Agilent 34401A repair help?

[Runco990]

First of, I apologize for asking for help as my first question.  I have been reading the forums for many years and never realized that I did not sign up for a log-in.  I very rarely ask questions, I just soak up knowledge by reading and keeping my mouth shut.

Anyway...

I acquired an Agilent 34401A recently and am stumped by it.  Disclaimer:  I fix audio gear but I am not a "test gear" guy, so while I can understand guidance and use it, I am not well versed in how digital meters actually work.

So the problem is:  Error 611, 612, 615, 617, 619.  When I cold start the meter and run self test Error 612 was NOT there, but comes back shortly.

The meter reads DC, AC, OHMS and CURRENT.  But all ranges and readings are a little low, by the same amount. 
For example:  5vdc will show 05.00047 on my good meter and 04.99609 and bouncing the last 2 digits.  This is consistent across ALL ranges as if the reference is bad, but it measures 7v and +10/-10 at the respective points.  Hmmm...

Power supply voltages measure within spec, and U201 was previously replaced.  As far as I can tell, all the Zeners are OK, as I can measure the correct voltage drop across them.  All caps APPEAR ok, and ripple is a few millivolts across each.
I DID need to clean the FRONT/REAR switch, as it was twitchy.  OK now.

So.. I'm a bit stumped.  I read everything I could find and checked what I could.  I can't seem to find a definitive fault.... I guess neither could the guy that sold it.  Upside is that it's a clean meter with an excellent display and it was cheap enough to take a chance.  The seller DID disclose all the error codes, but I bit off more than I could chew.  In the words of Jeremy Clarkson...  "How hard could it be?"

Can anyone give me some insight as to where to look next?

And as you can see I have a working and spot-on 34401A bought because of YOU guys and am very happy with it.  Was hoping for a second one without giving a kidney! 

[Mr. Scram]

I'm keen to hear from better informed people than me.

[Kleinstein]

There is a service manual available for this meter. The errors 612,615,617,619 are related to the ohms current source - so most of them may have a common reason.  Error 611 is about gain 100, though only a rather crude test.

So a simple test would be measuring the current in Ohms mode with a second meter.
The open circuit voltage in the ohms mode is what is checked, so one could measure this too - likely it would be too low.
Another simple test is to check if there is leakage at the input, by checking the input bias current.
Here the simple test is just reading the voltage with the 10 M input impedance enabled in something like the 100 mV range.
The more detailed test would be looking at the drift rate with something like a 10 nF PP capacitor at the input, starting from a few different voltages (e.g. some 9V, 5 V, -5 V and -9 V) in the 10 V high Z mode. No need to have very accurate readings, the order of magnitude matters. A good meter would be below some 50 pA and thus below some 5 mV per second. More than some 100 mV/s could cause trouble.

[Runco990]

I followed this:

I also have the paper service manual.  Printed out a few pages to make things easier to see.

My meter is producing the same currents in the video as I step through the ranges.  The current source seems to be working.
The meter is however unstable... even on ohm range it won't Zero.. it bounces around by .040 ohms plus/minus max.
Same on ALL ranges and Settings.  It's noisy.  How do you find that?  As it seems a very small voltage.

How warm should the reference get?  It's just barely warm to the touch and my flir camera shows nothing unusual either.

Eager to learn, that's for sure.  Never worked on meters, but was able to repair my HP 8566B spectrum analyzer in an hour.
Go figure....

[Kleinstein]

The LM399 reference is getting a little warm from the outside, but not very much, maybe some 60 C. The plastic cap is a reasonable good isolator. With the plastic cap it is hard to measure the temperature other than with an IR camera. Barely warm to a touch is OK, as the case is low thermal conduction.
It would be relatively rare for the heater in the reference to fail.

For 2 wire ohms no going to zero, the first candidate is the front / rear switch and maybe some relay contracts. There is a point to adjust the zero / subtract the offset.  +-0.04 Ohms in 2 wire mode does not sound so bad., though under idea conditions it may be better.  For an accurate test one should use 4 wire ohms.

So the current sources are working, but the meter still not working ? What is the open circuit voltage in Ohms mode (e.g. 1 K or 10 K range)  - this is what is checked for the errors 615,617,619.

[alm]

The meter is however unstable... even on ohm range it won't Zero.. it bounces around by .040 ohms plus/minus max.
Same on ALL ranges and Settings.  It's noisy.  How do you find that?  As it seems a very small voltage.

When you say +/- 0.04 Ohm noise on all ranges, does this mean 400 counts (last three digits on the display) on the 100 Ohm range, 40 counts (last two digits) on the 1 kOhm range, etc? Is this happening in both 2-wire and 4-wire resistance mode? With noise in the offset of a 2-wire resistance measurement, I would indeed think about series resistance in the switch or relay contacts, as Kleinstein mentioned.

The advantage of testing in 4-wire mode, is that you can test the current source and voltage measurements separately. You could try to short the input/source leads and feed the sense inputs a stable voltage between 0-100 mV (on the 100 Ohm range), which should produce a stable reading from 0-100 Ohm (exact value is not important). Or 0-1 V on the 1 kOhm range. If this is stable, and the current source is stable (you did not mention if it had any noise), then a 4-wire measurement should also be stable. If it's only noise in the 2-wire mode, then my money is definitely on excessive contact resistance in the path.

[Runco990]

The meter "works" in all ranges, but it's unstable and it continues to get the errors on self test.

Open voltage on Kohm is 7.88XX  where XX is constantly bouncing around. 
Open voltage on Mohm is 8.41XX bouncing.

This morning cold start NO error 612, but returns.

Something is unstable that affects the entire meter, not just one function.  DC/AC Current and Ohms "bounce around" and read equally low. 

This is my first foray into digital meters, so I have a learning curve ahead of me.  I still don't "understand" well enough how they actually work to systematically troubleshoot them.  But... eager to learn!

I'll start with the power supplies again and see if any are noisy.  I don't want to just change caps or parts undiagnosed.  My good agilent 34401A is rock steady.  Been using it to take readings.

[Runco990]

Oh... THIS is interesting...  I cleaned some flux residue off the current source area and now get Error 612,613,615,617,618,619.

Hmmmm.... something's loose!

[Runco990]

And... it's back to the old errors.    Must have been some alcohol still under a chip.

The output of U201 pin7 is noisy.  The Vreference is rock steady, as are the power rails.  U201 WAS changed.

Will noise cause self tests to fail???

[bdunham7]

Open voltage on Kohm is 7.88XX  where XX is constantly bouncing around. 
Open voltage on Mohm is 8.41XX bouncing.

How are you measuring those voltages?  What does the meter read in the various ohms and Mohms ranges if you have no leads connected?  Do any of your (DCV) readings change if you switch the impedance mode from 10M to 10G+?

[Runco990]

2 wire out from the sick meter to input on the good one.

OK, same DC volt readings on 10M or 10G.

Cleaned the board carefully with non residue cleaner and compressed air.  Back to 611,612,615,617,619.   Yet it measures all my precision resistors fine, though by the same amount low in all ranges.

Note:  I really hate surface mount crap.

[bdunham7]

Cleaned the board carefully with non residue cleaner and compressed air.  Back to 611,612,615,617,619.   Yet it measures all my precision resistors fine, though by the same amount low in all ranges.

Don't assume the two problems are related, especially if the meter has previously been worked on.  It does seem odd that you would get the errors that indicate a gross error, yet the unit more less works OK.  I haven't reasoned through your problem, so this is only semi-informed guessing--but if it is easy for you, can you try removing Q211 from the board and testing again?

[Runco990]

OK...  so the errors are mostly the ohms current source.  It is a little unstable, but I DO get all the outputs when I measure the current.  So they are present.  And it dawned on me... how does the meter KNOW they are present during self test?  Where to start there?  I want to confirm the meter can SEE the current source.  Looks like it CAN'T.  Relay??

One thing at a time, I guess.  No one said this would be easy.  Doesn't help that someone else was (of course) already here.  Should have known.

But if I figure this out with some help... it'll be worth it and I learned something.

Will try Q211...

[Runco990]

Q211 removed.  Have a hot air station.

Same errors.  Current sources are present, meter measures ohms.

[Kleinstein]

The self test is rather crude. From the description, it enables the current source in question and than measures the open circuit voltage that should be within 7.5 +-3 V. So a really crude test. With the measured voltages it is odd that you see the errors.

So there maybe something wrong, that may cause a delay or so.

[bdunham7]

OK...  so the errors are mostly the ohms current source.  It is a little unstable, but I DO get all the outputs when I measure the current.  So they are present.  And it dawned on me... how does the meter KNOW they are present during self test?  Where to start there?  I want to confirm the meter can SEE the current source.  Looks like it CAN'T.  Relay??

The manual has an excellent theory of operation section.  I'd recommend getting the online version from Keysight. 

Doesn't help that someone else was (of course) already here. 

A lot of assumptions have to be thrown out the window, especially if parts have been replaced.

[Kleinstein]

A possible candidate to look at may be L102. If this one is nearly open it could quite some trouble.  It would however also effect the 100 / 1000 V ranges.

[bdunham7]

OK, so I  have to go do something else, but I read through the relevant theory section and I notice that the test for error 611 simply has U101 connecting the (internal?) TSENSE to the DC100X section.  If the 100mV range works properly in normal use, error 611 would have to be U101 (or TSENSE, if it isn't actually internal to U101) or a logic failure causing U101 to not switch the TSENSE to MC when it should.  U101 is involved in all of those tests.  The intermittent nature of the problem makes me suspect solder issues underneath U101, but again, I'm just guessing.

[Runco990]

All right, I shall check into those and report back.  I too, have things to do.

Thanks so far. 

[Runco990]

OK, tested all coils, good.  U101 solder connections good.  Starting to think U101 may be faulty.

I'm also realizing that this is beyond my current understanding and this is one of those times ebay got the better of me.

I may have to just eat this one and put it on a shelf.  I have to work and can't put TOO much time into this.  Also struggling with the service manual, this digital stuff is just... confusing.  I have been fixing analog gear damn near forever, but THIS...  oh just forget it!

I cannot find the path to check the tsense to the ohms current source to test it.  It is unclear to me what voltages I should be getting on what pin of U101, etc.  What is xref everywhere, etc.

Sigh.  For once, I feel stupid. 

[bdunham7]

TSENSE doesn't go to an external pin so I'm assuming it is internal to U101.

I think 'xref' just refers to that portion of the circuit being 'in-guard' or floating referenced.

Does the meter read resistance correctly in 4-wire mode?  I'm stumped as well, the only thing I can think of that would consistently cause your errors but still allow the meter to work correctly would be if U101, U150 or a relay were slow to operate.  Relays might be more likely to release slowly, so K101 and K104 appear to release when doing the relevant self tests.  You might want to look closely at K101 and see if you can monitor its operation with a scope on the coil circuit and on the contacts.  K104 could be suspect as well, but K101 is a reed relay and that type of failure seems plausible.

Do you have a dremel tool, engraver or something like that that vibrates?  You might carefully put something against K101 to gently vibrate it while you run the power on tests.  If it is sticking, the vibrations may prompt it to release more quickly.  This seems a long shot and doesn't explain the TSENSE error, but what the hell, give it a try.

[Runco990]

Meter reads same in 4 wire mode.

One thing that bugs me is that the last 2 or 3 digits always jump around and the entire meter reads a little low.  My 2kohm standard for example reads 01.9906 kohms last digits constantly jumping around.  2k spot on and stable on my good meter.  I have 15 well aged General Radio sealed resistance blocks.

The meter "works", but I doubt it can be calibrated in this state.  It's too far off to be of use to me unless I use 4 digit mode.
"close enough" mode.

Driving me nuts.  I just do not fully grasp digital stuff yet.  Gimme transistors I can SEE any day.

As far as the "noise", I traced (so far) the jumpiness to U201 pins 5,6,and 7.  +/- 18v is steady so is Vreference.  U201 has been changed, twice by the looks of it.  Looks like the noise is going into U201 and amplified by it, but have yet to back-trace it.

Looked at everything under my stereo microscope, solder is good.  Freeze spray also yields no definitive results.  Neither does a thumb pressing here and there.  So I tend to rule out mechanical connection issues.

I'll see what I can do with the relays.  I own a Metcal system, so should be able to pull them out IF needed to replace.  I have already entertained pulling U102 to manually ohm it out.  Man, I PRAY it is good.  That one would be the deal breaker cost wise, if I can even get it.
U101... maybe...  But I can solder to that level with confidence.

Again, thank you for the help.  Digital meter "repair" is totally new to me.  I'll eventually "get it" but it's a learning curve. 

[Runco990]

Oh goodie!!!  After doing 4 wire and 2 wire ohms, now the meter has error 604 as well.

It's ok when you measure something, but as soon as you pull the resistor it starts clicking and displays 000.000 MOHM.
I can put a resistor in and it will read it.  I have to switch modes to get back to OVLD.  Great....

I bet the seller is jumping for joy to have discreetly unloaded this pile.  Can't even do a "not as described. "

Just realized it also does frequency.... quick test... THAT is spot on.  No problems there, no instability and accurate.

[Kleinstein]

The error 604 would be at the ADC itself, but may of cause also be a faulty signal getting to the ADC.
There is still a chance the amplifier part can be at fault. The self test parts are not very specific - don't take the simple test name to serious. more look at the description what is actually measured.

If experienced with the analog world, one could try tracing the signal from the input to the ADC input. So set the meter to non AZ mode and apply some clean, low frequency (e.g. 100 Hz range square wave) signal to the input. It should get well up to the ADC. One could also test with a DC signal (measure the amplifier output with the good meter): there may be a little offset (e.g. some 5-20 mV ?) and some drift, but not very much.

[Runco990]

Have to put it aside for a bit.  Will report back when I get back to it.  Real work beckons.