EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: BiOzZ on April 11, 2013, 06:32:17 pm
-
Me and my friend are having a debate ... he likes to think his scope is about as accurate as it gets when it comes to voltage measurement (not waveform viewing or anything like that but SPECIFICALLY measuring voltage with/without even a waveform) but i think its no more accurate than a $100 DMM when it comes to voltage measurement
we both have ~$500 entry level 25mhz digital scopes and i doubt his re-branded rigol is any better than my owon when it comes to voltage measurement (or any other $500 scopes)
i dont have any > $100 DMMs to test this with and i dont want to lug it in to work to test this
what do you guys think/know?
i cant picture it using any different type of voltage reference than a $100 DMM for there $500 price! and i have never even thought of having my scope calibrated nor did i really think thats necessary as i dont really use it for more than measuring waveform voltage 0.1v resolution ... and i just count divs not even going in to measure really! but if it is some how more accurate there than a $400 fluke than fuck ill use that! XP
-
'scopes are generally not very accurate when it comes to voltage measurements - there again some DMMs aren't that accurate either.
Almost always it's an 8-bit ADC in a 'scope input so the resolution is limited to 0.4% anyway but the usual accuracy of a 'scope input is 1-2% for voltage, plus there will be some fall off in response with high frequency signals.
A cheap DMM might only be specified as 1% accuracy so they're in the same ball park. More expensive DMMs are usually better - 0.1% or even 0.05% can be had for $100 (e.g. UT61E is specified as 0.1% for DC voltage),
Edit: stylistically it's just wrong to use the word "typically" three times in two sentences.... so I fixed that :)
-
yeah that was around what i suspected ... i was unaware they would use 8 bit adcs vs like 10 bit adcs but still thats what i expected ... atleast out of a digital scope
-
Most real time scopes only use 8-bit ADC so they can keep the sample rate up and process all that data real time. An oscilloscope is (typically) a general purpose instrument anyway and that is (typically) good enough. Typically :)
One thing they can do though since they take readings really fast is seriously over sample slow signals and then average to get "higher resolution".
-
Rhode and Schwatz have a 14-bit scope, I think. But if you have to ask the price...
-
Most real time scopes only use 8-bit ADC so they can keep the sample rate up and process all that data real time. An oscilloscope is (typically) a general purpose instrument anyway and that is (typically) good enough. Typically :)
One thing they can do though since they take readings really fast is seriously over sample slow signals and then average to get "higher resolution".
yeah than i forget the actual resolution of the screen and the fact that they have scaling ... 8 bit does the job fine than XP
-
I agree with what was already said, a scope is not a precision tool.
However, i have a 500$ scope (Rigol DS1052E) and a 100$ Multimeter (ELV branded CEM Meter) so i figured why not just test it.
Here you go.
Not sure if you can read it but the scope shows a max. V of 5.04V. Taking the Fluke 87V as a reference that's about 0.8% out.
I also threw in a cheap 15€ Meter and even that's closer. So i guess that settles it.
-
I took some measurements when testing my new scopes dc gain (a WaveJet) so the following are for the meter on my power supply, a DVM which cost about £40 from Maplin, my bench DVM 5 1/2 digits but not recently calibrated and finally the Oscilloscope:
| PS | DVM | BenchDVM | Scope |
| 100mV | 96.2mV | 96.199mV | 96.5mV |
| 500mV | 494.0mV | 494.17mV | 494mV |
| 1V | 0.996V | 0.99643V | 997mV |
| 2V | 1.992V | 1.9948V | 1.99V |
| 5V | 4.988V | 4.9936V | 4.99V |
| 10V | 9.97V | 9.9903V | 9.99V |
| 15V | 14.97V | 14.9865V | 15.0V |
| 20V | 19.96V | 19.9905V | 20.0V |
| 30V | 29.94V | 29.9852V | 30.0V |
Assuming that the bench DVM is the most accurate (as I say I've not had it calibrated so it might not be), the Scope does a bit better than the hand held DVM at higher voltages and a bit worse at the lower voltages but there is not much in it.
-
In theory, no, the best a regular digital scope can do is 0.4% resolution with it's 8 bit ADC.
So scopes are usually greater than 1% specified accuracy on the vertical ranges.
Hardly any scope maker gilds the lily on the vertical ranges.
-
| 15V | 14.97V | 14.9865V | 15.0V |
| 20V | 19.96V | 19.9905V | 20.0V |
| 30V | 29.94V | 29.9852V | 30.0V |
the Scope does a bit better than the hand held DVM at higher voltages
The scope cannot have better acuracy at those higher voltages when it has one digit less resolution than the DVM. XX.0 does not mean no error, it could be +/-0.05 off of zero and still read zero.
-
In theory, no, the best a regular digital scope can do is 0.4% resolution with it's 8 bit ADC.
So scopes are usually greater than 1% specified accuracy on the vertical ranges.
Hardly any scope maker gilds the lily on the vertical ranges.
The spec for the WaveJet is
± (1.5% + 0.5% of full scale)
so I guess the 0.5% of full scale is the lsb of the 8 bit converter. I did the measurements to check if it was within spec which it is, but I was quite pleasantly surprised how close it was to the bench DVM readings.
-
I think scope can still achieve decent resolution since they end up doing oversampling all the time. So the only weak part is when it comes to none repetitive signals where you can't average things, and the voltage refs used aren't as good compared to proper meters.
Personally I would say +-0.5% within a Fluke is pretty darn good for every day work and from psycho0815's test even the cheap ass Rigol aced that.
-
| 15V | 14.97V | 14.9865V | 15.0V |
| 20V | 19.96V | 19.9905V | 20.0V |
| 30V | 29.94V | 29.9852V | 30.0V |
the Scope does a bit better than the hand held DVM at higher voltages
The scope cannot have better acuracy when it has one digit less resolution than the DVM. XX.0 does not mean no error, it could be +/-0.05 off of zero and still read zero.
Having more digits decreases the rounding uncertainty, it doesn't necessarily mean more accuracy. Yes, you're right that it might mean 30.05V, but it also might be dead on the true voltage what ever that is. Rounding all the readings at 30V nominal to 3 sf the hand held DVM is reading 29.9V while the other three all agree at 30.0V. I'm sure a lot of cheaper DVMs have more digits than are justified by their intrinsic accuracy.
I'm not suggesting though that a scope is the right instrument to measure voltage, just that the accuracy is not far off that of a cheaper DVM.
Of course, as Spike Milligan put it, a stopped clock is perfectly accurate twice a day.
-
I think scope can still achieve decent resolution since they end up doing oversampling all the time. So the only weak part is when it comes to none repetitive signals where you can't average things, and the voltage refs used aren't as good compared to proper meters.
Personally I would say +-0.5% within a Fluke is pretty darn good for every day work and from psycho0815's test even the cheap ass Rigol aced that.
Hypernova is very right.
The dc accuracy of the scope is normally quoted for single point measurements which are 8 bit and only around 2% accurate. But the OP's question was regarding measuring dc voltage like a voltmeter so we can use averaging.
With 500 points on the screen and 256 point averaging on the WaveJet (for example), assuming the presence of some white noise, the number of effective bits is much more than 8, if the noise was white and evenly distributed then the number of bits would increase by (9+8)/2 giving a total of nearer 16 bits though in reality the noise floor of the scope would limit this. Also the error is decreased significantly.
I set my WaveJet for 256 point averaging and average reading of the dc wave form and set my power supply for as close to 1V as possible measured on the bench multimeter.
The bench multimeter reads 1000.45 mV and the average reading on the scope is 999.2 mV so the difference is only about 0.1% - much better than the single point accuracy of 2%.
If you're allowed to use saved waveforms then the averaging could take place over 500,000 or 1M points. Of course the scope's linearity and voltage references presumably become the limiting factor but on the WaveJet at least these seem pretty good.
The above measurements were made after not much warm up. Now the meter is reading 1000.07 mV and the scope is reading 1.000V. Perhaps scopes can make good dc measurements.
-
With 500 points on the screen and 256 point averaging on the WaveJet (for example), assuming the presence of some white noise, the number of effective bits is much more than 8
And therein lies the catch. Remove the noise and the ADC only outputs the same bit during the whole boxcar averaging cycle (for a clean DC in), so you don't get any advantage, you are stuck with 8 bit resolution.
-
I was surprised when I checked how good my scope (Agilent 3000X) is at dc voltage measurements. My calibrated 6.5 digit bench meter says the output from my power supply when set to 1V is 1.000285V, and the scope says 1.0000V (occasionally flickering up to 1.0005V).
Just in case that measurement was lucky, I then tried 20V. The DMM says 20.0117V, and the scope says 19.997V - an error of 0.07%. Moreover, it's perfectly capable of resolving voltage changes which are imperceptibly small on the actual waveform display.
Interestingly, the DVM option is much less precise than using the Avg-FS (average, full screen) measurement.
-
Rhode and Schwatz have a 14-bit scope, I think. But if you have to ask the price...
No they don't, they just have 8bit scopes and they are called Rohde & Schwarz, not Schwatz. They have special probes with integrated DVM functionallity, but that's it. DVM and DSO functionallity are two separate modules.
Most of the better 8bit scopes are specified to have a DC accuracy of 2%. Also the praise to be 12bit Agilent DSO9000 series because the just use an 8bit adc and oversampling along with boxcar averaging to clean up their signals. This increases the resolution but not accuracy.
The real 12bit DSO LeCroy HDO6000 series is instead specified to have a DC accuracy of 0.5%. A scope will never replace a DVM.
-
My analogies ..
Scope = speedometer
(http://upload.wikimedia.org/wikipedia/commons/thumb/5/5d/Ford_Mondeo_MK3_ST220_-_Speedometer_(light).jpg/200px-Ford_Mondeo_MK3_ST220_-_Speedometer_(light).jpg)
DVM = Ruler
(http://upload.wikimedia.org/wikipedia/commons/thumb/1/17/Measuring-tape.jpg/220px-Measuring-tape.jpg)
-
Me and my friend are having a debate ... he likes to think his scope is about as accurate as it gets when it comes to voltage measurement (not waveform viewing or anything like that but SPECIFICALLY measuring voltage with/without even a waveform) but i think its no more accurate than a $100 DMM when it comes to voltage measurement
we both have ~$500 entry level 25mhz digital scopes and i doubt his re-branded rigol is any better than my owon when it comes to voltage measurement (or any other $500 scopes)
i dont have any > $100 DMMs to test this with and i dont want to lug it in to work to test this
what do you guys think/know?
I have been investigating that for some certification work. A good scope (A-brand) is accurate to about 3% for frequencies up to 1MHz. Typically they can be better but no waranties. For measuring DC you can use a DMM which will get you sub 1% accuracy. However when it comes to measuring AC current or voltage all bets are off. Even the really high end multimeters don't go below 2 or 3%. It gets worse if you want to measure non-sinusoidal waves. Just read the specs and be amazed.
The bottom line is: for measuring AC a scope is just as good as a multimeter.
-
A DMM can definitely do better than 2% for ACV, though it depends on your definition of really high end. An average bench meter (the old Agilent 34401A) will do < 1% from 5 Hz - 100 kHz, and 0.06% from 10 Hz to 20 kHz for sine waves. Add 0.4% for crest factor up to 5, so about 0.5% from 10 Hz - 20 kHz. I'm too lazy to look up the specs for the 3458A, but I'm quite sure it can do much better since it can actually digitize the waveform and calculate the RMS.
Of course at 1 MHz DMMs tend to suck, if they can measure it at all. The input impedance starts to suck, for starters. A power meter would generally be the instrument of choice at higher frequencies.
-
With 500 points on the screen and 256 point averaging on the WaveJet (for example), assuming the presence of some white noise, the number of effective bits is much more than 8
And therein lies the catch. Remove the noise and the ADC only outputs the same bit during the whole boxcar averaging cycle (for a clean DC in), so you don't get any advantage, you are stuck with 8 bit resolution.
A signal with no noise? What's the odd of that? I think it's clear at this point that if you just need to measure average DC in a pinch a scope is more than good enough over cheap 1% DVMs.
Now, if you need to measure the voltage level at a particular spot on a 1ns pulse for that only occurs once in a blue moon, the 8bit is a killer, but it's not like a 6.5D DVM can do it either, so we are shit out of luck either way.
-
A DMM can definitely do better than 2% for ACV, though it depends on your definition of really high end. An average bench meter (the old Agilent 34401A) will do < 1% from 5 Hz - 100 kHz, and 0.06% from 10 Hz to 20 kHz for sine waves. Add 0.4% for crest factor up to 5, so about 0.5% from 10 Hz - 20 kHz. I'm too lazy to look up the specs for the 3458A, but I'm quite sure it can do much better since it can actually digitize the waveform and calculate the RMS.
Actually the 34401A is the best I could find. The rest is worse. Still the specs are not stellar and 100 times worse than when measuring DC.
@Hypernova: it is very easy to put a signal into a scope which doesn't have enough noise for the averaging algoritm to work properly. I have demonstrated that several times. For averaging to work you need at least 1LSB noise and that noise needs to be 'clean' as well.
-
Yeah, I suppose, but IMO that's pretty hard to come across in most ADC usages. I work around inverters so much that if you see a flat line out of the ADC it's probably shot.
-
A strange circumstance where noise is good!
As an experiment, if I compare averaging with the 20MHz limit on and that with the full BW of the scope set, the full BWis closer to the Bench DVM at 30V at least.
Bench DVM : 30.0025V Scope with 20MHz limit : 29.80V Scope with full BW : 30.04V
Yes - I do realise that using a scope to measure dc is not really sensible from the convenience and speed of measurement point of view, but it is quite a good way of testing the calibration of the scope and also it is good to get a feel of the general accuracy.
Basically it is fun playing around with the scope. :)
-
I know I shouldn’t post this because I going to end up looking silly, but too late now. Isn’t the accuracy of a scope reading dependent on the offset you set as well as the resolution/accuracy of the ADC within the scope?
E.g. If you add an offset and zoom into the point of interest don’t you get a more accurate reading than just having the full waveform on the screen. Doesn’t doing this increased the gain to the ADC and shift ADC’s 0V point up (or down). I’m definitely no expert on this but I thought the scopes worked by adding the offset voltage to the ADC input so that they could cover a larger range of voltages with greater accuracy.
-
I know I shouldn’t post this because I going to end up looking silly, but too late now. Isn’t the accuracy of a scope reading dependent on the offset you set as well as the resolution/accuracy of the ADC within the scope?
E.g. If you add an offset and zoom into the point of interest don’t you get a more accurate reading than just having the full waveform on the screen. Doesn’t doing this increased the gain to the ADC and shift ADC’s 0V point up (or down). I’m definitely no expert on this but I thought the scopes worked by adding the offset voltage to the ADC input so that they could cover a larger range of voltages with greater accuracy.
On the WaveJet you can set relatively large offsets so for instance you can have a +/- 10V offset at 50.2mV/div (at 50mV/div it drops to +/- 1V). Unfortunately from the point of view of measuring dc the offset itself is not much more accurate.
The spec for the DC Gain is +/-(1.5% + 0.5% full scale) while the Offset Accuracy is +/-(1% + 0.5% of full scale + 1mV). It does gain you something at around 10V if you measure it at 2V/div then the error is specified to be better than 230 mV, if you measure it at 100 mV/div then the offset error is 105mV and the measurement error is another 4mV so the total is 109mV.
But it is a good point as it gets around the 8 bit limit to some extent.
-
So by using the offset, even though you gain a lot more precision (i.e. you can zoom in on small spikes etc), you don’t gain as much on accurate because of the offset accuracy.
-
So by using the offset, even though you gain a lot more precision (i.e. you can zoom in on small spikes etc), you don’t gain as much on accurate because of the offset accuracy.
You actually gain quite a bit. Both the offset and the gain have an error term that is 0.5% of full scale but using the offset allows full scale to be much smaller so you gain there. The offset error is 0.5% less but there is an extra 1mV error so for voltages where 0.5% is greater than 1mV, i.e. greater than 200 mV you also gain on this.
The conclusion is that if your signal allows it, using the offset is a good idea though the mechanics is a pain - I've played with it on the WaveJet and the offset changes every time you change the vertical range. That is if you set -10V on the 1V/div range and change to 500mV/div then the offset drops to -5V and you have to readjust by knob twiddling to get it back to -10V. Perhaps Agilent and other scopes behave more sensibly in this respect.
-
OK,at DC & low frequencies,the DMM wins,but the digital voltage readout on DSOs will give you a useable reading way after your DMM has run out of frequency response.
Obviously,there are meters with wide bandwidth,but they tend to be specialist devices,like the old HP410C of analog days.
-
I know I shouldn’t post this because I going to end up looking silly, but too late now.
You're not end up looking silly, and it's a valid question.
Isn’t the accuracy of a scope reading dependent on the offset you set as well as the resolution/accuracy of the ADC within the scope?
E.g. If you add an offset and zoom into the point of interest don’t you get a more accurate reading than just having the full waveform on the screen. Doesn’t doing this increased the gain to the ADC and shift ADC’s 0V point up (or down). I’m definitely no expert on this but I thought the scopes worked by adding the offset voltage to the ADC input so that they could cover a larger range of voltages with greater accuracy.
This was true for some very old scopes (i.e. if I remember right the Philips PM3320 from 25 years ago did this) and as you said did affect the measurement accuracy. However, on most of today's scopes Vertical Offset is usually a software implementation and doesn't affect the measurement.
Edit: My above comment was wrong. You're right of course (and that makes your question even less silly), Offset accuracy does affect the measurement accuracy.
-
I've played with it on the WaveJet and the offset changes every time you change the vertical range. That is if you set -10V on the 1V/div range and change to 500mV/div then the offset drops to -5V and you have to readjust by knob twiddling to get it back to -10V. Perhaps Agilent and other scopes behave more sensibly in this respect.
No, most don't. The behaviour you see is certainly annoying but it's the standard behavior as it mirrors the Offset on an analogue scope.
-
Vertical position is usually software, but offset (if available) should be implemented in hardware. Otherwise it's pointless, since you're still limited by the dynamic range of the ADC.
The scope may have 0.5 % DCV accuracy, but don't forget that most 10x probes are +/- 1% attenuation accuracy, so the spec would only be valid for 1x probes.
-
Vertical position is usually software, but offset (if available) should be implemented in hardware. Otherwise it's pointless, since you're still limited by the dynamic range of the ADC.
In some of the scopes where I could/did check at this level I'm pretty sure this was implemented in software only. But these were quite old scopes, and as you say it's not a good solution as it limits the available dynamic range.
Edit: I checked a few Service Manuals of some newer scopes, and on all of them Offset is still acting as ADC BIAS. So I was wrong. Sorry for the confusion.
-
Vertical position is usually software, but offset (if available) should be implemented in hardware. Otherwise it's pointless, since you're still limited by the dynamic range of the ADC.
In some of the scopes where I could/did check at this level I'm pretty sure this was implemented in software only. But these were quite old scopes, and as you say it's not a good solution as it limits the available dynamic range.
Edit: I checked a few Service Manuals of some newer scopes, and on all of them Offset is still acting as ADC BIAS. So I was wrong. Sorry for the confusion.
I was about to comment on this but it crossed with the last three replies so I deleted my post.
What I was going to say was that the potential offset on the WaveJet at least is up to 50 times the full scale (+/- 10V at 50.2mV/div) so if it was just software the dynamic range would drop hugely.
Given that it changes with range it must be a D2A providing bias at the inside of the attenuators i.e. at the ADC as you say.
-
I was about to comment on this but it crossed with the last three replies so I deleted my post.
What I was going to say was that the potential offset on the WaveJet at least is up to 50 times the full scale (+/- 10V at 50.2mV/div) so if it was just software the dynamic range would drop hugely.
Given that it changes with range it must be a D2A providing bias at the inside of the attenuators i.e. at the ADC as you say.
The longer I think about the more I think I may have indeed mixed it up with vertical position.
I guess I'm getting old.
-
Looking at the WaveRunner service manual (the only one I could find on-line), the dc offset comes from a 16bit D2A.