Why doesn't my RMS multimeter agree with my Rigol 1054Z scope on the AC VRMS?

[Mike Meschi]

Here’s the circuit and my problem. My function generator is simply parallel connected to a multimeter and my scope. No load is connected and I am not using a scope probe, just a BNC connector with alligator clips. I was just checking to see if the readings on my scope for AC VRMS were the same as the AC VRMS readings on my multimeter. My multimeter is a true RMS meter. The scope is calibrated and has a 1M ohm input impedance. The function generator has a 50 ohm output impedance. So I set the function generator to put out 2 volts AC RMS (as read by my multimeter) at 500 Hz. (as entered on the function generator). Then I read the AC VRMS voltage on the scope and it does not match the multimeter.  Not even close. If the multimeter reads 2 volts AC RMS then the scope reads 500 millivolts AC RMS. The multimeter does read the correct frequency. If I crank up the frequency of the function generator to say 1K Hz., then 10K Hz., the difference in AC VRMS readings is lesser on the scope but the same on my multimeter. Same thing happens if I crank up the amplitude of the function generator. The readings are always different but never by the same amount. So I tried a different multimeter and get the same results. The waveform on the scope always shows the right frequency and is a perfect sine wave but never reads the same AC VRMS or peak to peak voltage as the multimeter. If I take the peak to peak voltage given to me by the scope and multiply it by .707 to get the RMS voltage,  it results in the same reading as the AC VRMS on the scope. Shouldn’t they be the exact same readings? What is going on? What is the best way to check the accuracy of the AC VRMS readings you get on the scope? It would seem the scope is reading incorrectly since both multimeters read the same.

[Fungus]

Some setting is wrong on the 'scope but it's impossible to say without screenshots.

eg. Is the 'scope in AC mode? Is the displayed signal centered around zero?

[AG6QR]

It does sound like the oscilloscope may have an issue.  If your 'scope has more than one channel, have you tried each of them?  Also, as a sanity check, try measuring a DC voltage with your 'scope. Something like an AA battery should measure somewhere around 1.2-1.5V.  Be sure the 'scope is set to DC coupling.

Something to be aware of is that a multimeter typically has a much lower bandwidth than an oscilloscope.  So at high frequencies, the response of the meter will fall off much faster than the response of the oscilloscope.  But this doesn't seem to be the cause of your particular problem, because it would cause the multimeter to read lower, not higher than the oscilloscope.  Still, it's not a bad idea to do an initial test at 60Hz, since any multimeter that measures true RMS ought to be able to handle mains power frequencies perfectly well. And an oscilloscope ought to handle it perfectly, as well.  Once that works, try higher frequencies.

[wasedadoc]

. If I take the peak to peak voltage given to me by the scope and multiply it by .707 to get the RMS voltage, it results in the same reading as the AC VRMS on the scope.

For a sine wave the peak-peak voltage is 2.828 times the RMS.  Or expressed the other way, multiply the peak-peak by 0.3535 to get RMS.

[alm]

The values should be pretty close (within 1% or so) as long as the signal is well within the bandwidth of both instruments. I agree with Fungus that we need much more information. Screenshots, models of the function generator and DMM, and a picture of the setup showing cabling would help.

My first guesses would indeed either be that the scope is set to AC coupling, which forms a high pass filter. Check the manual for the lower bandwidth limited when AC coupling is enabled.

What is the probe attenuation setting on the scope? Is it 1x or some other number?

[mawyatt]

One of the simplest and often overlooked waveforms for RMS evaluation is the squarewave. If one "thinks" of a squarewave as a bi-phase modulated DC voltage of V volts, which is simply multiplying by +-1, then the result is and RMS value identical to the DC value V before bi-phase modulation since multiplying by + or - 1 doesn't change the magnitude of the DC value.

Of course one must consider an ideal squarewave as having zero rise and fall times, and thus possessing an infinite bandwidth, however the harmonics fall off as 1/n^2, so quickly become small. Selecting a low squarewave frequency is advantageous here as the measurement instruments have finite bandwidths. Also one can't actually generate zero rise and fall times, and these finite times affect the result, however if quick relative to the squarewave period they tend to have a small effect.

RMS is the heating value of the waveform by definition and thus must contain all waveform artifacts in the measurement which includes the frequency content and the amplitude content.

One interesting squarewave type waveform is generated by a simple CMOS flip-flop (FF), which can produce a very accurate duty cycle of 50% at low frequencies when configured as a divide by 2. The RMS and average value of the output are identical, and these are exactly 1/2 the VDD supply ideally since the waveform transverses 0 to VDD volts. The small effect of finite rise and fall times can be factored in also as:

VRMS = VDD(sqrt(1/2 -rise/(3*P))), where equal linear rise and fall times are assumed, P is the output waveform period, and zero output impedance is assumed for the FF.

Quite some time ago we built a DMM test and verification device based upon this concept (used discrete P and NMOS devices to lower the FF output Z), and VDD was from a Precision 5.000V reference.

Here's the link:

https://www.eevblog.com/forum/testgear/ac-rms-dmm-tests/msg3940957/#msg3940957

Best, 
 

[mendip_discovery]

If your meter supports it get it to do a signal path check.

Utilities > Calibration > Signal Path test. YMMV (your meter may vary)

Do this with the meter warmed up and it should just take a few moments. I have "fixed" a few scopes this way, I think it helps they often go back to default settings so any random button press might get lost.

[bdunham7]

I get the idea, but it is helpful in cases like this to be very, very specific.  We know what scope you have, but I'd like to know:

Make/model of function generator

Make/model of DMM.

Exact method used to connect the various devices together.

Try this if you can:

If you have a digital AWG, set it for 1kHz and 1V_RMS, or if it won't display RMS directly, then 2.8128V_p-p or whatever it takes.

Verify the output of 1V using your multimeter.

First reset your scope to default settings (it's in the menus somewhere)  then set it to 5µs/div, 500mV/div, probe to 1X and input coupling to DC, trigger to AUTO with the trigger level and channel level both right at zero (default should take care of that).  Connect the function generator to the scope and let us know what you see.  What you should see is an AC sine wave that extends up almost 3 divisions either way from zero.  Set your measurements to display both RMS and StDev (standard deviation) and if available, Cycle RMS and Cycle StDev.  Again, report what you see.

Photos/screenshots are always nice and easy to take with that scope IIRC.

[J-R]

I would set the function generator to the Hi-Z output setting rather than 50 Ohms in order to be able to read meaningful numbers from the generator.  That might give you an idea of which piece of test equipment is providing the false reading.

One potential issue with making a direct connection to the scope is the input attenuation setting.  Typically you'd have a 10X or 100X probe attached, but in this case you would need to tell the scope to use 1X.

[wasedadoc]

If you have a digital AWG, set it for 1kHz and 1V_RMS, or if it won't display RMS directly, then 2.818V_p-p or whatever it takes..

It is less than 1% error but that should be 2.828V_p-p.

[Kim Christensen]

From the manual:

11. Vrms: the effective value on the whole waveform or the gating area

Not sure how you'd change the "gating area"... bdunham7's idea of setting your scope to default settings and trying again is probably the easiest.

[TurboTom]

What's the firmware version of your scope? On my DS1000Z F/W 00.04.05.SP2, the RMS voltage reading is spot-on, verified with a multitude of waveforms, frequencies and voltages. Maybe some early firmwares had a problem with this? DC offset could also be a problem.

P.S. Welcome to the forum, by the way 

[Fungus]

One of the simplest and often overlooked waveforms for RMS evaluation is the squarewave.

Let's walk before running. 

[Fungus]

11. Vrms: the effective value on the whole waveform or the gating area

And that's another reason why we need screenshots - the DS1054Z works with what's displayed on screen, not from the incoming signal.

[G0HZU]

On my old HP/Agilent Infiniium scope there is the option for AC Vrms per cycle and that's the one I favour using for stuff like this. It ignores any DC offset in this case.
I just set an Agilent sig gen to 500 Hz output at 2 V peak and the Infiniium scope measured 1.411 Vrms for a single cycle as in the image below. This scope always seems to read ever so slightly low but it still does well for such an old scope.

The single cycle method is less risky than using all of the screen data as there often won't be an exact number of cycles being measured from the screen data.

My Keithley 2015 DMM measured 1.4137 Vrms and my AD2 measured 1.4146 Vrms as below.

I don't use the scope mode in the AD2 very much so I don't know how it computes Vrms. I suspect it uses the screen data plus a little bit but I can't be certain. There may be a way to make it measure a single cycle but I don't know how to do this. I tried various timebase settings and it gave a consistent result. I even tried a custom timebase that forced the display to show less than complete cycles but it only made a tiny difference. Much less than 1%.

My trusty old Racal 9300 True Vrms meter measured 1.422 Vrms. This is still within 1% of the other readings but this meter is really old and hasn't had a formal cal in many years.

[2N3055]

Quite a discussion about something without basic data: type of sig-gen and details on settings, and complete lack of ANY information on scope and it's settings.

[joeqsmith]

Quite a discussion about something without basic data: type of sig-gen and details on settings, and complete lack of ANY information on scope and it's settings.

I was guessing AI generated on many of the 1st time posters like this one.

[2N3055]

Quite a discussion about something without basic data: type of sig-gen and details on settings, and complete lack of ANY information on scope and it's settings.

I was guessing AI generated on many of the 1st time posters like this one.

You're probably right..

[Datman]

A month ago I wrote to Rigol:

"On the DS1054Z I noticed that RMS value is wrong if I apply a square wave and duty cycle if different from 50%! If I select AC for the input, it seems to be OK, but if I select DC the value is wrong and it doesn't change when I adjust the duty cycle!

In AC, the value changes if I change the frequency of the square wave, displaying 2 or 10 or 20 or more waves. The scope should calculate RMS only on complete waves...

Can you verify?"

They asked Zhou Lihui to verify...

[Datman]

I found something from 2015: 
http://www.sonsivri.to/forum/index.php?topic=60363.0

(...) "Not sure if I will be able to convince them about changing the way calculations are done to match what tektronix and agilent do, but I'll try"

[Aldo22]

I just wanted to check this on my scope and DMM.
With sine waves the measurements are quite consistent, but with square waves the scope seems to be more accurate than my AN870.

Attached is a screenshot of a 5Vpp / 80% square wave.

I think 2Vrms would be the correct measurement here (?) and 1.97Vrms is close enough.
But the An870 is already 100mVrms below that.
After scratching my head a bit, I reduced the frequency from 500Hz to 100Hz and it starts to match better.
This is probably where the rise time of the Aneng comes into play.
Good to know.