Another sick Keithley 2000

[jchw4]

Some time ago I got mechanically damaged k2000, which otherwise was perfect. It still reads 5.00000 V from my DMMCheck.
(k2000 reports last calibrated in 2002)

But yesterday I noticed that it's 1M range is broken both in 2W and 4W.

For example, 100k in 2W mode:

Code: [Select]

100k range: 100.7717k
  1M range: Overflow
 10M range: 00.10071M
100M range: 000.1007M

Shorts:

Code: [Select]

100k range: 000.0003k
 1M  range: -0.000006M
10M  range: 00.00003M

4W:

Code: [Select]

1.80 Ohm on 1M  : -0.000053 M
2.78 Ohm on 1M  : -0.000072 M
11.8 Ohm on 1M  : -0.000672 M
101.8 Ohm on 1M : -0.002190 M
1.000 k  on 1M  : -0.021508 M
9.967 k on 1M   : -0.214120 M
100 k on 1M     : Overflow
(values on the left measured using other ranges)

All tests pass.
Current source is perfect (measured with external meter).
TP104 seems to be correct.

What could it be? I have a feeeling that it's calibration memory.
Here  https://www.eevblog.com/forum/repair/first-time-on-keithley-2000/msg974309/#msg974309 another user shared calibration dump commands and their actual values.

Here are mine:

Code: [Select]

000000000262145
000000000393216
000000000196608
000000000262144
+2.74088039E-02
+2.74717795E-01
+2.74241546E-02
+2.74939096E-01
+7.83504339E-05
+2.74746629E-02
+2.75300742E-01
+8.12454578E-05
+2.74820894E-02
+2.75495528E-01
+2.74714668E-02
+1.92738696E-01
+7.83581243E-05
+2.73505044E-02
+2.74136247E-01
+1.37793306E-02
+2.76778882E-01
+9.98209300E-01
+9.52346951E-01
+1.00000000E+03
+9.99984500E+02
+1.00000500E+01
-1.00000700E+01
+1.00001400E+02
+1.00000000E+03
+1.00000000E+04
+1.00000000E+05
+1.00000000E+06
+1.00000000E-02
+1.00000000E-01
+1.00000000E+00
+1.99251556E-06
+1.88449784E-06
-6.65748151E-07
+1.41637118E+00
-1.41637561E+00
+1.41689936E-01
-1.44000000E-01
+2.48186119E-05
+1.72858864E-05
+5.29676357E-06
+3.29774682E-06
+3.18277699E-06
+9.17269493E-02
+9.99929726E-01
-1.93286724E-06
-2.00583545E-06
-1.94541648E-06
-1.97593550E-06
+1.17605155E-06
+1.41526718E-01
+1.42849062E-01
+1.42880911E-01
-6.65767966E-02
+9.17000000E-02
+1.19220998E-06
+1.18602474E-06
+1.43014682E-02
+1.41262068E-03
+1.41139930E-02

Comparing them I can see that one value is significantly  different:

Code: [Select]

300.000%
-0.537%
-0.261%
-0.712%
-0.403%
-80.024%
-0.480%
-0.216%
-79.150%
-0.664%
-0.360%
-0.586%
-0.288%
-80.008%
-1.071%
-0.791%
-1.431%
-0.824%
-0.179%
-4.247%
0.000%
-0.002%
0.000%
0.001%
0.001%
0.001%
0.000%
0.001%
0.003%
0.000%
0.000%
0.000%
78.899%
130.010%
-4495.288% (-6.65748151E-07   +1.51468621E-08 1111111111111111111111111111111110110101001100101011010111011101 110010100000100001110001001101)
-0.235%
-0.235%
-0.238%
0.000%
61.263%
41.971%
113.748%
186.489%
193.640%
0.090%
0.007%
-24.512%
-17.552%
-23.834%
-22.502%
-2.625%
-0.058%
-0.016%
0.001%
-104.660%
0.000%
104.454%
96.454%
-0.322%
-0.692%
-0.737%
(I am not sure that interpreting these values as float32 is correct, but it looks reasonable to me.)

So my value is -6.65748151E-07 while other users value is +1.51468621E-08 . 

Could it be cal EEPROM fault?

What else should I check?

[Kleinstein]

A negative Ohms readings may be a faulty cal constant. The value seem to be off by a factor -21.

A first point to test would be to measure the test current the meter uses for the tests with a 2nd working meter.

Another possible test would be to measure the "2 wire resistance" of a 1.5 V battery in the 2 M range. This should show how the measured voltage is converted to the result.

[jchw4]

A first point to test would be to measure the test current the meter uses for the tests with a 2nd working meter.

121GW displays 10uA in 100k and 1M ranges, and 0.7uA in 100M range. It matches k2000 specs.

Another possible test would be to measure the "2 wire resistance" of a 1.5 V battery in the 2 M range. This should show how the measured voltage is converted to the result.

Measuring Ohms of 1.00000V  source (Ronan X85 calibrator that I posted):

Code: [Select]

10MOhm: 1.63400 M
1MOhm: Overflow
100k: 99.1294 K

Increasing voltage actually makes ohms read more negative. -1.000098 MOhm matches 0.470060V on the input.

[Kleinstein]

This really looks like a bad calibration constant.  AFAIK the Keithley meters allow to calibrate individual ranges. So one may be able to get at least a somewhat working state back with cal ran for the 2M range.

[jchw4]

This really looks like a bad calibration constant.  AFAIK the Keithley meters allow to calibrate individual ranges. So one may be able to get at least a somewhat working state back with cal ran for the 2M range.

And I guess that EEPROM needs to be replaced ASAP.

[Kleinstein]

It is not sure that this is really a damage to eeprom. It could as well be just a soft error, that would be corrected with new written values.
Damaged EEPROMs are relatively rare.

The wrong values could be by an unintended call of a memory write - possibly by the former owner, or a general computer problem like a brown out. In this case even check sums would not help.

Ideally I would expect quite some error correction / checksums, so that corrupted cal data would be rate. The memory chip should have several times the capacity to allow redundant storrage.
At least a good check sum should be there to give a warning.
With computers ideal and real implementation are 2 things.