Author Topic: Is accuracy even that important?  (Read 1480 times)

0 Members and 1 Guest are viewing this topic.

Offline FriedMule

  • Frequent Contributor
  • **
  • Posts: 727
  • Country: dk
  • Can make even the simplest task look imposible.
Is accuracy even that important?
« on: June 12, 2021, 10:37:47 am »
Short story:
I have a couple of handheld DMM's and a couple of cheap hung low bench DMM's plus my "flagship" bench DMM, all showing different values in the micro voltages.
My power supplies does also show their values, who does not match any of my DMM's.
I am speculating on calibrating power supplies and hung low DMM's so they agree with my "flagship" DMM.

Question:
If I chose to calibrate my gear to agree with my flagship and my flagship may be a couple of micro voltage off, does that even matter?
If my gear is agreeing on the voltage, then should I get a fairly precise relative reading. I.e. 5V in is what my gear shows is 5V in and if I get 3.25V out do I know it's a fairly precise change of 1.75V in relation to input, ignoring it could be 5.013V in if measured by perfect gear and 3.263V out.
Even if I appear online is it not necessary so, my computer is on 24/7 even if I am not on.
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 11194
  • Country: au
Re: Is accuracy even that important?
« Reply #1 on: June 12, 2021, 10:44:08 am »
Question:
If I chose to calibrate my gear to agree with my flagship and my flagship may be a couple of micro voltage off, does that even matter?

Horses for courses.  It completely depends on what you are measuring.

If you are working with typical Arduino level circuits, then a cheap 31/2 digit meter is more than adequate.

If you are working with 24bit ADC or metrology circuits, then you'll need something a bit more high end.
 

Online Fungus

  • Super Contributor
  • ***
  • Posts: 12717
  • Country: 00
Re: Is accuracy even that important?
« Reply #2 on: June 12, 2021, 10:55:27 am »
The only important thing is to know your meters.

We all like having a lot of digits but let's face it: 1% accuracy is plenty for most work.

More digits are useful for measuring differences, eg. I just measured the voltage at the start and end of some 30AWG wires to see what the voltage loss was along them.

(0.13V if you must know...  :) )

Question:
If I chose to calibrate my gear to agree with my flagship and my flagship may be a couple of micro voltage off, does that even matter?

No.

a) It might matter psychologically but it won't make your circuits any better.

b) Messing with gear to adjust it isn't risk-free. You might make it worse.

c) Most of these things will vary with temperature. Trying to get non-temperature compensated gear to agree down to the last digit is a fools errand.
 

Offline Pfriemler

  • Contributor
  • Posts: 43
  • Country: de
Re: Is accuracy even that important?
« Reply #3 on: June 12, 2021, 10:56:40 am »
I know this :-)
I bought a cheap voltage standard, had my son at work compare it to a high precision meter (it was accurate) and made an accurate 10V out of all the readings between 9.94 and 10.1 volts as far as it went, just for peace of mind. Only one can not be calibrated so easily and shows 9.96V. That is less than 0.5% deviation. Related to a power supply, the influence of the output cables is already bigger.
As said, a good multimeter for the few important accurate data is enough. The rest is only for the optics.
« Last Edit: June 12, 2021, 10:59:14 am by Pfriemler »
once you do it right, it works :-)
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 11194
  • Country: au
Re: Is accuracy even that important?
« Reply #4 on: June 12, 2021, 10:59:12 am »
Separate to the topic of accuracy is resolution - yes, these are two different things!!!

This is where you have the ability to detect changes at the Least Significant Digits level.  The actual number displayed may not be completely correct, but if the numbers are going down, then you can be reasonably sure the value you are measuring is going down - and the magnitude of that change is reasonably correct.

There are times that this sort of measurement is quite useful - such as seeing the voltage sag on a power rail when a load is connected.  Whether that power rail measures as 12.01V or 12.03V isn't necessarily as important as seeing it drop by 0.5V - especially when, for example, it's supposed to be a regulated supply.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 9788
  • Country: de
Re: Is accuracy even that important?
« Reply #5 on: June 12, 2021, 11:19:11 am »
For many uses the accuracy of a voltmeter is not very important. More often it is about detecting small changes or just measuring votlage that are overall small (e.g. thermocouple).  Especially a scale factor that is slightly off can often be tolerated, though it can get complicated when comparing readings from different meter (e.g. 2 meters measuring the voltage on different side of a cable).


With lower grade DMM there is little to be done on calibration and an adjustment may not be possible.  Quite often an official calibration (like needed for some professional uses) is testing only, telling you the meter is still well within specs. An adjustment is usually not done unless the DUT is way off - in this case a cheap meter is usually marked as defect and disposed, as it would not be reliable even after an adjustment. In some areas it can be cheaper to by new instead of a calibration.

With a good, reliable meter and some suitebale sources, one could do the testing of the lower grade meters (more than a factor of 10 lower specs). So there is very limited use in sending some of the lower grade meters to calibration.
 

Online J-R

  • Frequent Contributor
  • **
  • Posts: 254
  • Country: us
Re: Is accuracy even that important?
« Reply #6 on: June 12, 2021, 11:52:02 am »
If you're just checking batteries and wall sockets around the house, accuracy may not matter.  1.5V battery OK, 1V not OK, 12V battery OK, 11V, not OK.  0V-120V-240V 50/60Hz...  Just the basics.

But I say if you have an interest in electronics, understanding the concepts and aspects of obtaining accurate measurements is a worthy exercise.  However, without a recent calibration certificate or a reference standard, making things agree is pointless.

The VREF-01 from Voltagestandard.com is $31+shipping.
Welectron sells the DMMCheck Plus for international customers.

If you can easily calibrate your other equipment after obtaining a reference, then go for it, otherwise just note the discrepancy and conditions.  For power supplies, the displayed voltage will frequently be invalid due to the voltage drop from the supply to the device unless you have a sense feature.  So connect a known-good DMM at the destination.  Probably you already know this, just checking...
 

Offline FriedMule

  • Frequent Contributor
  • **
  • Posts: 727
  • Country: dk
  • Can make even the simplest task look imposible.
Re: Is accuracy even that important?
« Reply #7 on: June 15, 2021, 03:58:37 am »
Am I right in understanding that accuracy is important for when the DUT is supposed to work in real life, while precision is more important while building the unit, as long I know what the real measurement is, can a precise DMM easily be way off?

The voltage reference is 0,02V or something, but is that accurate enough? I mean, my most crappy meter goes down to 0,001 micro volts?
Even if I appear online is it not necessary so, my computer is on 24/7 even if I am not on.
 

Online J-R

  • Frequent Contributor
  • **
  • Posts: 254
  • Country: us
Re: Is accuracy even that important?
« Reply #8 on: June 15, 2021, 05:13:42 am »
You will want to read up on counts/accuracy/resolution.  One of Dave's old videos is maybe an OK place to start if you haven't seen it yet:



Also keep in mind that the accuracy ratings can be different for each range or function of a DMM.

As an example, it's generally trivial for a budget DMM to resolve values of 0.1 mV while in the mV range, but not so much at 12V.  You would need a DMM that could display 12.0000V at minimum which would be a 120,000 count DMM.  In reality you would want at least a 1.2 million count DMM to get the extra digit past your target.  On top of that the DMM accuracy specs will be critical to note.  In some cases even the last THREE digits on a DMM display cannot be taken at face value without verifying with a known standard first (or even then maybe simply cannot be trusted ever).

The VREF-01 has a rated accuracy of 0.01%, which means it will output between 4.9995 and 5.0005 worse case.  In reality they are far better.  For example, my VREF-01 5V reference is at 5.00004 (notice the 4 zeros) after 15 months.  I also have the DMMCheck Plus and the PDVS2mini.  Having more than one reference provides a nice level of confidence.

As I've mentioned in other threads, something that tripped me up early on when getting back in to electronics was spec vs. real-world accuracy.  I read lots of reviews and looked at prices and said to myself I'll just buy the minimum or I'll get this budget model that everyone is raving about.  But I quickly realized many budget DMMs just barely met their specs and many of the higher quality DMMs exceeded them by a factor of 10.  That's where I'm at personally, but everyone has different needs/desires/budgets so you'll have to make your own call.
 

Online Fungus

  • Super Contributor
  • ***
  • Posts: 12717
  • Country: 00
Re: Is accuracy even that important?
« Reply #9 on: June 15, 2021, 06:02:19 am »
Am I right in understanding that accuracy is important for when the DUT is supposed to work in real life, while precision is more important while building the unit

No.

One is for measuring absolute values, one is for measuring differences.

can a precise DMM easily be way off?

Yes.

(eg. MESTEK DM91A)

The voltage reference is 0,02V or something, but is that accurate enough?

Meters are digitally calibrated so absolute accuracy of the reference doesn't matter. What matters is the stability.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 9788
  • Country: de
Re: Is accuracy even that important?
« Reply #10 on: June 15, 2021, 06:07:23 am »
It gets easier if the meter is also accurate and not just precise. If one knows about the limitations one can work quite well with a meter where you don't know the accurate scale factor. As a kind of extreme example I have my DIY high resolution ADC / voltmeter with some 6-7 digit resolution and presicion, but I only know the scale factor to some 1%.

There are few cases were you actually need accurate voltage reading. Even for the charging limit on Li Cells something like 1% accuracy is about sufficient if one stays on the safe side.

If one want's to exchange data with others or use multiple meters together it gets ugly if the accuracy is low. That is were accuracy matters.

A point sometimes overlook is also reliablity - so the probablity that the value is way off. Often this is more imporant than accuracy and calibration of meters is also about at least detecting those rare cases that a meter is way off. It does not help if a meter is supposed to be (number in the specs) very accurate, but has a significant probablity to be way off  (e.g. battery low but no warning).
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 3937
  • Country: hr
Re: Is accuracy even that important?
« Reply #11 on: June 15, 2021, 06:27:25 am »
I mean, my most crappy meter goes down to 0,001 micro volts?

Errr, you sure you have instrument that has resolution of 1 nano Volt ? 0,000000001 V ?

Apart from that, Kleinstein is, as usual, right. Absolute accuracy is important when you work with other people. It is same thing as standardizing other measurements, in other industries.

For instance, if I buy a 3mm pin, what actual size it will be? If I buy 3mm pin from Japan, and a 3mm drill from Germany, will the pin fit in the hole? Will it be too small or too big?
That was the whole point of standardization effort in industrial revolution.  Same here.

And same as mechanical problems, it will be important how accurate that 3mm is based on that what you do. Optical lens holder for accurate laser system will obviously have to be more precise and accurately made than garden tool.

In general, mostly you don't need better accuracy or resolution than what good handheld meters provide. When you do, you will know. It will be fairly obvious as a part of the problem you're solving.....
I'm a firm believer in not buying stuff for your lab (privately or work) that you won't use... There are few things you have to have (some decent handhelds, some kind of oscilloscope that covers kind of work you do, some power sources, some signal sources), all other equipment comes as requirement of your work...
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 14238
  • Country: gb
    • Having fun doing more, with less
Re: Is accuracy even that important?
« Reply #12 on: June 15, 2021, 11:08:06 am »
Am I right in understanding that accuracy is important for when the DUT is supposed to work in real life, while precision is more important while building the unit, as long I know what the real measurement is, can a precise DMM easily be way off?

A company I worked for sold an instrument with a resolution and stability of 0.001dB - and an accuracy of 0.1dB. Think about it.

The customers didn't care about the accuracy, but the resolution and stability were vital.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 
The following users thanked this post: 2N3055

Offline tv84

  • Super Contributor
  • ***
  • Posts: 2379
  • Country: pt
Re: Is accuracy even that important?
« Reply #13 on: June 15, 2021, 01:43:53 pm »
The customers didn't care about the accuracy, but the resolution and stability were vital.

That made me remind one of my teachers.
 

Offline FriedMule

  • Frequent Contributor
  • **
  • Posts: 727
  • Country: dk
  • Can make even the simplest task look imposible.
Re: Is accuracy even that important?
« Reply #14 on: June 16, 2021, 08:31:35 pm »
I think we are talking a bit about the same, do I think.
What I am aiming at is as long you know your gear is stable, precise and reliable while working on your project does accuracy not matter, it's only when your project has to shift hands, "talk" to products from others (i.e. battery) or is meant to act on real event, that accuracy is important.

My original question is mostly about when you are working on your project at home.

You talk about 0.01% as a high standard but what then about a 6½-8½ DMM, how do you use a 0.01% voltage reference.
Even if I appear online is it not necessary so, my computer is on 24/7 even if I am not on.
 

Online Fungus

  • Super Contributor
  • ***
  • Posts: 12717
  • Country: 00
Re: Is accuracy even that important?
« Reply #15 on: June 17, 2021, 06:25:31 am »
what then about a 6½-8½ DMM, how do you use a 0.01% voltage reference.

For that level of accuracy you need a reference with temperature control, ie. you heat it up to a known temperature inside the meter.

You won't get that in a handheld DMM because it would eat up too many batteries.
 

Online james_s

  • Super Contributor
  • ***
  • Posts: 16144
  • Country: us
Re: Is accuracy even that important?
« Reply #16 on: June 17, 2021, 06:43:16 am »
It totally depends on what you're doing.

Modern day instruments are generally very accurate, even cheap ones. A $10 DMM today will have better accuracy and precision than a high end VOM from the 1960s. For most applications it isn't something you even need to worry about anymore. If you need exceptional accuracy for something you'll probably know you need it.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf