HP3478A Calibration

[bigfathairyvika]

Damn... bought a HP3478A meter and as the battery condition was unknown I proceeded to follow the procedure to replace the battery.
I am pretty sure I followed it to the letter, but still I now have an uncalibrated meter!

I suppose my only choice now is to send it to a calibration center?
And probably fork out the required figure to get it recalibrated.

Thoughts and suggestions for a place to send it?

Mark

[g4wim_tim]

Hi,

I have one of these meters - a rather old model and a much newer model 34401A which was in cal.

I managed to cal the old one against the 34401A one by following the instructions, so I'm not sure I understand your problem.

I had to improvise for an accurate reference voltage but to be honest calibrating it didn't make much difference apart from removing the cal error message.

[G0HZU_JMR]

I think it may still be possible to retrieve the previous cal data if you did do the battery swap correctly. I recall seeing this on Eevblog a few years ago.

Ideally the cal data should have been saved via GPIB before the battery swap attempt but it looks like the meter sometimes declares itself uncalibrated even though the procedure was followed correctly and it still has the cal data intact. It may be a checksum thing and the meter declares itself uncalibrated. I think you need a GPIB cable to get the data out and a simple script to dump the cal data to a text file.

[bigfathairyvika]

Unfortunately I don't have a GPIB cable or interface card or pc to fit one in.
Also don't have another meter of comparable accuracy or precision.

I have emailed two calibration centers for a quote.

Or I suppose I could just take the hit and ebay it as uncalibrated and find a calibrated replacement.

Btw: should the transformer hum be noticeable or do I need to be diagnosing that as well?

[factory]

Seems to be a common problem with these, you need another battery or power supply connected while removing & replacing the battery, if not the cal data will be lost. Also have to be careful not to short out the battery/supply when doing this.

Be careful what the ask the cal place to do as "calibration" is only checking it against calibrated standards, these days they refer to "adjustment" for making any changes.

David

[veedub565]

Originally Posted by factory

Seems to be a common problem with these, you need another battery or power supply connected while removing & replacing the battery, if not the cal data will be lost. Also have to be careful not to short out the battery/supply when doing this.

Be careful what the ask the cal place to do as "calibration" is only checking it against calibrated standards, these days they refer to "adjustment" for making any changes.

David

Yeah I second that, if you send it to a calibration centre they will just do a performance verification. Only if it is outside of it's spec will they adjust it (usually at additional cost).
If it's inside it's spec and you want it adjusting to be bang on, then you need to ask for it to be optimised. Again, more money for that.

I would imagine if it's coming up uncalibrated, than adjustment is required in order to re-write all the cal values back into it.

[bigfathairyvika]

The thing is I connected a battery and diode across c510 as spec'd using an isolated the soldering iron. Removed the old battery and soldered in the new. Removed the battery and diode.

I've had a quote back for "Tracable recalibration of bench multimeters" at 119.40 inc vat and pickup and return.

I will ofcourse verify what that is.

[bigfathairyvika]

Regarding adjustment, there are no adjustments.
The meter is self calibrating against standards.
You do need to run through the calibration mode whilst applying said standards.

[factory]

No the "adjustment" is referring to what a cal lab would do these days to get something back to within the spec, it doesn't matter if it's done by hardware pots or compensation values stored in the battery backed RAM IC.

Old manuals may refer to things differently of course, but you need to be sure when dealing with cal labs, or you may end up with it back in the same condition with an expensive bit of paper stating how far out of spec it is.

It may be possible the RAM supply got shorted with the solder wire to the case, even if your iron was isolated.

David

[bigfathairyvika]

Originally Posted by factory

It may be possible the RAM supply got shorted with the solder wire to the case, even if your iron was isolated.

David

Ah! That is a possibility I didn't think of.
Having my solder running from a spool on a feeder , it may have rested on the case!
I won't do that again.

I will be phoning the cal center tomorrow to confirm pickup address , so will question them on what exactly I will be getting.

As you say , it will be a right pain in the rear if it is returned with a note saying "yes, it is out of cal"

I may contact Chris who I bought it from as it appears he does calibration as a job.

Mark

[Jez1234]

I once worked for a cal lab... License to print money basically and as much use as a chocolate fireguard!

No doubt as part of a bigger fleecing industry... many companies would not use test gear that had no in date cal certificate and sticker on it, and, to be clear, all that was generally done was a basic check that yes everything still works and is within the manufacturers stated tolerances. No attempt would be made to correct any error's or to optimise calibration. As gear was usually of good quality and as they wouldn't pay to check any gear they knew was faulty the end result was basically that we'd take in maybe 30 multimeters a day and charge £40 per unit to tell them that it still worked and put a sticker on it! A scam is what I'd call it...

[Radio Wrangler]

Even done properly, most people mislead themselves about calibration. At best it retrospectively justifies what the instrument was used to do in the preceding period. It gives no guarantee for the future... the thing could get dropped on its way home form its annual cal, and if anything shifted it may not be noticed until last year.

Intelligent users have to have someone analyse any failures to consider what effect it may have had on their previous year's production. Product recalls might be necessary in some circumstances.

David

[G0HZU_JMR]

In some cases an instrument has an internally automated calibration routine so it may have to satisfy this routine in order to turn off the uncal indicator on the display. I've never owned an HP3478A and haven't used one in about 30 years but it may have to satisfy a particular test/adjust routine for every meter function and range.

For example. my Tek spectrum analyser has an automated routine to calibrate its own internal 10MHz OCXO. I have run this a couple of times now and the analyser uses its own GPIB interface to control an external sig gen locked to a traceable reference. The only way to calibrate it is via this automated program that is contained within the analyser. It isn't as simple as a pass/fail ppm check. The analyser spends quite a long time tweaking its own 10MHz OCXO and re-measuring itself against the external reference until it is happy that its internal OCXO agrees with the external cal reference within something like 0.003Hz.

To make the uncal display flag go away I think the HP3478A has to be convinced it has been through a calibration routine and this is probably done over GPIB.

[Jez1234]

Call me a cynic... but it was just a "covering yer arse" scam really.... They knew the gear worked fine, and we knew it worked fine, but non of that mattered. All that REALLY mattered was they had a piece of paper saying it worked fine.
A top of the range brand new Fluke would not be used without said piece of paper but a battered old Avo 8 with a sticky needle and correct sticker and certificate could be used... madness... And don't get me started on PAT testing!

[factory]

We have an internal cal lab at work (we also used to cal stuff from other sites in the same company), yes failures were rare but they did fail equipment, some would be adjusted if needed, but for most of the it it was time to invest in a newer replacement.
Some equipment just ended up with a BER sticker as hardware repair could be too time consuming, costly, or we had no service information (common with more modern stuff), also if we had more than we needed of that type of test equipment it wouldn't get repaired or replaced.
Certainly some DMM's had simple problems such as loose input jacks, the cal lab also replaced blown DMM fuses, and replaced the battery if needed.
By the way we work on a lot of safety critical equipment, regular cal is very important, regular external audits happen too, if they find out of cal stuff it would be very bad for company reputation.

The users of the equipment wouldn't necessarily be able to notice if the equipment had drifted slightly out of cal. I've only noticed faulty equipment a couple of times, one was a TTI meter with strange results on certain resistance ranges and a Solartron DMM that failed on the mA range, this was only notice as it got connected the wrong way round, the reading changed by an unacceptable amount in the reverse direction, both ended up on the BER pile in the store room.
The less said about the Fluke scopemeter that someone shoved 110V up it's recharge socket the better, it got repaired by an external company at great expense only to end up dead again around a year later.

David

[bigfathairyvika]

Good news.
Meter recalibrated.

Came with before and after sheet.
All the tests before were well out of spec and a few were out so much that the decimal point would be a waste of time.
So, when it flashed up "uncalibrated" it meant it.


After, much better, within original HP spec and my voltage reference now reads fine.

[Chris55000]

Afternoon Mark!

Glad you've got it sorted!

I do have a fairly new Keithley 2100 that I'm fairly sure is accurate enough to "calibrate by comparison", (which the 3478A S.M. says is permissible if the comparison meter is accurate enough!), but it'd be fairly tricky to make up sufficiently stable sources in a home workshop!

For example, the 300V at 100kHz a.c. would need 4 "100V line" type transformers run in reverse from a power amplifier fed from a signal generator with a very fine resolution o/p amplitude control!

Main thing is you've got a reference sheet for your meter now and it should retain calibration for at least three years as a minimum!

Chris Williams

[Keith956]

Originally Posted by bigfathairyvika

I've had a quote back for "Tracable recalibration of bench multimeters" at 119.40 inc vat and pickup and return.

Hello Mark, can you say who you used for cal? I have a couple of HP34401A's I really ought to get calibrated - last time was over 5 years ago.

I got a quote from PASS which seems reasonable but wondered who other people have used.

[majoconz]

No attempt would be made to correct any error's or to optimise calibration.

Quite right. If the DUT is in tolerance, record the error, give it the big tick and send it on it's way. When it comes in next time, see if the error is still there and it's value. If the error is reducing or the same - no problem. If the error is increasing and heading for the tolerance but not there yet but will be before the next cal date, then tweak it to be in spec. Note all in the report and hope like hell that it won't be out the other side at the next cal date. History is very important - I've seen things spend their whole lives wobbling about from one side of the tolerance to the other but always manage to sneak in!