Datron 1281 Repair/Maintenance

[leighcorrigall]

...

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 of this could already be the short / cables / terminals.

...

The Datron 1281 FRNT short is simply a copper wire compressed on the binding posts in the following order I+, HI, LO, and I- (i.e., an upward C-shape).

As for Ch A and B, they are made of shielded 22 AWG silver-plated copper-stranded wire with PTFE insulation in twisted pairs. The connector pins are gold platted and soldered with regular Pb-Sn wire. They are similar to the xDevs article https://xdevs.com/fix/d1281/#diyrear. The spades are pure copper crimped, not soldered.

Attached is a series of figures that have been taken from https://www.eevblog.com/forum/metrology/datron-1281-repairmaintenance/msg4448029/#msg4448029 that show the Datron cycling through all three channels. Note that all measurements have a 25-moving average filter applied for legibility. There is a strong correlation between temperature, reversal error, and short offset.

[leighcorrigall]

I had a change of heart and decided to go after the linear regulators L7815CV (M901) and L7915CV (M902) again. They seemed too hot, even when the thermal paste was replaced and when the instrument was left to stabilize with the cover off. These components are still available and cheap so it was easy. I decided to include thermal pads between the heat sink and the PCB to dissipate the heat into the board more efficiently.

Under the recommendation of RigRunner, I disabled the option boards (see highlighted areas in blue of the attached DCV board layout). The purpose is to isolate the Datron from the rest of the analog boards and to reduce heat caused by the ±15 V regulators. The boards can be disabled by changing the jumpers and disconnecting the ribbon cables as follows:

AC converter board (Option 10): LK704 and PL105
Ω converter board (Option 20): LK702 and PL109
I converter board (Option 30): LK703 and PL107

Afterwards, I calibrated the DCV mode (-FS, zero, +FS) for each range (100mV to 1000 V) with the Fluke 5440B. The Datron seems to have a recurring issue with the ADC internal calibration (errors: 2008 and 2054/2056/2058). These errors seem to be persistent now. I went as far as clearing the NVRAM in an attempt to remove these ADC errors. Fortunately, RigRunner is rescuing me again. He suggests that C505 could be responsible. See sheet 5 of the DC assembly schematics that shows the 100nF capacitor acting as a bridge from the integrator section to the null detector with R545 (brown red black yellow brown - 1.2M 1%), R546 (brown grey red green brown - 18.2M 1%), and R547 (brown black black yellow brown - 1M 1%). A previous post already exists on this matter: https://www.eevblog.com/forum/metrology/datron-1281-repairmaintenance/msg4123972/#msg4123972

The DCV calibration constants have been stored and seem fairly stable. I quickly tested the reversal error and they are negligible even after a slight temperature change. I will know more once Datron 1281 passes the ADC, INT SRC CAL, and the FULL TEST LOOP. Then the real test will begin as a stability test over a few days. Eventually, I will get back to measuring the buffered voltage references with the Advantest while the Datron measures the Fluke 732A-1 (+10 V) reference.

[leighcorrigall]

...

--> Front terminal release: A rear button retracts the front terminal of the instrument. The terminal block is spring-loaded so that the posts extend outward after a second button press.

...

Members have asked me to inspect the front posts of my Datron 1281 unit.

There are some interesting features to point out:
- Guard and Ω Guard posts are gold plated
- I+, I-, Hi, and Lo are of a copper alloy (perhaps tellurium-copper, C145)
- The connections are made with a flexible PCB with gold contacts
- The post-assembly can be retracted for safe handling as discussed before

EDIT: After cleaning, the difference in post coatings could actually be that the guards show fewer signs of wear. Now it seems clear that bare copper was deliberate.

[Kleinstein]

Retracting the therminal block may be for more than safe handling: it also gives some shielding against air flow and thus temperature fluctuations - however at a possibly different mean temperature.

It makes sense to have gold on the guard terminals: those are not very critical in respect to thermal EMF, so not problem with gold there and more normal gold coated brass can be used. It depends on how the switching for the resistance function is, if the I+- terminals are also used for small voltages (e.g. for 4 wire ohm).

[leighcorrigall]

...

The Datron seems to have a recurring issue with the ADC internal calibration (errors: 2008 and 2054/2056/2058). These errors seem to be persistent now. I went as far as clearing the NVRAM in an attempt to remove these ADC errors. Fortunately, RigRunner is rescuing me again. He suggests that C505 could be responsible. See sheet 5 of the DC assembly schematics that shows the 100nF capacitor acting as a bridge from the integrator section to the null detector with R545 (brown red black yellow brown - 1.2M 1%), R546 (brown grey red green brown - 18.2M 1%), and R547 (brown black black yellow brown - 1M 1%). A previous post already exists on this matter: https://www.eevblog.com/forum/metrology/datron-1281-repairmaintenance/msg4123972/#msg4123972

...

After a frustrating experience with a UPS delivery, I got the replacement parts for the ADC. C505 caused the ADC CAL errors (2008 and 2058) as indicated by RigRunner in previous posts. Apparently, the dielectric characteristics of the 100 nF capacitor are very important for ADC performance. The polarity (orientation) is also important as indicated by the layout schematic.

I tested the following capacitors:
1) Nichicon PWS1633100KG FAILED ADC CAL in both orientations
2) Cornell Dubilier 930C2P1K-F FAILED ADC CAL in both orientations
3) Cornell Dubilier 730P104X9250 PASSED ADC CAL in the orientation pictured below

These results are not to say that the 730P104X9250 will always work or is the only candidate. Unfortunately, this capacitor has the lowest quality construction out of all the candidates tested. Sometimes it is about functionality I suppose. Since I purchased two additional PWS1633100KG capacitors, I may try them out later after I have resolved the thermal dependency problem with the DC voltage measurements.

In the process of waiting for a UPS order that was delivered damaged, I refurbished the following ADC components:
- rebuilt the entire read-rate board and underneath resistors
- updated R545, R546, and R547 assembly to prevent debris and vibration (likely overkill)
- replaced M503 through M505 and surrounding resistors
- replaced M507 support capacitors
- replaced M506, M508, and M511
- cleaned the pins of M509 with grit paper to ensure proper contact

[Kleinstein]

If you have found a working capacitor, there is no real need to check if another type also works.
The point where C505 and R545-R547 meet is the integrator input and thus a slightly critical node. The isolation sleve at the resistors is kind of parallel to the capacitors and could also have an effect.

Ideally, after changing C505 the trimmer R544 may want a new adjustment. The part around R544 is some circuit to compensate dielectric absorption (DA) effects of C505 to avoid INL errors.
The other capacitors may just need a different setting of C505. Not sure if there are good instruction on how to adjust R544 - expect this to be a bit tricky. Even if not intending to adjust R544 this procedure could give a hint if the setting is about right or maybe still off quite a bit.
I would expect the largest DA related nonlinearity for small votlages, like +-50 mV for the 10 V input range.  So it may be a good idea to test the linearity in this range. With the rather small voltages this looks feasable even without very special instruments.

I would be careful with replacing the CMOS switches. These can effect the ADC performance and new chips may very well use a slightly difference process and new may not be better in all aspects. As the switches are at the reference side I would not expect that much effect, but why take a risk ?

[shenxiaoming]

My 1281 AC board
CLA5532(M412) on is missing, it seems to be customized! Is it possible to DIY one?
?

[Kleinstein]

The chip should be some custom logic. So in principle it could work to replace it with a CPLD of some kind. I don't expect a super complicated content, though it looks mainly like a digital FLL / PLL.
It would still be quite some effort and CPLDs with 5 V outputs are rare by now.  Besides the FLL there seem to be a few extra ouput for things like error, maybe frequency range.
A frist point would be understanding the circuit part (much of the page in the plan). There could be the alternative to replace the complete FLL part with something more modern.

[chinapp]

If you have found a working capacitor, there is no real need to check if another type also works.
The point where C505 and R545-R547 meet is the integrator input and thus a slightly critical node. The isolation sleve at the resistors is kind of parallel to the capacitors and could also have an effect.

Ideally, after changing C505 the trimmer R544 may want a new adjustment. The part around R544 is some circuit to compensate dielectric absorption (DA) effects of C505 to avoid INL errors.
The other capacitors may just need a different setting of C505. Not sure if there are good instruction on how to adjust R544 - expect this to be a bit tricky. Even if not intending to adjust R544 this procedure could give a hint if the setting is about right or maybe still off quite a bit.
I would expect the largest DA related nonlinearity for small votlages, like +-50 mV for the 10 V input range.  So it may be a good idea to test the linearity in this range. With the rather small voltages this looks feasable even without very special instruments.

I would be careful with replacing the CMOS switches. These can effect the ADC performance and new chips may very well use a slightly difference process and new may not be better in all aspects. As the switches are at the reference side I would not expect that much effect, but why take a risk ?

yes,anyone can find the factory repair manual of the 1281 ?maybe fluke has the detail files

[Arhigos]

My 1281 AC board
CLA5532(M412) on is missing, it seems to be customized! Is it possible to DIY one?
?

I think so! I have old 8508 dmm and it has datron AC board fitted with a special fpga-adapter instead of your chip.

[leighcorrigall]

My 1281 AC board
CLA5532(M412) on is missing, it seems to be customized! Is it possible to DIY one?
?

I think so! I have old 8508 dmm and it has datron AC board fitted with a special fpga-adapter instead of your chip.

Thanks for sharing!

It would be great if you could provide more details so that others may benefit 🙂

[chinapp]

If you have found a working capacitor, there is no real need to check if another type also works.
The point where C505 and R545-R547 meet is the integrator input and thus a slightly critical node. The isolation sleve at the resistors is kind of parallel to the capacitors and could also have an effect.

Ideally, after changing C505 the trimmer R544 may want a new adjustment. The part around R544 is some circuit to compensate dielectric absorption (DA) effects of C505 to avoid INL errors.
The other capacitors may just need a different setting of C505. Not sure if there are good instruction on how to adjust R544 - expect this to be a bit tricky. Even if not intending to adjust R544 this procedure could give a hint if the setting is about right or maybe still off quite a bit.
I would expect the largest DA related nonlinearity for small votlages, like +-50 mV for the 10 V input range.  So it may be a good idea to test the linearity in this range. With the rather small voltages this looks feasable even without very special instruments.

I would be careful with replacing the CMOS switches. These can effect the ADC performance and new chips may very well use a slightly difference process and new may not be better in all aspects. As the switches are at the reference side I would not expect that much effect, but why take a risk ?

let's go to find something on this C505  .....

[leighcorrigall]

If you have found a working capacitor, there is no real need to check if another type also works.
The point where C505 and R545-R547 meet is the integrator input and thus a slightly critical node. The isolation sleve at the resistors is kind of parallel to the capacitors and could also have an effect.

Ideally, after changing C505 the trimmer R544 may want a new adjustment. The part around R544 is some circuit to compensate dielectric absorption (DA) effects of C505 to avoid INL errors.
The other capacitors may just need a different setting of C505. Not sure if there are good instruction on how to adjust R544 - expect this to be a bit tricky. Even if not intending to adjust R544 this procedure could give a hint if the setting is about right or maybe still off quite a bit.
I would expect the largest DA related nonlinearity for small votlages, like +-50 mV for the 10 V input range.  So it may be a good idea to test the linearity in this range. With the rather small voltages this looks feasable even without very special instruments.

I would be careful with replacing the CMOS switches. These can effect the ADC performance and new chips may very well use a slightly difference process and new may not be better in all aspects. As the switches are at the reference side I would not expect that much effect, but why take a risk ?

let's go to find something on this C505  .....

I am not sure I understand what you mean. We have already examined some of the replacement candidates:

https://www.eevblog.com/forum/metrology/datron-1281-repairmaintenance/msg4537493/#msg4537493

This is a factory-selected part. You can spend some time swapping out capacitors until one passes the self-test. If you do this enough and measure all of the capacitor properties, you should be able to identify the ideal properties for the capacitor. I have no interest in doing so. Instead, I found one that worked for my instrument and moved on. If you wish to continue learning about this integration capacitor, please share your results.

[chinapp]

yes,we could repalce the C505 to fix it,but if can figure out how to  adjust R544 ,this will be a good work ,perhaps we do not need to replace the C505 if adjust the R544 at a proper position

[Kleinstein]

According to the service manual R544 is actually replaced with an essentially identical circuit on an extera riser PCB (could be some layout issues or similar details).
The service manual does not show a test that would really check if R544 (or the new equivalent) is adjusted right. This part is to compensate dielectric absorbtion in C505. The ADC uses feedback in the run-up, that is similar to a dual slope cycle and this should not be very sensitive to dielectric absorbtion to start with. Much of the effect can also be linear and just a minor change to the gain that would not be an issue.
The case that could be effect the most from the R544 setting could be the fast 4.5 digit conversions, that seem to be using a kind of dual slope like mode.  Dual slope ADCs are somewhat sensitive to DA, though with a good capacitor and 4.5 digits just barely visible. The effect is some nonlinearity, reducing the gain for large readings. If there is more than just the nominal 4.5 digits of resolution available via the computer interface, it could be worth checking the linearity for the fast mode,
E.g. with readings at 0 , 5 V and 10 V.

For practical use the accuracy in the 4.5 digit mode is likely of low importance.

[chinapp]

you are right , from practice,I found that most of the AD CAL error is 2058 ,and  the results is that the 4.5 dig 5.5 dig is not at a good linearity,the 7.5 dig 6.5dig is good

[Mickle T.]

Did a ton of testing with Datron 1271 DA correction circuits to eliminate INL rises of up to 0.2 ppm from 0 to 5% FS. I concluded that the standard RC circuits have no effect.

[chinapp]

Did a ton of testing with Datron 1271 DA correction circuits to eliminate INL rises of up to 0.2 ppm from 0 to 5% FS. I concluded that the standard RC circuits have no effect.

you mean the R544 of 1281 has no effect to the 4.5 dig  linearity?

[Mickle T.]

Did a ton of testing with Datron 1271 DA correction circuits to eliminate INL rises of up to 0.2 ppm from 0 to 5% FS. I concluded that the standard RC circuits have no effect.

you mean the R544 of 1281 has no effect to the 4.5 dig  linearity?

No, I don't say that.
Do you really think anyone cares about the integral nonlinearity of an 8.5-digit device when using a 4.5-digit scale? 

[Kleinstein]

When the meter was new, the rel. high speed 4.5 digit mode may have had some value. Today there are plenty of better options for fast measurement.

The 4.5 digit mode is likely the one where the DA effect is most obvious and one may use it to adjust the compensation. The effect on the lower speed modes is expected to be much smaller and an adjustment there more difficult. Also testing the linearity at 4.5 digits does not need such a good other instrument / stable external source.
How much effect than remains for 7 or 8 digit mode is a bit hard to tell. It should be much less and may not be relevant at all.

[chinapp]

When the meter was new, the rel. high speed 4.5 digit mode may have had some value. Today there are plenty of better options for fast measurement.

The 4.5 digit mode is likely the one where the DA effect is most obvious and one may use it to adjust the compensation. The effect on the lower speed modes is expected to be much smaller and an adjustment there more difficult. Also testing the linearity at 4.5 digits does not need such a good other instrument / stable external source.
How much effect than remains for 7 or 8 digit mode is a bit hard to tell. It should be much less and may not be relevant at all.

yes,sometimes you can see obviously change at 4.5 digit when the AD can not pass the calibration  ,when 7.5 is 10.000013,the 4.5 digit maybe 9.995, though it should be 10.000.

from this we can  see how badly the AD is now, the C505  contribute much,but not all,maybe others (like the switch ?) I am working on one.....

but no workout.

[Kleinstein]

The Datron1281 already has quite some bodges to fix / improve things. Looking at the switches would open a new can or worms and chances are they already did quite some tweaks / testing. They already have 2 different but similar switches (DG308, DG211). Likely differences in the details make a subtile difference. Otherwise they would have simplified the BOM. Unless there is really a defect in that area, I would not touch them - new ones could already be different enough to cause trouble and ideally want a check of the INL.

The ADC is a bit old style and not really made for good performance at higher speeds (e.g. less than 10 PLC). It is not just the possible effect of DA and the compensation circuit where we don't know how to trim, but it is also the noise and just the time lost for the run-down part. So the 4.5 digit mode is more like an extra that may help in a few cases to identify a problem, but nothing made for precission. It is just not made for high speed, but for good INL with long integration.