Agilent 34401A Bubble on PCB ... can it be saved ? (SOLVED)

[The_Spectrum.A_idiot]

The full 18 V at the Q202 gate suggsts that the regulation tries to turn the fet off (it is a P channel FET and thus off with a positive voltage).

The voltages at the OP-amp inputs (pins 5 and 6) suggest that the output should go negative and not positive as observed. So chances are high that U201 B is bad.

Still having some voltage at the output side (Q202 pin 2)  of the curren source somewhat points to leakage at Q202 - so this part may also be damaged ( I would at least order or look for a replacement, if ordering an AD706 (or substitute).
This leaves the question if the protection is working OK. This could be  a shorted transistors, or maybe one with too low a break down votlage and also an open Q211. Off cause the reason of the damage could also be from just too much voltage for the transistors.

Dear Mr. Kleinstein  Thank you so very much for your input and help,


Started Loading the parts canon.

Q201 - MMBF4392LT1G VS MMBF4392

Q202-MMBF5461 - is unobtanium. I am trying to find something similar.

Q211 -  MMBF4117 VS MMBF4117A

U201 - AD706JRZ the closest candidate VS AD706JR

Q203-210 - MMBT6520LT1G VS MMBT6520L

CR203 - BZX84-C5V1-QR VS BZX84C5V1

CR201 - BZX84C4V7-E3-08  VS BZX84-C4V7


Lets hope I will manage to pull this off

Any input on parts selection is always welcome

[Kleinstein]

For the P-JFET the MMBF176 should be OK - one of the few P-channel fets that are usually available.

[The_Spectrum.A_idiot]

For the P-JFET the MMBF176 should be OK - one of the few P-channel fets that are usually available.

For the Q202 right?

[Kleinstein]

The mmbf176 would be my suggesting for Q202. There is not that much choice if p-channels JFETs anyway. The speed and thus to a large part the capacitance should not matter much. The mmbf176 has roughly a similar threshold and Idss to the MMBF5461. The voltage rating should be sufficient for the use case.

[The_Spectrum.A_idiot]

The mmbf176 would be my suggesting for Q202. There is not that much choice if p-channels JFETs anyway. The speed and thus to a large part the capacitance should not matter much. The mmbf176 has roughly a similar threshold and Idss to the MMBF5461. The voltage rating should be sufficient for the use case.

MMBFJ176 you mean?
https://eu.mouser.com/ProductDetail/512-MMBFJ176

Many thanks

[The_Spectrum.A_idiot]

Dear All

All the parts arrived ,

Now I would like your recommendation for the hot air station setup , how should I go hot and high output to remove the components fast ? or slow/ medium heat , medium flow and take my time and slowly ramp up the temperature until the solder liquefies ? Big , small nozzle ?

lastly I will kapton tape everything around that area and hope for the best , lots of flux , and hope for the best.

Let me know if you have any other tips and tricks how to deal with hot air .

Thank you in advance

[wraper]

All of that can be soldered by dual wielding soldering irons, soldering tweezers or using low melting point alloy. I would avoid using hot air on PCB that already popped as much as possible to minimize chances of further damage.

[The_Spectrum.A_idiot]

Dear All ,

We kind of made it , now I have 2 errors. 612 and 618.

replaced :
U201
Q202
CR201
Q202-210
Q201
Q211
CR203
Q202

What now ?

Thank you for your help in advance

[Kleinstein]

From the description the errors 612 and 618 are essentially the same, a problem with the 500 nA current source or 10 M divider.
A way to check is to measure the 500 nA current externally. If needed with a resistor and voltmeter.
Another point to check could be the resistor readings for the 10 M range (e.g. check 1 M and 4.7 M).

[The_Spectrum.A_idiot]

okay so shorting and measuring the μA output :

000. ohm range -999.4 μA
0.1k range - 999.3μA
00.1k range -99.46μA
000.1k range -9.81μA
0.1M range -4.83μΑ
00,1M range - 0.41μΑ
000,1Μ range -0,42μΑ

[The_Spectrum.A_idiot]

From the description the errors 612 and 618 are essentially the same, a problem with the 500 nA current source or 10 M divider.
A way to check is to measure the 500 nA current externally. If needed with a resistor and voltmeter.
Another point to check could be the resistor readings for the 10 M range (e.g. check 1 M and 4.7 M).

A 10K 1/4w 5% resistor measures 09.912 kohm on auto-ranging

[Kleinstein]

410 / 420 nA should still be just inside the acceptable range for the current source.
It is still be a bit odd that the current is that far off nominal. This could be leakage (e.g. not yet cleaned flux), or maybe extra bias from the OP-amp. If the issue is leakage it may get worse with higher voltage / higher resistance.  A test would be to check the linearity with a series connection of 1-2 M resistors in the 10 M range.

Another possible point of failure may be the relay connecting the 10 M divider. Is the 100 V range working reliable ? For a test, e.g. check something like a 9 V battery in the 10 V (high Z) and 100 V range and switch between the 2 ranges a few times.

[The_Spectrum.A_idiot]

410 / 420 nA should still be just inside the acceptable range for the current source.
It is still be a bit odd that the current is that far off nominal. This could be leakage (e.g. not yet cleaned flux), or maybe extra bias from the OP-amp. If the issue is leakage it may get worse with higher voltage / higher resistance.  A test would be to check the linearity with a series connection of 1-2 M resistors in the 10 M range.

Another possible point of failure may be the relay connecting the 10 M divider. Is the 100 V range working reliable ? For a test, e.g. check something like a 9 V battery in the 10 V (high Z) and 100 V range and switch between the 2 ranges a few times.

Used a 9V PSU
Tried every range on VDC and values don't change.

re measured μΑ on ranges after cleaning the flux and doing the range change.
Results :
000. ohm range -999.1 μA
0.1k range - 999.3μA
00.1k range -99.46μA
000.1k range -9.84μA
0.1M range -4.83μΑ
00,1M range - 0.36μΑ **
000,1Μ range -0,36μΑ **

A test would be to check the linearity with a series connection of 1-2 M resistors in the 10 M range.

How can I do thins without a reference resistor ?

[Kleinstein]

With little difference at 9 V this suggests a woking relay.

Getting less current after cleaning suggests that the cleaning may have made things worse, e.g. leaving moisture or more conductive film at the critical areas. Another possibility could be semiconductor leakage (e.g. at Q211) and a higher temperature for the 2nd test.
One could look at how the current changes with temperature: leakage in semiconductor usually goes up with temperature quite fast. Surface leakage from dirt on the PCB tends to get down with temperature or at least not up very much.


For a linearity test it does not need very accurate resistors. The deviation from the current is rather large, reaching some 10-20%. So the resistors only need to be reasonabel stable to maybe 5% between the tests. So more normal (e.g. 5 or 10% tolerance, though maybe not use cabon film) resistors should do and the exact value does not matter. It is just to reach a sum in the 8-11 M ohm range.

The basic version would be comparing 2 x 4.7 M in series to the sum of the individual measurements in the 10 M range.

[The_Spectrum.A_idiot]

With little difference at 9 V this suggests a woking relay.

Getting less current after cleaning suggests that the cleaning may have made things worse, e.g. leaving moisture or more conductive film at the critical areas. Another possibility could be semiconductor leakage (e.g. at Q211) and a higher temperature for the 2nd test.
One could look at how the current changes with temperature: leakage in semiconductor usually goes up with temperature quite fast. Surface leakage from dirt on the PCB tends to get down with temperature or at least not up very much.


For a linearity test it does not need very accurate resistors. The deviation from the current is rather large, reaching some 10-20%. So the resistors only need to be reasonabel stable to maybe 5% between the tests. So more normal (e.g. 5 or 10% tolerance, though maybe not use cabon film) resistors should do and the exact value does not matter. It is just to reach a sum in the 8-11 M ohm range.

The basic version would be comparing 2 x 4.7 M in series to the sum of the individual measurements in the 10 M range.

ok finally got around to get it sorted,

So with a 10K resistor at 10K range manual
I get -9.941,10 kohms
with 1x 4.42K @ 1% I get 4.429,57 kohms
with 2x 4.42K @ 1% I get 8.848,81 kohms
at 000.1K range I get
with 3x 4.42K @ 1% I get 013.053,2 kohms

Thanks

[Kleinstein]

The 10 K range looks like it works. To check if the sum really works Ok one would need to measure both resistors separately to really add the numbers. The expected difference that may show up would be in the 6 th maybe 5 th digit.

For the problem with the low current one would need to do the similar test with larger resistors, like 4.7 M or a chain of 6-8 x 1 M. The 10 M range where a larger error is expected.

[The_Spectrum.A_idiot]

The 10 K range looks like it works. To check if the sum really works Ok one would need to measure both resistors separately to really add the numbers. The expected difference that may show up would be in the 6 th maybe 5 th digit.

For the problem with the low current one would need to do the similar test with larger resistors, like 4.7 M or a chain of 6-8 x 1 M. The 10 M range where a larger error is expected.

Okay lets say hypothetically , all these tests show that we are kind of working, in certain / most ranges.

is there a chance to even calibrate and keep the instrument error free ? or this is just a sanity check to make sure that we are measuring true regardless if the instrument is not passing it's self tests ?

Do you think was a mistake that I've changed every components all at once ? and most probably some components are not "matched" per say ?

Just so that I would know IF i have to live with it or hope for a trustful future

Because is all about trust. If I can't trust it , I can't use it.

[coromonadalix]

that's why you use the meter self tests. if it fail you have a problem

and when you don't get any fails, the ultimate answer reside in dmm calibration,  and you'll know it is perfect

But i'm not sure how the self test may handle bad calibration, or out of calibration ??

and just a comment, for future(s) purchase(s) if any,    always ask to see an passed self test picture, or ask the seller to do it

[The_Spectrum.A_idiot]

that's why you use the meter self tests. if it fail you have a problem

and when you don't get any fails, the ultimate answer reside in dmm calibration,  and you'll know it is perfect

and just a comment, for future(s) purchase(s) if any,    always ask to see an passed self test picture, or ask the seller to do it

I agree. It was a kind of deceptive practice as he listed it as {tested good} and at the very bottom of the description, 'continuity and resistance not operable' ... and you guest it I found out to read the fine print before looking as good deal.

On his defence it was mentioned. I didn't read it.

Anyway lesson learned. Now back to trying to get this **** working.

I would like to thank everyone for their help and input I very much appreciate it !!

Regarding bad calibrations well I've changed components that were supposed to be paired and verified with other non paired but original close enough components.

whether or not a proper full calibration will make/force the meter to pass it's self tests as at the moment in essence I don't have any faulty components that I hope someone will let us know.

[Kleinstein]

The parts that were replaced don't need to be paired and most of them should not even effect the calibration / scale factors. From the changed parts it is only the offset of the OP-amp that has a small effect on the scale factor of all resistance ranges. The OP-amp bias and leakage currents can also effect the high resistance ranges (10 M + 100 M and a little 1 M) . The others should not be effected enough to get much of an error.

It looks like the PCB is not yet fully clean / dry. If there is some conductive flux residue or similar it can take quite some cleaning to get rid of the leakage.
Worst case there is leakage from the overheated PCB part or dirt inside the bubble.
one may have to close the case for the tests, so that light would not effect some of the SMD parts - some can be a bit light sensitive.

Changing all the parts was not such a bad idea. The OP-amp was very likely bad from the tests and the other parts are not that expensive and possibly damaged.


The voltage and current ranges should not be effected by the repair, unless there is PCB contamination.
There are a few additional tests that one can do with relatively little effort / instruments to check a few more points no checked so much in the self test.

[The_Spectrum.A_idiot]

The parts that were replaced don't need to be paired and most of them should not even effect the calibration / scale factors. From the changed parts it is only the offset of the OP-amp that has a small effect on the scale factor of all resistance ranges. The OP-amp bias and leakage currents can also effect the high resistance ranges (10 M + 100 M and a little 1 M) . The others should not be effected enough to get much of an error.

It looks like the PCB is not yet fully clean / dry. If there is some conductive flux residue or similar it can take quite some cleaning to get rid of the leakage.
Worst case there is leakage from the overheated PCB part or dirt inside the bubble.
one may have to close the case for the tests, so that light would not effect some of the SMD parts - some can be a bit light sensitive.

Changing all the parts was not such a bad idea. The OP-amp was very likely bad from the tests and the other parts are not that expensive and possibly damaged.


The voltage and current ranges should not be effected by the repair, unless there is PCB contamination.
There are a few additional tests that one can do with relatively little effort / instruments to check a few more points no checked so much in the self test.

okay , I will try to clean as much as possible under the IC and around it ( Isopropyl and contact cleaner might do the trick ? )

I guess giving the whole PCB an ultrasonic bath would be a big no. no I assume.

I will get more resistors and see if I can get the reset fo the tests done.

Thank you very much

[coromonadalix]

same here, you solved many errors,  maybe another cleanup with better products,  contact cleaner must not leave any residue or oil / lubricant

and maybe, just maybe  try to get  better resistors values  as kleinstein wrote to tests all possibilities, and slowly redo all the tests / current generator  etc ..

this meter has gone thru a lot,  be sure the front and rear switch is good too, it can play some tricks, there was some who had relay problems  on one coto  model ...

be patient and do things methodically, you'll get there,  and lucky you  schematics are known ...

[The_Spectrum.A_idiot]

You can say that again indeed this poor instrument had a very harsh life.
But yes step by step hopefully I can get it back to spec.

[Dr. Frank]

sorry to reply so late.
The current source still might have a problem gor the high Ohm ranges. That should be the 500nA source current, affecting both highest ranges.
Please apply a 1M or 10M resistor in 2W autorange  mode, and fill out the diagram, which shows the crucial voltages in this circuit, once with the correct ones as measured by me.
Most suspect is this dual OpAmp with very low bias current, which fails quite often.

Frank

[The_Spectrum.A_idiot]

sorry to reply so late.
The current source still might have a problem gor the high Ohm ranges. That should be the 500nA source current, affecting both highest ranges.
Please apply a 1M or 10M resistor in 2W autorange  mode, and fill out the diagram, which shows the crucial voltages in this circuit, once with the correct ones as measured by me.
Most suspect is this dual OpAmp with very low bias current, which fails quite often.

Frank

1M resistor : 0.9953mohms on fluke 289
on instrument :  0.959,20 =mohms

R201 4.993v
R202 6.972v

U201-A 3 - 6.972v
U201-A 2 - 6.972V
U201-A 1 - 3.8189V
+7REF - 6.972V

Q201-1 - 8.896
Q201-2 - 7.113
Q201-3 - 3.8189

CR203-A 1 - 13.974
CR203-A 3 - 18.810

1SRC Q202 - 1 - 8.920
Q202 - 2 - 5.965
R290 - Q202 - 3- 10.406
 
CR201-A -3 - 10.406
CR201-A -1 - 5.667

U201-B - 7 - 5.670
U201-B - 6 - 8.926
U201-B - 5 - 8.901

Thank you in advance