Products > Test Equipment
Keysight's new 34465A (6.5 digit) and 34470A (7.5 digit) bench multimeters
<< < (98/145) > >>
Dr. Frank:
Here are my measurements, 34465A, 4W OCOMP, NPLC100, no ACAL, warm up curve, from cold start, 14h.
It's done on VHP202Z, 10k reference resistor, with proper cables.

After 1/2h, instrument is mostly warmed up, then between 1h...14h there's a drift of about 2.5ppm.
The main spacing in the xls diagrams is 1 ppm.

Standard deviation, i.e. rms noise, measured over the last hour is about 0.13ppm

I assume, that there might be a switch mode power supply or similar in your room, or your resistor is not properly connected.


What's this strange resistor, you are using? Didn't find it anywhere.
Do you have a photo of your setup?

Frank


HighVoltage:
Just for completeness, here are some more screen shots from the DMM7510, as a statistical reference.
This time I was careful not to touch the cables.
What is just really stupid on the Keithley is the numbering to the left of the graph.
May be Keithley did not expect us to use it for precision measurements?

And yes, Dr.Diesel the graphing functions of these great meters are something one can get really fast used to and never wants to miss again.

Alex Nikitin:

--- Quote from: Dr. Frank on October 22, 2016, 03:35:46 pm ---
I assume, that there might be a switch mode power supply or similar in your room, or your resistor is not properly connected.


What's this strange resistor, you are using? Didn't find it anywhere.
Do you have a photo of your setup?

Frank

--- End quote ---

I don't have problems with other meters (HP3458A and Keithley 2000) in the same lab, and I don't have problems with voltage measurements on the 34465A. The resistor I've used (LT450C) is this one , in a metal enclosure and connected with short twisted pairs and gold plated beryllium copper plugs.

Cheers

Alex
Dr. Frank:

--- Quote from: Alex Nikitin on October 21, 2016, 11:27:03 am ---Here is a graph with a 2 hour run using my LTC450C 10K "standard" (measured by HP3458 about a week ago and by Keysight 34465A today) . Overnight variations were even larger - about 25ppm p-p. Is this a typical behaviour for the 34465A ?

Cheers

Alex

--- End quote ---

This size of fluctuation, about 1ppm, up to 25ppm, has nothing to do with "noise". If your instrument would be the root cause, this would really be a defect of its Ohm circuitry.
But I doubt that. At least, I can't imagine the electrical failure mode, which would create such a failure picture. The Ohm circuit is assumed to be the very same as the one inside the 34410/411, and we've never seen a similar failure. It's also quite improbable, that the 34465A as a new instrument would fail, or you know already, that you somehow damaged the Ohm circuit by external overvoltage; which is also lesser probable to cause a failure, by the instruments protection circuitry.

Your resistor assembly, anyhow looks as to tend more probably to fail:
It has no 4W connection, so you will have to make 4W measurements artificially on these two jacks.
I don't know the quality of your cables, but broken cables or oxidized connectors might also create such failures.

You have used 3x 100M as trimming resistors. This gives about 33ppm change of 10kOhm, and that's in the exactly same ballpark as the ~25ppm you observe as fluctuations. That alone makes me suspicious.
Resistors in the high MOhm values are usually unstable, especially these thin film or coal film types, and tend to fail fatally. (which type are these, actually??)
Maybe there's a cold solder junction, or a crack inside these resistors, which create these fluctuations.

Then you mentioned, that you drilled the cables. What kind of isolation do these have? Which cables are paired?
If you drilled Sense+ and Sense-, and your cables have a poor isolator like PVC, this will create big errors by leakage, on the order of many ppm, and also depending on the moving of these cables.
If these are isloated by TF, it's better, but otherwise, you have to separate cables, which are on positive and negative potential.

Then you said, that you measured this resistor OK with the 3458A, but one week BEFORE this measurement.
Therefore you can't be sure, that your resistor assembly did NOT go defect in between.
If you re-measure with the 3458A NOW, maybe it also shows this fault.

I only can hint to my own experience, being tricked by the resistor DUT itself.
Comparing Ohm measurements (on 100k) of 34465A vs. 3458A, which differed by about 30ppm also, I thought, that my 3458A would be defect!

In the end, it turned out that the DUT  showed a dielectric relaxation effect, which I discovered only by the different OCOMP timings of both instruments.

So, purely from a logical aspect, I propose that you look for the fault at the more probable resistor, than at the less probable 34465A.

Maybe you have another stable 10k resistor, which you can test. 1k and 100k resistors also might do the job, as it's only another range. A simple thin film 10k resistor would show a big, but smooth temperature drift, but no fluctuations.
Or you simplify the DUT by measuring the precision resistor 'naked', as I have done; and by simplification, I finally found the root cause.

Frank
Alex Nikitin:
Hi Frank,

Thank you for such a long and thorough post. I can say that:

1) I've observed these fluctuations first on an overnight run with a completely different resistor (10K UPW25 wire-wound), just plugged in with it's leads formed to fit with some force into the meter's sockets, in 2-way mode. I've suspected the resistor first and repeated the measurements with my reference in  4-way connection and got the similar variations, I also connected the UPW resistor in 4-way configuration by soldering the wires to it. Same results.

2) My reference resistor was modified after I took that picture but before if was measured again with the HP3458A. The chain of three small 100M resistors was replaced by a quality thin film 1% 100ppm/C 300M resistor.

3) As I've mentioned, if I run ACAL at a point where the measured value is 10-15ppm off, the meter gets to a correct value within 1-2ppm (usual ACAL variations).

4) There are no obvious sources of strong RF/EM interference in the lab, I work there all the time with fairly sensitive stuff (electometers and very high stability power supplies) without any problems. I didn't have this noise on the same two resistors with either HP3458A or Keithley 2000 in the same lab. I will run the tests again side by side when I get the HP3458A back from Keysight calibration lab.

5) I use Cat 6 network cable twisted pairs and beryllium copper spring plugs (which I found rather nice and inexpensive). I will take some photos on Monday.

So, I will double-check all my findings but right now it certainly looks like a fault in the 34465A .

Cheers

Alex

Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod