Agilent 3458A GPIB Data Aquisition software and options ?

[HighVoltage]

I finally bought an Agilent branded 3458A in hopefully good working condition and without the troublesome drift that some people have been talking about here on eevblog.
So far, the instrument has not arrived.

I will probably have this instrument calibrated by Keysight and then use it as a reference for my other DMMs and as a calibration bases for many 34401A, that I have in the field with customers.

Now I want to prepare for data acquisition with this DMM and compare it to my many other Keysight DMMs.
So, I looked around and found a few options.
I do have the original 82357B GPIB to USB adapter available.


BenchVue
Unfortunately, BenchVue does not support the 3458A


RF Scientific
This is a GPIB data logger software that was recommended to me by another 3458A user.
But, this software allegedly does not work with the original Agilent 82357B GPIB to USB adapter.
It requires the Prologix GPIB to USB adapter
http://prologix.biz/gpib-usb-controller.html
Can anyone here confirm, that this is true and that I can not use my Agilent adapter with the RF scientific software?

What other software options are there to collect data from the 3458A and also control the instrument remotely?
The RF Scientific software seems to be great, especially since it can be used to control multiple instruments at the same time.
It is just a bummer, that I can not use the Agilent adapter?

Any other ideas and recommendations?

[TiN]

How do u know if it's not drifty, if it was not received yet?

Also what's wrong with easiest linux-gpib + RPi way? Your hpak adapter should work just fine.
Sorry, site is temporary down so I can't bombard you with spamlinks yet.

Anyhow, welcome to no sleep no ppm club. That 3458 will suck up all your time now, be sure.

[carl_lab]

At the moment, I can't help you much, but did you try this?
http://rfscientific.eu/kontakt
The software looks nice,  and it is freeware.   

BTW: I have the same Agilent GPIB-USB adapter and want to log several 3458A over time to compare their drift.

...welcome to no sleep no ppm club. That 3458 will suck up all your time now, be sure.

No problem, if you can do all the testing at work... 

[HighVoltage]

How do u know if it's not drifty, if it was not received yet?

Also what's wrong with easiest linux-gpib + RPi way? Your hpak adapter should work just fine.
Sorry, site is temporary down so I can't bombard you with spamlinks yet.

Anyhow, welcome to no sleep no ppm club. That 3458 will suck up all your time now, be sure.

Well, what I meant to say was that it is "hopefully" not drifting. I heard that even newer models are drifting. So, fingers crossed until I am testing it.

Yes, I searched google and your website came up a lot, just a dead end right now.
There is a saying: "If you want to get something done, ask a busy person"
I am very busy with so much stuff but I am also looking forward, playing with this meter.
Hopefully this will solve my "two watches" problem.

When will your website be up again?

[HighVoltage]

At the moment, I can't help you much, but did you try this?
http://rfscientific.eu/kontakt
The software looks nice,  and it is freeware.   

I have the same Agilent GPIB-USB adapter and want to log several 3458A over time to compare their drift.

...welcome to no sleep no ppm club. That 3458 will suck up all your time now, be sure.

No problem, if you can do all the testing at work... 

You have more than one 3458A at work?

Email was sent to their support site, may be the answer is positive.
But since the driver for the Agilent 82357B GPIB to USB adapter is installed with the I/O library, I would expect this adapter to also be working easily.

[carl_lab]

You have more than one 3458A at work?

Yes. 
And they all seem to behave different... 

[joeqsmith]

Did you get one with the oven?  These are nice.

Because anytime I have need to use one of these, these tests are always custom. I roll the code in Labview using  the newer version of National Instruments Ethernet GPIB controller.   

[HighVoltage]

Did you get one with the oven?  These are nice.

Don't they all have an oven?
I think, - but I am not sure- , that the higher stability option 2 has only a differently set oven temperature?
But the LTZ1000 is heated in all of them, as far as I have read so far.

[G0HZU]

What other software options are there to collect data from the 3458A and also control the instrument remotely?

The easiest/best way is to create something in excel to log over GPIB or write a few lines of code in VB to do it.  Or use Python or something similar.

It looks to me like that RFScientific logger program just date stamps and logs each DMM reading and sticks the result in a spreadsheet every few seconds? Maybe it can log faster than this but that program looks like some kind of slow and boring DMM comparison/crosschecking program for people who have far too many bench DMMS and nothing better to do than log them against each other for hours/days at a time.

[carl_lab]

Did you get one with the oven?  These are nice.

Don't they all have an oven?
I think, - but I am not sure- , that the higher stability option 2 has only a differently set oven temperature?
But the LTZ1000 is heated in all of them, as far as I have read so far.

That's correct, option 002 has reduced temperature setting for lower long term drift.
(PS: I'm not sure about it anymore, maybe I was wrong...)

I think here are some specialists, who made a modification (changing a resistor) on their "normal" 3458A's to have the same effect.
I'd like to know, if this modification is as good as the manufacturers option 002 (maybe manufaturer also uses harder selected LTZ1000)...

[joeqsmith]

Did you get one with the oven?  These are nice.

Don't they all have an oven?
I think, - but I am not sure- , that the higher stability option 2 has only a differently set oven temperature?
But the LTZ1000 is heated in all of them, as far as I have read so far.

I have never had one apart or looked for a service manual.  I was guessing the high stability was a double ovenized setup but really don't know. 

[acbern]

The lower drift version opt 002 is just selected parts, no change in temperature. An old, well aged 3458A will have the same or better bahaviour by today. You just cannot claim it is low drift, unless you qualify that yourself. And opt 002 only improves DCV. You can get even better accurcy using a standard 3458A and doing relative measurements to a voltage reference such as the 732. Thats what I do anyway if I need to do precision measurements (better 2ppm abs).

[edpalmer42]

Some other software options are:

NI Labview Home ($50)

or

EZGPIB (free).  It's a Pascal-based interpreter with GPIB extensions.  One of the example programs is for an HP 3457.

I've also heard of PyVisa and pylt, but I don't know anything about them.

Ed

[TiN]

The lower drift version opt 002 is just selected parts, no change in temperature. An old, well aged 3458A will have the same or better bahaviour by today.

acbern is right on money. All three reference modules are different just thru selection. Selected best ones go into HFL (2ppm/y, ~1.7K$ from HPAK site for A9 reference module), second best are Opt.002 (4ppm/y, ~1k$) and standard are for rest (8ppm/y). They all running scorching hot 90-ish C via 15K/1K temperature setpoint resistors. I'm not aware of any other difference on A9 reference between the three, except the selection of LTZ1000A's. There is also important to watch out for ones with pink noise shots (random jumps in voltage, usually 0.3-1.5ppm scale). These can be detected by monitoring known clean and stable LTZ module or similar 10V reference.

I added 100K in parallel with 15K long ago , to bring temps down ~50c, as my max room temperature is around +35c tops. Lower temperature allow to reduce aging rate, which I decided worthy, since my meter is running 24/7 non-stop since repair was complete.

Beware that you will lose calibration after the mod, as reference output voltage shifts a lot.

EZGPIB (free).  It's a Pascal-based interpreter with GPIB extensions.

Yup, I used that one for DMM noise testing project we had while ago. I have program to run on 3458A for it as well.

Server will be back tonight, once I come back Taiwan from business trip and check on it. Murphy law gotchas, everything runs months once I'm home, and breaks second day I leave the country. 

[carl_lab]

The lower drift version opt 002 is just selected parts, no change in temperature. An old, well aged 3458A will have the same or better bahaviour by today.

acbern is right on money. All three reference modules are different just thru selection. Selected best ones go into HFL (2ppm/y, ~1.7K$ from HPAK site for A9 reference module), second best are Opt.002 (4ppm/y, ~1k$) and standard are for rest (8ppm/y). They all running scorching hot 90-ish C via 15K/1K temperature setpoint resistors. I'm not aware of any other difference on A9 reference between the three, except the selection of LTZ1000A's. There is also important to watch out for ones with pink noise shots (random jumps in voltage, usually 0.3-1.5ppm scale). These can be detected by monitoring known clean and stable LTZ module or similar 10V reference.

OK, then I have mixed up your modification with option 002, I'm very sorry!

I added 100K in parallel with 15K long ago , to bring temps down ~50c, as my max room temperature is around +35c tops. Lower temperature allow to reduce aging rate, which I decided worthy, since my meter is running 24/7 non-stop since repair was complete.
Beware that you will lose calibration after the mod, as reference output voltage shifts a lot.

The resistor you used additional, need to be special high stability type?
If you modify temperature setting, will a selfmade artifact calibration be enough to correct?
Or is a full factory calibration/adjustment needed?

Sorry for OT, but I could not find the original modification thread using the forum's search function.
So if moderators want to move modification posting(s) to modification thread, please give me the link to it...

[acbern]

When you change the resistor defining the temperature of the LTZ, a new artifact adjustment is necessary. This is anyhow the only way to adjust the meter (except for regular ACAL, which is important; it should be done more than just once a day when meter is in use, just to be sure). Now you have the issue that you formally should do a performance verification. This will confirm everything went right during artifact cal. There is a military paper saying performance verification after artifact cal is needed only every second time (to comply to certain mil requirements; forgot the link, but google will certainly find it). However, this will not be acceptable according to dakks and  probably not to any other national cal organization requirements.
For a home lab, different story. But if you go that route, you can as well do artifact call every month or so and have a very low drift instrument without mod.

[HighVoltage]

acbern,

This is very interesting about the different models:
So, let me summarize:

- Normal version: 8ppm/y
- Option 002: 4ppm/y
- HFL: 2 ppm/y (Was only sold by Fluke, right)

And all three versions are running with the high temperature in the LTZ1000?
And the only difference is for Keysight to select the lowest drift reference for the best HFL meters, after they have aged them for a long time in the factory, may be many month or years?

What about all these 3458A reference boards showing up on ebay?
Are those mainly rejected boards or how is it possible that so many of them are on the market?
Or is it possible to find a 2ppm version under those?

[Dr. Frank]

HighVoltage, congratulation to your 3458A!

The LTZ reference modification, 100k in parallel to the 15k, requires a low T.C. resistor, 3ppm/°C is sufficient.
This brings the oven temperature to about 65°C, like in the LT datasheet, which should be good for <2ppm/ year.

As the initial stability should have been < 8ppm/year, by HP specification, and the temperature reduction is about 30°C, even a factor of 8 better stability may be expected.
All these numbers are valid for continuous operation, 24/365.
If the instrument is switched off, there will be practically no drift.
Hysteresis might happen with the original references, see AN-18, which is also less probable at lower oven temperatures.


You have to take care that the interior does not heat up in total (TEMP?) to more than about 50°C.
This limits the maximum room temperature to < 35°C, the fan has always to be kept clean, and the instrument should not be stored in a closed rack.
In other words, this modification is for lab / metrology environment only.

It's not really known, what modifications were done in the Fluke HFL3458A version.
Maybe Fluke used their own voltage references, based on the LTFLU.


Frank

[acbern]

highvoltage,

most old and still good voltage reference by today, especially when operated at lower temp with parallel resistor mod, will most likely be equivalent in drift to the opt. 002. reason being, keysight selects them new and extrapolates drift. and references drift more when new.
those on ebay are a gamble, because their origin is unknown. never bought one.

in your case, what I would do, is to validate your instrument as-is (potentiall with the mod, but no upgrade to opt 002). you would need a stable, calibrated voltage reference (732/4910), or access to it to calibrate the unit and check the drift. According to PTB, if you do that three times, and have a drift history, you can use this as the drift rate going forward. Unless you bought a bad unit, I am sure you will be 002-equivalent in drift rate, especially if you do the mod. you need to establish this, as you use it for your customers, so you need some evidence.

or, if you want to spend the almost 1k (for a opt002 upgrade), i would rather invest this in a 732 or so.

[HighVoltage]

Dr.Frank
Thanks for these great explanations.

As you know, I have this external LTZ1000 reference that you qualified for me once.
Can I use this reference to analyze the drift of my 3458A, once I have it?
Or is there a better way, since I don't have a 732?

May be someone that has a HFL3458A at work, can do a teardown and find out if there really is a LTFLU reference installed, as is in the DMM7510 from Keithley.

acbern
This means that eventually all 3458A meters will end up to be equivalent to a new one with option 002.
Does this also mean, that those units with option 002 will eventually be like an HFL3458A
If so, the $1k investment in to an option 002 upgrade would be wasted and as you said, much better invested in to a stable 732.
But then I have the problem of finding a stable 732.

Thanks for all the good hints.




 

[acbern]

I cannot comment about the HFL as I am not aware about the details internally. But generally, finding a LTZ itself, that drifts 2ppm per year, is achievable. Now there are other parts within the 3458A that drift and cannot be compensated by an ACAL DCV. Therefore, HFL stability may be a stretch.
Anyway, going with a 732 is the better aproach, because validation is doable any time. When I do precision mesurements based on comparison to a voltage standard, so I dont care about the drifts of my 3458As. That way I can get around 2ppm abs. any time a year on 10V and 1V. Finding e.g. a good 732 is a challange though, trial and error to a certain extent. When I finally got mine, it was my third.

[TiN]

I don't think HFL DC reference was anything but preselected LTZ1000 A9. Why I think so?
Because HPAK site have part "03458-66529, PC Assembly-DC REF Tested 2 PPM,Orderable,   US$ 1931.00".
So if someone want 2ppm selected reference and has extra 2K - here they are

HighVoltage I'm back home, and site is up. I had messed up SSL-key on my nginx config 

[Dr. Frank]

Dr.Frank
Thanks for these great explanations.

As you know, I have this external LTZ1000 reference that you qualified for me once.
Can I use this reference to analyze the drift of my 3458A, once I have it?
Or is there a better way, since I don't have a 732?

Thanks for all the good hints.

HighVoltage,
By means of an external LTZ1000 reference, you will be able to check for the short -term drift of the 3458A, like described in AN-18, or like checking A3, what TiN has done.

A >24h datalog of the external reference with one initial ACAL, but none meanwhile, will reveal any excessive drift of U180.

A single external LTZ will not identify long-term drift rate, as you would be the man -with-two-clocks only.
A third reference, preferably a SZA263 based, would be necessary to decide, and probably would allow to measure the 3458As drift, if it would run continuously.

Frank

[TiN]

What about all these 3458A reference boards showing up on ebay?

A9 I bought from eBay (STD option) had ~1.2ppm pink noise and random jumps. After replacement of LTZ chip - that was solved.
Interesting enough - "bad" noisy chip was later checked on my KX reference board and it had no jumps. However direct comparison is not possible, as it was running different zener current and also had 400K compensation resistor, just like LTZ1000CH reference design, instead of no resistor for ACH chip (which is from 3458A's A9 board).

If you can't find datalogs in my repair log, I can post them here later.

For the setpoint mod I used wirewound card resistor, ~3ppm/K.

P.s. GPIB article for 3458A.

[Alex Nikitin]

Now I want to prepare for data acquisition with this DMM and compare it to my many other Keysight DMMs.
So, I looked around and found a few options.
I do have the original 82357B GPIB to USB adapter available.
...

Any other ideas and recommendations?

I'm using EZGPIB and the program from TiN's site for our HP3458A with the original 82357B GPIB to USB adapter. It allows you to set the measurement parameters and log the data. Works fine and sufficient for my needs at the moment.

Cheers

Alex

[HighVoltage]

Thanks for all your suggestions, I will try different ways, when I have my 3458A at hand.
I will probably not do a resistor change in the beginning, may be down the road....
First I will keep it running to test the stability and drift of the 3458A as it comes.
For that I probably need some high end cables and not my usual pvc or silicone insulated test cables.

[TiN]

You can use single-strand shielded copper UTP cable, like CAT6 Ethernet. It works just fine for DCV stuff.

[Dr. Frank]

..
First I will keep it running to test the stability and drift of the 3458A as it comes.
For that I probably need some high end cables and not my usual pvc or silicone insulated test cables.

What you need much more importantly, is a room with stable temperature. (+/- 2°C over the year, +/- 0.2°C over the measurement run).
Can you use a basement for that purpose?

Frank

[HighVoltage]

Can you use a basement for that purpose?
Frank

No basement where I live, because of high ground water levels.
But I have one room in the lab that has no windows and is pretty stable in temperature changes.

The 3458A arrived and I unpacked it.
It seems the meter was never used. No dust on the fan at all and protective screen was still on the display.

After a few hours of warmup, I hooked up my older GellerLabs 10V Reference (Not the SVRT) and look what is shows:

[TiN]

She's the virgin! Nice score. I bet some folks are seriously jealous now  .
But..... you broke EEVBlog rule - "Don't turn it on, take it apaaaart" 

[HighVoltage]

She's the virgin! Nice score. I bet some folks are seriously jealous now  .
But..... you broke EEVBlog rule - "Don't turn it on, take it apaaaart" 

Actually:
"I took it apart before I turned it on"
Interestingly, you need a Philips PZ2 and two different Torx tools to take the cover off.
I just removed the outer cover and left the analog shield in place.
Absolutely no dust inside.
You are right, it really seems to be a virgin!

[HighVoltage]

Even the fan blades are clean as new.

[ManateeMafia]

It's nice to see one in great shape.  Congratulations.

No basements here in Florida either but you can make do with late night runs when the outside temperature is somewhat stable.

[TiN]

Made in 2011 too, great.

Interestingly, you need a Philips PZ2 and two different Torx tools to take the cover off.

More over, that's actually all you need to fully take everything apart.
Perhaps hires photos of analog section are too much to ask? 
I'd understand if you decide not to take shields off.

Now let it run for a week and get that software and GPIB ready for some long term logging. I'd suggest running SN18 test too, to make sure A3 is good and stable  .

[HighVoltage]

Made in 2011 too, great.

Interestingly, you need a Philips PZ2 and two different Torx tools to take the cover off.

More over, that's actually all you need to fully take everything apart.
Perhaps hires photos of analog section are too much to ask? 
I'd understand if you decide not to take shields off.

Now let it run for a week and get that software and GPIB ready for some long term logging. I'd suggest running SN18 test too, to make sure A3 is good and stable  .

After a while I will remove the analog circuitry shield and take some high res pictures.
Right now I want to make sure it is stable and has no drift.
I read through SN18A, did a ACAL and wrote down my constant values

CAL? 72: 1.00543380
CAL?   2: 7.23733175

Will see, how they are after one week and after a few more ACAL procedures.
It is a surprise, how long the ACAL takes, like 10 min or so.
Compared to the 34470A, which takes only a few seconds.
 

[HighVoltage]

It's nice to see one in great shape.  Congratulations.

No basements here in Florida either but you can make do with late night runs when the outside temperature is somewhat stable.

Thanks,
I think the bigger problem in Florida is the high and constantly changing humidity. As far as I know, many houses in FL have complete air conditioning.
Would be nice to have a lab with temperature and humidity control.

[HighVoltage]

I just noticed, CALNUM? returns "1"
It seems, sometimes we can get lucky.

[ManateeMafia]

Maybe someday there will be a project to build a volt-nut environmental chamber big enough for a 3458A and a few DUT's? I am sure there are many of us that would benefit. Of course, having the room for one would be a bigger challenge.

[HighVoltage]

@HighVoltage: Congratulations !

Q.: How & where did you find a 3458A in such excellent condition ? (just kidding)
Good to see that this adventure has a happy ending ...

 

Thanks, Flinstone it was a good adventure.

It came also with this set of test cable.
It seems like it was hand made or may be it is a production cable of some sort.
The cables feel like teflon insulation, but I am not sure.
Does anyone here know of such a cable?

[joeqsmith]

VERY NICE!!!   Time to put that thing to use.

[TiN]

I know this was covered many times, but perhaps worth to give you few newbie tips on your box
You can run just ACAL DCV, it's about 1.5 minutes, if your measurement will be only DCV. That's enough for SN18 test.

Keep some space at rear of instrument to let it breathe. That fan filter require cleaning every 2-4 month, to keep temperatures good. You can measure internal temp (sensor near front of A1 PCBA) via TEMP? command. Usually it's about 35-38C with ambient ~24-25C.

To get best accuracy, you will want to run ACAL and keep temperatures within 0.3C. Most meters have DCV tempco about 0.2-0.4 ppm/K without ACAL.

If you measure 10K+ohm resistance, make sure you use AZERO ON, OCOMP ON and DELAY 5 to avoid measurement errors. Dr.Frank covered this very well. Keep in mind this setup makes measurements really slow, about 42 seconds per reading.

Backup your CAL data via remote GPIB. Just in case, you know. I had programs for that posted in GPIB article linked before.

Don't apply more than 750VAC, as meter is not rated for that. There is special H01 option for meters which have modifications to accept 1kVAC.

After you get idea what functions you need most, you can program them on numeric keypad as hotkeys. Really saves lot of time messing with clunky UI of 3458A. I have ACAL DCV, TEMP?, MATH NULL, MATH OFF, CAL? 72, CAL? 1,2, TRIG AUTO,TARM AUTO and TRIG SGL on mine.

[Dr. Frank]

HighVoltage,
welcome in the 3458A-Club, finally!

Your unit is from 2012, or younger.
Would you mind sending me the S/N, so a calibration date and 1st owner might be determined?
What did you have to pay for this nice, crisp Instrument?

As TiN already wrote, frequent ACAL is extremely important for this Instrument.
After a 2h warmup, and afterwards one initial ACAL DCV, this drift test (~24h) can be made.

If measuring 4W Ohm with OCOMP ON, DELAY 1 for 10k, DELAY 5 for 100k is necessary / sufficient for max. uncertainty.
Every other range needs 0.1s only.

I'm very interested, what your 3458A delivers for the LTZ, I have measured.

(I'm just sitting in front of mine)

best regards Frank

P.S.: For These DIY cables, please check, that the Isolation is made from highly insulating Plastics, i.e. PTFE, or else. PVC creates big measurment error on 10k and higher.

This can be done with a 10k or 100k resistor, either 2W or 4W mode.
Using 2 or 4 separate cables, which should not touch each other (High cables versus Low cables), this reading should not show a difference to the measurement with the DIY cables.

PVC insulation might creates Errors as high as sevarl ppm for 10k, and several ten ppm for 100k,

[HighVoltage]

OK, the 3458A ran continuously for 1 week now and the drift calculations, based on the SN18A, are 0.11 ppm
This is smaller than the 0.43ppm in the SN18A and seems to be acceptable.
But I have no comparisons.
What is a realistic number to be achieved on a good 3458A?

Then I took some first measurements with a Prologix GPIB to USB adapter and the RF Scientific software by Doktor Pyta
This combinations works well on my Win7Pro Lenovo Laptop

In addition to the 3458A, I hooked up 3 x 34401A in parallel to the same 10V Geller Labs reference.
And here are the results ...
I did not log the temperature over night but would expect a change of about 2 degree drop for the time of the measurements

My big surprise is the difference in the three 34401A meters
The two 34401A meters that behave the same are Agilent branded and may be 6 to 8 years old
The one 34401A that shows the data so differently, is Keysight branded and about 1 1/2 years old

Although the temperature also changed for the 3458A, any measurement changes are tiny, compared to the 34401A meters

Any comments are welcome

 

[Dr. Frank]

OK, the 3458A ran continuously for 1 week now and the drift calculations, based on the SN18A, are 0.11 ppm
This is smaller than the 0.43ppm in the SN18A and seems to be acceptable.
But I have no comparisons.
What is a realistic number to be achieved on a good 3458A?

Then I took some first measurements with a Prologix GPIB to USB adapter and the RF Scientific software by Doktor Pyta
This combinations works well on my Win7Pro Lenovo Laptop

In addition to the 3458A, I hooked up 3 x 34401A in parallel to the same 10V Geller Labs reference.
And here are the results ...
I did not log the temperature over night but would expect a change of about 2 degree drop for the time of the measurements

My big surprise is the difference in the three 34401A meters
The two 34401A meters that behave the same are Agilent branded and may be 6 to 8 years old
The one 34401A that shows the data so differently, is Keysight branded and about 1 1/2 years old

Although the temperature also changed for the 3458A, any measurement changes are tiny, compared to the 34401A meters

Any comments are welcome

Where's the temperature log (at least from TEMP?)??
Over this week, the drift of 0.11pm is without ACAL DCV in between?

All instruments will drift over temperature, they all have typical 0.5..1ppm/°C.

If the temperature is constant to about 0.2°C, you can expect drift/fluctuation of <= 0.2ppm/24hr, compared to an external LTZ1000 reference.

These cited 0.43ppm drift in AN-18 is much too high  a criterion for a "good" U180, as this is only the down-calculated annual drift of 8ppm/yr, to 24h.

So in summary, if you really observed 0.11ppm over one week w/o ACAL in between, and with constant RT, then your 3458A is REALLY FINE.

The 34401A are more sensitive to RT changes, and also show much more fluctuation and noise. They play in a totally different league, one order of magnitude worse than the 3458A.

Frank

[HighVoltage]

Where's the temperature log (at least from TEMP?)??
Over this week, the drift of 0.11pm is without ACAL DCV in between?

All instruments will drift over temperature, they all have typical 0.5..1ppm/°C.

If the temperature is constant to about 0.2°C, you can expect drift/fluctuation of <= 0.2ppm/24hr, compared to an external LTZ1000 reference.

These cited 0.43ppm drift in AN-18 is much too high  a criterion for a "good" U180, as this is only the down-calculated annual drift of 8ppm/yr, to 24h.

So in summary, if you really observed 0.11ppm over one week w/o ACAL in between, and with constant RT, then your 3458A is REALLY FINE.

The 34401A are more sensitive to RT changes, and also show much more fluctuation and noise. They play in a totally different league, one order of magnitude worse than the 3458A.

Frank

Thanks, Dr. Frank

This was just a quick-test to see, if the data aquisition works.
I just moved the instruments back in to my other room and will have temp data as well for the next test

Yes, it was without ACAL in between
Monday 10-10-2016, ACAL and recorded CAL?72
Run for one week
Monday 10-17-2016, ACAL again and recorded CAL?72 again
Difference for 7 days = 0.11 ppm

Yes, I am aware of the huge difference to the 34401A and the larger drift is no surprise.
What surprised me is that one of the 34401A instruments had a totally different drift than the other two.

[Alex Nikitin]

Correct me if I am wrong, however from your graphs it looks like the vertical scale is 0.5ppm/div on the first graph and 5ppm/div on the second. So your drifts are 10 times higher than you've stated - i.e. 1.1ppm for the 3458A + the reference .

Cheers

Alex

[HighVoltage]

Correct me if I am wrong, however from your graphs it looks like the vertical scale is 0.5ppm/div on the first graph and 5ppm/div on the second. So your drifts are 10 times higher than you've stated - i.e. 1.1ppm for the 3458A + the reference .

Cheers

Alex

NO, sorry ...
the 1.1ppm drift has nothing to do with the graph
This is a calculated drift, based on the recommendation in Service Note 18A of the 3458A and the internal comparison of the CAL?72 values within one week of continuous run.

The graphs are showing - on a completely different subject - the measurement differences between 4 meters
And I am not surprised that the two 34401A meters drift down so much down, compared to the 3458A
I am just surprised that one of the 34401A meters has a completely different drift.
 

[Alex Nikitin]

Correct me if I am wrong, however from your graphs it looks like the vertical scale is 0.5ppm/div on the first graph and 5ppm/div on the second. So your drifts are 10 times higher than you've stated - i.e. 1.1ppm for the 3458A + the reference .

Cheers

Alex

NO, sorry ...
the 1.1ppm drift has nothing to do with the graph
This is a calculated drift, based on the recommendation in Service Note 18A of the 3458A and the internal comparison of the CAL?72 values within one week of continuous run.

The graphs are showing - on a completely different subject - the measurement differences between 4 meters
And I am not surprised that the two 34401A meters drift down so much down, compared to the 3458A
I am just surprised that one of the 34401A meters has a completely different drift.

Ah, OK ! The 34401 meters show a surprisingly high drift though, about 10-15ppm /week ...

Cheers

Alex

[TiN]

SN18 over 22 days on mine was -0.03ppm/d.

I monitor CAL constants data all the time though, even during normal datalogs. It takes a second to get CAL? values.
If you want run own checks on my data : everything is here: https://dev.xdevs.com/issues/1285

[Dr. Frank]

Here are my stability measurements. At first, the 10V gain factor over several months, and a 16h stability measurement on a virgin LTZ1000 (Andreas design).

As everybody knows, I switch my 3458A off, when not in use.
Anyhow, the AN-18 short term drift refers very probably to a defective U180, which will drift, even if the instrument is not powered--
Also, my 16h, or longer, stability measurements on LTZ1000 references also show stabilities on the order of < 0.2ppm over their time period.
There's usually no drift in one direction, it's more a random walk / fluctuation visible.



Same goes for the CAL? 72 query, over months, I see no drift over time, nor over temperature (which I would expect), only fluctuations of about +/- 0.2ppm from ACAL to ACAL.



So, obviously, I'm really blessed with an absolutely stable 3458A (U180!), the new LTZ1000 is also pretty stable from the beginning, there are much less spikes, no popcorn noise, and the basement is always at nearly constant room temperature (21.5 +/- 0.5°C). That's volt-nuts' paradise!

Frank

[plesa]

Congratz on almost new 3458A!

For data acquision with 3458A (long term logging) is best RaspberryPi 2/3 with GPIB adapter (NI GPIB-USB-HS) and Python.
With script log to .csv
I like live logging on TiN website and my idea is to run FTP on Raspberry if you needs access data and run Apache on RPi to visualize data by D3.
Like two graphs ( chart and histogram )
https://bl.ocks.org/mbostock

And also visualize the humidity and temperature measured by BM280 connected to RPI.

[HighVoltage]

I assume (believe) HighVoltage is now satisfied with the stable solution he has for all his DMM Agilent/Keysight units.

Actually, you are right.
At this time, I am completely happy with the Free Software provided by RF-Scientific and the Prologix GPIB to USB adapter.
It works so well, I can hook up many of my Agilent meters at the same time.

So far I have only played with three 34401A meters in addition to the 3458A.
But all I can say is, that this software works great!
It is highly recommended for anyone that wants to get data out of the 3458A quickly and without programming and still have control over the commands sent to the 3458A.

I will keep the 3458A running for one to two weeks now and see if it will change of stabilize the CAL:?72 10V DC constant.
It really seems, this 3458A was bought by the first owner and never used.

It seems, once in a while we can get very lucky, buying something used, even a high end instrument.

[HighVoltage]

This is a repeat of the previous test, when the 34401A meters had such a large drift.
This time, all four meters stayed turned on 24 hours at about the same temperature before I started the data collection.

Geller Labs SVRT 10V reference
3458A  (100 NPL)
3 x 34401A (100 NPL)

It seems the 34401A needed a much longer warmup time and now have stabilized.
Nice to see how rock solid the 3458A is in comparison.
These are measurements over a 20 hour span.

I have ordered a GPIB adapter for my 34470A, to include this instrument in to the same way of taking measurements, with the RF Scientific Datalogger Software.

 

[quarks]

Is it possible to move this to metrology?

[Kosmic]

Better now than ever I guess

[TheNewLab]

I assume (believe) HighVoltage is now satisfied with the stable solution he has for all his DMM Agilent/Keysight units.

Actually, you are right.
At this time, I am completely happy with the Free Software provided by RF-Scientific and the Prologix GPIB to USB adapter.
It works so well, I can hook up many of my Agilent meters at the same time.

So far I have only played with three 34401A meters in addition to the 3458A.
But all I can say is, that this software works great!
It is highly recommended for anyone that wants to get data out of the 3458A quickly and without programming and still have control over the commands sent to the 3458A.

I will keep the 3458A running for one to two weeks now and see if it will change of stabilize the CAL:?72 10V DC constant.
It really seems, this 3458A was bought by the first owner and never used.

It seems, once in a while we can get very lucky, buying something used, even a high end instrument.

$135 !! choke..choke 
https://www.amazon.com/UGSimple-USB-to-GPIB-Controller/dp/B00RGM88JK/ref=cm_wl_huc_item
How about this one, would it work??

[TheNewLab]

Better now than ever I guess

  

An old thread, and I ooked at only the first and  most recent post

1. has it been moved?
2. time to head for the pub. Guinness time!

[HighVoltage]

$135 !! choke..choke 
https://www.amazon.com/UGSimple-USB-to-GPIB-Controller/dp/B00RGM88JK/ref=cm_wl_huc_item
How about this one, would it work??

I don't think I would trust this amazon USB to GPIB adapter.

The new price for the Keysight 82357B - GPIB-Controller is € 634,27 (Including Tax) in Germany.
So, $ 135 is a bargain for a real 82357B

And watch out for the fake offers on ebay.
https://www.eevblog.com/forum/testgear/flood-of-new-agilent-82357b-gpib-usb-adaptors-on-ebay-the-real-deal/

[IanJ]

$135 !! choke..choke 
https://www.amazon.com/UGSimple-USB-to-GPIB-Controller/dp/B00RGM88JK/ref=cm_wl_huc_item
How about this one, would it work??

I don't think I would trust this amazon USB to GPIB adapter.

The new price for the Keysight 82357B - GPIB-Controller is € 634,27 (Including Tax) in Germany.
So, $ 135 is a bargain for a real 823587B

And watch out for the fake offers on ebay.
https://www.eevblog.com/forum/testgear/flood-of-new-agilent-82357b-gpib-usb-adaptors-on-ebay-the-real-deal/

I have one of those cheap 82357B adaptors (100 USD at the time) bought via Ebay and it seems to work fine with my 3458A.

Ian.

[TiN]

those cheap 82357B

They work okay in Windows environment, but often always flake out with linux-gpib on Pi during datalogging.

[TheNewLab]

those cheap 82357B

They work okay in Windows environment, but often always flake out with linux-gpib on Pi during datalogging.

IS anyone using windows inside a virtual box on Linux? If so, how well do the lab apps work? I have become so comfortable with Linux...I really want to just drop Mac and Windows...I er, wish..

[2N3055]

those cheap 82357B

They work okay in Windows environment, but often always flake out with linux-gpib on Pi during datalogging.

Mine works just fine. And I use it only from PI /linux-GPIB.
I never logged for 6 months but did for 45 days max, usually for weeks all without glitch....

To be honest that is a second one, I got as replacement. First one did just what you said, it would work fine for some time and then just stop. And it was random, sometimes after few days, sometimes after few hours... I returned and second one just works fine.