Calibrating 3458A?

[Keith956]

I have been very lucky in finding a 3458A recently. It's in impeccable condition with all the original manuals etc. and was bought by the seller's father's business in the mid 90's and used to verify calibration standard resistors the business made. So far I've just confirmed it passes self test and gives sensible readings (within 0.5mV of a 34401A in cal).

What I'm wondering is whether to get it calibrated - I have no idea how much Keysight charge for a standards lab level cal here in the UK but I expect it to be expensive. So what do others here do? Intended use is for lab reference and to calibrate the other  34401A's I have.

[Grandchuck]

My 3458A was purchased used several years ago.  Like yours, it was pristine with all the manuals and accessories and in its original Agilent shipping box.  Presumably, it had not be used for 9 years or so.  At first, it seemed a bit drifty.  After a lot of use it now seems to be settling down.  If yours is similar, you might want to give it a lot of use before having it calibrated.

[tszaboo]

I have been very lucky in finding a 3458A recently. It's in impeccable condition with all the original manuals etc. and was bought by the seller's father's business in the mid 90's and used to verify calibration standard resistors the business made. So far I've just confirmed it passes self test and gives sensible readings (within 0.5mV of a 34401A in cal).

What I'm wondering is whether to get it calibrated - I have no idea how much Keysight charge for a standards lab level cal here in the UK but I expect it to be expensive. So what do others here do? Intended use is for lab reference and to calibrate the other  34401A's I have.

As I recall, you only need a 10V and 10KOhm to calibrate it. And yes, sending it to a cal lab is expensive, they need a Fluke 734 to make the calibration.

[IanJ]

Hi,

Last time I was charged about £700+VAT from Keysight over a year ago. I'm about to call them again for costs as I'm due in March.
There are different levels (and price), 3 if I remember. Mine was to UKAS standard.

Remember, they don't necessarily make any adjustments. If it's within spec they just detail to you the numbers. Up to you after that.

Note: A few years ago they identified my front/rear switch was giving inconsistent results in one of the modes if I remember. Luckily I didn't need that mode and I was fine without repair until the I replaced it prior to the following years re-cal.

Ian.

[Kleinstein]

The adjustment only needs a 10 V and 10 K reference, but a proper calibration (legal certificate),still needs the full high grade calibrator to generate all the test voltage to do a full functional test of all the ranges.

=.5 mV difference to a 34401 is quite a large difference, unless tested in the 1000 V or 100 V range.

For calibration of a 34401 one would still need a stable source to generate the test signals the 34401 wants. The 3458 alone is not enough.

Keysight has some problems doing business with private customers at least in Europe. So one may have to look for a different lab (e.g Fluke) in the UK.

It may be a good idea to frist use the DMM for some time, so it was on much of the last 4-6 weeks before sending it in for calibration. Some units may show some drift if not used for a longer time. Another point is doing functional testing before: e.g. look at input bias, ADC drift (SN18) and check the date on the battery backuped RAMs. A drifting ADC is bad - it gets even worse directly after calibration.

[mendip_discovery]

The meter is possibly still well within specifications but you cant be sure unless you get it checked. I Personally prefer meters with documentmented history.

Plenty of labs in the uk that can calibrate it, not so many can adjust it but there will be some about. I would suggest finding one with 17025 accreditation (UKAS) and local to you so you can drop it off as the ppms dont like the delivery firms.

Not sure how much luck you will have with keysight but if you run into trouble I can see if I can arrange it to be a subcon job through the lab I work at. I don't have a good enough uncertainty to be looking at your meter but as an UK company I can send it to them.

[martinr33]

Your first step is to follow the process in this service note:
https://xdevs.com/doc/HP_Agilent_Keysight/3458A/service/3458A-18A.pdf

Given the age of the unit and potential downtime, you should run the test over a couple of weeks of power on, longer if you see drift. There are two objectives:
 - getting the reference settled. Per KS, long power downs can cause a reversible shift in the reference value. They recommend a 6 week minimum "warmup".
 - Seeing if you have a drifty ADC.

If you have a drifty ADC, the fix is a new board - if you can get one. However, for your calibration processes, the long term stability may not be an issue.

Finally, DC calibration is a relatively quick process, and requires just 10V, 10k, a good short, and some care around thermals, warmup and so on. As Kleinstein notes, verification is a much bigger deal.

So if you can find someone with good hobby 10V/10k willing  to help you, you can get well along the path. Or at least, the start of it...


 

[Bill158]

Congratulations on getting a top of the line DVM!
Don't forget about the age of the NVRAMS.  If they have never been changed then before anything else new ones must be installed.  BUT just installing new NVRAMS without copying the old ones and then programming the data into the new ones will loose all of the previous calibration data.  If your NVRAMS are original in your 3458A then you are running on borrowed time already.  The batteries in the NVRAMS are rated for 10 year life.  Yours could be as much as 25 years old.  Open the instrument and look at the date codes on the installed NVRAMS.  If you send your 3458A into KS then they will take care of this.  I am not sure if your "corner" cal lab will change them.  But the NVRAMS in your 3458A are probably soldered in so you need to unsolder them from the PCB and take them out, install IC sockets, and then plug in the new ones.  Search EEVBLOG for information on how to read the original NVRAMS before you remove them.  There are programs out there that can read the NVRAMS through the IEEE 488.
Bill

[Keith956]

Yep the NVRAMs are 94/95 vintage. I'll take a look for the info on how to read/change them.

[Dr. Frank]

Hello,
Maybe you also want to become a volt-nuts soon?
You might make or buy your own Volt and Ohm references, and compare these with references from other volt-nuts, in your country, maybe at low uncertainty.
The 3458A is no standard, by no means.
Therefore, for amateur use, you might use such references to check or to unofficially calibrate your 3458A. If you like, you could also adjust it to these references.
Here in Germany, there's user ap, from ab-precision, who might offer such services.

Frank

[Keith956]

Maybe you also want to become a volt-nuts soon?

Haha Dr Frank, you may be right! A 10V reference is next on the list, but so many choices! 

Keith

[IanJ]

FYI,

Keysight UK just got back to me with cal cost - £783.

Ian.

[Keith956]

Thanks Ian. I'll follow the advice here and let it run for a few weeks before getting it into cal.

Which level is that for? The basic 'Keysight Calibration' or all the way up to 'Keysight Standards Lab Calibration'?

[IanJ]

Its for UKAS Accredited Calibration, the standards lab level.

From memory, the basic Keysight level cal was a couple hundred quid cheaper. But, since I use mine for calibrating my products and producing my own cal certs i need the UKAS level.

Ian.

[Dr. Frank]

Hello Ian,
that sounds interesting.
Have you found a way to be able to emit your own, official calibration certificates?
Do you have an accreditation, or sort of?

Frank

[IanJ]

Hello Ian,
that sounds interesting.
Have you found a way to be able to emit your own, official calibration certificates?
Do you have an accreditation, or sort of?

Frank

Keysights website is a little confusing (as I was!)......so I called Keysight UK just now, here's the 3 levels. No prices as their pricing system was down.

1. Keysight Cal.

2. UKAS

3. UKAS Standards level, tighter tolerances/spec.

1st one is the cheapest.
I normally get 2nd one......per the price I gave earlier.
The guy told me that not many people go for the 3rd one.

Other than that, i just produce my own cal certs when i cal my products. Actually, its all automated via my GPIB app (control loop and spits out a Word doc), but i dont issue anything UKAS formally/legally in any capacity if that what you mean.

Ian.

[Keith956]

I've managed to read the NVRAMs using Ian's WinGPIB. Now the question is: what to replace them with?  DS1220Y-150 (obsolete now) or DS1220AD-150+ (Farnell have them in stock)?

And can anyone recommend a cheap programmer that will can program them using the saved data?

[Dr. Frank]

The difference between the different types was the voltage compliance, and speed, afair.

The circuit is fine with any of these as long as speed is 150ns or faster. So probably the AB is also ok, of course the -100 types.
Maybe you check if your 5V digital supply is really within +/- 5%.
Get what is the cheapest, i.e. the DS1220AD-100+, and don't forget to order two 32kx8 as well, i.e. DS1230xy which are about double the price of the 1220.
Please order precision DIL sockets for these ICs.

Maybe you want to add 128kB Ram, i.e. 4x 32kx8, 62256  SRAMs, available cheap for <2€/EA.

Then I recommend to update the firmware, to 8 or 9, as older versions still contain a lot of bugs.
6 or 1 chip EPROM images can be found elsewhere. If you have already the single 27C4002 version, get a new EPROM, plus DIL socket.
The older EPROMs, 6 chip version,  might not be available any more and they then need to be erased first.

All programming can be done with the cheap TL866 II from XGecu.
I recommend ordering from the original distributor, and to add the complete adapter set.

Frank

[Oldtestgear]

Keysight UK just got back to me with cal cost - £783.

They charged me £600 inc VAT in November 2021 for a standard ISO17025 qual. That really is inflation unless you specified a different qual. level.

Phil

     

[martinr33]

Regarding replacing the nvm on the 3458A...

Follow Dr. Franks' lead on this. A little more detail on desoldering below.

You will have to desolder the existing memories, and the code EPROM(s). This is not so easy if you have not done it before.

You will need a solder sucker; some solder; and maybe a bottle of flux (a small brush bottle is fine).

 - paint the existing solder with flux. This makes the solder more mobile.
 - Melt it with the iron, to the point where the whole pin is in molten solder. This takes a couple of seconds after you first see it melt.
 - Pull the solder with the solder sucker. Be quick.
 - If the pin is not loose in the hole, poke it with the tip of the iron until it breaks away from the side of the hole. If you push it just right, it will be loose in the hole when you are done. This is important.
 - if a pin does not come clean, add some solder and flux and try again. Adding a little solder helps draw out the residuals. You can also add alittle solder to the pin before you try to desolder it, this approach can help.

When you are done and all the pins are loose, the chips should pop out with minimal assistance. Sometimes, I use a flat blade screwdriver to gently twist off the chip - but BE CAREFUL.  If they do not, check the pins again. The risk is that you will pull the copper out of the PCB, which is a nuisance. 3458A boards are very robust, but you still need to be careful.

Also, after programming there is a chance that the chips will be marginally corrupted and throw an error. If that happens, just reprogram the part.

[martinr33]

@IanJ:

I thought that the KS calibration/maintenance multiyear  contracts were quite reasonably priced. Might be worth a look.

[Keith956]

Out of interest, does the data in the cal NVRAM change over time? I read the cal ram at the beginning of the month, then I did it again yesterday. Several bytes in groups (presumably float values) have changed.

Considering the cal ram is dated 94/95 (like most of the other ICs) so is over 25 years old, it's remarkable that the nvrams are still working. I get the impression this 3458A was not used a lot - the cal count is 1, the seller told me it had never been calibrated since new.

 Martin - I have a Hakko fr-301 which I've used to desolder 28pin DIL packages before, hopefully it will be straightforward to remove the cal ram and the settings rams and replace with socketed ones. I've got a TL866 programmer so can flash the new ones and verify them.

Dr Frank - the EPROM (the board is the one with the single EPROM) is version 9 already.

[Dr. Frank]

Out of interest, does the data in the cal NVRAM change over time? I read the cal ram at the beginning of the month, then I did it again yesterday. Several bytes in groups (presumably float values) have changed.

Considering the cal ram is dated 94/95 (like most of the other ICs) so is over 25 years old, it's remarkable that the nvrams are still working. I get the impression this 3458A was not used a lot - the cal count is 1, the seller told me it had never been calibrated since new.

Dr Frank - the EPROM (the board is the one with the single EPROM) is version 9 already.

Hello Keith,
no, the NVRAM data do not change over time. Only if the internal back-up battery is depleted, it will lose its content, but then the 3458A will display a checksum error.
The CAL nvRAM contains the calibration, Autocal and maybe some setup constants, see calibration manual, command CAL?.
Therefore, whenever you do an ACAL, its content will change.

Yes, the batteries might hold for 15 years or longer, but may fail all of a sudden, so you are safe already, as you have downloaded the content of the CAL nvRAM. Seems that somebody serviced the µP PCB already. Would be nice if you could provide a picture of the whole PCB. From about 1995 onwards, the 40k reference resistor R207 was changed to a VHP101 type, which is much more stable than the initially used one. Maybe you have a look and a picture on that, also. This resistor might be bent backwards a bit, but it's uncritical to align it upright, carefully.

I Thank - Frank

[bsdphk]

no, the NVRAM data do not change over time.

That is not quite correct...

The calibration constants only change when calibration procedures are followed, but the NVRAM is used to store a couple of undocumented variables which change outside calibration context.

The `CAL?` command has at least the following undocumented uses:

        # Undocumented CAL? queries
        # CAL? -1       21249=NUM_SAMPLES
        # CAL? -2       861=BURST2
        # CAL? -3       1149=BURST1
        # CAL? -4       11=NREADINGS
        # CAL? -5       2010=SMPL_RATE
        # CAL? -6       1=SSMP_INC
        # CAL? -7       3.54147092E-05=PERIOD
        # CAL 2437      16611, DEFEATS valid 2437
        # CAL 2941      0, DESTRUCTIVE OVERLOADS valid 2941

The last two of which are stored in the NOVRAM

[Keith956]

Dr Frank, here are some photos. Looks like the Vishay 40k resistor is there (R207). The IC datecodes are nearly all '95.

Keith