Never Trust Other People's Equipment!

[AndyC_772]

To me, "calibration" implies adjusting to spec - not simply measurement (unless the DUT is already in spec).

From the UKAS accredited laboratory we use:

"Calibration" = measuring the output of an instrument under known conditions and recording its output.
"Adjustment" = making changes to an instrument in order to make its output match the conditions it's actually measuring.

A reliable instrument can be calibrated every year without ever changing how it reads, and that's important. It means measurements you take with it today are directly comparable with ones you made last year, and the year before that, and the year before that etc.

If an instrument is adjusted, then its readings today are not directly comparable with those taken before adjustment was carried out. It may, however, be more accurate than it was previously, in absolute terms.

Which you need depends on your application. Normally when you send away (say) a DMM, it will come back calibrated - but only adjusted if it was actually out of spec to begin with. Even then, some labs wil ask first.

[tautech]

To me, "calibration" implies adjusting to spec - not simply measurement (unless the DUT is already in spec).

From the UKAS accredited laboratory we use:

"Calibration" = measuring the output of an instrument under known conditions and recording its output.
"Adjustment" = making changes to an instrument in order to make its output match the conditions it's actually measuring.

A reliable instrument can be calibrated every year without ever changing how it reads, and that's important. It means measurements you take with it today are directly comparable with ones you made last year, and the year before that, and the year before that etc.

If an instrument is adjusted, then its readings today are not directly comparable with those taken before adjustment was carried out. It may, however, be more accurate than it was previously, in absolute terms.

Which you need depends on your application. Normally when you send away (say) a DMM, it will come back calibrated - but only adjusted if it was actually out of spec to begin with. Even then, some labs wil ask first.

Yes well, Cal technicians I know here in NZ look at instruments a little different.

First its performance is verified and if meets manufacturers specs a Cal cert is issued.
If performance/accuracy doesn't meet manufacturers specs adjustment is made until it does and then a Cal cert is issued.
Yes....it's all in the wording and not some magic voodoo process but a simple check and then adjustment if necessary.

[IDEngineer]

Right. Definitely splitting hairs on word definitions, but I doubt any self-respecting Calibration Tech would issue a "Calibration Certificate" for something that was merely measured and left out of calibration.

Maybe I'm just old school, but "calibrated" to me means the device is in spec - whether it was already there or had to be adjusted to get there.

[Fretec]

Maybe I'm just old school, but "calibrated" to me means the device is in spec - whether it was already there or had to be adjusted to get there.

This is the only meaning of "calibrated" I ever knew - it was checked to comply with the specs and it does, if it had to be adjusted it will be mentioned.

[fourfathom]

frankly it's not an issue in the real world. As I said, I've yet to see the lead-acid battery that accidentally generates 2X is chemistry's rated voltage. And our spec sheets are very clear about the max input voltage.

EDIT #2: I just finished characterizing our standard protection circuit for sustained high voltage conditions. We can safely tolerate a sustained overvoltage of ~18VDC before the tranzorb's conduction climbs enough to matter (above 10's of milliamps), and then the limiting factor is only internal heating of the tranzorb (proper device operation continues). That's fully 50% over the nominal rating of 12VDC, and 30% above the typical alternator voltage of 13.8VDC. If someone's on-vehicle regulator is permitting a sustained output voltage 30% higher than alternator nominal, ours won't be the only device that fails. In production there's always a limit to how much you can spend on protection against extreme cases while remaining commercially viable. I think we're OK where we are with 30-50% margin.

FWIW, marine batteries often see higher charging voltages than you find in automotive applications.  You may see over 15V sustained during equalization, and over 14V at normal high charge rates.  Your 18V seems safe enough, but cars and boats do usually have different battery environments.  Of course, some boats have 24V, 36V, or 48V systems as well. 12V is by far the most common though.

[rsjsouza]

Right. Definitely splitting hairs on word definitions, but I doubt any self-respecting Calibration Tech would issue a "Calibration Certificate" for something that was merely measured and left out of calibration.

Maybe I'm just old school, but "calibrated" to me means the device is in spec - whether it was already there or had to be adjusted to get there.

Perhaps the Power supply had one of those NCR (No Calibration Required) stickers? I see several of them on power supplies around the lab in my company.

[srb1954]

Right. Definitely splitting hairs on word definitions, but I doubt any self-respecting Calibration Tech would issue a "Calibration Certificate" for something that was merely measured and left out of calibration.

Maybe I'm just old school, but "calibrated" to me means the device is in spec - whether it was already there or had to be adjusted to get there.

Perhaps the Power supply had one of those NCR (No Calibration Required) stickers? I see several of them on power supplies around the lab in my company.

I think that whoever thought up the idea of a No Calibration Required sticker was a penny-pinching accountant. What they really mean is that "this device should be checked or calibrated but we can't be bothered doing it".

Most of the instruments I have seen with these stickers really do require some form of calibration but it is left up to the user to somehow verify for themselves that the equipment performance is within spec. This is sometimes difficult to do when every piece of test equipment available to you has an NCR sticker on it.

[IDEngineer]

Perhaps the Power supply had one of those NCR (No Calibration Required) stickers? I see several of them on power supplies around the lab in my company.

I didn't see one on any of the six outside surfaces of its enclosure, and I definitely looked. While I was holding it I was tempted to throw it across the room but it would have probably gotten a final laugh by straining my muscles - as you might imagine, it had some heft.

[IDEngineer]

FWIW, marine batteries often see higher charging voltages than you find in automotive applications.  You may see over 15V sustained during equalization, and over 14V at normal high charge rates.  Your 18V seems safe enough, but cars and boats do usually have different battery environments.  Of course, some boats have 24V, 36V, or 48V systems as well. 12V is by far the most common though.

Noted, thanks. Although the battery charging blocks (take in the alternator and emit to one or more batteries) seem to stay below 14V in this industry. Most of these watercraft have a helm-mounted digital voltage display and I've never seen one hit 14V, always 13.X with the engine running. And those displays track our own voltage measurements with independent meters. I suspect they are simple diode-isolated designs so both our numbers may be accurate and simply separated by a ~1V diode drop.

[bdunham7]

To me, "calibration" implies adjusting to spec - not simply measurement (unless the DUT is already in spec).

Definitely not a given.  Not all devices are adjustable, so calibration often involves recording the measured value. Others may have an adjustment or calibration constant procedure, but rarely or never need adjustment unless certain parts are replaced.

I  think the missing links here are the terms 'broken' and 'diagnosis'.  If your first notion after finding a device out of calibration is to 'adjust' it, even if it has a nice assortment of fiddly trimpots, then you should have your fingers slapped with a ruler by an angry nun.  You need to at least think through why it would be out of spec.

One thing that comes to mind in the case of your device-eating power supply is that by either a defective part or defective design, it may have excessively high voltage until some minimum load is put on it.  Imagine a simple system, designed only with large loads in mind, with a rectified and filtered supply voltage of 29 volts at open circuit, followed by a linear regulator at 13.8 volts, but with a 100R resistor bypassing the regulator.  Under any significant load, you have 13.8 volts, but your DMM shows 29 and your devices 'felt' about 29 volts.  Suppose there was a calibration procedure and it called for measuring the voltage and ripple at 1A and 50A, and nothing more.  It would pass 'calibration' and might never have any issues when used only with circuits with substantial draws--like simulating a boat electrical system or charging large batteries.  And then you come along and....zap!

[bdunham7]

I think that whoever thought up the idea of a No Calibration Required sticker was a penny-pinching accountant. What they really mean is that "this device should be checked or calibrated but we can't be bothered doing it".

It is a talisman to ward off evil ISO9000 auditors.

[rsjsouza]

Right. Definitely splitting hairs on word definitions, but I doubt any self-respecting Calibration Tech would issue a "Calibration Certificate" for something that was merely measured and left out of calibration.

Maybe I'm just old school, but "calibrated" to me means the device is in spec - whether it was already there or had to be adjusted to get there.

Perhaps the Power supply had one of those NCR (No Calibration Required) stickers? I see several of them on power supplies around the lab in my company.

I think that whoever thought up the idea of a No Calibration Required sticker was a penny-pinching accountant. What they really mean is that "this device should be checked or calibrated but we can't be bothered doing it".

Well, to be perfectly honest, it could be penny pinching or engineers just like me that want to really leave a simple variable PS alone and not bother me with compliance questions, procedure and red tape... But that's just me.

[fourfathom]

FWIW, marine batteries often see higher charging voltages than you find in automotive applications.  You may see over 15V sustained during equalization, and over 14V at normal high charge rates.  Your 18V seems safe enough, but cars and boats do usually have different battery environments.  Of course, some boats have 24V, 36V, or 48V systems as well. 12V is by far the most common though.

Noted, thanks. Although the battery charging blocks (take in the alternator and emit to one or more batteries) seem to stay below 14V in this industry. Most of these watercraft have a helm-mounted digital voltage display and I've never seen one hit 14V, always 13.X with the engine running.

I'm more familiar with sailboats where we may have a large "house" battery bank charged by high-current alternators and a multi-stage "smart" charger.  With this setup we try to get as much juice into the battery as we can, running the engine as seldom as we can.  We will occasionally run a higher-voltage equalization cycle.  There will often be a separate starter battery, with some sort of charge-sharing mechanism, but most of the charging current goes to the house bank.  The charging profile is much different than you might see on a small powerboat, which usually has a smaller alternator and a regulator similar to that in an automobile.

The above is for lead-acid batteries (flooded, gel, AGM, or foamed carbon).  Obviously Li batteries are coming into more general use and these will have still different voltage profiles.

Regardless, in a 12V system I think 15 or 16V should be a safe maximum sustained voltage to design for.

[srb1954]

I think that whoever thought up the idea of a No Calibration Required sticker was a penny-pinching accountant. What they really mean is that "this device should be checked or calibrated but we can't be bothered doing it".

It is a talisman to ward off evil ISO9000 auditors.

You could well be right on that.

[Fretec]

I think that whoever thought up the idea of a No Calibration Required sticker was a penny-pinching accountant. What they really mean is that "this device should be checked or calibrated but we can't be bothered doing it".

It is a talisman to ward off evil ISO9000 auditors.

You could well be right on that.

He is, otherwise the not-calibrated unit would present an issue.

[tszaboo]

Right. Definitely splitting hairs on word definitions, but I doubt any self-respecting Calibration Tech would issue a "Calibration Certificate" for something that was merely measured and left out of calibration.

Maybe I'm just old school, but "calibrated" to me means the device is in spec - whether it was already there or had to be adjusted to get there.

Perhaps the Power supply had one of those NCR (No Calibration Required) stickers? I see several of them on power supplies around the lab in my company.

I think that whoever thought up the idea of a No Calibration Required sticker was a penny-pinching accountant. What they really mean is that "this device should be checked or calibrated but we can't be bothered doing it".

Most of the instruments I have seen with these stickers really do require some form of calibration but it is left up to the user to somehow verify for themselves that the equipment performance is within spec. This is sometimes difficult to do when every piece of test equipment available to you has an NCR sticker on it.

I had PSUs calibrated, and the only thing I can say: Maybe I should've bought a second power supply and stick a "not calibrated" sticker on the first one.
Now we have 1 PSU which is calibrated externally, and that's what we use for traceavle measurement, for everything else, I wrote a stupid 1 page calibration manual. You only need a DC load and an Oscilloscope BTW for that. And the original manual asked for a 6.5 digit DMM, for good measure. Pun intended.

[Mr. Scram]

Right. Definitely splitting hairs on word definitions, but I doubt any self-respecting Calibration Tech would issue a "Calibration Certificate" for something that was merely measured and left out of calibration.

Maybe I'm just old school, but "calibrated" to me means the device is in spec - whether it was already there or had to be adjusted to get there.

I'm pretty sure you'll get into a fist fight with the owner of something like a 3458A sooner or later if you start adjusting equipment without them asking for it. You only adjust equipment if and when asked.

[IDEngineer]

Not trying to fight with anyone. But if I wanted a cal lab to just report to me a device's present status, I'd ask for something like an "evaluation". If I sent it in for "calibration" I'd expect it to be in spec when it came back. But I accept that mine may just be the commonly accepted definition and not the strict definition of the word in the context of metrology. That said, I bet the majority of people - even in Engineering - would be startled if "calibration" didn't yield a device that met its specifications.

Here's an honest question: When a device has a Calibration sticker with a date on it, what exactly is being communicated? I've never seen a sticker that said anything like "Measured to be out of specification by 0.1% on the following date". It's just "Last calibrated on {date}". If all they did was determine that the device was out of spec, how are you supposed to know that from the sticker and its date? On the other hand, if they only give it such a sticker when it's within spec, doesn't that imply that "calibrated" means "in spec"?

[rsjsouza]

In the lab that I work, which I suspect reflects others, an equipment with a valid calibration means it is still within the manufacturer's accuracy specifications. If the manufacturer says their product operates with an accuracy of +/-0.005% or +/-5% or +/-50% of a standard reference, that is what you shouldexpectt when using it.

Many products are taken out of the lab and go to maintenance or are decomissioned in case they are "out of spec".

[0culus]

Not trying to fight with anyone. But if I wanted a cal lab to just report to me a device's present status, I'd ask for something like an "evaluation". If I sent it in for "calibration" I'd expect it to be in spec when it came back. But I accept that mine may just be the commonly accepted definition and not the strict definition of the word in the context of metrology. That said, I bet the majority of people - even in Engineering - would be startled if "calibration" didn't yield a device that met its specifications.

Here's an honest question: When a device has a Calibration sticker with a date on it, what exactly is being communicated? I've never seen a sticker that said anything like "Measured to be out of specification by 0.1% on the following date". It's just "Last calibrated on {date}". If all they did was determine that the device was out of spec, how are you supposed to know that from the sticker and its date? On the other hand, if they only give it such a sticker when it's within spec, doesn't that imply that "calibrated" means "in spec"?

I think it's more of a traceability thing. If some question about a measurement comes up, you can conclusively prove that the instrumentation was in spec. Doing it super regularly is probably more of a CYA thing from legal (for both the vendor suggesting you do it regularly, and you at the company having it done regularly). 

[IDEngineer]

 It's been asserted that "calibration" is merely measuring what the device does - not adjusting it to spec. So if you find an in-date calibration sticker, could that mean the device was merely measured on that date with no guarantee of being in spec? After all, that's all a "calibration" sticker says based on this thread's proposed definition.

If that's true, a sticker is meaningless. I've never seen one that states how far OUT of spec the "calibrated" device was measured to be. So all it tells you is that someone measured it on a given date. You know nothing about what was actually discovered.

[bdunham7]

It's been asserted that "calibration" is merely measuring what the device does - not adjusting it to spec. So if you find an in-date calibration sticker, could that mean the device was merely measured on that date with no guarantee of being in spec? After all, that's all a "calibration" sticker says based on this thread's proposed definition.

If that's true, a sticker is meaningless. I've never seen one that states how far OUT of spec the "calibrated" device was measured to be. So all it tells you is that someone measured it on a given date. You know nothing about what was actually discovered.

A calibration sticker, without the actual calibration certificate, is technically pretty worthless.  If you were audited, for example, you might have to go back to the calibration lab for a copy if you didn't have it.

That said, for most common things you would think of having calibrated, like a typical DMM, the usual calibration service would indeed be verifying that it is within the manufacturers specifications and adjusting it where possible.  This is a pretty standard service and for most customers and most labs with most service-bench DMMs, a unit that is out of spec and cannot be adjusted for one reason or another will get a 'REJECT' sticker.  So if you see a Fluke or HPAK bench DMM with a calibration sticker from Fluke or Keysight that is less than a year old and the meter appears undamaged and operating properly, you can be pretty sure it is within the published specs.  Well, actually not quite even that because the calibration procedure doesn't necessarily cover the performance limits.  So  you'd need to study both the specs and the calibration procedure itself to be as sure as possible.

The problems occur when you try to apply this principle too widely without actual, specific knowledge about what happened, which means reading the actual certificate.  And, in some cases, deciding whether to believe it. 

https://www.eevblog.com/forum/metrology/can-we-believe-a-calibration-certificate/msg3141840/#msg3141840

EDIT:  Also wanted to mention, 'calibration' and 'calibration sticker' are not synonymous.  Devices like precision standards, Weston cells and the like may have labels where data is recorded or may say 'see certificate' or 'see log'.  I would not expect an ordinary sort of bench service device with a plain calibration sticker with no info except a date and expiration to not be calibrated to manufactures specs if done by an outside lab.  However, I'm sure that there are exceptions to even that expectation--I just can't recall any right now.

[0culus]

Presumably there is paperwork on file. Case in point, my HP/Agilent 8664A sig gen. It was seen by Agilent Calibration in 2013, and the sticker expired 3 years later. I called Keysight's cal folks up out of curiousity, and they dug up the full paperwork on it and sent it to me. It was exactly what you'd expect...detailed the uncertainties of various things and whether or not it met the original manufacturer specifications.

Any other definition of "calibration" is pure , IMO

[0culus]

I will add another curiousity to the pile though...the HP 53310A modulation domain analyzer. These instruments are often found on the used market with cal stickers and seals...which is really much ado about nothing as anyone with a reasonably high quality 5V reference source can do the calibration. That is the ONLY external signal needed!

[Berni]

The purpose of the calibration sticker is to be able to quickly identify if the equipment you are using is still within its calibration period, that is IT.

The calibration lab that put the sticker on has the full report on file, so if you are there costumer you can call them up and have them dig it up for you according to the equipment serial number.

So what use is that piece of paper that goes with the sticker? It was a pretty damn expensive piece of paper after all! Well if shit hits the fan and someone threatens you that your products are out of spec you can show them a calibration report that your equipment is in spec. Especially if you have two reports, one before the use of it and one after. This definitely proves that the gear was in spec at the time the measurement was taken since its extremely unlikely it would have drifted off and then drifted back in time for the calibration. But if say you are a multimeter manufacturer and need to calibrate your own meters, in that case you might even use the numbers off the calibration sheet of your expensive transfer standard multimeter to compensate for its error. Who cares if it shows 4.9999871 V when you give it 5V, its within spec and if it shows 4.9999876V on the next calibration you can be sure you can trust it down to almost the last digit. You know when its showing 4.999987x that is bang on 5V according to the cal labs fancy pants NIST traceable standard.

A calibration sticker does NOT automatically mean it is in spec. It just usually means it is in spec, but does sort of grantee its not way off. When the cal lab notices something being out of spec they will notify the costumer for further action. If it slowly drifted slightly out of spec they might calibrate it anyway and just keep that in mind to avoid loosing cal traceability. But if something shows way out of spec then they know there is something wrong with the equipment so they need to repair it or decommission it and it will not get a calibration sticker (but likely instead a big red fail sticker). If you don't need past traceability and its slightly out of spec then you can ask the call lab to also adjust it (this is never done without your permission).

So to sum it up. A calibration sticker on its own mostly means the equipment was in good working order on that day. People using the equipment will very rarely test the equipment for proper functionality or drift, but equipment carrying calibration stickers is checked out regularly by trained professionals to make sure it is performing as it should.