Ok, it was a pretty nasty measure but I finally got it.
Since the target is to test a drift on the R measured, I needed a way to check not only his value over time but also the current generated by the 34401A.
Here my setup:
I have the 34401A measuring a precision 90R in parallel with a 34461A to check the voltage across it (related to the current generated by the 34401A DUT).
Basically the voltage measured by the 34461A divided by 90R will provide us the current sourced by the 34401A. The coax cables will minimize the noise pickup.
Now I had the 34461A USB connected to the PC, and and the other 34401A connected as well using GPIB or RS232.
I had two 34401A units (GOOD #1 and GOOD #2) fresh from calibration no problem to use as reference to compare with the, according to the calibration lab, bad one (BAD).
Of course the units were properly warmed up and the measure time was over one hour period.
Here the setting used for the R measures:
Auto Range: On
Auto Zero: On
Measurement: Resistance (2W)
NPLC: 10
Range(Ohm): 100
and for the current ones on the 34461A:
Auto Null Value: FALSE
Auto Range: On
Auto Zero: On
Input Impedance: 10 MOhm
Measurement: DC Voltage
NPLC: 10
Null State: FALSE
Null Value(Vdc): 0
Range(Vdc): 0.1
Let´s dig into the results...
GOOD #12W R measured:
Statistic [Ohm]
Mean 90.07426003
Standard Error 3.79745E-06
Median 90.074236
Mode 90.074083
Standard Deviation 0.000375295
Sample Variance 1.40846E-07
Kurtosis -0.493118701
Skewness 0.283102244
Range 0.002418
Minimum 90.073072
Maximum 90.07549
Sum 879755.2978
Count 9767
Largest(1) 90.07549
Smallest(1) 90.073072
Confidence Level(95.0%) 7.44379E-06
Current (90mV=1mA)
Statistic [A*90]
Mean 0.09014677
Standard Error 2.22337E-09
Median 0.090146746
Mode 0.090146714
Standard Deviation 2.22526E-07
Sample Variance 4.95176E-14
Kurtosis -0.075682427
Skewness 0.473640809
Range 1.4409E-06
Minimum 0.090146106
Maximum 0.090147547
Sum 903.0001985
Count 10017
Largest(1) 0.090147547
Smallest(1) 0.090146106
Confidence Level(95.0%) 4.35824E-09
GOOD #22W R measured:
Statistic [Ohm]
Mean 90.07989073
Standard Error 3.09024E-06
Median 90.079946
Mode 90.079946
Standard Deviation 0.000302496
Sample Variance 9.15041E-08
Kurtosis -0.177549751
Skewness -0.477571608
Range 0.0019
Minimum 90.078848
Maximum 90.080748
Sum 863145.513
Count 9582
Largest(1) 90.080748
Smallest(1) 90.078848
Confidence Level(95.0%) 6.05752E-06
Current (90mV=1mA)
Statistic [A*90]
Mean 0.088696667
Standard Error 2.62486E-09
Median 0.088696721
Mode 0.088696711
Standard Deviation 2.62263E-07
Sample Variance 6.87817E-14
Kurtosis -0.557971289
Skewness -0.449007757
Range 1.5119E-06
Minimum 0.088695812
Maximum 0.088697324
Sum 885.4588291
Count 9983
Largest(1) 0.088697324
Smallest(1) 0.088695812
Confidence Level(95.0%) 5.14525E-09
BAD2W R measured:
Statistic [Ohm]
Mean 90.02526044
Standard Error 5.78764E-06
Median 90.025232
Mode 90.025119
Standard Deviation 0.000564228
Sample Variance 3.18353E-07
Kurtosis -0.456379127
Skewness 0.228340502
Range 0.00326
Minimum 90.02379
Maximum 90.02705
Sum 855600.0752
Count 9504
Largest(1) 90.02705
Smallest(1) 90.02379
Confidence Level(95.0%) 1.1345E-05
Current (90mV=1mA)
Statistic [A*90]
Mean 0.089503874
Standard Error 2.02854E-09
Median 0.089503873
Mode 0.089503842
Standard Deviation 2.02641E-07
Sample Variance 4.10633E-14
Kurtosis -0.520592125
Skewness -0.068937736
Range 1.3154E-06
Minimum 0.08950316
Maximum 0.089504475
Sum 893.159158
Count 9979
Largest(1) 0.089504475
Smallest(1) 0.08950316
Confidence Level(95.0%) 3.97634E-09
CONCLUSIONSWe need a side to side comparison.
2W R MeasurementsHere the yellow values are the only one concerning me. Still not a huge drift in my eyes. The BAD shows a negative drift decreasing the R value. The GOOD 1 also does that, but the gradient is less than the BAD one.
Current MeasurementsPretty much here everything is acceptable, since the unit cal lab reported the drift only in the 2W R mode, I was not expecting a problem with the current:
Maybe the current generated by the BAD one is suspiciously pretty nervous and irregular in the graph.
Next steps0) Check/Understand the Spec. Tolerance (let´s do tomorrow. Is already 00:51 AM)
1) Dig into the schematics and isolate the part of the circuit which can cause a drift only in 2W R mode
2) Perform a 4W R 1 hour test as above to verify the drift is not there in 4W
3) Check the 100 Ohm Range current generation, check if current is constant in 1K Range
3) Open the unit and look the circuit for bad joints on the part of circuits founded in point 1 and 3
I don´t know if I will fix this unit, it seems a nasty slightly out of spec. malfunction.
On the side: EMI CRAP?I had the bad idea to work under the bench using my 12V impact drill during the 1 hour measurement and here what I got on a good one! (left current, right R measured)
Even when I turned on my monitors from sleep, I noticed similar jumps. I decide to measure when I was not at the bench during night time (everything off, just the PC and the DMMs).