Please help, FLUKE 8920A Multimeter Repair

[Harry_22]

No just touch it by hot iron through termal paste.

[daisizhou]

Does it operate as shown in the picture below?To put it simply, the high-temperature heat is transferred through the soldering iron head to the C18 shell.
Then observe the changes in the 8920a display?

[Harry_22]

This is not C18. C18 is a smaller one. Don't overheat it.

[daisizhou]

OK.I will test the capacitor with the arrow pointing to it using your recommended test method.

I am doing this test while waiting for the 8920a to display "700" flashing, or "mv, v" flashing alternately.Because only in this way the changes will be more intuitive

Thanks

[Harry_22]

You can do this test now by monitoring the readings.
Could you film this.

[daisizhou]

I tested it, and it seems to have no effect.

It's night here, the light may not be very good.

Do I need to wait for the fault to recur before repeating this test?

By the way, I found that my 8920a host is missing this PCB. Is this normal?

[Harry_22]

Seems you have no ground connection of 8920.
Please do the following:
1. Connect metal case to the ground.
2. Connect oscilloscope to TP1
3. Push Low Range button
4. Switch on 8920
5. Check 8.6 volt at TP1
6. Touch C18 by hot soldering iron.
7. Observe oscilloscope

[daisizhou]

I took apart the electrical outlet panel from the wall,Found that it only has a "blue" neutral wire and a "red" live wire, and no yellow ground wire.So I don't know about your suggested metal shield case grounding, how should I connect it

Also, as suggested by "blackdog", replace these red tantalum capacitors,Because these tantalum capacitors will fail after long-term use.According to my experience, most tantalum capacitor failures are caused by breakdown and short circuit, and this is the first time I have encountered leakage current.
Do you think I need to replace them all?

[Harry_22]

There is a whole topic regarding how to do grounding correctly. All equipment requires grounding. You can use a nearby metal base or pipe.
I do not recommend replacing all capacitors. It is not for nothing that tantalum is used in medical equipment. Change only the critical 1uF in both amplifiers.

[daisizhou]

Ok, I'm going to spend some time looking for a real ground test point.

My original plan was to replace tantalum capacitors with 35V or 50V withstand voltage specifications, not electrolytic capacitors.I understand that tantalum capacitors have special functions.Because these red tantalum capacitors have been in operation for 20-30 years,Isn't replacing them in advance the best option?

Do you mean to just replace the C18 35V 1UF tantalum capacitor with a 35V1UF or 50V1UF tantalum capacitor?Other tantalum capacitors remain in their original state

[Harry_22]

Let you repair it first.
I would ask you to successfully complete C18 heat test to prove my Multicap simulation.

[daisizhou]

Ok, I understand.I will wait for the fault to reappear and connect a suitable ground test point

[daisizhou]

Hi Harry_22.

I would like to ask if a real ground test point is absolutely required?
I searched several locations,Due to the use of new materials, many pipes are currently made of non-metallic materials such as PVC, PPR....      There are also some old pipelines that have been severely corroded. I am not sure whether these pipelines are really connected to the earth.

I plan to test the 8920a again in 12-16 hours.Do you need the TP1, R40, R41, R42, and C24 test point voltages (measured by a digital multimeter)?
Also, I still use a soldering iron to bake C18 capacitors and take videos, right?

[Harry_22]

Hi! We need only TP1 to test amplifier A offset. It should be around +8.6 volt. After you try to heat C18 and observe oscilloscope. It is expected that the line goes to zero. Please film this.
Regarding grounding are you shure that there are no yellow-green connection anywhere  at your location. Try to find ground bar outside your building.

[daisizhou]

I re-tested, but unfortunately the fault did not recur.

I found a ground test point though, I'm not sure if that works.
If there is anything that needs improvement, please let me know. I'm still waiting for the fault to reappear and test it again.

[Harry_22]

Dear Daisizhou. With grounding the background noise has noticeably decreased.
It seems that your well-heated soldering iron scared off our malfunction and now it may not appear soon.
You can go another way. You remove C18 and measure its resistance with a multimeter. It is up to you.

[daisizhou]

OK。I plan to do it in 2 steps.
The first step is to buy a brand new 50V 1UF tantalum capacitor.
The second step is waiting for transportation,Since shipping takes a few days until new capacitors are received, I perform one last test to see if the fault can be reproduced and measure the resistance of the old capacitors.

Because it won't be a waste of time

[daisizhou]

Tested again today, the good news is that the fault reappeared, the bad news is that I haven't tested the heated C18 yet, the fault disappeared again, I took part of the video。
I seem to have discovered a pattern，That is, the longer the power outage lasts, the longer the failure will last.
It seems that it is not that the capacitance of C18 is reduced, but that there is a possibility of C18 leakage current.


Sometimes the fault of closing the shielding shell can be maintained forever,When you open the shielding cover or connect the TP1 test point, the fault will disappear automatically. It’s more serious on rainy days.Watch the video below.

[daisizhou]

I received the 50V 1UF capacitor and tested it for LCR.
Next I am going to disassemble C18 and replace it with a new 50v 1UF one

[daisizhou]

I replaced the C18 capacitor.
and measured the old C18 and it looked like it was normal

[Harry_22]

Please measure the leakage by multimeter.

[daisizhou]

Is this how it is measured?
I tested the multimeter wire resistance

[Harry_22]

Seems it's Ok.
Now you need to wait and check the further behavior of 8920a... wait and check. It's such a malfunction.

[daisizhou]

Ok, I'm going to keep testing this for a while.
Frankly,The old red tantalum capacitor has a better series equivalent resistance than the new yellow tantalum capacitor.6.197 ohm(YELLOW), 2.532 ohm(RED)

[Harry_22]

Don't be guided by this resistance. Not everything is so simple.
Please see the measurement of 1.5uF capacitor and 9.8k resistor in parallel.

PS
You had no bias on tantalum during measurements. May be this is a reason of high Rs.