Datron 1281 Repair/Maintenance

[TheSteve]

I don't recall ever getting the TL866A/CS to work for programming FRAM. It would always mess up the first byte of each block. The TL866II Plus worked fine though.

[rigrunner]

Just looking through my notes and with the original TL866 I selected device DS1225 and it worked fine,
With the later TL866 I used DS1230 and selected the range.

[Dr. Frank]

I don't recall ever getting the TL866A/CS to work for programming FRAM. It would always mess up the first byte of each block. The TL866II Plus worked fine though.

If this is the case, I may help PeLuLe, as I have the TL866II PLUS, and could lend it to him.

Frank

[leighcorrigall]

Hello everyone,

I am working on addressing a tricky TCV problem that I have had with my instrument. It seems to have a TCV of about 1 ppm/K while measuring a Fluke 732A 10V (selfcorr off, resl8, filter off, fast off). I am wondering if anyone else has experienced this problem and would share their experience.

Rigrunner and myself have been trying to diagnose the problem for a while now and we suspect that the TC is coming from sheet 4 (references and signal switching). Specifically, the circuit around M406 (TP403 = 0.9 ppm/K) and M407 (TP404 = 1 ppm/K). These TPs should track each other well (~0.1 ppm/K difference), and I would suspect that they should also have a very low TC (not the case for my unit).

The voltage references measured at TP405 and TP406 are very low (< 0.1 ppm/K). The junction between R417, R1418, and Q401 is ~0.3 ppm/K. I have also examined the TCR of R409 and R414, which I believe to be very low (< 0.2 ppm/K).

What is important to note is that when CAL? is run before each measurement, the instrument has < 0.1 ppm/K TC:

Code: [Select]


print("self calibration enabled...")
dmm.write("ENBCAL SELF")

print("CAL? ...")
dmm.write("*CAL?")
time.sleep(300) # wait at least 1 min for DCV and 12 mins for all options
   
print("self correction enabled ...")
dmm.write("SELFCORR ON") # toggle SelfCAL corrections

[ ... include CAL? in the measurement loop]

The low TC when looping CAL? suggests that the M403 circuit and sheet 6 (self cal multiplier) are working very well. It also suggests that the ADC is working well.

Can someone with a working Datron 1281 measure the TC of TP403 and TP404, then share the results with us? Please note that the instrument should be run with the settings presented above (selfcorr off, resl8, filter off, fast off). Any other TC references would be helpful too. The TC can be measured with and/or with the cover.

Thank you very much!

[Mickle T.]

I encountered the same problem in Datron 1271. While the temperature coefficients of the LTZ1000 reference module and the input amplifier (A=1) are less than 0.1 ppm/°C, the temperature drift during cold start and +10 V measurement is up to 8 ppm, and the temperature coefficient of the multimeter readings changes from about -0.8 ppm/°C at the beginning of warming up to +0.3 ppm/°C after reaching the steady-state thermal mode.
I determined the reason for this behavior of the device: nonlinearity and mismatch in the temperature dependences of the resistance Rdson of the ADC switches. As a result, compensation of switch resistance via VREF_COMP has no effect.
By changing resistors R422 and R423 in the compensation loop, it is possible to correct the final temperature dependence, but only its linear part and only in one direction. So after experiments, I rejected this idea.

[leighcorrigall]

I encountered the same problem in Datron 1271. While the temperature coefficients of the LTZ1000 reference module and the input amplifier (A=1) are less than 0.1 ppm/°C, the temperature drift during cold start and +10 V measurement is up to 8 ppm, and the temperature coefficient of the multimeter readings changes from about -0.8 ppm/°C at the beginning of warming up to +0.3 ppm/°C after reaching the steady-state thermal mode.
I determined the reason for this behavior of the device: nonlinearity and mismatch in the temperature dependences of the resistance Rdson of the ADC switches. As a result, compensation of switch resistance via VREF_COMP has no effect.
By changing resistors R422 and R423 in the compensation loop, it is possible to correct the final temperature dependence, but only its linear part and only in one direction. So after experiments, I rejected this idea.

Hi Mickle,

Thank you for bringing this to my attention. Yes, we seem to be sharing a similar problem...

1) If I activate the Self Corrections, the problem is significantly reduced to the point that the instrument is close to TC specifications on the 10 V range when comparing the instrument to Fluke 732As.
2) If Self Corrections is off, the instrument has a TC of approximately 1 ppm/K even with the cover on and with a full warm up. Clearly, there is something not working correctly.
3) The +V_ref_COMP and -V_ref_COMP appear to have a TC of about 1 ppm/K, but track each other well (< 0.2 ppm/K).
4) I was able to rule out the V_refs as suspects. The critical resistors in the REF AMP and REF BUFFER seem fine.
5) Since the Self Corrections seems to do a good job, I believe the ADC is working reasonably.
6) INT. SRCE CAL. and FULL test pass.

Have you made any progress? I am about to begin another attempt at diagnosing the problem.

My setup will include:

For 10 V range measurements:
3458A and Keithley 7012 (10x channels)

For voltage comparisons:
Keithley 7182 and Keithley 7168 (8x nV channels)

[Mickle T.]

Hi Mickle,

Thank you for bringing this to my attention. Yes, we seem to be sharing a similar problem...

1) If I activate the Self Corrections, the problem is significantly reduced to the point that the instrument is close to TC specifications on the 10 V range when comparing the instrument to Fluke 732As.
2) If Self Corrections is off, the instrument has a TC of approximately 1 ppm/K even with the cover on and with a full warm up. Clearly, there is something not working correctly.
3) The +V_ref_COMP and -V_ref_COMP appear to have a TC of about 1 ppm/K, but track each other well (< 0.2 ppm/K).
4) I was able to rule out the V_refs as suspects. The critical resistors in the REF AMP and REF BUFFER seem fine.
5) Since the Self Corrections seems to do a good job, I believe the ADC is working reasonably.
6) INT. SRCE CAL. and FULL test pass.

Have you made any progress? I am about to begin another attempt at diagnosing the problem.

My setup will include:

For 10 V range measurements:
3458A and Keithley 7012 (10x channels)

For voltage comparisons:
Keithley 7182 and Keithley 7168 (8x nV channels)

Unfortunately, I couldn't find a simple solution to this problem. In the end, I traded the Datron 1271 and Solartron 7081 for the Advantest R6581.

[Rax]

Hopefully I'm not hijacking Leigh's tread by asking if anyone's aware of an SM (schematic/board layout) with the older type of digital board? In my digging around, I could only find the newer (?) board being documented. I enclose a pic of mine.

I've been running into some "NV RAM WRITE FAILURE" (error 2022) issues while running self tests and self calibration and need to dig deeper. For instance, I've been trying to identify the TPs on this version board with mixed results. Also, which ICs exactly can cause this error. The battery seems to be OK at 3.7V (though certainly ancient).

Thank you all for your insights.

[leighcorrigall]

Hi Rax,

You are more than welcome to 'hijack' this thread. The more information we gather on this instrument, the better off everyone is.

Another Datron 1281 appeared in the mail today. The digital board looks similar to what you have (pictured).

EDIT: There are plenty of reference pictures of this older PSU/Digital board on xDevs:
https://xdevs.com/doc/Datron/1281/img/dig_top.jpg

The board has date codes as early as 1984 that I can see. The DCV board averages around 1987. This is a quick observation.

EDIT: The firmware version is "890144-01.07"

When I have more time to think about this, I can certainly try to help you with your problem.

Tonight, I plan on running the CAL DUMP programs before I replace all the capacitors.

Leigh

[leighcorrigall]

I encountered the same problem in Datron 1271. While the temperature coefficients of the LTZ1000 reference module and the input amplifier (A=1) are less than 0.1 ppm/°C, the temperature drift during cold start and +10 V measurement is up to 8 ppm, and the temperature coefficient of the multimeter readings changes from about -0.8 ppm/°C at the beginning of warming up to +0.3 ppm/°C after reaching the steady-state thermal mode.
I determined the reason for this behavior of the device: nonlinearity and mismatch in the temperature dependences of the resistance Rdson of the ADC switches. As a result, compensation of switch resistance via VREF_COMP has no effect.
By changing resistors R422 and R423 in the compensation loop, it is possible to correct the final temperature dependence, but only its linear part and only in one direction. So after experiments, I rejected this idea.

Thank you for this. I believe I have identified the problem and I will try to summarize it soon.

A total of three people, including myself, have experienced this problem and I believe it has to do with the matched R513 R514 set. Very difficult to identify.

[Rax]

In terms of options, I wonder if there's a good sense of modules interchangeability between different models (or even within the 1281 across years of production). For instance, it seems to me the 1081 AC module (also option 10) looks pretty similar to the 1281 (the headers for connectors maybe not so much?...). The 1061/1071 doesn't seem to be similar.

I'm looking for an option 10 for the unit I have here and I'm wondering on the range of choices I have.

[leighcorrigall]

Hello again, Rax,

The digital board you have, and the one I just acquired, is an older model that uses two pairs of EPROMs (4x 512 kbits). The newer boards use a single pair of EPROMs (2x 1 Mbits). Unfortunately, removing the daughter boards to install 1 MEG EPROMs won't get you far. There is a pin-compatibility issue.

Before you begin, I would highly recommend backing up all of the data. The NVRAM can be dumped with the attached scripts. The NVRAM can be copied remotely over GPIB without disassembly. Please note that they are untested. I had to make some quick modifications to the previous NVRAM dump script because firmware version 1.07 communication is terribly sensitive. I am leaving for vacation soon and I do not have much time to revise. There is another tested version earlier in this thread:

https://www.eevblog.com/forum/metrology/datron-1281-repairmaintenance/msg3876473/#msg3876473

Next, you should copy the EPROMs. I would like to have a copy of version 2.03 (from August 16 1988) if possible. I see that you have them.

From the attached pictures, I will name out the order of the memory from left to right:
- TC5564PL-15 is responsible for the calibration constants, the serial number, and the line frequency (i.e., NVRAM). It relies on a backup battery (1/2AA 3.6V).
- 84256-12LL is static RAM (SRAM) for basic I/O and is hidden beneath the daughter board.
- 410408-A daughter board that hosts 290159 (TMM27512) and 290160 (HN27512) firmware.
- HD68000P8 microprocessor.
- 410408-A daughter board that hosts 290164 (TMM27512) and 290165 (HN27512) firmware.
- 84256-12LL is static RAM (SRAM) for basic I/O and is hidden beneath the daughter board.

Using the firmware available on the xDevs website, you can optionally upgrade:
https://xdevs.com/fix/d1281/#firmware

Some firmware will allow you to use old calibration data. However, in my case where the instrument was calibrated with version 1.07, I was unable to reuse the data. In this case, a blank TC5564PL is needed or an FM1808 adaptor board will have to be prepared either with a universal programmer or by going into the special calibration menu and 'clearing the NVRAM' (CLRMEM ALL, Wavetek 1281 User's Manual page 5-106). If you do this through the menu, then no serial number will be presented. To correct this, you can send a command called SERIAL which can be found on page 5-105.

I would suggest reviewing Appendix A (4-A5) where it shows a list of Corruptions (errors 2013 through 2022) and Non-volatile RAM Checksum Errors (errors 2110 through 2113).

I believe your error (i.e., 2022) means that you need to replace or upgrade your NVRAM as it is bad. Luckily, you can make a FM1808 adaptor board and replace it (pictured as a yellow board). The FM1808B is still available from your favorite electronics distributor. Remember to remove the battery before installing!

I hope this helps,

Leigh

[Jay_Diddy_B]

Hi group,

I am looking at doing a FRAM conversion on my Datron 1281. I have looked at the FRAM adapter on the xDevs website. This uses the FM1808 32k byte FRAM memory. I am reasonably sure the FM16W08 8k byte FRAM memory should work.

Schematic



Model



I plan on programming this with a Xgecu T48 programmer. I should not need an adapter, unless the programmer pulls pin 26 low.

Do you see any problems with this approach?

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

I have written a program in vb.net that uses the AR488 to extract the NVM content using the DUMP? command. I have compared my binary file with some of those on the forum and I think that I have got it right.

Can somebody confirm that the memory map is:



The last section looks like this:



18201 is the serial number of my Datron 1281

Does this look right?

Regards,

Jay_Diddy_B

[leighcorrigall]

Hi Jay Diddy,

The size of IPZ, SECN, and PRIM change depending on the firmware. My (unverified) dump? script above accounts for this.

Leigh

[Jay_Diddy_B]

Leigh,

I based my code on the Python script in reply 27 of this thread. Does this script work? or does it only work with some firmware revisions?

I have SW revision 3.12

Thanks!!

Jay_Diddy_B

[leighcorrigall]

Leigh,

I based my code on the Python script in reply 27 of this thread. Does this script work? or does it only work with some firmware revisions?

Thanks!!

Jay_Diddy_B

Refer to another more recent entry "reply 161".

[Jay_Diddy_B]

Hi group,

I can report a little progress on this project. The PCBs have arrived and I have assembled one:





I had little error with the footprint for U1, but I was able to make the ic fit the PCB.

I can read and write the FRAM on the adapter using a XGecu T48 programmer using the Ramtron RM16W08 settings.

When I am confident that I have read the data out of the meter I will update the group again.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

I can report success with my FRAM conversion.
Here is a picture of my FRAM adapter installed in the 1281:




I used Leigh's algorithm and coded in vb.net and then adapted the program to use the ar488 interface.



The FM16W08 adapter was programmed with the extracted binary data using a XGecu T48 programmer.
The NVM memory were read after installing the FRAM in meter, the contents matched.

Thanks to all those who have contributed to this topic.


Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi,

I received a request for the Gerber files for this project.

Regards,

Jay_Diddy_B
 TC5564PL_FRAM_R2.zip (42.26 kB - downloaded 17 times.)

[IanJ]

Nice work on the Fram board, always great to see our precious instruments data well protected.

Just a wee note……The thin pin on those double male headers is the ‘solder tail’ per the datasheet, so as I read it you would solder that side to your new board.
If you mount them upside down they can be a bit weak in an IC socket.

https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/6642/350_Datasheet.pdf

Ian