3458A Issues

[plesa]

HP 3458A is quite popular on this forum and I would like to ask for your experience with this meter.
Like how long you have it?
What is your meter drift between calibrations?
7 days drift (according to service note http://literature.cdn.keysight.com/litweb/pdf/3458A-18A.pdf)
Did you replaced any component in this meter or board(s)?

[HighVoltage]

I am kind of new to the 3458A and over the first couple weeks I did the CAL 72 stability test as described in the SN18A.
And the drift was 0.11ppm per day over 7 days.

My 3458A was built in 2012, sold in 2013 and never really used, may be not used at all.
So, right now I have it ON all the time and will wait a few more weeks or month and then I will have it calibrated officially.

[plesa]

I forgot to add my CAL 72 stability test data, in 7 days it is fluctuating -0.023 to -0.011 ppm/day. Meter internal temperature is fluctuating 0.5°C.
The difference between calibrations was -0.9 ppm/6 months for Vref and  +1.26ppm for Rref.
Unit 3458A/002 (4ppm).

[ManateeMafia]

This is from a 3458A OPT 002. Of note, the A1 board was replaced before they sent it back but this is AS RECEIVED data.
I don't have any CAL?72 data. I sent this meter in for cal within a year of buying it.

Here are my 10V DC range results from Keysight.
MIN ----MEASURED---MAX
2014  Jun
 9.9999349   9.9999921   10.0000575

2015 Jun
 9.9999292   10.0000089 10.0000518

2016 Sep
 9.9999431  10.0000112  10.0000657

[TiN]

Data from 31 May 2016 to 20 July 2016 (21 day span)
Average delta -0.055 ppm.

[Alex Nikitin]

I'm getting the 3458A Opt 002 today or tomorrow back from a calibration at Keysight UK and the calibration results are available on their site already. The meter was previously calibrated in January this year, however the calibration data were taken by a different company using the Fluke 5500 (with 50ppm uncertainty  ) .The good news is that the meter passed the calibration without any problems. The errors on the resistance ranges are very small (essentially below 0.5ppm from the expected value for 100 Ohm, 1K and 10K, 1.1ppm for 100K , 2ppm for 10 Ohm, 3ppm for 1M and 5ppm for 10M). Errors for DC voltage are very consistent across the ranges - the meter reads about 2.2ppm low on 1V, 10V and 100V, and 3.5ppm low for 1000V, which might be an indication of a reference drift since the last adjustment (plus an error in the 10V reference used at that time) .

I am planning to run CAL? 72 check over next few weeks.

Cheers

Alex

[plesa]

Data from 31 May 2016 to 20 July 2016 (21 day span)
Average delta -0.055 ppm.

I performed everyday measurement and it looks like over 10 days there is strong convergence to 0.005 ppm / day in last 6 days.It takes 2 weeks to settle down.
I will probably add sequence for DCV ACAL into the logging script ( every time is internal temperature change by more than 1°C and every 24h )

I am planning to run CAL? 72 check over next few weeks.

We needs to agree how to collect and interpret data. Daily DCV gain followed by DCV ACAL?

[Alex Nikitin]

We needs to agree how to collect and interpret data. Daily DCV gain followed by DCV ACAL?

OK, I'll try to do that (except for weekends though, as the 3458A is at my work lab).

Cheers

Alex

[dacman]

We have several at my day job.  I'd say we've had at least one for maybe 12 years.  (Our original one was a used Hewlett-Packard, and we still have it and it works just fine.)

The worst one drifts about 3 ppm a year (for dcV).  Some drift less than 1 ppm in that time.  (I really do not track Ohms drift, although I could look it up.)  We have a 3458A/002 and it doesn't do any better than some of the non-002 units.  Most all would pass the 002 specifications if the artifact were performed once a year.

We don't track the (7 day) A3 drift.  (Maybe we should.)

If one is OOT (after calibration) or giving errors, we will usually send it to Keysight.

Keysight likes to replace the calibration RAMs no matter what the problem is (I'm not sure why).  The battery is in the RAMs, and therefore, to replace the battery the RAMs need to be replaced.  We did send in one unit that would not pass at Zero Vdc (even after performing CAL 0) and the only thing they did was replace the calibration RAMs.  (The unit would pass every other check.)

If a multislope rundown convergence error occurs, it is most always due to the A3 card (which is where the A/D converter is located).  Error codes that tag along with this error will indicate what the 3458A was trying to do but may seem random because lots of functions use the A/D converter.  If you replace this card yourself, then I would recommend inspecting the fiber-optic connections for foreign objects.

I've repaired several other things myself, such as replacing the display or terminal board or broken pushbuttons and associated hardware.  (The six binding post terminal board is one unit, and is the same on both the front and rear.)

[branadic]

A bigger issue is a nice working software, doing all the essential communication and stuff everyone with a 3458A needs. The solution using a Raspberry Pi is not the golden way in my opinion. I wonder why noone has yet programmed a good software with all the stuff needed including graphical displaying of trends and data saving.
But the things needed sound like a Matlab based GUI program cold do the job.

[plesa]

A bigger issue is a nice working software, doing all the essential communication and stuff everyone with a 3458A needs. The solution using a Raspberry Pi is not the golden way in my opinion. I wonder why noone has yet programmed a good software with all the stuff needed including graphical displaying of trends and data saving.
But the things needed sound like a Matlab based GUI program cold do the job.

I'm using 3458A for monitoring voltage and do not want to PC running only for this. So Raspberry is good option. And what if you have prepared website on RPI, where you will only set the commands of you can choose from some Python scripts and run them over web interface? for example. What do you think?
Ammount of 3458A users here is quite large and by this we can make platform commonly shared and easily available for all ( RPi+ almost any GPIB interface including Prologix)
https://d3js.org/ provide quite nice graphs including histograms on web interface, logs can be accessible over FTP.
To RPi you can easily connect BME280 humidity/pressure/temperature sensor, like TiN have on xDevs.
Nice GUI I can program in Labview, but it will require PC/Computestick or similar PC compatible solution.

If one is OOT (after calibration) or giving errors, we will usually send it to Keysight.

How many repairs did you have with 3458A?

Keysight likes to replace the calibration RAMs no matter what the problem is (I'm not sure why).  The battery is in the RAMs, and therefore, to replace the battery the RAMs need to be replaced.  We did send in one unit that would not pass at Zero Vdc (even after performing CAL 0) and the only thing they did was replace the calibration RAMs.  (The unit would pass every other check.)

That's strange. In case they want to prevent loosing data I will expect to swap boards and replace Dallas chip based A5 board with most recent A5 boards revision with FPGA.
Or your instruments can affected by http://literature.cdn.keysight.com/litweb/pdf/3458A-20.pdf Who knows...

[dacman]

If one is OOT (after calibration) or giving errors, we will usually send it to Keysight.

How many repairs did you have with 3458A?

My guess is about half the units get repaired (parts replaced) every ten years.  Some of the repairs are due to damage.

Keysight likes to replace the calibration RAMs no matter what the problem is (I'm not sure why).  The battery is in the RAMs, and therefore, to replace the battery the RAMs need to be replaced.  We did send in one unit that would not pass at Zero Vdc (even after performing CAL 0) and the only thing they did was replace the calibration RAMs.  (The unit would pass every other check.)

That's strange. In case they want to prevent loosing data I will expect to swap boards and replace Dallas chip based A5 board with most recent A5 boards revision with FPGA.

I may have overstated that one, but Keysight has replaced (what they call) the calibration RAMs apparently due to what was printed on them, and I don't know if it's because of the date code or due to a troublesome batch.  And I also do not know what they replaced them with.
 

[Alex Nikitin]

Here are my results over last 20 days (not measured daily, but 11 points of CAL?72 taken). If I take just end points over 20 days period, the drift rate is good at 0.016ppm/day. If I take the extremes at the same internal temperature  with 8 days between these points I get 0.27ppm/day .

Cheers

Alex

[plesa]

Thanks Alex.
Join to RPi logging activity and you can have such graph prepare without Excel Just open webpage.
https://www.eevblog.com/forum/metrology/raspberry-pi23-logging-platform-for-voltnuts/
On my meter trendline has slope +0,01ppm/day

[CalMachine]

I feel like most of the new agilent and keysight labeled 3458a's are fairly cheap compared to when HP produced them. 

[TiN]

HP had MSRP about 6600$ back in 90s which is about same as todays 9800$ for base config. Options like 002 or HFL can add few K$ on top .

[Pipelie]

luckily, my 3458 looks stable for now.

[TiN]

What is xxx value on the graph? If it's CAL? 72 constant, and you follow SN18 procedure for test, than it's not stable at all..

[Pipelie]

What is xxx value on the graph? If it's CAL? 72 constant, and you follow SN18 procedure for test, than it's not stable at all..

Hi ,TiN

I did  follow SN18 procedure for test, attach the data.  drift/ppm is the result of 3458 stability test.
xxx is another  10V  Voltage standard,