus and XGPro. I bought the TL866II Plus specifically for the GALs because my other universal programmer isn't that universal after all. Attached is a compressed file I used for the EPROMS and GALs organized into the socket designations. SHA256: 835656a4939d1b52c8f5afac1887858536df383ed417720487bb1cf1da4a9730 (checksum)

Another feature of the XGPro software package is that it can test some of the components through Device -> Logic IC Test, which can be rather helpful for isolating the program. I was able to rule out the 74245, 74374, 74165, 7404, 74260, and 74165 chips this way. The ROMs can be simply reprogrammed and then verified that they match a predetermined data dump.

At this point, I have adjusted my Datron to a well-determined Fluke 732A 10 V output and have not noticed a change in value since yesterday. The measurement error range is maybe +/- 0.2 ppm, at most, and is likely caused by thermal drafts. My next plan is to perform a fully automated test that will include a front input short (i.e., zero volts) and a reverse polarity test of a reliable voltage standard using the rear inputs (ChA and ChB) over several weeks at 23 °C. 

[TiN]

It was 5720A, not 5700A :-). I thought it was confirmed that problem saving calibration constants was related to jumper in incorrect position on digital board, no? Also I don't think constant values were randomly changing, but rather reset to stored NVRAM values once unit was reset/enter/exit calibration procedures. Since we did that multiple times trying to "troubleshoot" non-storing values, it may "felt" like random act of ppm pixies..

Also I'd like to use this chance to reiterate that no instrument can be deemed properly operational after even minor repairs until it is actually went thru full "as-received, adjusted, as-returned" calibration and service adjustment procedures. Imagine sending this 8.5 digit meter to Fluke, paying $2000 for calibration, and getting it back with all data lost  .

[leighcorrigall]

It was 5720A, not 5700A :-). I thought it was confirmed that problem saving calibration constants was related to jumper in incorrect position on digital board, no? Also I don't think constant values were randomly changing, but rather reset to stored NVRAM values once unit was reset/enter/exit calibration procedures. Since we did that multiple times trying to "troubleshoot" non-storing values, it may "felt" like random act of ppm pixies..

Also I'd like to use this chance to reiterate that no instrument can be deemed properly operational after even minor repairs until it is actually went thru full "as-received, adjusted, as-returned" calibration and service adjustment procedures. Imagine sending this 8.5 digit meter to Fluke, paying $2000 for calibration, and getting it back with all data lost  .

I forgot what option the 5700A was from memory, so I didn't get too specific.

Option 70 may have been related to my Datron symptoms, but perhaps not. It might have been coincidental that removing U205 and disabling the option jumper improved the situation. The instrument might have been impacted by the ~1400 km road trip to and from NJBH. After more 'long-term' observations, the measurements were still not consistent with Option 70 removed. The entire chip swap has shown the greatest amount of consistency so far. I will need your help next week to set up a long-term performance study to determine if the bugs have been squashed  

By 'random', I meant that the values were not what was expected, given a known input. I agree, the instrument likely reverted back to a previous calibration constant once it detected a fault somewhere. What was troubling was that Datron gave no warning about its decision, which caused a great amount of confusion.

I sent my Advantest in for calibration and found out the hard way that the instrument doesn't meet specifications and I definitely want to avoid that outcome for Datron. Lessons learned.

[TheSteve]

I feel the need to ask which jumper is used to enable option 70?
Did you try switching back to the stock nvram?

[leighcorrigall]

I feel the need to ask which jumper is used to enable option 70?
Did you try switching back to the stock nvram?

Rigrunner has determined the purpose of the jumpers:

...

TL203 Option 70 enable/disable. Position A is disable.

...

The storage has remained FRAM since the update. So far, it does not seem to be the issue because the memory can be overwritten. A more complete picture will come about after I perform a long-term drift study on the 10 V range.

[TheSteve]

It sounds like you've already swapped the main parts but scoping the write enable lines for the nvram may be useful if you continue to have issues.

[leighcorrigall]

...

My hypothesis was that the digital board had some sort of malfunction in storing calibration constants. Although potentially unrelated, I should have taken more care in diagnosing the scrambled screen issue described in the initial post. I thought that the board might have been shorted because the card slot was damaged and the protective sheet under the digital board was a piece of tape and some card stock. Well, that was only part of the problem.

In a retaliatory manner , I decided to replace all of the socketed components on the digital board to address the problem (what a beautiful design feature of the older Datron models). This is because I was unable to repeat the issue with a simple cause-and-effect process and I figured it wasn't that expensive to replace these components anyways.

...

My next plan is to perform a fully automated test that will include a front input short (i.e., zero volts) and a reverse polarity test of a reliable voltage standard using the rear inputs (ChA and ChB) over several weeks at 23 °C. 

Studies
1) 2022 06 11 to 2022 06 22 (11 days) Experiment - Before Digital Board Chip Replacement
- ChA connected to Fluke 732A-2 (+10.0002207 V)
- ChB connected to Fluke 732A-1 (+10.0001320 V)

Note the 6 skew data points on the "stability evaluation" figure.

2) 2022 08 26 to 2022 09 28 (33 days) Experiment - After Digital Board Chip Replacement
- ChA connected to Fluke 732A-1 (+10.0001320 V)
- ChB connected to Fluke 732A-1 (-10.0001320 V)

No skew observations for over a month of data logging.
The spread of the data is greater because there are triple the amount of days in this study.
Datron 1281 temperature coefficient of voltage between 19.5 and 24.2 C °C: α = -0.342 ppm/°C, β = 0.256 ppm/°C^2

Note:
These rough parameters were computed by assuming the Fluke 732A source was independent of temperature.
The temperature sensor used could have an offset up to a few °C relative to other sensors present during the data collection.
I am assuming that these voltage coefficients are associated more with the 13 to 18 °C specifications.

Conclusion
The Datron 1281 does not produce skew data anymore after replacing the digital board chips. I have also not witnessed a scrambled screen anymore. I suspect the PALs were going bad and possibly some other chips. Note that the FRAM modification presented above was left alone and is not suspected to be the problem. I also removed Option 70 and placed a wire jumper like other Wavetek models.

My only concern now with the Datron 1281 is that it might have a relatively large temperature dependence compared to the specifications. If there are members with quantitative results to compare with, I would appreciate it if they provided feedback.

The next task will be to completely replace all tantalum capacitors on each board.

Thanks!

[Micke]

I did an experiment a while ago to investigate the TC of my Wavetek 1281 vs a borrowed Keysight 34470A.

Conditions:
1) Measuring 10VDC from a LTZ1000 placed in temperature chamber set to 34C (only able to heat) in a separate room from the DMM´s, shielded RG316 cable from LTZ1000 to DMM´s.
2) Both DMM in the same room
3) The temperature logged is the room with the DMM´s
4) Started a 2000W heating fan to heat up the room with DMM´s.
5) 1281 in 8.5 digit mode, WinGPIB samples temperature and 1281 every 25s. Data from 34470A saved to USB memory, tried to match its trigger time to match 1281. Did not succeed to let WinGPIB to log both instruments.

My setup and measurements might have flaws, but my results were TC of +0,2ppm/C for 1281 and -0,6ppm/C for 34470A.
The 1281 is more long time stable, did a 139h test, the 1281 tracking ambient temperature perfectly, while 34470A drifting slightly upwards during time.

[leighcorrigall]

While stuck at home sick, I took the time to examine the performance of my Datron 1281. Members such as RigRunner and MartinR33 have suspected that my Datron troubles are not over and I want to find the root cause of this mystery problem.

If you compare the  2022 08 26 - Datron 1281 closed case example.png (380.65 kB. 2970x1980 - viewed 119 times.) figure with  2022 10 25 - Datron 1281 open case study.png (579.25 kB. 2979x1989 - viewed 102 times.), there doesn't seem like a whole lot of difference between when the Datron 1281 top cover has been removed or when it is closed. This to me is suspicious.

I then decided to compare the lesser Advantest R6581T with the Datron 1281 using a Fluke 732A-1 (+10 V) as a common voltage reference  2022 10 27 - Datron 1281 and Advantest R6581T performance comparison.png (590.94 kB. 3116x1989 - viewed 100 times.). Clearly, the Advantest R6581T has won the stability award which tells me that there are still some serious problems with my Datron! 

The plan moving forward:
1) Measure the -7.2 V references at TP405 (M408) and TP406 (M409) with TP102 ('MECCA') as common. This can be accomplished with the Advantest R6581T now that I know it is reasonably stable and temperature insensitive.
2) Measure the buffered voltage references at TP403 and TP404 with TP102 as common.
3) Measure the -35 V and +35 V supply voltages as recommended by RigRunner. He suspects that the designer balanced these supplies with R904 and R906 to enhance the performance of the instrument. The resistors are both labelled "FSV" which may be an abbreviation for 'factory selected value'. The 35 V supplies influence many critical areas of the DCV board. None of the other inside guard regulators have been trimmed in this way. My DCV board shows the most amount of abuse in this area from a past owner. Perhaps this could be an area to work on first.

All future plans will start measuring after the instrument has reached a steady-state condition with the cover off and after the cover has been reinstalled. See  2022 10 26 - Datron 1281 open to closed case reversal error.png (565.82 kB. 3121x1989 - viewed 95 times.) for an example of when I placed the top cover on the Datron 1281 while continuously measuring a Fluke 732A. These conditions may provide insight into how voltage measurements behave as a function of temperature. I wish I could do more than one measurement at a time with the Advantest R581T. This may take a while.

[alm]

I wish I could do more than one measurement at a time with the Advantest R581T. This may take a while.

I realize this is yet another project to build and verify, but one way around that would be to build a low-thermal-EMF scanner like this to allow a single DMM to measure multiple signals. I would imagine it's really useful to see relations between the voltages. An alternative, if you have lower resolution meters that can log, could be if you have a stable voltage source of nearly the same voltage, like a divided 10V from 732A, and have a lower-resolution meter measure the difference, kind of like a differential voltmeter. I've used this trick in the past to use 6.5 and 5.5 digit meters to monitor power rail stability.

[Kleinstein]

Having quite some drift in the trun over error is suspicious. This could be a linearity problem, but also an offset problem. Different from the R6581 meter the 1281 uses a kind of chopper stabilized amplifier and am ADC that is made to have relatively low drift even without an auto zero mode.
Ideally the chopper part has no drift, but at the µV level and below, there is a chance for some problems.
So I think a check of the offset drift / stability would be a good idea. The 100 mV range would show more sensitivity to amplifer offset drift and less from the ADC part. So it could help to localize the offset drift.

For the longer time tests with the 10 V on the drift one chould also check on how much of the drift is corrected with a ACAL call.

I don't think the +-35 V supply should have very much influence it mainly for the supply to less critical amplifiers. A point were is may have an effect is with some JFET switching, but don't think gate leakage there would be that critical. The much more critical supply would be the +-15 V supply (it is used e.g. for quite some CMOS switches and the supply can effect charge injection and on resistance).
There are also 3 sets of resitors that effect the gain factor for the 10 V range:
1) the gain setting resitors, that should be acting as a divider by 2 for the 10(20) V range. I somewhat doubt they would automatically switch between an extra range with 10 V max and the 20 V max range, as this would cause discontinuity problems near 10 V.
2) The 7 to 10 V gain for the ADC reference. (not very much effect, but still possible error source).
3) The resitors at the ADC input

A temperature effect that can increase from a partial defect are leakage currents. A part could have developed higher than normal leakage current and such leakage is often quite temperature sensitve. However leakage in most cases would give more like an offset and not an effect on the gain.

An indirect effect can be the fan speed / air flow.

The tests so far still run with a relatively small variation in temperature. A larger temperature swing could show a clearer effect.

[leighcorrigall]

...

The plan moving forward:
1) Measure the -7.2 V references at TP405 (M408) and TP406 (M409) with TP102 TP101 ('MECCA') as common. This can be accomplished with the Advantest R6581T now that I know it is reasonably stable and temperature insensitive.

...

 2022 10 27 - Advantest R6581T stability according to Fluke 732A-1 (+10 V).png (761.76 kB. 2979x1989 - viewed 101 times.) is an example of the Advantest R6581T [AZERO, 100 NPLC] measuring a Fluke 732A-1 (+10 V) for 33 hours taken from the previous post https://www.eevblog.com/forum/metrology/datron-1281-repairmaintenance/msg4489450/#msg4489450. During that time, the ambient temperature decreased from 24.2 to 22.1 °C (ΔT = -2.1 °C). The measurement error changed from 0.13 to -0.38 ppm (Δε = -0.51 ppm). Perhaps the R6581T has a temperature dependency of 0.25 ppm/°C. It would also be beneficial to get a sense of the measurement dependency while the temperature increases as well. Another indication of performance would be measuring the Advantest while the FRONT is shorted. If only I had a low thermal scanner or a secondary Datron. 

Using  2022 10 27 - Advantest R6581T stability according to Fluke 732A-1 (+10 V).png (761.76 kB. 2979x1989 - viewed 101 times.) as a control, it is now possible to roughly evaluate the Datron 1281 (DCV, FRNT: short) voltage references (M408 at TP405 and M409 at TP406):

•  2022 10 28 - Datron 1281 M408 stability according to Advantest R6581T.png (680.86 kB. 2979x1980 - viewed 118 times.) is an example of the Advantest [AZERO, 100 NPLC] measuring TP405 for 12 hours. During this time, the ambient temperature increased from 21.8 to 22.6 °C (ΔT = +0.8 °C). The measurement error increased from 0.5 to 0.9 ppm (Δε = +0.4 ppm). Perhaps the combined temperature dependency of the Datron (M408 at TP405) and the Advantest is approximately +0.5 ppm/°C.  2022 10 28 - Datron 1281 M408 stability according to Advantest R6581T.png (680.86 kB. 2979x1980 - viewed 118 times.) and  2022 10 28 - Datron 1281 FRNT short.png (590.37 kB. 2979x1989 - viewed 103 times.) were measured simultaneously.
•  2022 10 30 - Datron 1281 M409 stability according to Advantest R6581T.png (792 kB. 2979x1989 - viewed 92 times.) is an example of the Advantest [AZERO, 100 NPLC] measuring TP406 for 30 hours. During this time, the ambient temperature increased from 22.5 to 23 °C (ΔT = +0.5 °C). The measurement error increased from -0.13 to 0.25 ppm (Δε = +0.38 ppm). Perhaps the combined temperature dependency of the Datron (M409 at TP406) and the Advantest is approximately +0.75 ppm/°C.  2022 10 30 - Datron 1281 M409 stability according to Advantest R6581T.png (792 kB. 2979x1989 - viewed 92 times.) and  2022 10 30 - Datron 1281 FRNT short.png (607.03 kB. 2979x1989 - viewed 94 times.) were measured simultaneously.
  

Comments:

• I believe that most of the temperature dependency must be attributed to the Advantest. The time required for the measurement to drift as a function of temperature is small according to the figures, which suggests that the Advantest is responsible. Another way of interpreting this hypothesis is to think about the Datron voltage references as being surrounded by a blanket of hot air without circulation. The Advantest uses a fan and has plenty of air vents.
• The M408 is connected to the reference inverter circuit, which would annihilate most of the temperature-dependent error when compared to the M409 buffered reference source. See sheet 4 of the D.C. Assembly Schematic. At this stage, I do not suspect the Datron voltage references to be responsible for the significant temperature dependencies demonstrated in the previous post.
• https://www.eevblog.com/forum/metrology/datron-1281-repairmaintenance/msg4489450/#msg4489450 shows that the Datron deviated by +3 ppm when a temperature change of -1.5 °C occurred over 29 hours (i.e., 2 ppm/°C) while measuring the Fluke 732A-1 (+10 V). I would expect that if the Datron voltage references were the immediate problem, the Advantest measurements would have detected something greater.

As a side note, measurements made by the Advantest at both TP405 and TP406 changed according to the measurement made by the Datron (e.,g. cycling each channel [FRNT: short, ChA: +10 V, ChB: -10 V] causes slightly different results (~= 1 ppm) measured by the Advantest).  2022 10 28 - Datron 1281 M408 stability performace while cycling channels.png (1476.52 kB. 2979x1989 - viewed 100 times.) is an example. I suspect this effect is normal.

EDIT: Note that the voltage range is set to 10 V for all studies, including the FRNT short.

[Kleinstein]

I agree that the temperature effect shown in the last plots seems to be mainly from the R6581T. There are considerably more resistors inside the DMM than for the relative simple 7 V to 10 V step in the Datron meter. The capacitive inverter should be pretty stable and not cause much error.

There is some offset effect at the DA1281 from temperature changes (looks a bit like the dT/dt as the main factor), but this is relatively small (<1 µV range), though still not very good. Part if this could already be the short / cables / terminals.

The effect of the measurement performed by the Datron meter on it internal measured reference could be an effect of the layout / placement of the ground test point. If this would be a true effect on the references it would be bad for the linearity.