Test Equipment Anonymous (TEA) group therapy thread

[Specmaster]

It’s never quite that simple. There is an important scenario you need to consider which is invalidation of measurements. If you find your calibration fails then the certainty of measurements that have been made since the previous calibration should be called into question. At that point the device is usually returned to the client until their investigation is complete. If the instrument is found to have violated the worst case specification drift then it’s usually sent back to the manufacturer for repair of in warranty or scrapped if not. If it’s expected then it may be shelved as “for indication only” and used for casual work that doesn’t require traceable measurements, or send for adjustment.  Or in the case of where I worked chucked in a pile of suspect stuff and thrown in a skip about 2 years later because it was cheaper to replace it than play crap tennis.

Regarding the calibration process, this is mostly automated. The reason the cal houses turned down the 3478A is probably because they don’t have any programs for their kit. The RS guys probably downloaded a PDF from somewhere random, poked at it like some monkeys around a monolith and gave up.

That still does not make sense, if calibration does not mean making adjustments, and that the meter has to be returned to the manufacturers for repairs to be made if economical to do so or scrapped etc, then why do the manufacturers bother to build in "Cal Enable" switches and provide detailed instructions on how to make the adjustments to bring the instrument back to being within specification and therefore calibration?

These instruments were never intended to be in the hands of hobbyists, they were all at the time classified as laboratory / professional grade instruments.

In the case of the 3478A, I successfully managed to re-calibrate it after losing the cal data from the RAM, so that it did not disagree with either of my Flukes (8840A and 8842A) or my, at the time brand new Brymen 867S and have also since managed to re-calibrate my BlackStar 4503 after it suffered massive contamination on the circuit board from its battery backed RAM when its fitted battery spewed its guts everywhere. Both cases involved using the Cal Enable function and following procedures in the SM.

[bd139]

The cal enable switch is to stop dum dum end users screwing up the instruments. Same as why the put the trimpots on the inside on older kit 

Have a look at the more recent instrument service manuals - they are structured as "calibration", "verification", "adjustment" etc.

[VK5RC]

Re calibration, I am on bd's side.
All instruments drift and do not ever give truly accurate readings, what counts is that you know how far out the instrument can be.
You end up with the errors on errors problem with the Calibration lab having the biggest headache. Even a brand new instrument is not a lot of 'accuracy type' use until its drift and error range can be established by multiple calibrations over time - I think this is the main reason why most instruments don't get adjusted as the calculations of error and drift go crazy.
If you look at the specs of the 3458 a supposedly 8 1/2 digit meter, it is only spec to 8ppm/yr ie ~5 digits of real accuracy BUT  with repeated calibration cycles confidence can approach the 8 digits. The 8 1/2 is really its transfer spec ie 1 hr or so!
Rob

[Cerebus]

My only remark on the calibration issue: If "calibration" really means "calibration and adjustment" as some keep on insisting on thinking, despite many sage and experienced words to the contrary, what does a cal lab do when you send them a steel ruler to calibrate? How would they adjust the markings on the ruler? (Before anyone asks, yes you can send a steel ruler to be calibrated. I've even used a steel ruler with a cal sticker on it.)

[tautech]

Calibration is not magical but does encompass a number of steps.

Accuracy verification.....against manufacturers datasheet.
Adjustment..........if required to meet manufacturers spec.
Cart cert creation listing instruments used and their traceable Cal certs.

With sufficient precision in the calibration process instrument drift can be quantified.

Adjustment is a whole nuther rabbit hole where in old instruments it was done with physical adjustment whereas the modern instrument use closed case adjustments where a PC script controls the calibration instruments and adjust the reference points in the instrument being calibrated......a totally automated process.

The few instruments we've supplied to Cal labs get regularly checked for verification of accuracy and nothing more as I've never been asked to provide the Cal scripts that allow for closed box automated calibration.

[beanflying]

Incidently just rewatched Dave's video on the 3478A and from 10:30 on he talks about the battery backed RAM and if you disconnect the battery or short it out you will lose the calibration constants  and will then need to spend more money then you paid for the meter to get it traceable calibrated once again. So in his view, having a meter calibrated does indeed involve having adjustments made to the meter to return it to the design specifications again and not just verifying and noting any drift that the user should take into account when using said meter. So Dave and I both share the view that calibration is not a case of checking against references and recording the differences, but actually includes making the necessary adjustments to make the meter reflect the references applied to it.   

I think if you asked Dave for his current perspective (not 2013 Dave) and in particular after visiting the Keysight Cal Labs several years later you will get a different answer. In the case of the 3478A, 34401A etc they all have a 24 Hour spec. IF one or more ranges don't meet this spec and only IF you agree after being notified they will perform an adjustment on that range or ranges. The only other time an Adjustment would be done would be if the OWNER requested it.

If it meets the 24 hour Spec then it will be issued a Certificate of Calibration.

[mnementh]

Mini rant: Kubernetes is Pandora’s box incarnate. No one knows entirely how it works. No one knows entirely what it does. But the outcome is nearly always mayhem and chaos. Much like “The Cube”, it is written by lots of people who independently concentrate on their little bits of the machine unaware of each other and the sum of the parts resulting in the destructive potential for the cumulative pile of mistakes in each part. From synchronising state across several partitions, through Rube Goldberg chains of events, struggling through eventual inconsistency and landing on mutant vocabulary they have succeeded only in removing understanding and determinism from computing. When all hell breaks loose you cannot reason about fixing it because of the inane complexity because you can’t communicate the problem nor rationalise it due to the non determinism.

At this point I am going to sit back, buy test gear to play with and watch it all fucking burn and hope I get paid to pretend I give a shit.

If this is entire thing means nothing to you then you are lucky bastards.

And med, nice score. Even if it cost you a tyre

For a moment there I thought you were talking aboot Windoze and AD; then I noticed that not once did you curse PowerShell...

mnem

gimme gimme gimme !!!

Okaaayyy... here's one of your very own...

mnem
*toddles off to do... something, even if it's wrong...*

[Kosmic]

My only remark on the calibration issue: If "calibration" really means "calibration and adjustment" as some keep on insisting on thinking, despite many sage and experienced words to the contrary, what does a cal lab do when you send them a steel ruler to calibrate? How would they adjust the markings on the ruler? (Before anyone asks, yes you can send a steel ruler to be calibrated. I've even used a steel ruler with a cal sticker on it.)

This is very true. My ESI RV722 voltage divider is working exactly like that. The divider is calibrated but no adjustments are possible on the instrument. Instead a chart like this is produced for each decades.



You then use the chart to adjust your measurement after the fact. Technically you could also do that with your multimeter and the calibration report. Instrument adjustments are really just convenience.

[Runco990]

I'm not a big poster, more of a professional lurker...  But here's a Rubidium standard I scored.... for free!  2 new Lead acid gel cells later and some cleaning..... 

[beanflying]

I'm not a big poster, more of a professional lurker...  But here's a Rubidium standard I scored.... for free!  2 new Lead acid gel cells later and some cleaning.....

Its ok we allow lurkers and Test Equipment perverts too  Nice add on to your gear too 

[VK5RC]

Hi Runco, Efratom is a great bit of gear, can't beat that price either👍

[med6753]

That ugly discussion of "calibration" has reared it's ugly head again. I chose to stay far away. 

Anyway, I pulled the Type 1A1 plug-in out of the Type 547 to give it an inspection and cleaning/dexoit. I'm going to install it in the Type 535A and see if it works. Even though this is a 50MHz plug-in and the Type 535A is only 15MHz B/W it will work.

Also, the transistors for the Type 321A will indeed arrive Saturday. For once USPS is on time.


 

[tautech]

Anyway, I pulled the Type 1A1 plug-in out of the Type 547 to give it an inspection and cleaning/dexoit.

Now the 547 is suddenly 2kg lighter !
Pull the timebase next and you might be able to carefully manage the mainframe on your own.
Mind that back Jack ! 

[bd139]

My back is why I didn’t buy that 547 here a while ago 

Nice to watch someone else playing with one though 

[McBryce]

Incidently just rewatched Dave's video on the 3478A and from 10:30 on he talks about the battery backed RAM and if you disconnect the battery or short it out you will lose the calibration constants  and will then need to spend more money then you paid for the meter to get it traceable calibrated once again. So in his view, having a meter calibrated does indeed involve having adjustments made to the meter to return it to the design specifications again and not just verifying and noting any drift that the user should take into account when using said meter. So Dave and I both share the view that calibration is not a case of checking against references and recording the differences, but actually includes making the necessary adjustments to make the meter reflect the references applied to it.   

Even Dave admits that he sometimes makes mistakes in his videos.

McBryce.

[mansaxel]

My back is why I didn’t buy that 547 here a while ago 

Nice to watch someone else playing with one though 

There's an OK-looking (though not shown powered on) 549 over here, collection only. Currently at 1000SEK ($115ish). I've bought a 400H VTVM from that seller. He's buying estates and selling them, claiming zero TE knowledge, but there's been a steady trickle of vintage TE from him. Probably got a mother-load estate that he's parting out slowly.



It's 330km away, so a full day road trip. I think I'll pass.

Also, a 5245L turned up from another seller. That's more like it.

watchingly,

[bd139]

Ooh nice 

No discord tonight. My mother had a cardiac ablation done yesterday so I’m over there this evening until late. Seems to have worked though so far.

[beanflying]

Ooh nice 

No discord tonight. My mother had a cardiac ablation done yesterday so I’m over there this evening until late. Seems to have worked though so far.

Good News but aging Mothers are still a nuisance     Might see if I have the urge to wet the bed and need to get up for Pain pills and Coffee in the morning....

[AVGresponding]

It’s never quite that simple. There is an important scenario you need to consider which is invalidation of measurements. If you find your calibration fails then the certainty of measurements that have been made since the previous calibration should be called into question. At that point the device is usually returned to the client until their investigation is complete. If the instrument is found to have violated the worst case specification drift then it’s usually sent back to the manufacturer for repair of in warranty or scrapped if not. If it’s expected then it may be shelved as “for indication only” and used for casual work that doesn’t require traceable measurements, or send for adjustment.  Or in the case of where I worked chucked in a pile of suspect stuff and thrown in a skip about 2 years later because it was cheaper to replace it than play crap tennis.

Regarding the calibration process, this is mostly automated. The reason the cal houses turned down the 3478A is probably because they don’t have any programs for their kit. The RS guys probably downloaded a PDF from somewhere random, poked at it like some monkeys around a monolith and gave up.

That still does not make sense, if calibration does not mean making adjustments, and that the meter has to be returned to the manufacturers for repairs to be made if economical to do so or scrapped etc, then why do the manufacturers bother to build in "Cal Enable" switches and provide detailed instructions on how to make the adjustments to bring the instrument back to being within specification and therefore calibration?

These instruments were never intended to be in the hands of hobbyists, they were all at the time classified as laboratory / professional grade instruments.

In the case of the 3478A, I successfully managed to re-calibrate it after losing the cal data from the RAM, so that it did not disagree with either of my Flukes (8840A and 8842A) or my, at the time brand new Brymen 867S and have also since managed to re-calibrate my BlackStar 4503 after it suffered massive contamination on the circuit board from its battery backed RAM when its fitted battery spewed its guts everywhere. Both cases involved using the Cal Enable function and following procedures in the SM.

This a fair point, in that it seems a more accurate description for these switches would be "adjustment enable".

[bd139]

I wish it was that. Wouldn’t be having this discussion now then  .

A point to note though with some of the instruments. The 3478A and most closed case cal units tend to have the ability to read the cal ram over GPIB when cal is enabled. Part of a more detailed cal process, and full check out of the instrument, is comparing the last recorded cal data with the current one to make sure no one has futzed with the meter or any calibration issues are hardware failure related.

[tggzzz]

Ooh nice 

No discord tonight. My mother had a cardiac ablation done yesterday so I’m over there this evening until late. Seems to have worked though so far.

Good News but aging Mothers are still a nuisance     Might see if I have the urge to wet the bed and need to get up for Pain pills and Coffee in the morning....

Yeah just carted my mother to the hospital in an ambulance.

[Peter_O]

After leaving the electronics hobby at age of 18 and having some test gear contacts at research centers as a programmer I left the stuff and went with consulting where no equipment is needed. 

Now I remember these old times and started to fiddle with electronics again and - sure enough - bought some absolutely neccessary test gear.  :-)

Of course I needed a precision PSU and so I got this beautiful old HP6114A.

One of the voltage switches needs cleaning and a 1µF capacitor made a loud chirping noise.


So far I was able to claim this device to be neccessary.

But with a HP5335A I might be touched lighty by the syndrom discussed here. 

I bought it, because it remembers me of those old times, has beautiful big LED numbers and lots of buttons. It is in full working condition but for my needs the DVM and Oscilloscope capabilities would have been sufficient. So its totally useless. The 1-pound Pabst easily outperformes the noise of all other test gear together, except of the Lambda PSU whis is in it's own league noisewise.

Because I like the LEDs lighted up, I replaced the Papst with a rubber mounted Sunon and the noise ist now some 20 db lower. Some glue to fix the power button and some rubber feet (not original - shame to me) added and the HP5335A now adds nicely to the Background of lab pictures.

Because there are a lot of homeless 5335 on the market I have no bad conscience to give one of them a new and respectful home.

[bd139]

Ooh nice 

No discord tonight. My mother had a cardiac ablation done yesterday so I’m over there this evening until late. Seems to have worked though so far.

Good News but aging Mothers are still a nuisance     Might see if I have the urge to wet the bed and need to get up for Pain pills and Coffee in the morning....

Yeah just carted my mother to the hospital in an ambulance.

. Hope all goes well.

[med6753]

Ooh nice 

No discord tonight. My mother had a cardiac ablation done yesterday so I’m over there this evening until late. Seems to have worked though so far.

Good News but aging Mothers are still a nuisance     Might see if I have the urge to wet the bed and need to get up for Pain pills and Coffee in the morning....

Yeah just carted my mother to the hospital in an ambulance.

. Hope all goes well.

Yes, agreed. Hope all goes well.

I had a terrible nights sleep. After pretty much pain free yesterday the spasms returned with a vengeance last night. Still hurting a little this morning and can't sit here for long periods. I doubt I'll be on Discord later today.   

[Robert763]

My only remark on the calibration issue: If "calibration" really means "calibration and adjustment" as some keep on insisting on thinking, despite many sage and experienced words to the contrary, what does a cal lab do when you send them a steel ruler to calibrate? How would they adjust the markings on the ruler? (Before anyone asks, yes you can send a steel ruler to be calibrated. I've even used a steel ruler with a cal sticker on it.)

This is very true. My ESI RV722 voltage divider is working exactly like that. The divider is calibrated but no adjustments are possible on the instrument. Instead a chart like this is produced for each decades.



You then use the chart to adjust your measurement after the fact. Technically you could also do that with your multimeter and the calibration report. Instrument adjustments are really just convenience.

Important points well made by Cerebus and Kosmic.
The same applies to a lot of electrical primary standards like resistance, inductance and the older voltage references. These have no pratical means of routine adjustment.
As has also been said any actual adjustment (as opposed to changing a look-up table or polynomial) will almost certainly affect the stability of the item. Modern closed box calibration of indicating instruments is just automation of the calibration table / graph. In the vast majority cases modern "closed case"adjustment" is not adjustment it is just look-up table changes. About the only exception is the few designs that use "digital potentiometers" basically FET switched resistor arrays with NVRAM. Modern "adjustment" isn't adjustment it's the old calibration chart update hidden in the instrument. Modern users are hidden from the complexities and correct use of terminolgy gets blurred. It's like one of my pet peves, toxic / posionous. Toxins come from orgainisms. Lead is not toxic and ricin is not a poision but common usage has made the two interchanable.

One of my previous day jobs was Quality Manager / Chief Inspector of a fully accredited instrument calibration lab.