Agilent 3458A - I joined the club

[TERRA Operative]

I just recently picked up my first 3458A, it's an Agilent unit, with serial US2802xxx

I found it in a test equipment shop in Akihabara, and with a quick email negotiation, managed to get it for 550,000yen, down from the original quote of 600,000yen+tax (for an original total of 660,000yen). Not a bad discount, especially as it has been factory calibrated in 2021 and 2022, so I'm in the process of trying to get my hands on those reports from Keysight.

The only problem with it is apparently a slightly high reading in the -1A range, but it's close enough that a cal might be able to fix it. Unless anyone thinks I should have a poke at it before I sent it to Keysight for calibration and adjustment? (Once my bank account recovers.....  )

So far, I have given it a clean, including cleaning and oiling the filthy fan. Ordered a set of HM62256BLP-10 RAM chips from a local seller to upgrade to Option 001, and started making a list of stuff to buy.
I'll also be ordering a few FRAM adapters and parts to ditch the Dallas NVRAM's (I already backed up the calibration constants twice), and I'll socket U110 and update the firmware to v9 (I picked up a 40-pin socket today)

I'm also thinking of replacing the fan with a Papst 612NLE, which seems to be the closest replacement to the original 612L I can find, and replacing the electrolytic and any RIFA capacitors (pending checking for any).


The display is a bit worn as expected for it's age, I may or may not bite the bullet with a new VFD from China... Same with a new front panel plastic from Keysight. The plastic on mine is in pretty good condition, but I did have to apply a little bit of superglue on the corners, so these two parts are on the 'wish list'.

It agrees perfectly with my 3-year-since-last-cal 34461A on 10V and 30K that I tried so far. Until now the Fluke 731B was only adjusted to my 34461A, but now I can really dial it in.

So, feel free to educate me on using this instrument and any other things I should be looking at while I'm poking around, and also how to measure drift and performance and stuff, I have no idea but am willing to learn! 

[srb1954]

Congratulations on joining the club.

When looking for RIFA capacitors also check the mains inlet filter. If it is a pre-2000 Schaffner unit that may also contain exploding capacitors and it is best to replace it to avoid the risk of the awful stench the filter produces when those internal capacitors explode.

[TheDefpom]

Very nice, i'm a bit envious, I would love a 3458A, I currently only have the Advantest R6581T, which is not bad, but not great either.

[Kleinstein]

Reading low on the 1 A range is a bit odd. With the ACAL procedure there is no additional calibration factor that is applied. So the error there would be some kind of error in the ACAL procedure, like some thermal effect or not so perfect ground current compensation. Another possible source could be a relatively high TC for the shunt and self heating effect at the shunt.

As a test it may be worth to look at the CAL constants that ACAL calculates for the 1 A range for a few repeated ACAL runs. When done at somewhat variable temperature one could also get an idea on the TC of the shunt.

Before sending the unit in for calibration, I would use and test it for quite some time. It takes some time to recover from longer idle time. A point to test is also the ADC drift according to the SN18 service note. So read the CAL72? constant (and ideally also the temperature) after doing ACAL over a few weeks.

[Dr. Frank]

Congratulations!
About 3400 €, that's a very good price, provided that its U180 is good.. but vintage 2000 (like mine, but from HP) should be good in this aspect, anyway.

The 1A range is auto-calibrated  only, it can't be adjusted by an external CAL (which would only adjust/calibrate 10VDC and 10kOhm).
It's strange that it didn't pass the verification, could have been a problem with the calibrator.
Has this cal report been provided by Agilent, or by an external lab?
The limits look odd, as they are not centered, they are asymmetric.. strange.
I have several test reports from other instruments, I'll have a look what's going on.

My instrument still has OK capacitors, and the RIFAs seem so be still fine .. the Schaffner did not blow up yet..
So, before replacing thoughtlessly, check visually and measure capacitance and ESR first. 

Don't you have any possibility to access any precise 10V and 10kOhm from another volt-nuts at your location?
Calibration @ KS is unnecessarily expensive, only because it's a 3458A, and always be aware, that no DMM is regarded as a "standard".
Setting up your own bank of 10V/10kOhm standards would be better, technically and economically.

You need better cables for precise measurements, like PTFE with copper lugs, and you need a lab with constant room temperature, otherwise you can't make any use of the 3458As stability and uncertainty.

Frank

PS: Don't worry about the asymmetry of the 1A range.. that seems to be an offset, which HPAK probably never properly addressed in their calibration/validation routine. I looked into other reports directly from HP, Agilent and Keysight, and this asymmetry is not as prominent as in your case.

Especially the limits of + and - polarity are extremely deviating in your case.

It's not mentioned in the manual, but you always should use a Null before measuring currents, to achieve highest uncertainty.
Same goes for 100mV and 1V ranges; these perform better than specified, if you take care about offsets before mking the real measurement. 

Another hint: The test report you show here, is this status "As received", or "As Completed"?

I have a test report from Keysight, probably Penang, where this 1A range was nOK (i.e. U = Undetermined) before calibration, but good afterwards, although this could only be judged after an ACAL beforehand.

[chilternview]

When looking for RIFA capacitors also check the mains inlet filter. If it is a pre-2000 Schaffner unit that may also contain exploding capacitors and it is best to replace it to avoid the risk of the awful stench the filter produces when those internal capacitors explode.

Indeed, the mains filter went on mine back in the summer. Not only a stink but gooey black gunk came out.

[coromonadalix]

for the vfd  you could have  fine dust on it who could slightly mess with the clarity  ...
do a full checkout for the famous rifa and other ac input parts who could blow,   

for the supposedly a3 board drift   if your lucky as you say and you dont have it   nice catch     


and you can refer to this thread too  for some parts ....   sure it goes beyond that  loll

https://www.eevblog.com/forum/metrology/3458a-worklog/

[aronake]

Congratulations!

It may not have been running much for the last couple of years since they were calibdated. So keep it running for a couple of weeks (a couple of month better), do recurring ACALs, ideally connect a stable voltage reference and track DC readings and CAL72 values. Best if done in somewhat stable temperature. In case you plan to hand it in at Hatchioji, best to make sure that things have stabilized properly. Some of my 3458a have needed month to get CAL72 to settle in at a low and more constant drift (one almost a year!). And it may be that even its been so long since they were running that the LTZ1000 needed to settle in to. But I never got to measure that.

It helps to do above measurements with a python script, which is not too complicated. What seems to be the better and more stabe GPIB adaptor is Keysight/Agilent E5810.

I got some VFDs form China. They work very well! And new nice shining VFDs are always nice to look at!

As pointed out here previously, depending on your intended usage, there may be better options than a full official keysight calibration.

[IanJ]

So, feel free to educate me on using this instrument and any other things I should be looking at while I'm poking around, and also how to measure drift and performance and stuff, I have no idea but am willing to learn! 

You are on the right tracks......you've backed up the CalRam and a FRAM conversion is imminent.
Once you have had it powered up 24/7 for a few weeks then start logging CAL? 72 in order to evaluate the U180 for any drift. If you come out the otherside of that all-good then you got a real bargain.
Attached is a spreadsheet template I created you can use that has the calcs built in.

Mine is a 2000 vintage and with a new Keysight front panel, fully FRAM converted.
It's overdue a Cal at Keysight which I plan to do in the coming months.
I have a 3245A that I fitted a new VFD (Ebay) to and it's perfect......in fact I plan on swapping it with the 3458A VFD as they are identical.

Ian.

[aronake]

It may also be an idea to see how CAL72 change after ACAL with temperature differences. Below 0.1 ppm/C is good, above 0.2 ppm/C start to be a bit problematic. Datasheet allow for a bit more though.

[TERRA Operative]

Thanks all for the info so far, greatly appreciated!

I have some good volt and resistance references, but they haven't been characterised to the required extent for this instrument.
I have a bunch of vishay resistors, a Fluke 731B, a voltagestandard.com 10V reference, some Yokogawa calibrators, etc, etc, but the best they are adjusted to is my 34461A (The voltagestandard.com reference is on it's original cal).

So, a proper 3458A calibration is on the cards to give me a baseline to work from once the dust settles. business expenses etc.


Tonight's work was replacing the Rifa capacitors (I'll grab a TDK line filter from Akihavara on Friday, my Shaffner has a date code of 0036, so meh, I'll replace it to be safe), and socketing the NVRAM chips ready for FRAM, and also socketing the U110 EPROM so I could do a firmware upgrade.
The firmware upgrade went smoothly, so now I'm on firmware 9.2, so that's nice.
I also took a direct dump of the contents of the Cal NVRAM, and it matched perfectly with the data I sucked out the GPIB port using WinGPIB, so happy days (Thanks Ian! Good software! )

I've also taken a list of electrolytic capacitors to buy, I'll replace them while I'm inside as they are all over 20 years old now. I'll use United Chemicon KZN etc series where possible for long lifetime.
There is a big fat axial cap, 8300uF 35V that might take some searching to find a replacement (Mouser don't have much at the moment) so I'll test that one and see how good it still is.


Hopefully the Option 001 RAM chips arrive tomorrow, and they'll go straight in.

So far, so good.



I also remembered I have a Matsunaga SVC-1010 1kVA (10A @ 100V) autotransformer type line voltage stabiliser. I wonder if it's worth using that to make the line voltage nice and stable during longer data logging periods?
Google translated link to SVC-1010 data

[MiDi]

Welcome to the club!
At the bottom are all parts listed that should be replaced: https://www.eevblog.com/forum/metrology/3458a-worklog/msg2098195/#msg2098195

It may also be an idea to see how CAL72 change after ACAL with temperature differences. Below 0.1 ppm/C is good, above 0.2 ppm/C start to be a bit problematic. Datasheet allow for a bit more though.

There are a couple units I know that have around 0.3ppm/°C TC, that is absolutly in spec (0.5ppm/°C TC) and good.
Mine has rather high 0.34ppm/°C TC and U180 drift.

[TERRA Operative]

Great, thanks for the list, it helped fill in a few blanks, so now I have a full shopping list on mouser.
(Rifa's are done, Line Filter is coming from Akihabara)

Here's what I'll buy:

Capacitors:
Ref Des	Brand	Value	Part No.	Series
A2C702, A2C704	Vishay	22uF, 63V	MAL202138229E3	021 ASM
A4C7, A4C8	Chemi-Con	2200uF, 50V	EKZN500ELL222ML40S	KZN
A4C9	Chemi-Con	15,000uF, 35V	EKMQ350VSN153MR35S	KMQ
A4C17	Chemi-Con	22uF, 50V	EKYB800ELL220MF11D	KYB
A6C1, A6C11	Chemi-Con	680uF, 63V	EKZN630ELL681ML20S	KZN
A6C8	Cornell Dubilier	10,000uF, 35V	109TTA035M	TTA
A6C200, A6C201	Chemi-Con	47uF, 100V	EKY-101ETD470MJC5S	KY
A6C202	Chemi-Con	330uF, 63V	EKZN630ELL331MJ25S	KZN
A6C203	Chemi-Con	220uF, 50V	EKZN500ELL221MJ16S	KZN

Rifa Capacitors:
Ref Des	Brand	Value	Part No.	Series
A6C15, A6C16	Kemet	2200pF, 350VAC	R41BF122050T0K	R41

Option 001:
Ref Des	Part	Part No.
A5U123, A5U124, A5U125, A5U126	32K x 8 CMOS SRAM	62256

Misc:
Ref Des	Part	Brand	Part No.
B1	Fan	ebm-papst	612NLE
B1	Connector	Molex	50-57-9402
B1	Pins	Molex	16-02-0104
A5U132	8K x 8 FRAM	Infineon	FM16W08-SG
A5U121, A5U122	32K x 8 FRAM	Infineon	FM18W08-SG

I've also just installed Option 001, which seems to work with no errors. Pic below showing the socketed chips and Opt 001 installed.

[TERRA Operative]

Well, there's the reason I had to put some superglue on the front panel.... 
(And yes, I did put the strap on upside down, meh, will fix it soon when the covers come off again).

Time to make a list to sent off to Keysight for a few spare parts.

[chilternview]

I can't recall the exact price Keysight charged for a front panel but it was somewhere in the region of £110 when I ordered it for one of mine a while back.

[aronake]

I can't recall the exact price Keysight charged for a front panel but it was somewhere in the region of £110 when I ordered it for one of mine a while back.

When I contacted Keysight they didn't want to sell one to me, but rather send in the whole multimeter for "repair".

[IanJ]

That's what Keysight are doing now.....almost across the board for 3458A parts by all accounts........
I must have hit it lucky about 5 or 6 years ago they sold me a new front panel and a front/rear switch.

Ian.

[laichh]

Hi TERRA Operative,

Congratulation & welcome to the Volt-Nut club, $3.5k+ for a used 3458A is really a bargain.

Like many had pointed out, there's no specific calibration adjustment to the DCI 1 A range. In fact, all functions & ranges except DCV 10 V, Res 10 kΩ, & ACV, are ACAL corrected. In your case, likely you've got a slightly higher TCR 1 A shunt (R213 on A1 board).

If you could get yourself a good 100 mΩ shunt like https://www.digikey.com/en/products/detail/vpg-foil-resistors/Y07340R10000F9L/4223414, get it kludges in & you should be good to go.

[TERRA Operative]

Well, I got a quote from Keysight for a bunch of parts, so I'll sent the purchase order back soon. Seems they might actually sell me the parts directly with some luck.

I replaced the line filter in the meantime. The new one was a slightly different size, so a little filing of the plastic on the filter, and a trim of the rear panel sticker (The sticker cutout is smaller than the panel cutout, and it's all installed.

No Rifa or Shaffner smoke for me!


I also noticed the Pass/Fail output cutout on the rear (visible in the attached photo). Does anyone have any info on that? I haven't found anything so far in my searches.

[Dr. Frank]

Well, you might be lucky that you're dealing directly with KS in Penang/Malaysia, where the manufacturing and main Engineering support for the 3458A is nowadays located.

Volt-nuts here in EU (or as well in U.S.) had to contact either local support, and they request that you send your defect 3458A to them, and would not sell any spare parts anymore since a few years, already.

This pass/fail empty connector might be an option, which was never officially available, like opt. 01/02, therefore, might have been a request by electronic manufacturing companies.We use many 3458A in our factory, to precisely measure component values, or calibrate something on the PCBs at those elevated temperatures in the manufacturing halls. (That's the main application of the 3458A, not in a lab in first place). We are steering the 3458A over GPIB, therefore, such a pass/fail is not needed.

Anyhow, I can imagine, that other manufacturers would need this function, if they produce something (like components, e.g. resistors), and would like to automate the selection process w/o PC.
I had a brief look into the schematics, but could not discover any unused port output on the main CPU board, A5.

The Trig In / Trig Out jacks are interfaced to the A/D board A3 via an optical link on A6, outgard power supply.   

If I remember correctly, the PTB of Germany (or maybe the NIST) used this empty jack for a synchronous 10MHz output signal from A3, via an optical link, to exactly synchronize the high speed sampling for precise AC measurements. You can find several unpopulated component places on the new A3 board (i.e. the very first version of this redesigned PCB), where one could assemble a gate IC, and an optocoupler, to bring the A/D clock (CK10 from U131B) to the outside.

FRank     

[TERRA Operative]

The delivery guy came with a parcel from Mouser today.

So the new fan is installed, and all the electrolytic capacitors are replaced except A4C17 which I forgot to grab from Akihabara last week, so I'll grab that on Monday on my way into work.

Nothing else to report so far.

[TERRA Operative]

Just bit the bullet and ordered a replacement VFD from ebay.
Working through the back and forth with Keysight to order parts now too.

[IanJ]

Just bit the bullet and ordered a replacement VFD from ebay.
Working through the back and forth with Keysight to order parts now too.

Here's my 3245A that I fitted an Ebay VFD to about 2 years ago. Uses the same VFD as the 3458A.
My 3458A VFD is still pretty good, no plans to change it out anytime soon (I have an Ebay VFD sitting ready for it though).

My HP 3458A (manu. 2000) also sports a new Keysight branded front panel I bought from Keysight UK about 5 years ago.

Ian.

[TERRA Operative]

eeewwwwwwwww.....

Something gross happened here. Luckily it washed away with soap and water, so now the binding posts are all clean.
I also gave the front panel PCB and the buttons a good clean too.


I also found a 2A fuse in the current binding post holder, so that tells a bit of a story about the current range being a bit out....
I've stuck a 1A fuse in and once the instrument is all back together and settled over a few days of runtime, I'll give it a self calibration and hope for the best.
I can't measure if it will come good or not but it's only a tiny bit out, so a proper adjustment hopefully might bring it good, maybe...



I also found the rubber bung where the wires pass through the chassis to the front binding posts was as brittle as Bakelite, so it had to be cut to allow me to remove the binding post assembly, so now I'm on the hunt for something that will fit.

The parts manual states it's a "GROMMET-RND 0.5-IN-ID, 0.625-IN-GRV-OD" (12.5mm ID, 15.9mm OD) part for a 1.5mm panel. Shouldn't be toooo hard to find here in metric land I hope...

[EDIT] Found grommets from Misumi Vona (A machine parts supplier here in Japan), should arrive Saturday. I also threw in a few plastic GPIB port dust covers too just for fun.

[aronake]

eeewwwwwwwww.....

Something gross happened here. Luckily it washed away with soap and water, so now the binding posts are all clean.
I also gave the front panel PCB and the buttons a good clean too.

Ha! Very Japanese! Important things are clean! How about you bring the 3458a next time you go to onsen? Even for a viking descendant like me, used 3458a can be quite gross before being cleaned up...

How cooperative has Keysight been in selling spare parts?