HP 3458a thermal / temporal drift

[martinr33]

My 3458a had fairly nasty temporal drift on it (yes, the ADC board, and 3458as have had this problem forever). I replaced the ADC board, and the drift pretty much disappeared.

I've been logging my 10V reference for a few days now (the ref doesn't drift any meaningful amount). I have bad cabling (cheap) but that is not so much of a concern here.

Where I am going is, it looks like thermal compensation - at least at 10V - should be pretty straghtforwards. I've meddled with the axes below to show how measured voltage varies with the reported internal temp. There's minimal drift, and pretty good tracking between the two variables. The voltage scale is 2.5uV/division, so there's about 12uV drift across 8 degrees C. So about .15PPM per degree.

And yes, I twiddled the scales to get good alignment.

[TiN]

I'm not sure what I'm looking at in the graph, sorry.

[Kleinstein]

The graph is pretty clear, its voltage and temperature versus time.

Having a 0.15 ppm/K TC before ACAL is not bad. In theory there could be some compensation / adjustment, but AFAIK there is not provision for such compensation in the 3458.
Under normal lab conditions the temperature should not change that much and ACAL should be called a few more times with so much temperature change.

[MisterDiodes]

Run ACAL when temp changes +-1°C or at least once every 24 hrs - or more often as req'd.  We have the operators do that at least once every shift during complete test system checks.

Not sure if OP is complaining or ??

[dl1640]

It looks the DMM is not in lab condition (air condition room) if TEMP? drift that much, however the tracking of temp and volt are nice.

[martinr33]

I am still thinking about where this goes. My goal is to figure out how to get more accurate readings by compensating for temperature and drift. I also wonder about the consistency of ACAL. I need to run a number of them and see what the spread is.

Obviously, ACAL corrects for any drift in  the ADC. I just ran it, and the reading dropped by about 1uV. That's over 5 days, so I might assume that drift on this unit's ADC is about .02ppm/day. Or not - a bunch of that could be calibration uncertainty. Successive calibratoins have some uncertainty in the last digit.

However, I don't see how ACAL can compensate for temperature-driven changes across the entire analog front end.

An implication of this would be  that the meter should be operated at the same internal temperature as it was calibrated at. The internal temperature seems to track the voltage shift well. This means that you have to pay attention to how the unit is located. Inside an equpment rack, it could be quite a lot warmer than  the room in which it operates.

[MisterDiodes]

Check your 3458a manual and the docs at Keysight for proper operation of the tool - there is much more information regarding how ACAL works, and why you run that once every 24 hrs (minimum) or when temperature changes +- 1 deg C.  Don't ever wait 5 days to run it in one measuring session!!   ACAL will compensate for some short term component drifts but it is not a fix all for major temperature swings and long term drift; and doesn't do anything for the drift of the VRef module of course.

Remember, the use of ACAL reduces errors even if you've moved more than the standard +-5°C from the meter's official calibration temperature (Tcal).  Here are some examples of measures used with and without ACAL, and a guide for looking at uncertainties the farther you are from Tcal:

http://literature.cdn.keysight.com/litweb/pdf/5965-4971E.pdf

Additionally check Art of Electronics 3rd edition for another good description of how the 3456a / 58a front ends and their ADC's work.

[martinr33]

Thanks for that. Page 12 of the data sheet has the calculations. It does indeed show that ACAL eliminates all temperature-based deviations from the system.

However - with +/-6uV across my whole temperature range, that is almost within the datasheet spec for +/- 1 degree. These systems have a lot of headroom in their specs.


Given my difficult environment, I am looking to characterize temperature and time related offsets. Autocal's a bonus.

I'll upgrade my connections and see how autocal affects the reading at various internal temperatures. I may be measuring some cable effects here, so I will take care of that.

[dl1640]

Given my difficult environment, I am looking to characterize temperature and time related offsets. Autocal's a bonus.

Given such difficulty the internal temperature is that dynamic you should characterize the relationship of temp drift vs reading drift as time goes by. Maybe forget about ACAL if you do not need any gap bewteen your measurements because even ACAL DCV need about 2 minutes to complete and your enviroment is dynamic so ACAL will not help much. However since only 10V source (internal temperature maintained constant) is characterized the outcome may be limited. I am also interested how you are going to characterize the cabling effects, maybe some thermocouple effects is concered if measure DC.