Agilent 3458A repair

[franklin]

I see I’ve overlooked several things on the schematic; boot is one of them. I’m still learning (I hope). I understand that CR10 is a constant current diode (regulator?); to me it looked just like another diode although the symbol is different.

I think I don’t have the necessary overview to understand how U8A influences the measurement but are you suggesting I should FORCE the U15 gate to -20 V by a direct connection to -20 V?

[Dr. Frank]

Hello franklin

Well, I'm also still learning from your case.
Yes, CR10 is a constant current diode, or FET. As the 10V channel is able to digitize @ 100kHz, this boot has to follow as fast as this, so they needed a fast pull-up component also.
 
We still have not identified the failure, so at the moment it's simply looking for oddities and then trial & error.
To my understanding, U8A should not be engaged if you are using AZ OFF, and the level should be a straight zero volt. U7D is strange as well.

The outputs of the comparators are Open Collector, so it should be possible to force their output to the negative supply voltage.
Maybe you use a 1..10k resistor for safety reasons, and take care not to produce a short to other pins..

Do you have an oscilloscope to check the steering signals on all of the comparators and FET switches?

Frank

[Kleinstein]

When reading low DC volts Q15 / U8A should be set to off (that is some -20 V from U8A). The low voltage at the input (though not all the way to 0 ) also suggests that the 4094 sends the command to off. Seeing 7.8 V at the output suggests that U8A is actually not sinking significant current, as this is voltage read and likely the voltage from the BOOT buffer.
So the readings at U8A alone (inputs and output) highly sugggest that U8A is broken.

If in AZ mode Q15 would still be off. The AZ mode usually (maybe except for reading the internal reference) uses Q11 / Q12.

For the current ranges and 100/1000 V Q15 is on - so these ranges could work OK even if A8A is broken.

There is a gate resistor for Q15, so there is no real risk in typing the output of A8A to the -20 V supply of the chip. There may be a  extra error message on boot up though.  I doubt one would need a switch to enable the connection only after boot up. The test would be more to see if there are additional defects.

One could even do a test with just an extra LM339 just put directly on top of (in parallel)  the broken one. Chances are this could fix the error. With open collector outputs the 2 chips would not interfere with each other.  The comparator is only used as a kind of level shifter.

[franklin]

Hi.
Things have been moving VERY slowly here for a couple of days but I will try to catch up.

I’m sorry, I don’t have an oscilloscope and as of now I cannot borrow one either. I’m not aware of what to look for regarding these steering signals. As for the LM339, I don’t have one at the moment but will try to get one. First, I will try to connect Q15 gate to -20V through a 10 k resistor and see what happens.

Just applied -20V to U8A pin 2 (easier to access than Q15) and 3458A display jumped to 10V.

[Dr. Frank]

If you order parts, also buy some CD4094, as these are most suspicious, due to the probable temperature error you've once seen. You may also order appropriate precision sockets, because the 339 as well as the 4094 are not inside the analoge circuit, only on the digital side.

These specific strange voltages you see with your DMM may be caused by short 'normal' pulses, so an oscilloscope would tell you directly, what really happens.

That way, it's really very slow to analyse and track down the fault.
Frank

[franklin]

I’ll see what I can do about the scope. Just tried the -20V to Q15; applied -20V to U8A pin 2 (easier to access than Q15) and 3458A jumped to 10V. About the 4094, LM339 and sockets, I’ll get that as soon as possible.

[Dr. Frank]

Wow, first sign of the root cause! 
Please try the similar with U7D, i.e. connect its pin 12 to pin 13, I think...

[Kleinstein]

The voltage at U7D looks a little odd too, but more like not turned on all the way. The control signal from the 4094 is high, so the comparator should be off at the output. It is only that the voltage is only 3.8 V and not all the way to 7.8 / 10 V. The lower voltage could still be due to gate current towards the 1 V level and a relatively high resistance FET for Q22. So this could as well be OK if U8 works.

[franklin]

A7D (AUTOZERO OFF, RANGE 10VDC): Connected pin 12 (-18V) to pin 13: no reaction on 3458A display; should there be some reaction?

[Dr. Frank]

Not necessarily. Could have been..
This only excludes a fault in this comparator.
We're still looking for the sink of these 220uA, aren't we?
maybe another test, carefully cool down slightly the U8 and the shift register, whether you can reproduce this supposed temperature error
Frank

[Kleinstein]

The path for the 220 µA should be through Q15 (that should have been off) and Q22 (intentionally on) to the 1 V test-signal.
So it is possible that the comparator U8A is the only fault.
When ordering parts I would still get at least a 2 nd LM339, some 4094 and maybe replacement electrolytics for the supply.

With U8A still turing on Q15, I would have expected a change when Q22 is turned off. When Q15 is already forced off, there should be essentially no change (maybe some pA range change in the input current).

[Dr. Frank]

A7D (AUTOZERO OFF, RANGE 10VDC): Connected pin 12 (-18V) to pin 13: no reaction on 3458A display; should there be some reaction?

Maybe I have misunderstood your reply.

Have you also applied the 10V at the input, when doing this test?
Does the 3458A read these 10V correctly, when you connect pin 12 to pin 13? 

[franklin]

First, I must correct this: “A7D (AUTOZERO OFF, RANGE 10VDC): Connected pin 12 (-18V) to pin 13: no reaction on 3458A display”.
There IS a reaction and display show 10V! I must have been unconscious when doing this test.

Also cooled down U8 and U9 but only a fraction of mV change so I feel I can rule out the temp. problem. I guess what happened in the beginning was only a coincidence.

With a limited knowledge and overview of the 3458A, I’m starting to think this:
1.   There is no current leak through Q26/Q27 pulling the 10V down.
2.   The current (220 µA) must go through Q13 somehow. Something on the output of Q13 might be pulling voltage down.
3.   Applied -20V to U8A pin 2 again; BOOT (see CR10) raises to 10V plus some µV (haven’t tested that before) and U11A pin 2 raises to 10V.
4.   U8A pin 5 is 0.15V and U7D pin 13 is 3.8V; strange. I’m not at all aware of what voltages should be in this case.

3458A read these 10V correctly, when I connect pin 12 to pin 13

[Kleinstein]

Q13 is turned on as it is supposed to, so nothing unusual there.
Q15 is supposed to see -20 V at the gate to be turned off. However U8A is bad and Q15 is on instead.
Q22 is supposed to be on (control from U7D) and thus sends the 220 µA towards the 1 V auxiliary voltage. In addition with so much current, there is some gate current at Q22 and this brings the voltage at U7D to the slightly unusual 3.8 V level instead of some 7.8 V.
Under normal operation Q22 is not supposed to be on with much more than 1 V at the boot signal - so the extra gate current would not be there.

Applying the -20 V to U8A out / gate of Q15 turns of Q15 and things are OK (at least for low DCV).
Applying the -18 V to U7D out turns off Q22 and things look OK too, though having slightly more leakage currents.

So chance are high that U8 is the culprit and this could be the only fault.

[franklin]

Q22; do you mean Q21?

[Kleinstein]

Q22; do you mean Q21?

I meant Q21 - the picture with the voltage readings is not very legible.

[Dr. Frank]

I have drawn the failure current in the schematic.

Q15 and Q21 are both closed (active), so the input voltage appears across R8, which usually is the attenuated output of the temperature measurement sensor.

 Q21 therefore should definitely be open (turned off) for all modes, apart from TEMP?
I'll check this, and the sensor output voltage later.
You may check whether U9 causes the problem, by applying a 470 Ohm resistor from pin 4 (Q0) to its pin 8, GND. That should pull the comparator input below 2V, and shut off Q21, if the comparator U7D is ok.
That something is not correct with this channel, is supported by the fact that TEMP? delivers 7°C only. Maybe the 4094 is defective and scrambles the serial data, or has problems with certain configurations.

Q15 might also be off when AZERO OFF, but maybe that's the case only when the 3458A is put in a DIGITIZING state, to remove any unwanted capacitance.
This 0.15V input on U8A, if it's really a steady state, i.e. no short pulses, should also switch Q15 off, so I will check these levels also.

Frank

[Kleinstein]

If Q15 is off, here is no real problem of having Q21 on. It is a little odd as there would be gate current, but nothing dramatic. Q21 would give the MUX part below Q15 at least a defined voltage in stead of letting it float. So there may even be some though behind it.

Having Q15 during low voltage DC measurement would be not only for less capacitance, but also for less leakage. So it absolutely makes sense to have Q15 off.

The error in the temperature measurement is still a good point to check.

[Dr. Frank]

Surprise, surprise!

It's working the other way round:
Q15 is (always) off, separating the lower MUX part from the signal path.
AZ is obviously realized with Q23, Q24, as there is no difference with AZERO ON.
pin5 of U8A is exactly zero Volt, like all comparators U8A-D and U7A-C.
Then this modus operandi explains, why 100V and DCI are working correctly.

Q21 is always on, obviously its Gate diode is conducting in forward direction. U7D, pin 11 has 5.0V signal. The T-sensor U21 is NOT powered on its pin1, obviously only switched on briefly for TEMP?
Therefore R8 (13.5k) pulls the MUX bus towards zero, plus this odd level at, and current  through the Gate of Q21.

Please check accordingly with a 470 Ohm resistor between U9 pin 12, and GND on pin 8, whether the shift register is faulty, or U8A.

I can't imagine, what happens, if U8A draws excessive current out of output Q6 (pin12), whether this could also influence the level of its Q0, or what might be the failure mode of this shift register.

But I'm pretty sure, that either of both ICs is causing the trouble, very probably U9 U8 is it.

Frank

PS: Yes, in both DCI and 100V range, Q21 is turned off, i.e. 0V at U7D, Q15 is turned on, +5V at 8D, and either U7A / Q18, or U8C / Q16 is turned on, 5V at either comparator input.
So you can also check these voltages, whether U9 delivers these correctly.

[Kleinstein]

The 100/1000 V range could be a way to check the lower MUX with Q16-Q21.

Az switching in most cases should be with Q12 and Q11 alternating ON/OFF.  Q11 has most of the "negative sides" to choose from and Q12 is the switch with all the extra effort to keep charge injection / switching peaks small. It also has the extra precharge path with the BOOT signal.  So I would expect additional short pulses with Q10 on too.

[franklin]

U9:   470 Ohm resistor from pin 4 to pin 8 (U7D, p11);    3458A displays 10V
470 Ohm resistor from pin 12 to pin 8 (U8A, p5);    3458A displays 10V
Checked TEMP? for both cases above:     Still about 6-7 ⁰C

U21:   Reacts to cooling. Voltages: se enclosed schematic

U7A, p5 (to Q18): 0V
U8C, p9 (to Q16): 0V

By the way, haven’t been able to provide for a scope yet.

[BU508A]

How about renting one?

Example:
https://www.4test.no/Oscilloscopes/Economy+1000X-2000X/Oscilloscopes/

[franklin]

Not so easy; KORONA you see.

[Dr. Frank]

The signals seem to be static, so an oscilloscope is not required anymore.
The change of U21 is irrelevant at the moment, why only 7°C is measured, we will see later, when the MUX is working correctly again; seems to be a timing problem with the faulty U8A / U9, maybe.

This one "470 Ohm resistor from pin 12 to pin 8 (U8A, p5);    3458A displays 10V" I did not expect, as I supposed that U8A is defect and tries to pull U9 towards 5V.

Anyhow, I see no possibility anymore from further exterior measurements to decide whether U9 or U8 is defect.

Maybe if you measure these voltages, as requested  :
For DCI mode: U7A, p5 (to Q18): ?
For 100V DCV mode: U8C, p9 (to Q16): ?

So best would be if you replace both ICs, please.
Have you already ordered, and do you have appropriate desolder-station?

Frank

[BU508A]

@Frank

I have a question about these U7 / U8 (LM339): which one do you recommend?

Here is a list from Mouser:
https://www.mouser.de/Semiconductors/Amplifier-ICs/Analog-Comparators/_/N-cib1w?P=1z0z63x&Keyword=LM339&FS=True

Thanks.