FLUKE 8920A RMS AC meter failure needs help

Can anyone help me?

If "not using" time makes a difference then I first would check all electrolytic capacitors. Not with a DMM but desoldered using an LCR meter measuring C, D (tan d) and DC leakage.
If the oxide layer of the caps is not 100% bad and reform after some power-up time, they will fail permanent in the future.

This sort of intermitted failure can be very hard to find, for most because if the unit does not fail, you will not find a fault. Sometimes doing the calibration helps because if it is a small fault that is sometimes causing problems but not always, it often does show up during performance checks and/or calibration. I often found problems in scopes this way. But only do this if you have a sound knowledge about electronics and have all the recommended gear. If you then find a step that does not react like it should, watch the schematic to see what components are used for that step.

Op zaterdag 27 april 2024 08:15:51 schreef fred101:
If "not using" time makes a difference then I first would check all electrolytic capacitors. Not with a DMM but desoldered using an LCR meter measuring C, D (tan d) and DC leakage.
If the oxide layer of the caps is not 100% bad and reform after some power-up time, they will fail permanent in the future.

This sort of intermitted failure can be very hard to find, for most because if the unit does not fail, you will not find a fault. Sometimes doing the calibration helps because if it is a small fault that is sometimes causing problems but not always, it often does show up during performance checks and/or calibration. I often found problems in scopes this way. But only do this if you have a sound knowledge about electronics and have all the recommended gear. If you then find a step that does not react like it should, watch the schematic to see what components are used for that step.

Thank you for your reply.
I checked the filter capacitors corresponding to the +15v, +5v, and -15v power supply parts. C208?C209?C210?The C210 capacitor was a bit old, so I replaced it with 35v 680uf.C208 and C209 are normal after LCR testing. The D value is 0.2498-0.2289 and the capacitance value is 199-206uf.And I also measured the waveforms corresponding to +5v, +15v, and -15v.They all look normal.
But the fault still exists,Please watch these 2 videos
https://youtu.be/UmSWiNHtUiQ
https://youtu.be/E9kwZJmgdj4
I don't know where the TP3 pulse waveform comes from,At present, I am only stuck here at TP3 and have not continued to check. I am lost.

The signal at TP201 and 202 are to high. They are sense lines and come from U1. So I would first focus on U1 and measure around that IC (f.i. the output of U4a ) to find the source of the pulses that are leaking through (?) to TP201/202.

If the input on the sense lines is to high the input is to high and flashing 700 could mean that the input signal could be to high. What are you feeding the meter and in what range.

Op zaterdag 27 april 2024 11:13:19 schreef fred101:
The signal at TP201 and 202 are to high. They are sense lines and come from U1. So I would first focus on U1 and measure around that IC (f.i. the output of U4a ) to find the source of the pulses that are leaking through (?) to TP201/202.

If the input on the sense lines is to high the input is to high and flashing 700 could mean that the input signal could be to high. What are you feeding the meter and in what range.

8920a input port, I always connect a 50 ohm load resistance.
I followed the troubleshooting checklist and I checked TP201-TP202?Then I checked TP3.The waveform is as captured in the video above.
https://www.youtube.com/watch?v=UmSWiNHtUiQ
Following the checklist, next I need to check TP1.I have not yet tested the TP1 due to the need to open the metal shield cover.
I don't know if my checking direction is correct.

Op zaterdag 27 april 2024 11:13:19 schreef fred101:
The signal at TP201 and 202 are to high. They are sense lines and come from U1. So I would first focus on U1 and measure around that IC (f.i. the output of U4a ) to find the source of the pulses that are leaking through (?) to TP201/202.

If the input on the sense lines is to high the input is to high and flashing 700 could mean that the input signal could be to high. What are you feeding the meter and in what range.

This is a real picture of U1.
Do you mean you need to check the waveform diagram of TP5?

Op zaterdag 27 april 2024 08:15:51 schreef fred101:
If "not using" time makes a difference then I first would check all electrolytic capacitors. Not with a DMM but desoldered using an LCR meter measuring C, D (tan d) and DC leakage.
If the oxide layer of the caps is not 100% bad and reform after some power-up time, they will fail permanent in the future.

This sort of intermitted failure can be very hard to find, for most because if the unit does not fail, you will not find a fault. Sometimes doing the calibration helps because if it is a small fault that is sometimes causing problems but not always, it often does show up during performance checks and/or calibration. I often found problems in scopes this way. But only do this if you have a sound knowledge about electronics and have all the recommended gear. If you then find a step that does not react like it should, watch the schematic to see what components are used for that step.

I tested TP3, TP4, TP5 waveform video, please check
https://www.youtube.com/watch?v=rj4nOueJUaE

The "700" status TP3 waveform is displayed. Unfortunately, the TP4 and TP5 waveforms are not obtained.
https://www.youtube.com/watch?v=iDLoAJ05TK8

Normal state TP3 TP4 TP5 waveform diagram
https://www.youtube.com/watch?v=jfN7OpkL5sU

Op zaterdag 27 april 2024 11:13:19 schreef fred101:
The signal at TP201 and 202 are to high. They are sense lines and come from U1. So I would first focus on U1 and measure around that IC (f.i. the output of U4a ) to find the source of the pulses that are leaking through (?) to TP201/202.

If the input on the sense lines is to high the input is to high and flashing 700 could mean that the input signal could be to high. What are you feeding the meter and in what range.

I tested TP1 TP2 waveform,Please note that I did not connect a 50 ohm load.
Because testing TP1 and TP2 requires removing the shielding shell.
The shielding shell was removed, resulting in the load resistor being unable to connect to the access point.
I still haven't found the fault point. I look forward to your help. Thank you.
https://www.youtube.com/watch?v=Y3bslfXznCM

Op dinsdag 30 april 2024 15:54:35 schreef clowns:
[...]

I tested TP1 TP2 waveform,Please note that I did not connect a 50 ohm load.
Because testing TP1 and TP2 requires removing the shielding shell.
The shielding shell was removed, resulting in the load resistor being unable to connect to the access point.
I still haven't found the fault point. I look forward to your help. Thank you.
https://www.youtube.com/watch?v=Y3bslfXznCM

It looks like this triode is self-oscillating,Does anyone know how to adjust it?

It could be because the capacity of your probe. Do you have it on 10x ?

Just put a signal on the input and follow it through the meter. That is often a more easy way to find the problem or in what area it goes wrong.

Op woensdag 1 mei 2024 07:46:18 schreef fred101:
It could be because the capacity of your probe. Do you have it on 10x ?

Just put a signal on the input and follow it through the meter. That is often a more easy way to find the problem or in what area it goes wrong.

I connected a 50 ohm load resistor to the input,In the video you can see
https://www.youtube.com/watch?v=fvrIU8fIPfs
https://www.youtube.com/watch?v=3elmkGOjpp8

Looks like Q3 Q4 Q5 Q6 is normal,Amplifier A becomes the biggest suspect

Why do you ask help if you do not answer questions or try suggestions.

You constant talk about the 50 ohm input but I suggested to feed it with a signal. And I asked about the settings of your probe. That has nothing to do with the 50 ohm termination of the input. Here is why:

If you feed no signal you can not see where it goes wrong. Just follow it with your scope and 10x probe. A 1x probe causes to much load and the added capacitance can cause oscillations.
There will be a point in the signal path where it goes wrong and this way you can find it. Besides that, Fluke advises to inject a signal.

If I find a part that does not function like it should, I remove it and test it on a breadboard or try a known good replacement as a test. It will not say "i'm bad", so you should test it outside the meter to be 100% sure.

[Bericht gewijzigd door fred101 op woensdag 1 mei 2024 11:02:25 (15%)]

Fred,

Clowns is een Chinese clown die eerder op soortgelijke wijze op EEblog heeft zitten vissen.

Daar kwam hij, ondanks aangeboden hulp van Blackdog, niet verder.

Dus heeft hij nu hier z'n hengeltje uitgeworpen om verder te vissen

groet! Gertjan.

Op woensdag 1 mei 2024 10:59:05 schreef fred101:
Why do you ask help if you do not answer questions or try suggestions.

You constant talk about the 50 ohm input but I suggested to feed it with a signal. And I asked about the settings of your probe. That has nothing to do with the 50 ohm termination of the input. Here is why:
[bijlage]
If you feed no signal you can not see where it goes wrong. Just follow it with your scope and 10x probe. A 1x probe causes to much load and the added capacitance can cause oscillations.
There will be a point in the signal path where it goes wrong and this way you can find it. Besides that, Fluke advises to inject a signal.

If I find a part that does not function like it should, I remove it and test it on a breadboard or try a known good replacement as a test. It will not say "i'm bad", so you should test it outside the meter to be 100% sure.

Sorry, I forgot to specify, I have been using a 10X scope probe(Because 10x is the default setting).
I noticed that Fluke recommends using a 1v AC signal input.If you want me to input a 1v AC signal, I can follow your instructions, but I don't think it makes practical sense.Because I use a 50 ohm load resistor, the interference fluctuation still exists.it's obvious.Then I will input a 1V AC signal and shoot a new video.
Please forgive me, the fault does not always exist. I have been trying to reproduce the fault phenomenon, which takes time.

Sorry again

Op woensdag 1 mei 2024 11:30:54 schreef miedema:
Fred,

Clowns is een Chinese clown die eerder op soortgelijke wijze op EEblog heeft zitten vissen.

Daar kwam hij, ondanks aangeboden hulp van Blackdog, niet verder.

Dus heeft hij nu hier z'n hengeltje uitgeworpen om verder te vissen

groet! Gertjan.

Sorry, I've been trying to repair the 8920a.I'm a beginner
I'm sorry if it bothers you

Op woensdag 1 mei 2024 10:59:05 schreef fred101:
Why do you ask help if you do not answer questions or try suggestions.

You constant talk about the 50 ohm input but I suggested to feed it with a signal. And I asked about the settings of your probe. That has nothing to do with the 50 ohm termination of the input. Here is why:
[bijlage]
If you feed no signal you can not see where it goes wrong. Just follow it with your scope and 10x probe. A 1x probe causes to much load and the added capacitance can cause oscillations.
There will be a point in the signal path where it goes wrong and this way you can find it. Besides that, Fluke advises to inject a signal.

If I find a part that does not function like it should, I remove it and test it on a breadboard or try a known good replacement as a test. It will not say "i'm bad", so you should test it outside the meter to be 100% sure.

As per your suggestion I did the full troubleshooting checklist test and below is the video of the test.
I tested TP201 and TP3 waveforms,Please watch carefully, the second half of the video is when 8920a returns to normal state from fault state.
If you need TP1 and TP2 waveforms please let me know as it is inside the metal shield and requires lead output to test.
I think the fault seems to have nothing to do with the signal input,If you have any better suggestions please tell me
Thank you for your help
https://www.youtube.com/watch?v=i7wkUX8ld8Y

First, clean all those switches, many problems with measurement gear that use push switches are "repaired" by cleaning the switches.

The checklist states "Test the voltage between TP201 and 202" not "at 201 and 202". So use a differential probe or set your scoop up for differential measurement using 2 probes (see manual scope, or find a video about using your scope probes as a "poor mans differential probe") You now measured TP201 to ground.

The video is much to long, 2 minutes just to show your probe is nuts.

You should not skip 6 - 16 because the meter does react to the input. Not the correct reading but it does react.

In the second half it still reads the wrong voltage, at what time stamp is shows correct 1Vrms ?

Like said before, inject a voltage and follow it with your scope. First TP1 and then TP2 and then TP3.

Think about what you do and if what you see is correct, measuring without knowing/thinking is useless. Take the schematic, read about how it works and go step by step through it instead of mindless probing. This is not a beginner project, that is why almost nobody reacts, even on EEVBlog. Think about parking it until you know more about electronics, this way it will take like forever, I have hobbies too. I love to help but not if it takes me hours and I need repeating things. This is already complex enough if you know what you are doing

A repair like this can take a lot of time, measuring, thinking, checking parts or things, constant doing checks if you do things right. Desoldering parts to check them etc. If I measure TP1 and do not get a realistic result I measure everything in between the input and TP1 and if that does not lead to clearness I desolder the opamp and test it on a breadboard, test the resistors, check tantalums for DC leakage (like blackdog advised but you did not do), check semiconductors etc. Sometimes I copy part of a schematic to a sim to get values outside testpoints (Or calculate them by hand)

In case of intermitted faults like here, I stop measuring testpoints until I checked every solder joint and capacitor desoldered.

Talking someone through a repair, who does not know what he is doing and does not do what you ask is consuming to much time and is very frustrating.
TV repairs you can do from videos without having a clue, because the people who repair them a lot, know what goes wrong in certain models. For this meter you must know what you are doing, are able to read schematics and understand how things work at component level.

Hoop txt Fred in het kader van de nl China betrekkingen doe je goed werk.

Op donderdag 2 mei 2024 11:31:34 schreef fred101:
First, clean all those switches, many problems with measurement gear that use push switches are "repaired" by cleaning the switches.

The checklist states "Test the voltage between TP201 and 202" not "at 201 and 202". So use a differential probe or set your scoop up for differential measurement using 2 probes (see manual scope, or find a video about using your scope probes as a "poor mans differential probe") You now measured TP201 to ground.

The video is much to long, 2 minutes just to show your probe is nuts.

You should not skip 6 - 16 because the meter does react to the input. Not the correct reading but it does react.

In the second half it still reads the wrong voltage, at what time stamp is shows correct 1Vrms ?

Like said before, inject a voltage and follow it with your scope. First TP1 and then TP2 and then TP3.

Think about what you do and if what you see is correct, measuring without knowing/thinking is useless. Take the schematic, read about how it works and go step by step through it instead of mindless probing. This is not a beginner project, that is why almost nobody reacts, even on EEVBlog. Think about parking it until you know more about electronics, this way it will take like forever, I have hobbies too. I love to help but not if it takes me hours and I need repeating things. This is already complex enough if you know what you are doing

A repair like this can take a lot of time, measuring, thinking, checking parts or things, constant doing checks if you do things right. Desoldering parts to check them etc. If I measure TP1 and do not get a realistic result I measure everything in between the input and TP1 and if that does not lead to clearness I desolder the opamp and test it on a breadboard, test the resistors, check tantalums for DC leakage (like blackdog advised but you did not do), check semiconductors etc. Sometimes I copy part of a schematic to a sim to get values outside testpoints (Or calculate them by hand)

In case of intermitted faults like here, I stop measuring testpoints until I checked every solder joint and capacitor desoldered.

Talking someone through a repair, who does not know what he is doing and does not do what you ask is consuming to much time and is very frustrating.
TV repairs you can do from videos without having a clue, because the people who repair them a lot, know what goes wrong in certain models. For this meter you must know what you are doing, are able to read schematics and understand how things work at component level.

Thank you for your reply.
I have completed the most basic inspections, including: cleaning of mechanical switches, replacement inspection of electrolytic capacitors, and rough measurement of tantalum capacitors.It looks like they are not malfunctioning

About TP201, TP202 voltage:The TP202 test point and GND are directly connected,GND here I refer to the shielding shell.In other words, I measure TP202 and metal shield case, their resistive zero ohms.That’s why I chose a metal shielding case

6-16: Do you mean (Low-Midband check, Low level DC check, High Frequency Response check,...TP206+15V, TP207+5V, TP208-15V) these checks?If so, I don't know what it looks like for them to work properly

Shows correct 1Vrms:In the oscilloscope, when you see that the pulse waveform disappears, then the 8920 host is normal.I never knew where the pulse waveform came from.I'm guessing it's coming from amplifier A?similar to self-oscillation,It looks like an amplifier circuit,This is just my guess

The fault occurs when the machine is turned on for the first time after being shut down for a long time.In other words, if after it returns to normal, I turn it on or off again, it will be normal and the measurements will be accurate.
So I'm very confused,I don't know where to check,So I’ve been asking everyone for help

Op donderdag 2 mei 2024 11:31:34 schreef fred101:
First, clean all those switches, many problems with measurement gear that use push switches are "repaired" by cleaning the switches.

The checklist states "Test the voltage between TP201 and 202" not "at 201 and 202". So use a differential probe or set your scoop up for differential measurement using 2 probes (see manual scope, or find a video about using your scope probes as a "poor mans differential probe") You now measured TP201 to ground.

The video is much to long, 2 minutes just to show your probe is nuts.

You should not skip 6 - 16 because the meter does react to the input. Not the correct reading but it does react.

In the second half it still reads the wrong voltage, at what time stamp is shows correct 1Vrms ?

Like said before, inject a voltage and follow it with your scope. First TP1 and then TP2 and then TP3.

Think about what you do and if what you see is correct, measuring without knowing/thinking is useless. Take the schematic, read about how it works and go step by step through it instead of mindless probing. This is not a beginner project, that is why almost nobody reacts, even on EEVBlog. Think about parking it until you know more about electronics, this way it will take like forever, I have hobbies too. I love to help but not if it takes me hours and I need repeating things. This is already complex enough if you know what you are doing

A repair like this can take a lot of time, measuring, thinking, checking parts or things, constant doing checks if you do things right. Desoldering parts to check them etc. If I measure TP1 and do not get a realistic result I measure everything in between the input and TP1 and if that does not lead to clearness I desolder the opamp and test it on a breadboard, test the resistors, check tantalums for DC leakage (like blackdog advised but you did not do), check semiconductors etc. Sometimes I copy part of a schematic to a sim to get values outside testpoints (Or calculate them by hand)

In case of intermitted faults like here, I stop measuring testpoints until I checked every solder joint and capacitor desoldered.

Talking someone through a repair, who does not know what he is doing and does not do what you ask is consuming to much time and is very frustrating.
TV repairs you can do from videos without having a clue, because the people who repair them a lot, know what goes wrong in certain models. For this meter you must know what you are doing, are able to read schematics and understand how things work at component level.

I found that it seems to be caused by the load resistor.
The 8920a always fails when I turn it on after connecting it with a 50 ohm load resistor,When I disconnect the 50 ohm load resistor, the fault no longer occurs.Here are my recent findings and I will continue to test.
Magical malfunction phenomenon.I took pictures of the actual load and drew a simple schematic diagram

I read the 8920a user manual, and it does not prohibit starting with a load resistor.At least the manual doesn't specifically emphasize that this is wrong.