Keithley DMM6500 repair

[E-Design]

Alright, seems you have good connection up to the shunts and previously, you measured each of the shunts individually as correct so lets check from the bottom of shunts back out.

So, measure the 5 parallel (R225 to R229) from the side not connected to R230 measure that node out to LO terminal. It should look low like a low value again (it just contains the 0.1 ohm shunt and low resistance wires, traces.)

See what that shows us.

I measured the resistance from the parallel resistors to the LO input terminal and got a value of 0.123 as expected.

OK as a brief summary..
1) you have removed all parts suspected of having damage, and have measured expected values from each of the shunts- though discolored / scratched, they are measuring good.
2) Relays appear not damaged, Diode bridge seems not damaged (from previous) -
3) With the temporary wire bridge forcing all shunts in the path (10K ohms), you measure  AMPS to shunts then shunts to LO seems correct. Not sure why AMPS to LO doesn't see the sum of those 2 measurements.

At this point, there isnt much else left in the circuit (CR20 and CR6), you could keep hunting for damage
OR
you could start putting parts back in (known good) - Starting with a new U52 or suitable stand-in.

[drummerdimitri]

Alright, seems you have good connection up to the shunts and previously, you measured each of the shunts individually as correct so lets check from the bottom of shunts back out.

So, measure the 5 parallel (R225 to R229) from the side not connected to R230 measure that node out to LO terminal. It should look low like a low value again (it just contains the 0.1 ohm shunt and low resistance wires, traces.)

See what that shows us.

I measured the resistance from the parallel resistors to the LO input terminal and got a value of 0.123 as expected.

OK as a brief summary..
1) you have removed all parts suspected of having damage, and have measured expected values from each of the shunts- though discolored / scratched, they are measuring good.
2) Relays appear not damaged, Diode bridge seems not damaged (from previous) -
3) With the temporary wire bridge forcing all shunts in the path (10K ohms), you measure  AMPS to shunts then shunts to LO seems correct. Not sure why AMPS to LO doesn't see the sum of those 2 measurements.

At this point, there isnt much else left in the circuit (CR20 and CR6), you could keep hunting for damage
OR
you could start putting parts back in (known good) - Starting with a new U52 or suitable stand-in.

CR6 has a voltage drop of 0.520 and OL in reverse polarity so it should be fine.

CR20 I assume is a 4A full bridge rectifier? I removed it from the circuit to test it under load to make sure it's in working order.

Couldn't find the datasheet for this model. It is made by ON semiconductors but the QJ16 model number is not showing up on the web.

Is it safe to assume this will work on 220 V AC for testing purposes?

[Kleinstein]

Testing the rectifier does not need hight current / high voltage. It is more about having all the diodes present (diode test from the DMM) and leakage current.

For the resistance measurements, there could be interference for some bootstrapping of the protection. I am not sure of the DMM6500 uses this, but it is common in higher grade meters. So there is a high Z buffer  buffer the voltage at one side of the rectifier and than apply it through 2 resistors to the + and - pins of the rectifier. If this amplifier is broken, it can interfere with the lower current ranges.  (the good thing is this rectifier would usually be a not so critical, special type). One would notice such an error a voltage (e.g. up to +-0.5 V, less in the higher ranges) present between the amps terminals.

Reading too high a resistance from the amps terminals could indicate a broken trace. Is the bridge really at the correct position ? It should be connected to a shunt and a input terminal (likely through the fuse and maybe a relay).

[drummerdimitri]

Testing the rectifier does not need hight current / high voltage. It is more about having all the diodes present (diode test from the DMM) and leakage current.

For the resistance measurements, there could be interference for some bootstrapping of the protection. I am not sure of the DMM6500 uses this, but it is common in higher grade meters. So there is a high Z buffer  buffer the voltage at one side of the rectifier and than apply it through 2 resistors to the + and - pins of the rectifier. If this amplifier is broken, it can interfere with the lower current ranges.  (the good thing is this rectifier would usually be a not so critical, special type). One would notice such an error a voltage (e.g. up to +-0.5 V, less in the higher ranges) present between the amps terminals.

Reading too high a resistance from the amps terminals could indicate a broken trace. Is the bridge really at the correct position ? It should be connected to a shunt and a input terminal (likely through the fuse and maybe a relay).

In that case the rectifier is working fine out of circuit.

Maybe it would be helpful so probe the AC and DC side of the rectifier in circuit?

[Kleinstein]

In circuit the 2 DC side pins of the rectifier are connected with the diode and likely 2 resistors for the bootstrapping. with nothing connected to the inputs the voltage should be close (e.g. less than +-5 mV) to the COM terminal. If there is no bootstrapping the DC side would be kind of semi-floating only connected through the diodes in the rectifier.

The AC side of the rectifier should go to the terminals, with the fuse in between at one side.

[E-Design]

Yes, leakage is generally the issue. The 65xx does guard out that leakage provided its not excessive (from damage)
The condition of the instrument right now has U52 removed which is what drives the guard. So in circuit right now, it wont be guarded.
Thats probably ok for now while we complete finding all the damaged devices.

I am not 100% sure if that bridge could survive a direct connect to 220V, but external wiring impedance and the rather fast fuse opening time, it very well might have survived it.

The bridge is a GBU4A
http://www.vishay.com/docs/88614/gbu4a.pdf

[drummerdimitri]

Yes, leakage is generally the issue. The 65xx does guard out that leakage provided its not excessive (from damage)
The condition of the instrument right now has U52 removed which is what drives the guard. So in circuit right now, it wont be guarded.
Thats probably ok for now while we complete finding all the damaged devices.

I am not 100% sure if that bridge could survive a direct connect to 220V, but external wiring impedance and the rather fast fuse opening time, it very well might have survived it.

The bridge is a GBU4A
http://www.vishay.com/docs/88614/gbu4a.pdf

So do I need to replace U52, U42 and Q70 before we proceed with the troubleshooting?

[Kleinstein]

If the bridge rectifier still has low leakage it should have survived the surge. If at all the lowest current range would suffer from a little leakage.

For Q70 one could test is outside - chances are high it is Ok anyway. I see currently no need to put it back in. It is likely a JFET similar to J107 and likely used to the switch for the 10 mA range.

One could try to check U52 outside the board and if available use a replacement. It depends on the circuit how critical U52 performance is. If just for the guard and maybe bootstrapping the diode bridge it would be a rather non critical OP, one could as well replace. What type is U52 ?   This could give a hint about the circuit used.

For the start I would leave out U42 and keep the bridge to kind of manually force the 10 µA range if I understand it right.

The sometimes negative resistance readings in some cases earlier point to some leakage current from the active circuit, like from U52 or from the guard trace.  If U52 and the rectifier removed I would expect the guard trace around U42 to be now floating, with no more connection to the main circuit.

[E-Design]

So, if we are content with no leakage or bridges, then you can proceed with re-populating.
You have also checked the Zener CR6 for leakage?

When re-populating:
U42 - you should replace with new one - this is likely damaged.
U52 - you should replace with new one - it could be damaged. Suggestion to test it outside circuit is ok, but I am concerned  might be difficult for you. The damage can be small input current leakage. You can build a low leakage setup and get the right equipment, but it may just be less hassle to replace it.

Q70 - I agree can leave out for now, it just a range switch.
When U42 and U52 are back in, it will be enough to tell you if you have found all the damage. Remember, both U42 and U52 can have suitable stand in parts. What do you have on hand?

After re-population, give the area a good cleaning with proper solvents to get rid of all that flux residue and contamination.

[drummerdimitri]

So, if we are content with no leakage or bridges, then you can proceed with re-populating.
You have also checked the Zener CR6 for leakage?

When re-populating:
U42 - you should replace with new one - this is likely damaged.
U52 - you should replace with new one - it could be damaged. Suggestion to test it outside circuit is ok, but I am concerned  might be difficult for you. The damage can be small input current leakage. You can build a low leakage setup and get the right equipment, but it may just be less hassle to replace it.

Q70 - I agree can leave out for now, it just a range switch.
When U42 and U52 are back in, it will be enough to tell you if you have found all the damage. Remember, both U42 and U52 can have suitable stand in parts. What do you have on hand?

After re-population, give the area a good cleaning with proper solvents to get rid of all that flux residue and contamination.

When you say leakage, what exactly do you mean and how do I check for this?

I always test components out of circuit to make sure they are working properly.

The parts are not available locally so I will need to buy them online. Can you recommend a store that provides cheap shipping?

Also, what is the exact part number for that amplifier and Q70 component so I can find a suitable replacement for them?

Thanks.

[coromonadalix]

A leakage could be :

Partially failed parts  : ic's,  transistors, capacitors   

Surface conaminants like : solder residues, solder resins, solder flux

In multimeters, oscilloscope, rf equipemnts  and many other sensitive instruments, a small leakage can cause many problems, all your electronic parts would be ok / brand new etc ... and you could still get wrong values (voltages, impedances etc ..)


I work on some Rf boards around 27 mhz, small flux residues on pcb give me lots of problems, the flux we use is very strong, if we left it uncleaned it can corrode some parts overtime, sometimes i have to clean the pcb's 2 or 3 times to be sure its devoid of any contaminants, we rince the board with specially treated / filtered water.

On the same board we use a relay, if the relay develop a resistance (or the leakage we talk) on the contacts on some tests points well see strange voltages values, they can't be calibrated nor compensated,  we have discovered fake and counterfeit relays due to this behaviour.


@drummerdimitri   thats why on a earlier thread post i wrote you to clean your board properly, if the meter has an very high input impedance on some range(s) it could be affected by residues / contaminants, a very small solder short etc...

And E-Design  suggested the same thing ...

[Kleinstein]

U42 is supposed to be an DG444.  There are quite a few similar parts that should be OK too (e.g. DG202B,DG212B,DG441, ADG....). The switch should not be used dynamically so that speed and charge injection should not matter, mainly R_on and leakage. One may get one of these locally.

I very much doubt that Q70 is bad. It should be a JFET with a relatively low R_on and not too much gate leakage. One could measure R_on, even if it may be damaged. I would expect something like J107 or  PN4391/2 , but other low resistance ones should work too.

For U52 I don't know, but the markings should tell, what type. Chances are one could do a test with a more common type, so one could be reasonable sure there are no other broken parts before ordering.

[E-Design]

Yes leakage can mean contamination on the board... this is true.. when all components are put back in, the OP should clean the area up where he was working as a good practice.
The leakage I had in mind for this effort though is due to semiconductor damage - seems more likely to me given the unfortunate circumstances.

I agree that the Q70 is probably not bad.. leave it out for now or put it back in, wont matter - its for 1 particular range. It is a J105 Jfet.

U42 is DG444 but agreed could be any of those other alternatives. If you try an alternative, please ensure the logic control polarity is the same.. these analog switches are usually offered in both flavors.

U52 is an LMP2021 - a low ib, zero drift amplifier.

There is a lot of sense to putting in on hand parts to see if you can get it working - but if you have nothing on hand and have to order.. might as well order the right parts. There is high availability on these.

I cant suggest any particular place for you to order from with low cost shipping - thats up to you and where you like to shop.

Build a parts list, post it up here, we can tell you if it will function as replacements.

[JimKnopf]

How does the story ends? Did you fix your device?