Finally 7.5 Digit - My "new" Keithley DMM7510

[Mickle T.]

While researching the service mode for a solution to the ADC nonlinearity problem, I compiled a "Hall of Fame" for the DMM7510 and DMM6500 developers. The color scheme was chosen arbitrarily. But the order of the names and the breakdown into color subgroups are exactly as they are in the secret firmware menus of the devices.

[exe]

So I just left my dmm on with the reference connected to it measuring my reference. You can see data in attachment.

One bug or feature I found: in my measurement script I disable screen. However, it seems after one hour the screensaver kicks in and... switches screen back on to low brightness  When I touched the screen, it went black(. So, it  seems screensaver restored previous brightness setting on touch.

I disabled screensaver to overcome this.

PS 1ppm is ~6.9uV on the plot.

[Martin72]

This clearly demonstrates the effect of temperature on the rest.

[exe]

This clearly demonstrates the effect of temperature on the rest.

Is this a good thing or a bad thing?) I tried to find copper binding post for a reasonable price and I failed. So, rolling with what I have. Just covered the whole setup with a piece of clothes.

[Martin72]

Everything hinges on the temperature; it has the greatest influence of all.
That said, a deviation of +/- 1 ppm over 35 days is quite remarkable.

[exe]

I asked Claude to apply correction for tempco. It seems internal temperature is the best predictor for drift (top plot). External temperature also predicts well, but slightly less good. Using both external and internal correction doesn't improve anything.

I'll also try to correct for atmospheric pressure to see if it gives anything. I suspect it might. At least there is some anecdotal evidence I read somewhere.

[exe]

Correction for atmospheric pressure didn't do anything useful, the R^2 dropped from  0.779 to 0.768.

PS I applyed quadratic fit, now R^2 is 0.787. Cubic fit didn't improve anything. So, if you squint a little bit, the plot will be +-0.5ppm, I'm happy, though I was hoping +- 0.1ppm. A bit too optimistic, I know. But every voltnut dreams of getting a "golden" sample of equipment, right?)

[exe]

I played with second-order temperature correction a little bit, and I'm quite impressed how well it works. It seems to me that average reading is well within +-2uV (measuring adr1399, in module TEN399), except the first week when reference or DMM was slowly drifting down.

Compensated measurements: faint blue is raw data, the red line is a sliding average over 500 measurement.
Uncompensated data: grey lines, goes off the chart.

Tempco is about +1.8uV/C.

The gap on the plot is due to rpi stopped working for unknown reason. The "big" dip in the middle of the plot is presumeably due to cleaning at home done by my cousin, though not 100% sure.

[Mickle T.]

Model DMM7510 Firmware Revision 1.7.18
https://www.tek.com/en/support/software/firmware/model-dmm7510-firmware-revision-1718-and-release-notes

OVERVIEW
Version 1.7.18 provides several critical fixes.
CRITICAL FIX
Reference number: CESK-3 20
Symptom: Lengthy scan s can hang when exporting data to USB after each scan.
Resolution: This issue has been resolved.

Reference number: CESK- 506
Symptom: If a single -channel scan is aborted and the data is saved, on ly the first point of the scan is saved to the USB drive on export.
Resolution: This issue has been resolved.

Reference number: CESK- 507
Symptom: If the file name is set to use the date and time, instead of saving a new file every time you run a new scan, the file is overwriting the existing file.
Resolution: This issue has been resolved.

Reference number: CESK- 508
Symptom: When you export scan data in channel columns, the column headers include an extra comma which causes the headers to be off by a column.
Resolution: This issue has been resolved.

Reference number: CESK- 566
Symptom: Updating the system time on the instrument to align with the PC causes the data -returned to display an incorrect channel and the timestamps jump.
Resolution: This issue has been resolved.