Author Topic: How acccurate should digital scope voltage measurements be?  (Read 7796 times)

0 Members and 1 Guest are viewing this topic.

Offline Chris Wilson

  • Frequent Contributor
  • **
  • Posts: 991
  • Country: gb
  • Race car engineer, dog lover, hoarder.
How acccurate should digital scope voltage measurements be?
« on: January 29, 2012, 11:02:41 am »
I have a newly acquired Fluke / Philips PM3380B Combiscope CRT scope that does both analogue and digital, 2 channel. If switched to digital on screen numeric voltage readings for DC differ by about 0.02 volts per channel, using the same probe and dry cell voltage source, just changing probe from channel 1 input BNC to channel 2, and both readings differ by about 0.1 to 0.2 volts from both my (cheap and banged about) hand held DMM's, and reading these voltages on my USB scope.  The scope has an auto cal function, which when run after a two hour warm up helped, but there's still the above difference. Frequency read outs seem to match my USB scope and my signal generator digital readout. What would a calibration firm use as a voltage source to rely on? Which begs the question how does one calibrate the calibration meter? ;)

Thanks.
Best regards,

                 Chris Wilson.
 

Online Rerouter

  • Super Contributor
  • ***
  • Posts: 4222
  • Country: au
  • Question Everything... Except This Statement
Re: How acccurate should digital scope voltage measurements be?
« Reply #1 on: January 29, 2012, 11:12:44 am »
in a pinch, you could buy a reference regulator, e.g like a 2.5V one with a tight initial accuracy, and use that to provide a far better calibration than what you would normally have on hand,
 

Offline Bored@Work

  • Super Contributor
  • ***
  • Posts: 3932
  • Country: 00
Re: How acccurate should digital scope voltage measurements be?
« Reply #2 on: January 29, 2012, 11:16:18 am »
Oscilloscopes aren't precision instruments. I wouldn't be surprised if your oscilloscope is specified with +/-3%. I am too lazy to look it up, but for sure you aren't too lazy to look it up in your oscilloscope's manual, are you?
I delete PMs unread. If you have something to say, say it in public.
For all else: Profile->[Modify Profile]Buddies/Ignore List->Edit Ignore List
 

Offline wkb

  • Frequent Contributor
  • **
  • Posts: 540
  • Country: nl
Re: How acccurate should digital scope voltage measurements be?
« Reply #3 on: January 29, 2012, 11:26:06 am »
Just consulted the manual of a Tek 2465A which is not a cheapie scope:

For channel 1 & 2:

Quote
In the +15 degrC to +35C range: Within +/- 2% at any VOLTS/DIV setting for  a four or five division signal centered on the screen.

deltaV accuracy (using cursors over the the entire graticule aread): +/- (1.25% of reading + 0.03 div + signal abberations)

I like the deltaV one.  Signal abberations?
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29264
  • Country: au
    • EEVblog
Re: How acccurate should digital scope voltage measurements be?
« Reply #4 on: January 29, 2012, 11:27:15 am »
Both analog and digital scopes have always been only a few % accurate at best, it's just the industry norm. Around 3% is a very common figure.
With an 8bit ADC most digital scopes have, anything less than about 0.5% would be pointless of course.
As for analog, it's the same reason why you rarely see analog multimeters better than 1% (several % common), it's pointless given the visual interpretation required on the analog scope (or screen graticule)

Some of the analog Tek (and maybe others) scopes had real multimeter coupled to the front end, so often got better performance than the scope part.

Dave.
« Last Edit: January 29, 2012, 11:28:59 am by EEVblog »
 

alm

  • Guest
Re: How acccurate should digital scope voltage measurements be?
« Reply #5 on: January 29, 2012, 02:10:38 pm »
Even if you ignore the ADC/CRT, there's still the front-end to account for, and it's not typically optimized for great accuracy. Note that the 1-2% spec you might expect to see on a DSO only applies at DC, it gets worse as the frequency goes up. At its rated bandwidth, you might expect it to be up to 30%. Few scope vendors publish even a typical frequency response, let alone a guaranteed max. at anywhere but DC and its rated bandwidth.

Unless your DMM is way out of spec or very crappy, expect it to be more accurate than your scope, at least at DC and for 50/60 Hz sinusoidal signals.
 

Offline Chris Wilson

  • Frequent Contributor
  • **
  • Posts: 991
  • Country: gb
  • Race car engineer, dog lover, hoarder.
Re: How acccurate should digital scope voltage measurements be?
« Reply #6 on: January 29, 2012, 04:56:17 pm »
OK, thanks for this, I had wrongly assumed an instrument as expensive as these were new would be more accurate than the average elderly Taiwanese DMM, I am still learning, apologies for the naive post.  I am pleased though,  as it means a power supply I was looking at for ripple and voltage is OK if I go by my two DMM's voltage readout, and not the DC voltage the scopes numerical digital readout gave me ;) I have been carefully through the Fluke manual and the datasheet for it, and it makes no mention of voltage readout accuracy, perhaps unsurprisingly!

I am off to look up about voltage reference regulators, having already seen Dave's blogs on the Rubidium oscillators for frequency standards.
Best regards,

                 Chris Wilson.
 

Offline Lesolee

  • Contributor
  • Posts: 10
Re: How acccurate should digital scope voltage measurements be?
« Reply #7 on: January 29, 2012, 05:24:14 pm »
It's a common mistake to think that expensive = accurate. I have been designing test equipment for 30 years, and digital storage oscilloscopes for 20 of those 30. AC measurements are inherently less accurate than DC.

When you made your measurement were you perchance using a 10Meg 10:1 scope probe? That is only accurate to 1% on its own. Then you have the 1M input impedance uncertainty of the scope, maybe another 1%. We haven't even got to the "accuracy" spec of scope yet! Then did you measure at full scale or 1/10th full scale? You have a linearity error as well as a gain error you know. When somebody glibly says the scope is "3% accurate" that is oversimplified  ::)

The most accurate thing to measure DC with is a DVM (as stated in an earlier post). Scopes are typically calibrated at 1kHz.
 

Offline logictom

  • Supporter
  • ****
  • Posts: 336
  • Country: au
Re: How acccurate should digital scope voltage measurements be?
« Reply #8 on: January 30, 2012, 12:03:43 am »
I was wondering this myself. I am measuring a signal and was trying to work out the best method of measurement.
I know that an oscilloscope wouldn't be as accurate as my DMM, just switching between ranges gave significant difference on a couple 100mV signal I was measuring.
For logging we have a Vbox but it simply states it's accuracy as 400uV (3uV resolution). I understand accuracy when stated as a percentage but does this mean it's accurate to within 0.4mV over the entire range?
« Last Edit: January 30, 2012, 12:49:08 am by logictom »
 

Online amspire

  • Super Contributor
  • ***
  • Posts: 3638
  • Country: au
Re: How acccurate should digital scope voltage measurements be?
« Reply #9 on: January 30, 2012, 12:26:55 am »
Just to add to Lesolee's comments, even if you have calibrated for perfect DC accuracy with the current oscilloscope and probe combination, to get 1% accuracy for AC voltage levels needs a frequency response flat to within 0.1dB.

Now if you can get an audio amplifier to be 0.1 dB flat from 20 Hz to 20KHz, you have done extremely well.  To get oscilloscope and probe combination to be accurate to 0.1dB from DC to 100MHz or more would be something absolutely amazing.  Even a modest 1dB error amounts to a 12% voltage error.

Richard
 

alm

  • Guest
Re: How acccurate should digital scope voltage measurements be?
« Reply #10 on: January 30, 2012, 01:11:19 am »
For logging we have a Vbox but it simply states it's accuracy as 400uV (3uV resolution). I understand accuracy when stated as a percentage but does this mean it's accurate to within 0.4mV over the entire range?
What's a Vbox? A link so we know what you're talking about wouldn't hurt. 400uV accuracy should apply to all ranges (if it has any form of range switching), but it's also possible that they only quoted the (best) spec for the lowest range and omitted the spec for the other ranges. Typically accuracy specs consist of a certain part of the value and a certain part of the range. Accuracy for analog VOMs is usually specified as percentage of full scale. DMMs are often specified as percentage of value plus percentage of range (or number of least significant digits). Some also have an additional absolute term, for example to account for lead resistence. You can add more terms, eg. to account for temperature or time since last calibration, to make the model more accurate. You can also ignore terms because they're negligible or because they don't match your marketing claims ;).

For example, the DC accuracy spec from a random Tek DSO manual I have handy is +/- (0.02 * abs(reading - offset - vertical position) + offset accuracy + 0.15 div + 0.6 mV). The 0.02 gain error is increased by 0.00025/°C above 30°C ambient. As Lesolee remarked, you should add the uncertainty in input impedance and probe attenuation if you're using a 10x probe. It's not as bad as it sounds since uncorrelated random errors add up as the square root of the sum of squared errors, so the contribution of the smaller values is reduced. In this case, the error of the scope (about 4% for an 8 V signal at 1V/div) dominates and the 1% errors of the attenuating probe are insignificant, so the 3% claim was not too far off ;).
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 9489
  • Country: us
Re: How acccurate should digital scope voltage measurements be?
« Reply #11 on: January 30, 2012, 01:53:00 am »
Just as a data point, I did a quick check on my Rigol with a 5 V reference source. As you can see it settled at the right voltage. (It read the same on Ch 2 as well.)

However, on this vertical scale the precision is 0.04 V so waving my hand near the probe could make the voltage jump around between 4.96 V and 5.04 V. Since full scale is 8 V, that would give an accuracy of about +/- 0.04/8 = +/- 0.5% on DC at this scale.



I'm not an EE--what am I doing here?
 

Offline logictom

  • Supporter
  • ****
  • Posts: 336
  • Country: au
Re: How acccurate should digital scope voltage measurements be?
« Reply #12 on: January 30, 2012, 01:46:31 pm »
What's a Vbox? A link so we know what you're talking about wouldn't hurt. 400uV accuracy should apply to all ranges (if it has any form of range switching), but it's also possible that they only quoted the (best) spec for the lowest range and omitted the spec for the other ranges.

Thanks Alm, I edited it to include a link but for clarity here it is again:
http://www.velocitybox.co.uk/index.php/en/peripherals/modules/66-analogue-input-module.html
http://www.racelogic.co.uk/_downloads/vbox/Manuals/Input_Modules/RLVBADC03_Manual.pdf
It does +-50V but doesn't mention range switching, I haven't been permitted to open it up to have a look.
I don't have much experience in taking measurements - if using ~60cm cables from output being sensed and input of logging module what sort of error from noise/voltage drops would I be likely to see - measured voltage 0-5V but trying to record differences at mV.
 

Offline olsenn

  • Frequent Contributor
  • **
  • Posts: 993
Re: How acccurate should digital scope voltage measurements be?
« Reply #13 on: January 30, 2012, 02:23:36 pm »
Scopes are great for a quick eyeballing of a value... you can easily distinguish 5V from 0V for TTL logic etc. and most importantly, unlike even the most expensive multimeters, the AC bandwidth is very high (100MHz for the Rigol DS1102E etc) so your peak-to-peak measurements will be decent even at high frequencies (Fluke 87-V only goes to 20KHz for example).

If you need 5+ decimal places of precision/accuracy, then you'll need a benchtop multimeter!
 

Offline Chris Wilson

  • Frequent Contributor
  • **
  • Posts: 991
  • Country: gb
  • Race car engineer, dog lover, hoarder.
Re: How acccurate should digital scope voltage measurements be?
« Reply #14 on: January 30, 2012, 02:46:52 pm »
logictom: Are you using the Vbox for anything exciting like motorsport? Just curious as that's my passion, cheers.
Best regards,

                 Chris Wilson.
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: How acccurate should digital scope voltage measurements be?
« Reply #15 on: January 30, 2012, 05:30:37 pm »
Scopes have their best accuracy towards the high end of its scale, as you've pictured.  But it can vary depending on vertical gain too, from issues with the vertical amplifier, independent of the amps frequency response.   The published figures should be worse case, and at least, Rigol 1052e does live up to its spec sheet, at my last evaluation.

Try a variable PSU, adjust the voltage down while watching the output with your DMM, and compare it to the Vout reading of the Rigol.  When its as small as possible but still measurable, adjust the Rigol's vertical gain to bring that DCV to full scale, and compare again.

Changes in environmental conditions can cause offsets too; for best accuracy before use, at least on the Rigol, do a self-cal before use and insure the scope is warmed up 30+min.
 

Just as a data point, I did a quick check on my Rigol with a 5 V reference source. As you can see it settled at the right voltage. (It read the same on Ch 2 as well.)

However, on this vertical scale the precision is 0.04 V so waving my hand near the probe could make the voltage jump around between 4.96 V and 5.04 V. Since full scale is 8 V, that would give an accuracy of about +/- 0.04/8 = +/- 0.5% on DC at this scale.


Best Wishes,

 Saturation
 

Offline tekfan

  • Frequent Contributor
  • **
  • Posts: 385
  • Country: si
Re: How acccurate should digital scope voltage measurements be?
« Reply #16 on: January 30, 2012, 05:33:35 pm »
If you need 5+ decimal places of precision/accuracy, then you'll need a benchtop multimeter!

And even very expensive benchtop multimeters can usually achieve only 1% accuracy at their high frequency limit (usually less than 2MHz). That's due to the True RMS converter limitations.

If you really want exact measurments at high frequency you need a True RMS voltmeter with a thermal RMS converter (eg. HP3403)
One can never have enough oscilloscopes.
 

Offline logictom

  • Supporter
  • ****
  • Posts: 336
  • Country: au
Re: How acccurate should digital scope voltage measurements be?
« Reply #17 on: January 30, 2012, 08:28:01 pm »
logictom: Are you using the Vbox for anything exciting like motorsport? Just curious as that's my passion, cheers.
Nothing that exciting I'm afraid - the kit was (I'm assuming) originally bought for logging on vehicles but I'm using it as it is the only thing available to me for logging. I'm working on a sensor for an exhaust gas recirculation valve but the company has a small range electric vehicles which the kit is mainly used on as well as some custom work we've done for others (Liberty Range Rover ;))

I'm not convinced, however, that the Vbox is the best tool for the job. What do others use for logging? I need to log up to 5 thermocouples and 8 analogue outputs, 0-5V DC signals.
 

Offline chscholz

  • Regular Contributor
  • *
  • Posts: 85
Re: How acccurate should digital scope voltage measurements be?
« Reply #18 on: January 30, 2012, 08:50:39 pm »
DC accuracy is spec'ed  at +/- 0.5% of Full Swing for a LeCroy HRO 12 bit scope.


Oscilloscopes aren't precision instruments. I wouldn't be surprised if your oscilloscope is specified with +/-3%. I am too lazy to look it up, but for sure you aren't too lazy to look it up in your oscilloscope's manual, are you?
 

Offline Chris Wilson

  • Frequent Contributor
  • **
  • Posts: 991
  • Country: gb
  • Race car engineer, dog lover, hoarder.
Re: How acccurate should digital scope voltage measurements be?
« Reply #19 on: January 30, 2012, 08:51:44 pm »
logictom: Speaking from the automotive, and specifically the race world, probably the best data logger is the Motec ADL, but it's costly and probably OTT for what you need. I use a GEMS DA99-L2 logger for data logging via CAN from my Motec M800 ecu. It will take a few analogue inputs, but is specifically really of use via a CAN interface. They do analogue to CAN converters though. The Motec ADL has specific add ons that will have K type thermocouple inputs, as well as a host of spare analogue and digital one. However, their excellent Interpreter software is most definitely motorsport biased. it will take GPS input though. I am sure there will be some logger more suited at a better price, but if you want links to the Motec loggers or the GEMS ones ( I have three LA99-L2 units and want to sell one) just ask.

Motec are at http://www.motec.com.au

GEMS at http://www.gems.co.uk

Cheers.
Best regards,

                 Chris Wilson.
 

alm

  • Guest
Re: How acccurate should digital scope voltage measurements be?
« Reply #20 on: January 31, 2012, 12:11:29 am »
However, on this vertical scale the precision is 0.04 V so waving my hand near the probe could make the voltage jump around between 4.96 V and 5.04 V. Since full scale is 8 V, that would give an accuracy of about +/- 0.04/8 = +/- 0.5% on DC at this scale.
This could be entirely consistent with a 3% accuracy on that range, since accuracy is often specified in 99% or 95% confidence intervals, the majority will be much closer, especially at room temperature (assuming normal distribution). It's also common to see DMMs with an uncertainty of 100 least significant digits agree down to 10 LSDs or so. The spec on that DMM and reference is probably also worse than the 0.01% error displayed here.

Thanks Alm, I edited it to include a link but for clarity here it is again:
http://www.velocitybox.co.uk/index.php/en/peripherals/modules/66-analogue-input-module.html
http://www.racelogic.co.uk/_downloads/vbox/Manuals/Input_Modules/RLVBADC03_Manual.pdf
It does +-50V but doesn't mention range switching, I haven't been permitted to open it up to have a look.
I don't have much experience in taking measurements - if using ~60cm cables from output being sensed and input of logging module what sort of error from noise/voltage drops would I be likely to see - measured voltage 0-5V but trying to record differences at mV.
400 µV accuracy on 50 V is pure fantasy, that would rival a good 7.5 digit bench DMM. Getting just a voltage reference that accurate will cost some serious bucks. From the description it appears to have ranges from +/- 200 mV to +/- 50 V. They quote 16-bit of resolution at 200 mV and 24-bit at 50V. The resolution will probably be much worse after you account for noise. They might be able to meet the 400µV spec at +/- 200 mV, but I expect the +/- 50 V accuracy to be much worse. If they actually claim 400 µV accuracy for 50 V values, then I'd ask for a copy of their calibration procedure and equipment list. This has to be fairly impressive to meet the 8 ppm spec.

DC accuracy is spec'ed  at +/- 0.5% of Full Swing for a LeCroy HRO 12 bit scope.
That's slightly better than a basic 3.5 digit DMM. Hardly qualifies as precision, although it's much better than most other scopes. I believe Tektronix had a series (11k?) that could achieve a fairly decent accuracy.
 

Offline Chris Wilson

  • Frequent Contributor
  • **
  • Posts: 991
  • Country: gb
  • Race car engineer, dog lover, hoarder.
Re: How acccurate should digital scope voltage measurements be?
« Reply #21 on: February 01, 2012, 08:29:12 pm »
Just as a data point, I did a quick check on my Rigol with a 5 V reference source. As you can see it settled at the right voltage. (It read the same on Ch 2 as well.)

However, on this vertical scale the precision is 0.04 V so waving my hand near the probe could make the voltage jump around between 4.96 V and 5.04 V. Since full scale is 8 V, that would give an accuracy of about +/- 0.04/8 = +/- 0.5% on DC at this scale.



There have been some great and in depth answers here, again I am slowly learning, thanks everyone for the trouble you have gone to.

 I was looking at the little voltage reference board that you are using in the photo, what do you think of them? I have seen them on Ebay. Would you recommend one for the price? As it is I have no means of knowing if my two bench DMM's and my two hand held ones are anything like accurate, other than comparing one to another.  I need to look at buying an ESR meter, some sort of cheap voltage reference, and I would like a resistance box, probably like the one Dave stripped down in one of his blogs. I would also like a slightly more function rich signal generator. The difficulty is whether to go for things that are used and which probably cost thousands new, from Ebay, and risk buying something faulty and too complex to attempt a repair on, or something from China new that may be less sophisticated but SHOULD just work...
Best regards,

                 Chris Wilson.
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 9489
  • Country: us
Re: How acccurate should digital scope voltage measurements be?
« Reply #22 on: February 01, 2012, 09:07:41 pm »
I was looking at the little voltage reference board that you are using in the photo, what do you think of them?

The board in the picture is the DMMCheck device from here:

http://www.voltagestandard.com/

You get a voltage check, a current check and some resistance checks. I was in the same position as you, having no good way of telling how accurately my meters were reading. With this little board I at least have one absolute point of reference to compare against.

Recommended.
I'm not an EE--what am I doing here?
 

Offline Chris Wilson

  • Frequent Contributor
  • **
  • Posts: 991
  • Country: gb
  • Race car engineer, dog lover, hoarder.
Re: How acccurate should digital scope voltage measurements be?
« Reply #23 on: February 01, 2012, 09:21:48 pm »
I was looking at the little voltage reference board that you are using in the photo, what do you think of them?

The board in the picture is the DMMCheck device from here:

http://www.voltagestandard.com/

You get a voltage check, a current check and some resistance checks. I was in the same position as you, having no good way of telling how accurately my meters were reading. With this little board I at least have one absolute point of reference to compare against.

Recommended.

Right, thanks, just ordered one, his postage rates are far better than any on Ebay. I am about to order an ESR tester, the Peak ones on Ebay are dearer than buying direct from the manufacturer 50 miles away from me. Thanks for the link, should be here in a couple of weeks I guess. Cheers.
Best regards,

                 Chris Wilson.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf