Agilent 3458A repair

[franklin]

Hi. This is my first time on this forum so I may be making some mistakes.

I just got an Agilent 3458A very cheap and I knew it had an error. The unit is from 2013. After switch-on this happens:

Switch-on: ERROR 110   CALIBRATION REQUIRED -- ACAL

ACAL:   ERROR 205   HARDWARE FAILURE, TEST VALUE OUT OF RANGE: 108
           ERROR 210   CALIBRATION REQUIRED
           ERROR 101   CALIBRATION ERROR

I left the meter ON for a few hours with a 10 VDC reference connected and it showed approximately 7.8 V. At some instance it started to display correct voltage (10 V) which lasted for about 30 minutes when the display reverted to 7.8 V. Reversing input leads to -10 V displays -9,25 V.

I’ve checked the most obvious, the voltages to PCB A1 and they are +5, -18.5, +18.3, and -9.9 VDC. I’m thinking the error is caused by A1 (A1 manufactured 2013) after reading TIN’s work but I’m very much in need for advice where to start. I have only a DVM as tool to help. DCI and OHMS are not working properly, that is the values are not right. I believe there must be a common denominator. By the way, meter temperature is 7 degrees; obviously way off.

I’d be grateful if someone could put me on the right track.
Thanks

[TiN]

What about self-test?

[franklin]

The selftest at switch-on displays this:    

TESTING RAM
ACAL DCV, OHMS R
TESTING HARDWARE
ERR is on, ERR 110, “CALIBRATION REQUIRED – ACAL”

[Kleinstein]

There is also a dedicated self test to be called. This is more extensive than the power on test.
Expect a few more error messages than from the failed ACAL .

[franklin]

Thanks for the answer. I don't think I've come across this dedicated self-test in the manuals; I may have overlooked that. Is there a procedure for this self-test?

[coromonadalix]

check this youtube done by " the signal path "......  very good infos

[franklin]

I think I misunderstood the thing about SELFTEST. Hopefully I’ve done this right this time. The self-test goes like this:
ROM TEST
RAM TEST
SLAVE TEST
TEST DCV 10V (duration about 3-4 seconds)
SELF TEST FAILED

ERROR 205, Hardware Failure, Test Value Out of Range: 108

I’ve downloaded the CLIP but the copy is not very clear and the writing is rather small. I’m not at all sure where to start. I’m guessing PCB A1 but don’t have a clue where. For those that may be interested in pictures I’ve enclosed images of the boards. The DMM is as far as I can see just like new both outside and inside. There was no dust on the filter or the fan blades so I’m wondering has it ever been used? I’ve not discovered any scratches or other signs of mishandling so I’d say it’s in “as new” condition. The display is bright and for those interested, I’ve swapped the snaphat batteries on A5 from 2001 (2,88 V) to 2019 (3,40 V) just to be safe. See images.

The last calibration was in 2015 so I suppose it has been in service at some time and in an environment with little dust. When the DMM is fixed (if it’s possible) I would end up with a “brand new” 3458A. But for now, I don’t know where to look for any problem components.

[Dr. Frank]

There is a button on the front panel, labeled TEST.
Please press the blue button and then this TEST button for extended test.

Have you ever attempted to perform an ACAL ALL?

PS: seems to be an intermittent error, maybe a temperature fault of a component.
Try different NPLC settings, especially short aperture like 10..100µs, if this problem persists. Also check 10V in 100V range, and lower anges (1V, 100mV).
You may track the DC signal (10V) through input circuit to the multiplexer, DCV low voltage input (sheet 4/5), then beyond the MUX towards the input amplifier sheet 3/5, TP100, and further on to ADC board.. there exist two differnt paths, a slow, and a fast one, depending on aperture.
Do Ohm mode and ACV sync modes work?


PPS: Sorry that I stitched that schematic so badly, years ago, but there was nothing better available.
You can zoom the pdf, though.
 
Frank

[franklin]

I’ve just tried the extended test. Here’s the result:
ROM TEST
RAM TEST
SLAVE TEST
TEST DCV 10V (duration about 3-4 seconds)
SELF TEST FAILED
ERROR 205, Hardware Failure, Test Value Out of Range: 108

ACAL ALL
ERROR 205, Hardware Failure, Test Value Out of Range: 108
ERROR 210, Calibration required
ERROR 101, Calibration error

DCV
1 VDC
Range AUTO, displ. 068…
Range 1 V, displ. 068…
Range 10 V, displ. 068…
Range 100 V, displ. 1.0..

10 VDC
Range AUTO, displ. 7.77….
Range 10 V, displ. 7.77….
Range 100 V, displ. 10.000 00..
Range 1000 V, displ. 10.000 00..


DCI
Seems to work fine (disregard the March 7 text)

ACV
12 VAC (the only voltage available) works fine

OHMS
Range AUTO, does not displ. OV.LD (with no input), displ. stuck on 21 kOhm
Range, manual, does displ. OV.LD (with no input)
Does not measure correct values of resistors

I will track the DC signal as you suggested later on.

About the CLIP: I’m actually not complaining just stating facts as I see it (my vision is not the best) and it’s the only CLIP I’ve found so I thank you for that.
 

[Dr. Frank]

So obviously you are already tracking down the fault.. it's very close.
It will be at / before the MUX, i.e. problem with the driver or the FET of the 0.1 / 1 / 10V channel, which is used for Ohm also.

DCI, ACV, 100V/1kV are all using different ones, as you can see for yourself on the schematic.


PS: I forgot the smileys  when I mentioned the stitching of the schematics.. no problem whatsoever, I'm always happy  when it's of use.

I have to drive right now to get some cheap gasoline, and I'll be back later for further analysis.

Frank

[Dr. Frank]

Please check first, if DCV 10V reads something useful, when AZ OFF.
Then check, if in DCV 10V mode, the input draws excessive current, should be about -2pA only.
Then check if -14.7 and +14.7V of the input protection circuit is ok.

Apply U_in =10V at input in DCV 10V mode.
If AZ ON, then use a scope to check U11A, its steering voltage pin5, should be 0 or +5V, then FET driving voltage pin 2, should be -20V  or about U_in, everything  in comparison to U8C in 100V range, which seems to be ok. If AZ Off, you should measure steady values with a DMM there, i.e. +5V for pin5, about U_in for pin2.

I suspect, that either the driver U4 is faulty, (temperature effect) or CR10 (?), this CC diode.

Maybe there was an overload which damaged the input protection, therefore the check of these +/- 14.7V.

Frank

[franklin]

Sorry about the delay on my part. Just recovered the PC after a crash.

AUTOZERO OFF.
Instrument displays 7.8 V (+10 V input to instrument).
Excessive current: I can measure down to 10 nA but not 2 pA. However, it seems ok since there is no obvious change in the measurement when connected or not.

+14.7 V and -14.7 V are both measured to +15,0 V and -14,9 V. It seems there are two zeners, CR6 and CR8 (not quite sure about that) but I don’t know if this increased voltage is significant (maybe not).
At U11A, pin 5 is +5 V and at pin 2 it’s 7.8 V (same as display; +10 V input to instrument).
When in 100 V range, U8C pin 9: +5 V and pin 14: 0 V
When in 10 V range, U8C pin 9: 0 V and pin 14: -20 V.

I’ve located U4 (16-pin right?) and CR10 on the print but not in the schematic. I guess that there is a system/logic in the schematic which I’ve not yet fully recognized. What would I expect to measure at U4 and CR10?

[Dr. Frank]

Hello franklin!

This pA measurement is done as follows:

put a short on the input jacks (short clamp or cable)
Set DCV, 10V (fixed range), AZERO OFF
FIXEDZ OFF (default)  => 3458A has GOhm input imepdance by default
remove short and watch the display, reading should go slowly downwards, at a rate of about -0.01V/sec. The input bias is about I_b ~ 200pF x dU / dt

explanation: The input has a capacitance of 2 x82pF (C22, C23), plus some pF from the PCB, so about 200pF. The bias / leakage of the input MUX / amplifier @ AZERO OFF will charge this capacitance with typically -2pA  and so you can easily estimate this bias by observing the rate of the voltage.

If this input path had been damaged, i.e. Q26, Q27, or Q13, you would probably see an increased rate, maybe also of another polarity.

These -14.9 and + 15.0V are fine.


Another possible failure mode is the breakdown over temperature of one of these CD4094 shift registers.
These steer all kind of FET switches, and my unit, which was then 9 years old, had a similar failure message as yours, but only over 45°C inner temperature (on a hot summer day), because of a defective 4094.


U4, pin 11 directly steers the comparator, pin 5, and so switches FET Q13 on and off.
These signals seems to be ok, I assume you have measured with a DMM and AZERO OFF.

If AZERO ON, Q13 is continuosly switched on and off, as Q17 will make alternatively zero measurements.

When in 100V range, with 10V applied, you should measure about 0.1V at pin 14, because this is the 10V divided by 100. Sorry to have described that incorrectly.
-20V  at pin 9 is fine, you should see also -20V at U11A, pin 2, when in 100V range.

CR10 is this strange component with an arrow inside a circle, sitting directly at a the output of U11A, and serving as a fast pull up current source for the OC of U11As output. 

The voltages seems to be ok.

So now please apply 10V in DCV 10V range, and measure the voltage at label "DCV", or at this input side of R17, with another DMM. There you should see these 10V.
Please measure  also behind R18, i.e. at the two points indicated in the schematic by the green arrows.
Watch the display of the 3458A also, when you measure at these two points.

If that reading is not ok, then we will have to inspect the input relays circuit, especially K3, which feeds the input voltage to R17. These relays also sometimes fail.

Frank

[franklin]

Hi Frank

I did find the CR10 finally; I was just looking for the symbol for a CR and to start with, didn’t recognize this as a diode.
pA measuremenet.
I might be doing something wrong here since there is no -0,01V/s change; it’s about -7 µV in 15 minutes.

Now, the DCV input was connected to 10V (see schematic); 8.89V after R17 (display shows 7.79V and then drops 40 mV when measuring); 7.79V after R18 (display shows 7.79V and then drops 70 mV when measuring). About R18, can’t see the designation on the PCB but it’s adjacent to R17 so I guess I’m right. You will of course notice that the voltage 7.79V is the same as the display. Looks like there is an excessive current draw of about 220 µA. Should I suspect Q13 or Q26/Q27 for pulling down voltage?

How come that you are so familiar with this instrument and have all this experience if I may ask such a question?

[Kleinstein]

All 3 paths are possible: Q26, Q27 and Q13. As a first test one could measure the gate current at Q13, though this is only one relatively unlikely path.
The fault is not necessary with one of the fets, the actual fault could be bind Q13 too, so the current drawn somewhere there.

220 µA are enough that this should be visible as a voltage drop across Q13.  The on resistance should be in the 20-80 Ohms range - so a measurable voltage if the current flows there.

[Dr. Frank]

Hello franklin,
something is drawing the input towards 0V, not towards these +14.7V, neither -14.7V.
As the input is @7µV after 15min, that's another indicator for leakage to ground.

It's quite improbable that this happens behind Q13, because the 100V range works correctly, and there is a 9.9MOhm resistor in series with the input, instead of only 10kOhm.

You could at first check, if Q13 is leaky, by measuring the voltage across its gate resistor R13.
220µA * 2k = 0.44V should be well visible.
The idea to measure across Q13 from S to D is also fine, but I don't expect anything here.

You can do another test, to rule out that the fault is behind Q13: What happens in current mode DCI, if you connect nothing and switch from highest to lowest range, i.e. 1A down to 100nA?

Is there an overflow, or 220µA, or simply near zero reading?

If all that gives no result, then try something odd:
The only components which could draw the voltage exactly to GND are these 82pF capacitors C22 and C23. From the direction of the current flow, C23 would be suspect.

multi layer ceramic capacitors can be mechanically damaged => cracks, maybe also by over voltage, and then they might also have high leakage currents.
So what happens, applying 10V again,  if you tap gently on C23, or cool it down a bit with ice spray? 


PS: Well I've got an education as an electronic technician in a German Air Force calibration lab 40 years ago (already  ) where I repaired and calibrated hundreds of all kind of multimeter gear, mostly hp.
That habit I kept at university, where we also had a warehouse full of old, abandoned instruments, which I repaired plenty. So I learnt a lot especially from the old hp schematics.

I also bought 'my' first hp3458A in 1989 for my PhD studies, but I did not have the CLIP at that time.
When I finally got my really own 3458A in 2009, I dug deeply down into the CLIP at that time, and  also thanks to TiN, I think I have now understood most of the schematics of the 3458A.

Same story goes for the 34401A, which I also bought in 1989 for the university institute, and also another instrument myself, as a reward for my diploma. That 30 years old instrument still works fine, and I was able to help others to repair their 34401A remotely.

Frank
 

[franklin]

Hi.

Q13: there is a 11 mV drop between S and D if I have measured correctly (AUTOZERO OFF, 10V range and 10V input.

Q13 gate resistor R 13: drop about 6 µV

Current mode DC, 2358A displays: 1A range: 30 µA;    100 mA range: 3 µA;    10 mA range: 0,5 µA;    1 mA range: 40 nA;    100 µA range: 12 nA;    10 µA range: 1 nA;    1 µA range: 0,1 nA;    100 nA range: 30 pA   

Ceramic capacitors, 10 V applied, 10 V range: tapping or ice-spraying C 23 does not affect voltage at R 17 and R 18

[Dr. Frank]

So it's necessary to search for one of the FET switches not working properly.
Please check all comparators, if they receive the correct steering signals , and if they all produce these -20 / + U_boot or GND levels for driving their respective FETs, i.e. U7A-D, U8A-D, U20A-D, U11C-D.

I don't see any other possibilities, where this current should be sunk.

Frank

[Kleinstein]

11 mV over Q13 and some 270 µA would be some 40 Ohms - about right for JFET switch like 2N4392.
6 µV over R13 is too little to explain the current.

So the problem is likely to the right of Q13.
A point to check would be TP10, the voltage to control most of the FET gates.
Q14 would be possible candidate for failure, as it is relatively exposed to the outside.
A possible failure mechanism could be JFET gate at something like Q14 to  Q15 to break down at some 28 V (- 20 V at the gate and +8 at the input).

Checking the current ranges is a good idea, however this is with a very small voltage. The problem seems to be more like with higher voltage only.  This could also explain why 10 V in the 100 V range is working. This would lead to some 0.1 V behind Q13.
So I see nothing suspicious from the current readings.

For a test one could also check if -10 V work.

[franklin]

I tested with reversed polarity (-10V) and to sum it up, display on 3458A shows:
-10 input: -9.2….V   .      +10V input:7.8….V.
So, there is a difference but I can’t see why it should be.

For U7A-D, U8A-D, U20A-D and U11C-D I’ve made the measurements and hopefully got it right. For the values, see a copy of schematic 4 (of 5) enclosed.

[Kleinstein]

The signal at U8A looks suspicious. This should be more like -20 V at the output.  So there is a good chance that U8 is broken - which is good as this is a pretty standard part. One could douple check with a scope or the DMM in AC mode to see if the odd reading is some AC effect, but it does not look like, as the 7.8 V are the expected on level.

With more current flowing the JFET switches may not be fully symmetric anymore. This can lead to the asymmetric error.
Also the broken LM399 may still be able to sink a limited amount of current and thus still have some effect with a -10 V input.

[Dr. Frank]

To me, U8A is strange, these 0.6V instead of exactly zero, as well as U7D, that it has 5V, as I would expect that it should be zero, but I don't know, why the 1V output, divided from the temperature sensor is used during DCV measurement.

So their respective steering Shift register U9 could be faulty also.

Do you see any activity or oscillations on the respective outputs of U9?
Could be, tat you measure these voltages only due to some multiplexing going on, despite AZERO OFF (I assume, that you measured in that state).

If it wouldn't be that big, sensitive PCB, I'd say at this point, just replace the 4094 and the comparator chips.

Maybe you can do very carefully another test, if the 3458A then reads 10V.
 As these comparators are OC, it's safe to short pin2 of U8 to pin 12 (-21V), to shut off  Q15, and to short pin13 of U7 to its pin 12 (-18V) to shut off Q21.

Frank

[franklin]

Thanks for all the info. I have a question about CR10 and BOOT; I’m not aware of what BOOT imply here but guessing it might be a “sort of”  ground. The reason for asking is if I understand schematics page 4 correctly, CR10 cathode is connected to A11 pin 2, +7.88 V in my case. There is only a few mV drop over CR10 so I’m thinking it’s not OK. What do you think?

[Dr. Frank]

Hello franklin,
there's a Guard Ring around many of the different FET switches and resistors, i.e. that whole dotted line as indicated, it's usually used for cancelling any leakage currents on the PCB.
This guard is held at the signal voltage  behind the MUX chain by  amplifier Q7A, Q7B, and U12.
I suppose that hp therefore called this 'boot' like bootstrapping.
So it's a virtual voltage follower of the measurement signals, but you can draw some current out of it, because of U12, in contrast to OpAmp virtual grounds (i.e. at - input in inverting configuration, or + input of non-inverting configuration) where you shall not draw any current.

I can not decide, whether CR10 is ok, or not, out of your observation. Reason is again, that its anode is connected to this virtual boot potential, therefore latter just follows the output of FET Q13, so the cat bites its tail. As this FET seems to be ok, a defective CR10 would either fully shut off this FET, or would fully open it, and that seems not to be the case.
Frank

   

[Kleinstein]

As far as I understand it CR10 is a constant current diode, so actually a selected JFET with gate connected to source. So essentially no voltage drop when the switch is on is correct.

I support the idea of using a wire bridge at U8 to bring the gate of Q15 to -20 V. One may have to use a switch for this, as the turn on self test could need Q15 to be turned on.