Yet another sick HP 34401A - No more errors, but still acting weird.

[slash14]

Hi All,

I bought a used HP 34401A with some faults. When performing the self tests, almost all errors were reported, from 608 to 622
First thing, I got the Service Manual, started to probe power rails: bingo! CR305 was broken (open), meaning no +15VDC. It was the only missing power rail.
I used a bench supply to feed the missing +15VDC rail.
Still, the meter was showing errors 612 and 618 (all related to the 500nA current source), and sometimes, the error 614 would appear randomly.

For errors 6112 and 618, this forum quickly drew my attention to U201, although I was a little skeptic since the other current sources were fine.
I got myself a LM353 as a replacement (the only CMS my retailer had in stock that could replace the AD706).
Instead of using my external bench power supply, I replaced the broken CR305 with a non CMS component, at least temporarily.

Today, the meter fully passes all its tests!
I compared measures with another 344401: they're pretty close from each other for each function (VDC, VAC, IDC, ohm 2W (I didn't test 4W) and freq/period. 

But ! 

I'm still noticing a weird effect:
- on VDC function, with no test lead connected, the meter keeps displaying random figures, around 0.220V (sometimes much higher !) .
If I connected a test lead in the LO input, same behaviour
If I connect a test lead in the HI input, and that I touch the tip of the test lead, the displayed voltage drops to 0.3mV, but still very noisy.
- on VDC, shorting the HI to LO input gets me to 0.0003mV (AFAIR)
- on 2W ohm, in AUTO range, the meter sets on the highest range and displays some random but stable figures sometimes around 20Mohms, sometimes about 80Mohms.
Again, shorting the inputs or putting a known resistance in between HI and LO seems to provide an acceptable reading.
- on ADC, putting a test lead into the HI input (yes, the HI input, where you're supposed to measure VDC or ohms) gets the display to 0.221A
- shorting the LO to the current input gets the display to almost 0mA.

This behaviour affects both the rear and the front inputs, and playing around with the front/rear switch have no effect.

Looks like a small current leaks from the HI to LO input, disturbing the meter with the lowest current source, hence displaying erroneous value in the ohm function.
I also noticed that on VDC function, once I stop shorting the HI to LO, the displayed values quickly increase from 0V to .... I don't, it keeps increasing.

I gently taped the components to find a potential bad solder joint (more particularly resistors R104 to R111, which are prone to bad joints as far as I read from this forum). No luck so far.

Any ideas?

Thanks for your help.

[Dr. Frank]

Hello,
in the long run, you should try to get an AD706, as the LF353 has too high input offset and offset drift, so accuracy of calibration and the stability of this calibration of the OHM modes are affected.

If you let the jacks open in DCV, and I assume that you let the instrument run with 10M constant input impedance, it should display up to about 300 µV only, equivalent to 30pA max. bias current. Please check, that you have no strong EMI sources in your room, (like bad or defective switch mode power supplies) which could disturb the open, sensitive  circuit... connecting a single cable to the Hi input and getting weird display might be an indicator for that.

You instead measure 220nA with no cables connected, so there might be something wrong around the MUX IC U101. Check + / - 18V supply of that IC.
What is displayed in  the manual 100V DC range, with no cables connected?


DCI should give solid zero reading, and not be affected at all by Input Hi signals (but again, EMI disturbance??), so what is displayed when you switch the different ranges, again with no cables connected?

Do you get correct measurement in 2W and 4W Ohm mode, when measuring reference resistors (or compared to other DMMs), or do you see strong deviations, especially in the MOhm ranges?

Frank

PS: What do you mean with CMS component, is it SMD? So you use a through hole / leaded diode instead?

[slash14]

Hi Dr Frank,

Thanks for trying to help.

First video: unit has been power up just before making the video. As you can see, starting from the lowest range, there's a measurement of 150mV, slowly dropping, maybe converging to 0.0000 (I didn't have the patience to wait till it happens), but way too slow! As you'll notice, the measured value drastically changes when moving to 100V range, as well as retunfging to 10V range (2.65 ratio).


Interestingly, same video, later (once DMM has warmed up a little I'd say). This time, ratio is 3.82 roughly.


This is a video I shot to demonstrate the effect of having (or not having) a test lead in the HI input when DMM is setup to VDC manual lowest range.
No lead : about 120mV
With lead: 10mV
Touching the opposite tip of the teste lead: 1.5mV


Finally, what happens when measuring resistors (no test leads connected). In AUTO range, relay keeps clicking. On manual range, the DMM keeps showing OVLG from the lowest range up to 10Mohm (hwich is fine). But the DMM seems to acquire something on the 100Mohm range


The power supplies of U101 look ok (at least within the service manual acceptable range) and stable.

The unit still passes all its tests.
I've been thinking about EMI issues too. So I did the same measurement with the PCB shield back on, as well as with the DMM cover on. Same behaviour.
The previous videos were shot with the meter all guts open (no PCB shielding nor DMM cover).

My bad, I meant SMD indeed (CMS means the same, in french )

[Dr. Frank]

That's not looking fine. There are bias or leakage current on the order of nA.
Most suspect is U101, the MUX IC, which might be damaged.

Have a closer look on the top and bottom side of the PCB, around this IC, and the different relays K101 - K104, if there any signs of poured fluids, which create leakage paths from the relays coil Vcc to  their switching contacts.

If that's clean, I fear there's no other root cause possible, than a defective U101.

Frank

[slash14]

Hello Dr Franck,

Thanks again for trying to help.
Sorry for the lack of update.

I made a few new measures (with a simple battery, and a few random resistors I had on hand), once unit has been warming up for 30 minutes roughly.
The sick meter is on the top.
It still displays a 6mV (45mV when cold) offset when no lead is connected in the HI input. The meter no longer change its relay when on switching back and forth the 100Mohm / 10Mohm ranges (in auto range).

https://youtu.be/jzgx4XtLK_Y
Looks pretty consistent to me.

Another attempt to insert test lead in the HI input (auto range, DCV).
https://youtu.be/nI3Nmhs94bs

I've replaced one zener (I had a though hole fitted, now, replaced with a SMD one).
Ohm function may slightly be affected by the fact I'm using a LF353 instead of an AD706 (I'm waiting for an AD706).

Next step:
- I'll maybe attempt to replace one or two other Zener diode 3.3V (although voltage rails seem fine).
- I'll try to unsolder U201, clean the board underneath, and reflow U201. Before I ayyempt this last tricky step, any last idea that would help me troubleshoot this unit?

[4thDoctorWhoFan]

Since you replaced U201, make sure you cleaned the area very well.  Any residue left can cause problems especially if flux or cleaning chemicals got under the main chip next to U201.

Anyway, just a suggestion.

[slash14]

Hi All,

Quick update:
I've received a new AD706 so I replaced the LF353 I put temporarily.

I also asked a colleague of mine to unsolder U101 (as advised by DrFranck) where he spotted that something could have spilt below (my colleague's full time job is to fix electronic boards, so he has all the proper equipment, the knowledge and know how, etc etc. ...).
He cleaned the PCB as well as U101.
Put U101 back into place.
And guess what: meter is working much better!

The meter is no longer sensitive to the presence of a test lead in the HI input.

I did a quick test with this meter, as well as another one, and my good old Fluke 87 III (set to 4 digits precision). Altought the 3 meters do not give exactly the same figures, they're all close from each other (too bad I forgot to take pictures).

Yet another 34401A back in business!

Thanks Dr Franck and 4ThDoctorWhoFan for helping!