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

[Dr. Frank]

and the other currents, 5µA and 10µA?

Does it now pass the test?

That means, DIP 8 is obsolete, and you have used an SOIC / DIP adapter? Or was it an SO package initially? I don't remember..
This type is correct, anyhow

Well I managed to solder it on .. regardless if it was a different package ... pinout was the same so I just made due.

That's sub optimal. Please have a closer look for possible contamination or solder bridges, with your nice microscope

[The_Spectrum.A_idiot]

and the other currents, 5µA and 10µA?

Does it now pass the test?

That means, DIP 8 is obsolete, and you have used an SOIC / DIP adapter? Or was it an SO package initially? I don't remember..
This type is correct, anyhow

Well I managed to solder it on .. regardless if it was a different package ... pinout was the same so I just made due.

That's sub optimal. Please have a closer look for possible contamination or solder bridges, with your nice microscope

Thank you for getting back to me ,

Well I will have a look , but I don't have much hope. ( it was pretty flush with the PCB )

If that ends up to be a dead end. what are other possibilities ?

Thanks

[Dr. Frank]

you might replace U201 with another sample, maybe you created an ESD damage when assembling the new component.
And as said, I assume you can use the instrument with a fresh calibration in most modes / ranges.

[The_Spectrum.A_idiot]

you might replace U201 with another sample, maybe you created an ESD damage when assembling the new component.
And as said, I assume you can use the instrument with a fresh calibration in most modes / ranges.

Okay , hopefully I will be able to pull it of in an ESD safe environment ,

I bought off ebay original agilent part number parts. Hopefully by next week they will be in my possession , If I can't find a solution for this I will most probably just swap the original part and then send it for calibration and hope that they can calibrate it in fault mode. ( I think they are not accepting devices for calibration and certification if they fail any of the self tests ... am I correct ?

Thanks

[The_Spectrum.A_idiot]

  IT PASSED !!!!

After re-checking every solder joint , every flux residue , anything and everything on the PCB.

3 things.
Dr.Frank - Thank you for everything for all the valuable advice and helping me every step of the way !!! the IC being the wrong footprint and pointing that out put me into thinking !!
Kleinstein - You were right and I was wrong there was some pesky flux hidden under the IC so contact cleaner and IPA and passing KIMTECH wipe under the IC for 1h it finally worked

Everyone else that spend time helping me out and pointing out obvious things that I didn't know and helped me through eveything step by step I would like to tell you a huge thank you !!

Now off to calibration house !!! Hopefully they won't need and arm and a leg

Thank you in advance for your help , for your time , and your patience !!!

The Spectrum

p.s.: The Ohms values :

Results :
000. ohm range -997.0 μA
0.1k range - 997.0μA
00.1k range -99.44μA
000.1k range -9.98μA
0.1M range -5.00μΑ
00,1M range - 0.52μΑ
000,1Μ range -0,52μΑ

[coromonadalix]

[coppercone2]

is that guy on the left George washington??

[The_Spectrum.A_idiot]

is that guy on the left George washington??

No.

[coromonadalix]

wow    totally UN-apropriate to bring  this  here 

is this how  you end your repair thread ?

[The_Spectrum.A_idiot]

You are right.It was wrong.

When I am happy sometimes I go above and beyond.

Sorry about that.

[coppercone2]

its not a bad analogy how it hides under ceramic and prevents a system from functioning and its hard to get out

[Kleinstein]

I would still wait a little before sending the meter out to calibration. One point is waiting for settling after the disturbance from soldering. There is likely PCB stress and a disturbed humidity and this can cause some drift for the next few days / weeks.  Just after a repair there is also a slighly increased chance to get another defect, a little like with a brand new meter.

Another point is first doing a few more checks to see if the meter really works. There may be other, less visible defects / problems.
A relatively easy part to check is the input bias current, e.g. in the 10 V DC range. This can be checked from the rate a low leakage capacitor of some 10 nF charges or discharges.

Another point can be the drift during warm up. Excessive drift could point to a hidden defect. Here one may need some stable reference, so not so easy a test.

[The_Spectrum.A_idiot]

Another question for the experts ,

I spoke with 2 labs today.
1 was recommended by keysight
1 is a local lab.

The recommended keysight lab said that will verify the instrument and all its functions , will give a certificate of calibration. BUT if any of the functions have any deviation will give a limited certification with a deviation factor. I would like to know if this is standard OR I didn't convey my thoughts to the guy at the other end of the line correctly?

In my mind calibration means that I am forcing the instrument to measure a known thing and replace whatever value its measuring with the proper one.

so IF I am measuring a 1M standard that is 1ppm and the instrument is 40ppm and the instrument measures 0.98M then by calibrating the instrument means I am making modifications via the software to force the meter to get the measurement from 0.98M to 1M and adjust whatever values are wrong in its memory to account for age and wear and tear , then I am testing the linearity of the instrument. If all goes well means the correction was successful.

As far as I understand what they provide is just a proof testing that indeed the instrument is measuring what is real +/- the allowed deviation or error from the datasheet. And just say that this instrument is measuring the values as close to reality. Nothing else.

Is this what calibration means in the real world ?

The local lab asked me to email them the datasheet and instrument brand and model and they will get back to me. (most probably won't be able to verify it leat alone calibrate it in the way I have it in my mind)

Let me know if there is a better way to ask what I have in mind.

Thank you in advance for your time

The Spectrum.

[coromonadalix]

if the instrument normally pass self tests, you should be able to get a full certification

BUT   yes it could pass self tests (i dont know the precision of the internal tests done)    and maybe fail  if it is at the "out of tolerance" limit or goes beyond that ...

and yes you could still get a limited certification,  AND accept or refuses it,  but fee's can or could be charged to you either way  ...  (Here in my country they work like that, if it pass or fail)


the best would be find some people who can check it  BEFORE doing certifications,  maybe in the TEA thread some people could help ?

yes  some wrote, the best would be  to run the meter for a few days and try to see  if it deviates,  even a good cleaning could still leave traces ...  re re and re check ...   this meter has gone thru a lot  ... i mean a lot ...  you have to be sure

the  meter is a good piece of precision,  if it goes well, it may or will be precise for many years,  i had 3x  of them (1 xformer blown, 1 dim vfd, and one fully recapped),   the later received a full certification (but frankly it was still in the specs after a decade) ..

if the meter do and does precision measurements,  you need to get it certified,  or if its a "toy" or non precise stuff and nothing depend or need the be certified BY YOUR meter,  you can use it like this, but you'll never know  unless doing some good checks  with precision parts in the 0.5 %  0.05%  ...    i have Fluke 0.013% precision resistors  ...

if a local lab ask you specs  mmmmmmmm  i would not be sure of their capacity ... 
normally you need  Fluke calibrators or else  in the 7 or 8 digits of precision

[The_Spectrum.A_idiot]

if the instrument normally pass self tests, you should be able to get a full certification

BUT   yes it could pass self tests (i dont know the precision of the internal tests done)    and maybe fail  if it is at the "out of tolerance" limit or goes beyond that ...

and yes you could still get a limited certification,  AND accept or refuses it,  but fee's can or could be charged to you either way  ...  (Here in my country they work like that, if it pass or fail)


the best would be find some people who can check it  BEFORE doing certifications,  maybe in the TEA thread some people could help ?

yes  some wrote, the best would be  to run the meter for a few days and try to see  if it deviates,  even a good cleaning could still leave traces ...  re re and re check ...   this meter has gone thru a lot  ... i mean a lot ...  you have to be sure


if a local lab ask you specs  mmmmmmmm  i would not be sure of their capacity ... 
normally you need  Fluke calibrators or else  in the 7 or 8 digits of precision

I will run it the next month , But I need to get some precision resistors , I have a GBID ( that interface ) to capture the readings ( it even has serial ).
And maybe use 4W and 2W measurements to test said resistors.

The most difficult would be for voltage and current ( as I have no idea what can be that stable while measuring it , a battery is out of a question )

Because they said their stuff is still out on calibration aboard.

But my question was if this is normal to just verify the instrument and not adjust it via its calibration menu etc etc ...

Anyway slowly I will get it done somehow. I will email keysight too what is their normal procedure in case of non conformities ...

if a local lab ask you specs  mmmmmmmm  i would not be sure of their capacity ... 
normally you need  Fluke calibrators or else  in the 7 or 8 digits of precision

My thoughts exactly ! That's why I will send it abroad.

[coromonadalix]

it is normal or not  upon  what use you intend to do with it

precision means calibration / certifications at 1 or 2 years intervals

basic use means, you do basic or non precise checks   but not undergo the cals certs

but at some point  you will need to be sure of this meter and it's capacities, at 6.5 digit,  it is not a toy ... do you accept to use it  not calibrated an live with uncertainties ....

i personally would not,  it is a 0.035% precision meter if i recall ...  or more in the 0,01% in resistances  ....

[The_Spectrum.A_idiot]

it is normal or not  upon  what use you intend to do with it

precision means calibration / certifications at 1 or 2 years intervals

basic use means, you do precise checks   but not undergo the cals certs

but at some point  you will need to be sure of this meter and it's capacities, at 6.5 digit,  it is not a toy ... do you accept to use it  not calibrated an live with uncertainties ....

i personally would not

As soon as I do it once and I know that is ok, then I can start to trust it.

I can repeat the cal every 5 years or so just for my sake , I am not a professional but I want to have confidence in my readings.

Thanks we have a similar mind set

Thank you in advance

The Spectrum

[Kleinstein]

If not needed for some professional use the question is if one really needs a formal calibration. Form most hobby use one may not really care that much, though a good test to get confidence is still wellcome.
The usualy calibration is just the testing - often without giving the actual readings, but just a pass / fail. The change of the cal constants in the meter is called adjustment and this may cost extra and is rarely done, as long as the meter is still in specs. In my oppinion one should consider readjustment, or a marking on the meter if it comes close (e.g. 75%)  to the accuracy limits, as there is also a drift component and the calibration should give confidence that the meter will meet the specs for the next 1 or 2 years.
The 34401 are usually quite stable and an re-adjustment is rarely needed.

The repair in the ohms section should not have effected the amps and voltage part. The resistor part can be effected a little, from the offset of the OP-amp that was changed. Up to 100 µV offset relative to 5 V would be 20 ppm contribution from the offset and 2 OP-amps in the circuit. For the 10 M range this would be 10 x higher with only 0.5 V ref. voltage for the 2nd OP-amp. In addition the bias / leakage current could be relevant there too. So there is a chance that the 10 M range can be out of spec.

The self test is not very strict and does not cover all aspects. From the service manual the last 2 tests that caused an error have a test limit of +-20% for the current - so not at all strict. So it may not trigger on a lesser problem that can still make the meter to fail. Testing to better than 10% is not that hard.

One may not need super accurate parts to do some testing considerably stricter than the self tests. Some of the tests also cover things not covered by the self tests and often not checked during calibration. A few tests that can be done relatively easy are:
1) input bias with a low loss, low leakage capacitor (e.g. 1-10 nF PP or PS type)
2) sum of 2 resistors to check the linearity in the resistance mode  (can be effected by leakage).  With some care in the environment, relatively common 100 ppm/K parts can be sufficient. For 4.7 M+4.7 M even lower grades can be OK.
3) drift during the warm up, especially for zero voltage and resistors
4) a turn over test with a 9 V battery or similar. This may need some care to keep the battery undisturbed and repeats to check.
5) A check of the noise with a short.
6) Check some resistors (e.g. 10K, 100K, 1 M) in 2 ranges each.
7) zero resistor reading
AC reading in a square wave with accurate amplitude (e.g. CMOS gate with 50/50 ratio)
Doing the tests is a way to get familiar with the meter too.

[The_Spectrum.A_idiot]

If not needed for some professional use the question is if one really needs a formal calibration. Form most hobby use one may not really care that much, though a good test to get confidence is still wellcome.
The usualy calibration is just the testing - often without giving the actual readings, but just a pass / fail. The change of the cal constants in the meter is called adjustment and this may cost extra and is rarely done, as long as the meter is still in specs. In my oppinion one should consider readjustment, or a marking on the meter if it comes close (e.g. 75%)  to the accuracy limits, as there is also a drift component and the calibration should give confidence that the meter will meet the specs for the next 1 or 2 years.
The 34401 are usually quite stable and an re-adjustment is rarely needed.

The repair in the ohms section should not have effected the amps and voltage part. The resistor part can be effected a little, from the offset of the OP-amp that was changed. Up to 100 µV offset relative to 5 V would be 20 ppm contribution from the offset and 2 OP-amps in the circuit. For the 10 M range this would be 10 x higher with only 0.5 V ref. voltage for the 2nd OP-amp. In addition the bias / leakage current could be relevant there too. So there is a chance that the 10 M range can be out of spec.

The self test is not very strict and does not cover all aspects. From the service manual the last 2 tests that caused an error have a test limit of +-20% for the current - so not at all strict. So it may not trigger on a lesser problem that can still make the meter to fail. Testing to better than 10% is not that hard.

One may not need super accurate parts to do some testing considerably stricter than the self tests. Some of the tests also cover things not covered by the self tests and often not checked during calibration. A few tests that can be done relatively easy are:
1) input bias with a low loss, low leakage capacitor (e.g. 1-10 nF PP or PS type)
2) sum of 2 resistors to check the linearity in the resistance mode  (can be effected by leakage).  With some care in the environment, relatively common 100 ppm/K parts can be sufficient. For 4.7 M+4.7 M even lower grades can be OK.
3) drift during the warm up, especially for zero voltage and resistors
4) a turn over test with a 9 V battery or similar. This may need some care to keep the battery undisturbed and repeats to check.
5) A check of the noise with a short.
6) Check some resistors (e.g. 10K, 100K, 1 M) in 2 ranges each.
7) zero resistor reading
AC reading in a square wave with accurate amplitude (e.g. CMOS gate with 50/50 ratio)
Doing the tests is a way to get familiar with the meter too.

Thank you very much for the pointers.

The reason I am leaning heavily towards calibration and adjustment is the sole reason that the instrument comes from an UNKNOWN environment with unknown calibration verification. No cert has accompanied the meter.

So this lack of traceability is the issue.

Plus as I said its all about trust.

Thank you in advance