HP/Agilent/Keysight 34420A NanoVolt/Micro-Ohm Meter repair

[plesa]

Multiple months has been under my bench waiting for repair, so lets started.
Meter is quite old ( according to serial number made 1995 ).
This meter is quite popular for ration measurement compare to 34401A it has low voltage preamplifier which enable 1mV DCV range and low ohm measurement with high precision.
- 7½ digits resolution
- 100 pV, 100 n? sensitivity
- 1.3 nVrms, 8 nVpp noise performance
- Built-in low noise 2 channel scanner
- Direct SPRT, RTD, Thermistor, and Thermocouple measurements
Thanks to absence of fan unit is clean after several years in operation.
Compare to competitor (Keithley 2182A Nanovoltmeter) has own current source which enable Ohms/RTD measurement.
Meter is still in active production (end of this year will be not possible to ship to EU - RoHS)
Price $4445 new one, $1538k unit exchange by Agilent Keysight. Used working unit can be found on eBay for quite high prices ( similar to 3458A), working unit in uknown state should be about $500, with calibration $700-1k.
Like 3458A it has been sold by Fluke as http://assets.fluke.com/manuals/hp34420xiseng0000.pdf for temperature measurement and calibration.

Initial inspection:
- Power on and passed self test
- VFD display is little bit dim
- Input connector has bend pins
- DCV channel 1 - not working
- DCV channel 2 - OK
- 2W Ohms - OK
- 4W Ohms - OK
- RTD - OK

Schematic: http://www.ko4bb.com/manuals/download.php?file=HP_Agilent/Agilent_34420A/Agilent_34420A_SCHEMATIC.pdf
User Manual: http://literature.cdn.keysight.com/litweb/pdf/34420-90001.pdf?id=1000003601:epsg:dow
Service Manual: http://literature.cdn.keysight.com/litweb/pdf/34420-90010.pdf?id=816987

When I tried to fix bend pin I found they are made from pure copper ( not like the regular gold plated LEMO pins ) and are quite brittle So I needs to order replacement.
Replacement we discussed https://www.eevblog.com/forum/repair/looking-for-the-input-connectors-for-agilent-34420a/
and after this excersise I can confirm that the connector housing is EVP.1S with custom made pins.
I do not want to stop repair by this, so I desoldered connector assembly and solder insted of it 50cm of PTFE insulated silverplated wires (made by Habia).

After repair meter can be handy for low EMF measurement and ratio measurement of LTZ1000 boards.
Any experience with DCV Channel 1 repair may be helpful.

[ManateeMafia]

I bought a 34420A that had a defective input channel. I cannot remember which one was defective, but I traced it straight to an optoisolator that wasn't functioning. Looking at the schematics, I think mine was U125. I could see the enable signal being applied, but the ground connection wasn't completing. I was able to jumper pins 4 and 6 to get the DC function working.

I would have to open it again to verify which one it was but I would start with those. The replacement was bought at Digikey, I will have to search through previous orders to find it if needed.

[plesa]

Thanks for tip, it is the same failure U125. When shorted pin 6+4 Channel 1 DCV reading is OK!!
If you can find replacement of HSSR8400 it will be appreciated.
It was much faster repair than I expected, now it will be little bit tricky to find connector replacement.

[ManateeMafia]

That's great. My repair was also quick and it took longer to track a replacement.

I ended up using this part http://www.digikey.com/product-search/en?keywords=LH1510AAB-ND

[plesa]

Thanks!!

[HighVoltage]

First time I see the inside of a 34420A.
Congratulations on the fast repair.
I am looking forward to some measurements you will perform.
Now you need a real SPRT sensor, like this one, to have one of the greatest thermometer.

https://youtu.be/zQ5_oJI1cJM

[plesa]

Now you need a real SPRT sensor, like this one, to have one of the greatest thermometer.

I was quite surprised that they are using 25 Ohm.

But back to the repair.
In service notes I found that there are lot of issues related to older version of firmware and communication.

http://servicenotes.literature.keysight.com/litapp/SearchSN.do?method=openExternalSNSearch&prodNum=34420A

 - RS-232 communication can be solved by replacing U700 ( Intel N83C51FA  with 34401 marking )chip, fortunately my unit is OK, otherwise it will require return to Keysight. Chip is in socket so I do not know why . Maybe that's why there is attached sticker on instrument.
BTW the version of this chip is the middle number in revision menu.

- Firmware is stored in U502 (27C020  in PLCC32, Keysight blank can be ordered 1818-5187) in my unit is version 3.0 (first number in revision menu).Chip is also stored in socket so I do not understand why unit needs to be returned to vendor. That's a big difference compare to 34401A where both chips are soldered to board, where I understand that it is not customer job.
Please send me PM if you have more recent version of firmware ( >4.0).

[plesa]

After replacing obsolete HP HSSR8400 by Vishay LH1510AAB channel 1 is working properly.
What I found as a new issue for solving is permanent overload on lowest 1mV range on both channels.
I make low thermal short and check it on 10mV range, readout is below spec. 120nV on 10mV range and on 1mV is still overload
So it must be connected to low voltage amplifier and to 10000 gain setting only.Gain in this range is switched controlled by ASIC U101, resistor networks is OK.

 

[Kleinstein]

I first thing to check is if the amplifier is somehow oscillating (ouput of U306). The feedback path is quite long and thus many places thing can go wrong.  Candidates for failure are U303C and U303D.

p.s. Funny how they made a low amplifier with an LM358.

[ManateeMafia]

Sounds like my recently repaired K2182. It had a bad resistor network. A couple of the resistors were open and all the low ranges were in overload.

Good luck. If you need a comparison measurement, let me know.

[plesa]

Thanks, analog switches U303, U304 (DG411,DG211) looks OK. Both resistor networks as well.So I will check oscillations.
Is anybody interested in posting old firmware?

[ManateeMafia]

I will check my versions and upload what I have to xdevs. It won't be until later today.

[ManateeMafia]

Bad news, my rev 5 eprom is soldered to the pcb. It looks like it may have been replaced at one time. Another one has rev 6. I will open it later as I was gathering data for the Fluke 8808A thread.

[ManateeMafia]

Second 34420A with Rev 6 eprom is also soldered to main board.

[plesa]

Thanks for check. It was similar on 34401A that some units has socket and some directly soldered EEPROM to PCB. Version 6 of firmware seems to be the most recent one. I will run lon term GPIB test and we will see if there will be some GPIB communication issues related to older firmware.

When I inspected units I found that someone tried to fix channel 1 issue and desoldered Q128/Q129 FETs, I correct it and clean PCB. And I removed wire connections U313 pin 2 and C451 bypass cap, which looks like from unsuccessful repair.
But it was mistake after putting the wire back the overload on 1mV range disappeared.

Another issue what I found was on channel 1 and 10V range on autorange, sometimes the readout was 20-50mV.
After replacing U120 HSSR8400 optocoupler by Vishay LH1510AAB everything works like charm.
To prevent other issue I replaced all opto-couplers, but two additional without further improvement.

On 2W ohm measurement seems to be offset 0.1 Ohm, but this I hope will be removed during calibration (still in spec) . 4W Ohm works ok ( 20uOhm stable readout, spec is 17uOhm).Quite good results for unit which was calibrated 20 years ago
With inputs shorted there is 48.5 uV on Channel 1 and 82uV on Channel 2 readout, which is according to spec OK.
 

[ManateeMafia]

This is good news. After a long burn-in, we can collect some data on your 34420A. I can also do the same on mine as my 3458A testing has collected enough data to show how they are drifting. 3 of 4 tested fine. The last one is almost as bad as TiN's high plains drifter.

[plesa]

I just assembled unit and overnight I would like to take first data set.

[plesa]

Data collected by BenchVue. Channel 1 noise 13.1nV, offset -8.5nV, Channel 2 noise 8nV, offset 28nV.
Also I receive confirmation from Lemo about connectors:
 
For cable FVN.1S.304.CLYC37Z price 50 EUR
meter input connector is EVP.1S.304.CLY price 40EUR
both with copper contacts not gold plated.

EZGPIB scpit for some reasons seems to not work with my unit,old firmware issue maybe.

[HighVoltage]

Did the 2W Ohm reading correct the offset after calibration?

[plesa]

Did the 2W Ohm reading correct the offset after calibration?

Yes, all offset corrected after performing Zero calibration. Now the offset in 2W Ohm measurement is 125 uOhm!!
Also the offset in DCV were corrected to -3nV on both channels. Currently I'm working on LabView program for data collection.
I must say that I'm impressed by this meter.

[alanambrose]

Re noise results - this seems to be the spec (rms at 10nplc / low impedance source):



That is, lowest ppm noise on 10V range. 2-3x on other ranges except 1mV which is ~30x higher.

And here are my early results - with a dodgy short as I don't have any proper connectors yet:



Interestingly, lowest on 100mV range, with other ranges <1.8x that. The 1mV range seems good. My 1V and 10V ranges look out of spec, curiously.

[alanambrose]

BTW anyone have an actual 34103A short that they don't mind taking the case off? I'm reluctant to pay $100 extra to have someone solder a couple of copper links into a FVN.1S.304.CLYC37Z Lemo. I was expecting this to be a crimp connection, but it looks like 'solder bucket' to me.

TIA, Alan

[Vgkid]

I believe in the Keithley info sheets, it says that they are soldered. I believe it was SN95AG5 solder, apparently what they seitched to after cadmium.

[alanambrose]

Ah yes thanks...

Model 2188 Low-Thermal Calibration
Shorting Plug: A low-resistance, low-thermal
shorting plug required to calibrate the Model
2182A Nanovoltmeter. Made of a LEMO connector
with its pin connectors shorted together
using copper wire and Sn/Ag4 silver solder.

p.s. for the Europeans, the best place I've found for 96/4 is:

http://www.reichelt.com/Solder/2/index.html?&ACTION=2&LA=2&OFFSET=500&GROUPID=557&SORT=artnr&SHOW=1

i.e. the Felder Sn96 Ag4 F-SW32

[alanambrose]

OK my first cut of the noise floor. My meter reporting firmware versions: 9.0-5.0-2.0 btw.

(edit) Added the correct analysis.

[plesa]

BTW anyone have an actual 34103A short that they don't mind taking the case off? I'm reluctant to pay $100 extra to have someone solder a couple of copper links into a FVN.1S.304.CLYC37Z Lemo. I was expecting this to be a crimp connection, but it looks like 'solder bucket' to me.

TIA, Alan

Of course. It is soldered. I cannot check the solder used, maybe later. For investigation I have also the input connector which is also soldered.

[alanambrose]

Ah thanks so much - all the contacts are joined rather than just the two pairs?

(edit) Duh, of course, for 4W ohms...

[Goff]

Hello,

I'm new to the forum here and

already have a difficult question.

Recently bought a "new" 34420A directly from Keysight Germany.

Question: The meter has an inherent noise,

it buzzes (not loud but not quiet),

can anybody confirm this?

Here are a few dates:

Keysight 34420A

- Green Display, white Keysight Logo

- Date of first Calibration: Jul 2016 in Malaysia

- Revision 10.0 - 5.0 - 3.0


Thanks in advance for any answer!

with friendly greetings from Germany,

Goff

[HighVoltage]

Hello Goff and welcome to the Forum.

The 34420A should not make any noise. The instrument does not have any fan (like the 34401A) and the noise that you describe might be resulting from a noisy transformer.

I would contact Keysight in Böblingen and I am sure they will exchange to instrument to your total satisfaction.
 

[plesa]

My meter is silent.Does not matter if new one or 20 years old one.
 Maybe the transformer shielding is not properly fixed.

[Goff]

Hello plesa,

thanks for the information / answer.

... HighVoltage - Yes, the complaint was made.

[MegaVolt]

Good afternoon.
Is there a description of the change history of firmware 34420a?
Has anyone seen on the network firmware files newer than 5-5-2?
Are there any instructions and successful experience updating the firmware?

I would be grateful for any information.

[HighVoltage]

My two 34420A have following revision numbers:

Agilent
SN:   MY42004xxx
REV: 10.0-5.0-2.0

Keysight
SN:   SG42000xxx
REV: 10.0-5.0-3.0

I would have expected a unit with higher SN to have a newer FW Revision.
However, I do not see any differences in the functionality of the two instruments.

[MegaVolt]

My two 34420A have following revision numbers:

Agilent
SN:   MY42004xxx
REV: 10.0-5.0-2.0

Keysight
SN:   SG42000xxx
REV: 10.0-5.0-3.0

I would have expected a unit with higher SN to have a newer FW Revision.
However, I do not see any differences in the functionality of the two instruments.

I met mention of a maximum belief of 10-5-5.

Do you have the opportunity to read the contents of the flash?

[branadic]

Sorry to pick up that old thread, but has anyone ever created a firmware dump? I'm particulary interested in the firmware Rev3.0.

-branadic-

[branadic]

I was sensitized to check my 10 V range with a battery powered LTZ reference to see if everything works as it's supposed to work. There was strange behavior reported by someone on a unit that was in for service. So I monitored the unit as suggested and the unit I have seems to do well. No excess noise, no strange jumping around, everything seems to be fine which indicates that my unit was a bargain.

Edit: The test was performed at 10 NPLC and the filter turned off.

Code: [Select]

hp34420A.write("CONF:VOLT:DC 10;:SENS:VOLT:DC:NPLC 10;:INP:FILT:STAT OFF;")  # 10V range, NPLC 10, Filter off
-branadic-

[HighVoltage]

Looks nice!

Now you need to do the same test on the other side of the spectrum in the nano volt rage with a stable 1 nV source.

[alm]

a stable 1 nV source.

I think that's called a short

[Kleinstein]

The stability looks OK. Much of the lower frequency noise is likely from the LM399 reference in the 34420 and thus the popcorn type jumps of some 0.4 ppm.
The 10 V range is a bit noisy, but that is a limitation of the ADC and there is not much that can be done there. This is not really a problem for the low ranges (1 mV and 10 mV), where this meter is good for.

A test at the other end of the range would be either a short, or a 1 mV/ 10 mV source to be near the end of one of the lower ranges.
The more interresting part would likely be the short and maybe some extra series resistance (e.g. 10 or 100 K) to see if there is drift / noise in the input current.

[HighVoltage]

a stable 1 nV source.

I think that's called a short

The Keithley 260 Nano Volt Source is your friend for such measurements.
Range: 0.01 nV to 1.11 V

I can dial in 10 nV stable on my 34420A with the Keithley 260.

Maybe we can call 0.01 nV a short.

[TimFox]

The last voltage divider stage in that Keithley source is made of copper, despite its temperature co-efficient of resistance, in order to minimize thermal emf.

[alm]

What's the noise like at 10 nV from the 260? The 34420a is specced for 1.3 nVrms, which I imagine is under absolutely ideal conditions re thermal gradients and cleanliness of connections.

I actually had the Keithley 260 in the back of my mind when I wrote it, but I was in doubt even that can deliver a stable 1 nV outside a physics lab.

[MegaVolt]

Since both resistors in the dividers are made of the same metal, their temperature coefficient is compensated.

The noise level is determined by the output impedance. On the nV range, this is only 1 ohm. This is very little noise.

[Dr. Frank]

Two years ago, I had a 34420A for repair on my bench.
Using the same configuration and settings as branadics, w/o filter, it showed random excessive noise, triple the usual level, i.e. rms 7.6µV vs. 2.6µV.
It had the same FW version 2.0.
I measured stability of its reference, at the output of the input amplifier, with different GPIB adapters, and so on, and in the end could not identify the root cause. With branadics measurement, I now can rule out a FW error.
The 2nd graph shows the calculated digital filter, like it would be implemented in the 34420A.
Frank

[HighVoltage]

What's the noise like at 10 nV from the 260?

I don't remember and would have to measure this again.

The Keithley 260 is well built inside for low EMF and low noise.
Look at these impressive copper connections..

[okti]

Here's my trusted old HP34420A at DCV10 NPLC10 FILTEROFF (source: LTZ1000) measurement fresh from the oven:

[okti]

Same meter, DC1mV NPLC10 FILTEROFF with short:

[branadic]

The 10 V range is a bit noisy, but that is a limitation of the ADC and there is not much that can be done there. This is not really a problem for the low ranges (1 mV and 10 mV), where this meter is good for.

Would you mind explaining why you think that the ADC is the limiting factor here instead of the reference? To put some numbers to it, LM399 is typ. 7 µVrms, max. 50 µVrms noise (values for LM399, not LM199 in LT datasheet). Take that times ~1.4 for the +/-10 V reference voltage in the unit, which is the typ. 10 µVrms that we are familiar with. The noise I see is smaller than that, indicating the noise of LM399 is smaller here.
Has there been any report to the noise limit of a multislope III ADC, like feeding it with lower noise reference such as LTZ, that I've missed?

-branadic-

[Kleinstein]

The lower frequency noise (after the 50 points filtering) is to a large part the LM399 reference, though there could also be some 1/f noise from the resistors.
The higher frequency noise seen for the single 10 PLC conversion looks quite high for a LM399 reference:
I have looked at my test with an external LM399 reference and LM399 ref. at the ADC and there I get around 6 µV_peak to peak in between the popcorn events even with 1 PLC and some 12 µV peak to peak including the popcorn noise / longer time scales. 2 x LM399 reference at both ends would be comparabele to the 7 V scales up to 10 V. I have some reference filtering, but this would no longer be effective for 10 PLC and only 1 reference.

The reportet noise for the 34401 that I found is somewhere around 2-2.5 µV RMS ( 12-15 µV peak to peak) for 10 PLC and I don't see much of the noise coming from the input stage, though there is a little. So this would explain much of the observed noise.
I would be easy to measure that noise with a short.

I have not seen much theoretical analysis on the 34401 type ADC noise, though I have just done some calculations on that. There are a few uncertainties, and my calculations still get a slightly lower noise ( some 1.4 µV RMS) from the noise sources I included. The noise is not just one large noise source, but more like a combination of many sources. From my analysis the worst offenders are the current noise of the OP27 in the integrator, higher frequency reference noise, jitter and quantization noise. In theory one could improve on the current noise and a little on jitter, but not with many other noise sources, especially not the quantization noise. 

There is one unknown in how the ADC is handling the first / last reference steps - that is the reference signal just before the auxiliary ADC is read. This part naturally will not be 100% effective and there is a chance that this factor is not perfectly treated in the firmware. This would be some extra discrete size errors, that can look rather random, though also changing with the input voltage. So some readings may be effected more than others.
Ideally there would be a procedure in the factory calibration to minimize this error part, but I am not so sure they actually do this and if so, that factor may also change over time.

[branadic]

I searched the web and found that report here: bbs.38hot.net/thread-69139-1-1.html where someone replaced the LM399 in 34401A by an LTZ1000 and noticed an improved noise behavior as well as improved t.c.. As 34420A is a somewhat pimped 34401A chances are that replacing it with a lower noise ADR1399 will slightly improve noise in 10 V range without to much hassle.

-branadic-

[Kleinstein]

The translated BBS38hot text is about a 34410, so a newer version with a completely different meter.

As far as I see it the largest noise part in the ADC is the current noise of the OP27. This competes with the input signal though a 100 K resistor and calculated back to the input the noise would thus be current noise times 100 K.  The 10 PLC AZ mode has an effective BW of 5 Hz because of the 2 conversions and  an effective frequecy of 2.5 Hz for the 1/f noise. The datasheet gives a current noise for the OP27 of close to 1 pA/sqrt(Hz) at 20 Hz that would translate to some 3 pA at 2.5 Hz with a square root of frequency relation.  However the DS says they measure the current noise with the resistor split over both inputs - which in my view is wrong: This suppresses the correlated part of the current noise and the uncorrelated current noise would only add as root. So at best the current noise figure is too small by a factor 1.4, and possibly more. So the current noise from the OP27 in the integrator would correspond to more than 3 pA/sqrt(Hz) *1.4*100 K = 420 nV/sqrt(Hz) of noise at the input. This would be 930 nV of RMS noise for the input (and possibly more from the correlated current noise part).

The 2nd largest noise part would be the auxiliary ADC with it's quatization noise (around 350 nV/sqrt(Hz) for the input). Here it is a bit unclear how noisy the ADC is and especially in how much the contineous changing input may effect the µC internal ADC. So this noise part may be also quite a bit larger.
The big unknown is how good the timing is when the auxilliary ADC is sampling:  so does it get the last run-up step fully included or only partially and may need an extra correction factor for the last step.

The 3rd largest noise I have in my list is the higher frequency ref. noise (could be some 150 nV/sqrt(Hz) for the input). The noise from jitter (likely mainly the 74HC4053) may reach a compable size.

The ADC mixes back some reference noise from the 50-150 kHz band back the the near DC band. There is some parial filtering from the relatively slow AD706 OPs, but not full filtering. There is a good chance that some capacitance or better capacitance with series resistance (similar to the suggested ARD1399 circuit) in prallel to the LM399 reference could reduce the higher frequency noise. So something like 1 µF+5-10 Ohms in parallel to the 7 V from the LM399. There is a chance this would reduce the reference noise on the 100 kHz frequency band. It is not the largest noise source, but also a relatively small change.