Electronics > Metrology

Best way to determine other 10K standard resistors base on one newly calibrated

(1/3) > >>

aronake:
Hi

I plan to have one of my 1 ohm and 10K ohm standard resistors calibrated at government standards and calibration laboratory. They will do 10k at 3 milliohm and 1 ohm at 0.3 microohm uncertanity at 95% level. So pretty decent. The lab is 10 minute walk away from where I live.

I would want to compare these to be calibrated standards as accurately as I can to other 1 ohm and 10k ohm standard resistors I have before I hand them in for calibration and after. How do I best do this?

Tools at hand that may be relevant are a couple of 3458a with 2 having certified calibration but these expired 3 years ago, a Keithely 7510 (in calibration, but i doubt the quality of the calibration), a Fluke 732b, a couple of DIY 10V LTZ1000, a LTZ1000 modified Yokogawa 7651 current/voltage source (which is very stable), a Keithley 2450 SMU.

My plan:
- Measure all resistors with all 3458a and the 7510 in best possible settings. As the 3458a are pretty far from the uncertanity of the calibration lab, resistances measured will mostly be seen as ratios against the to be calibrated resistors.
- Use the 732b as voltage source over the to be calibrated 10K vs all other 10K (one by one) and measure voltage over both and each resistor. I would think the 732b have current capability for this, but will check before I try.
- Use the Yokogawa at some suitable voltage and do same measurement as for with 732b for the 1 ohm resistors. It may be that the higher current capability of the Keithley 2450 may be better, but I think the Yokogawa is sufficient. It is quite a bit more stable

Then analyse all results that hopefully match quite well to determine absolute value of other resistors.

Reason to do measurements both before and after is to see that nothing have happened with the standard resistors while they were away. Also to test the measurement methodology so I can quickly do everything once I get the measured resistors back.

Do this sound good based on the tools I have or any suggestions how to do this better?

Bonus question: To what uncertanity could I expect to measure other resistors based on the newly calibrated with the proposed or some other better way of doing it (existing tools being the constrain)? I would do the comparison measurements same or the day after the resistors have been measured by the calibration lab. So not much of time drift. I will obviously be very cautions when bringing them home and keep them at 23 degrees in a sealed box, and bring them into my home lab which will then be at 23 degrees. So most of uncertanity ought to come from measurement methodology and limitations.






Sensorcat:
A thought on your bonus question: Do you already know the temperature at which the lab will do the calibration? In many calibration reports I have seen, the temperature is not necessarily 23°C, but only reasonably close, and the lab indicates the temperature at which the calibration was done, instead of maintaining exactly 23°C. This affects your plan to do the before-after comparison. Furthermore, doing the comparison one day after the calibration is not as good as doing it right after it, as drift starts fast and then fades out with time, such that the first day is already a large fraction of the drift you get over a 90-day period. You can see this in specs of some precision instruments with errors for 1h/1d/90d/1y etc.

bdunham7:

--- Quote from: Sensorcat on June 12, 2024, 07:30:25 pm --- Furthermore, doing the comparison one day after the calibration is not as good as doing it right after it, as drift starts fast and then fades out with time, such that the first day is already a large fraction of the drift you get over a 90-day period. You can see this in specs of some precision instruments with errors for 1h/1d/90d/1y etc.

--- End quote ---

But calibrating a precision resistor consists of just measuring it and recording the value.  How would that affect drift?

aronake:

--- Quote from: Sensorcat on June 12, 2024, 07:30:25 pm ---A thought on your bonus question: Do you already know the temperature at which the lab will do the calibration? In many calibration reports I have seen, the temperature is not necessarily 23°C, but only reasonably close, and the lab indicates the temperature at which the calibration was done, instead of maintaining exactly 23°C. This affects your plan to do the before-after comparison. Furthermore, doing the comparison one day after the calibration is not as good as doing it right after it, as drift starts fast and then fades out with time, such that the first day is already a large fraction of the drift you get over a 90-day period. You can see this in specs of some precision instruments with errors for 1h/1d/90d/1y etc.

--- End quote ---

Thanks for input.

The 10K resistor have a TC of around 0.04 ppm / C and the 1ohm around 0.1 ppm / C. So i would not be too concerned about +/- 1 degree, but of course the lower the better.

On drift, i would think in reality it is quite linear (or driven by evens like thermal shock, being dropped, overvoltage etc) and that it is growing much quicker in the beginning is mostly related to the statistic model used for drift. Also measure uncertanity will probably also be much bigger than 1 day drift. But still a good point in trying to do the comparison as soon as possible.

Alex Nikitin:
I would suggest using a stable current source and measuring the voltage ratio of two resistors connected in series for two polarities of the current, using the ratio mode on the HP3458A.

Cheers

Alex

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod