Are RMS multimeters a con?

[dom0]

True-RMS has come up when some manufacturers started to label their shit-meters with "RMS" while they didn't measure RMS at all, but measured average or peak value and scaling that with a factor for a pure 50/60 Hz sine wave.  Since those times people say RMS when they mean scaled average and True RMS when they actually just mean RMS.

Another entirely different aspect is bandwidth.
Some RMS voltmeters have a bandwidth specification like "10 Hz - 100 kHz", which obviously removes a DC component, since they only measure the RMS from 10 Hz to 100 kHz - that is, they filter the signal before measuring RMS. This is what most multimeters do.
Others are "DC - x MHz" and measure the RMS value from DC to x MHz.  This is what many dedicated RMS meters do.

And yet another aspect is the permissible crest factor. Some meters don't go very far here, like 3 or 4:1. Better meters can handle 10:1, some even more (mainly depends on the measurement method).

[c4757p]

A con?

No. </thread>

One might discuss whether it's useful.

[dom0]

Linked above:

So, if you are measuring RMS value containing with DC signal, but the meter is only AC coupled, the results can be dramatically wrong.

ugh, for christ's sake, RTFM. I can't stand people claiming wrong results rooted in ignorance of specifications.

[electr_peter]

TRMS vs RMS - pure marketing BS, no real difference. It is either TRMS or not. The real question is if it is AC only TRMS or AC+DC TRMS. Specs define such info.

[Wytnucls]

http://www.newark.com/pdfs/techarticles/fluke/whyTueRMS.pdf

[rs20]

http://www.newark.com/pdfs/techarticles/fluke/whyTueRMS.pdf

Ugh. This was already linked in #3, but more to the point I hate how it's packed full of the word "correct" when they really mean "RMS" (or "True RMS" if you like), as if TRMS is the one true answer to all questions. There's been some good justifications in this thread for why RMS is often what you want; but also a demonstration by the OP of a use-case which is better served by averaging (albeit even better served by a tool dedicated to the OP's purpose).

Writing datasheets that calls RMS "CORRECT" and scaled averaging* other methods "WRONG" with a little tiny footnote saying "* if you're interested in heating effects" is pretty poor form, and promotes ignorance IMHO. But I suppose, what can you expect from marketing material.</broken record sorry>

* Edited in response to coppice below

[coppice]

http://www.newark.com/pdfs/techarticles/fluke/whyTueRMS.pdf

Ugh. This was already linked in #3, but more to the point I hate how it's packed full of the word "correct" when they really mean "RMS" (or "True RMS" if you like), as if TRMS is the one true answer to all questions. There's been some good justifications in this thread for why RMS is often what you want; but also a demonstration by the OP of a use-case which is better served by averaging (albeit even better served by a tool dedicated to the OP's purpose).

Writing datasheets that calls RMS "CORRECT" and scaled averaging "WRONG" with a little tiny footnote saying "* if you're interested in heating effects" is pretty poor form, and promotes ignorance IMHO. But I suppose, what can you expect from marketing material.</broken record sorry>

RMS isn't the only useful measure of the voltage of an AC signal. For example, peak voltage is often required and a number of instruments can measure that for you. Even the average of the rectified voltage has real uses. However, scaling the rectified average to be like the RMS would be for a pure sine wave is simply a botch without any physical meaning. It is only produced because its easy to produce. Nobody would sit down and try to develop an instrument specifically to output this value.

[rs20]

RMS isn't the only useful measure of the voltage of an AC signal. For example, peak voltage is often required and a number of instruments can measure that for you. Even the average of the rectified voltage has real uses. However, scaling the rectified average to be like the RMS would be for a pure sine wave is simply a botch without any physical meaning. It is only produced because its easy to produce. Nobody would sit down and try to develop an instrument specifically to output this value.

Fully agree; despite the way my message made it look like I'm trying to defend scaled averaging in particular, I was really just trying to push the point that RMS is just one of many useful measures (a point which your response justifies much better than my message), and calling RMS the one "correct" measurement is just deceptive and simplistic.

Pedantically, the equations used by the OP would cancel out the scaling fudge factor -- but what is perhaps more interesting is this question: Scaling aside, how good is a cheap [scaled] averaging multimeter at actually getting the averaging right? I'm guessing they don't specify that in the datasheet. How linear is it, what's the bandwidth? How do they implement and integrate abs(x) in cheap circuitry? It's surprising that they scale an average-of-absolute value instead of similarly fudging a peak value, which I would guess to be easier to measure. If only I had a sig gen I'd be doing a few experiments on my multimeters right now...

[The Electrician]

However, scaling the rectified average to be like the RMS would be for a pure sine wave is simply a botch without any physical meaning.

Its meaning is that it is the RMS value of the applied pure sine wave.

It is only produced because its easy to produce. Nobody would sit down and try to develop an instrument specifically to output this value.

The people who designed this: http://simpson260.com/260-3/simpson_260-3.htm

developed just such an instrument.

[The Electrician]

Just for the sake of argument, what's the least contrived reason someone can come up with for a true RMS voltage measurement?

Lacking a specialized power factor meter, if one wants to measure the power factor of a load when the grid waveform is distorted (and all grid waveforms these days are distorted; flattened on top by all the non-PF corrected power supplies in computers, etc.) you need a wattmeter to measure the true power and an RMS measurement of the applied voltage and load current to determine volt-amperes.

[coppice]

It is only produced because its easy to produce. Nobody would sit down and try to develop an instrument specifically to output this value.

The people who designed this: http://simpson260.com/260-3/simpson_260-3.htm

developed just such an instrument.

So, you don't think they (and pretty much every other passive analogue multimeter maker) ended up with a meter like that because it was too hard to do a true RMS meter in those days?

[The Electrician]

It is only produced because its easy to produce. Nobody would sit down and try to develop an instrument specifically to output this value.

The people who designed this: http://simpson260.com/260-3/simpson_260-3.htm

developed just such an instrument.

So, you don't think they (and pretty much every other passive analogue multimeter maker) ended up with a meter like that because it was too hard to do a true RMS meter in those days?

I offered no opinion about whether it was difficult or not.  I responded to your claim that there was no physical meaning to the scaled reading of the average rectified value, and your claim that nobody would design an instrument to output this value.

[N8AUM]

Interesting thread. I barely remember back in the early 80's I had to start up the worlds largest 3 phase UPS system for IBM. After working out all the bugs I got the system going and IBM signed off on it, everybody is happy. Well, then started the random mainframe crashes. IBM is pissed! I would be to after spending millions! The mainframes power supplies had a 50 volt reverence derived from monitoring it's input voltage and frequency "POWER GOOD", if someone farted it would kill the mainframe. Turned out that load variations would change the harmonic content ever so slightly causing the whole problem. IBM finally woke up and increased their tight tolerance and lived happily ever after till the PC market wiped them out.

 bring back raised floors and vacuum tubes (valves) !       

[rs20]

I offered no opinion about whether it was difficult or not.  I responded to your claim that there was no physical meaning to the scaled reading of the average rectified value, and your claim that nobody would design an instrument to output this value.

Coppice's point is that the only reason it's designed that way is because True RMS is difficult. No-one would willingly design a scaled averaging meter unless budgetary or design constraints force them to, which makes it a strange thing to actually want/go out of your way to get.

[DAIRVINE]

Instead of RMS why not true power?
Instead of SQRT(average  (V^2)) or SQRT(average(I^2))
Please could meter manufacturers do
Average ((V*I))
Should work for normal DMMs with a common terminal and clamp DMMs.
Auto ranging both current and voltage does make it more complicated, but possibly just do it for the highest current range on cheaper meters?
True power is much more useful than true RMS. No need to do VA, as that is easy to do by hand.

Equation corrected.

[dom0]

All DMMs have only one ADC.

(Well, except for poly-phase meters, but those are actually handheld power analyzers and don't really qualify as a DMM)

[DAIRVINE]

 And measuring square-wave output inverters with a view to powering resistive toasters or incandescent light bulbs.

For this use case AVERAGE is more accurate than RMS. With a square wave the volts is in essence constant, so power is proportional to current. The load does not matter.

The same argument possibly applies to current into a bridge rectifier/capacitor when the voltage does not change much during current spikes, at least for large capacitors. So power would be approx AVG current *(correction factor*) peak voltage.

[DAIRVINE]

All DMMs have only one ADC.

(Well, except for poly-phase meters, but those are actually handheld power analyzers and don't really qualify as a DMM)

You only need one ADC. Do the average (V*I) before the ADC.
Though two ADCs may be better.

[dom0]

So you'll need an analog multiplier with reasonable linearity and accuracy for, say, 2000 counts or two analog frontends and ADCs. Either way: too little use, too expensive.

I am probably repeating myself here, but if you want to do accurate power analysis, go get a power analyzer. The right tool for the right job.

[The Electrician]

Instead of RMS why not true power?
Instead of average (SQRT (V^2)) or average (SQRT(I^2))
Please could meter manufacturers do
Average ((V*I))
Should work for normal DMMs with a common terminal and clamp DMMs.
Auto ranging both current and voltage does make it more complicated, but possibly just do it for the highest current range on cheaper meters?
True power is much more useful than true RMS. No need to do VA, as that is easy to do by hand.

This meter does it.  Dave even reviewed it:

http://www.gossenmetrawatt.com/english/produkte/metrahitenergy.htm

[rs20]

 And measuring square-wave output inverters with a view to powering resistive toasters or incandescent light bulbs.

For this use case AVERAGE is more accurate than RMS. With a square wave the volts is in essence constant, so power is proportional to current. The load does not matter.

Wrong. Since the absolute voltage of a balanced square wave is essentially constant, the squared voltage is also constant. This means that RMS, average AND peak voltage will all give the same answer and be equally accurate (modulo noise sensitivity). Except, most "averaging" meters are actually scaled average, and will give a meaningless number.

...
Instead of average (SQRT (V^2)) or average (SQRT(I^2))
Please could meter manufacturers do
Average ((V*I))
...

Just for the pedantic record, you probably mean sqrt(average(V^2)), etc. (I like to use the mnemonic RMS - Root Mean Square  )

[The Electrician]

I offered no opinion about whether it was difficult or not.  I responded to your claim that there was no physical meaning to the scaled reading of the average rectified value, and your claim that nobody would design an instrument to output this value.

Coppice's point is that the only reason it's designed that way is because True RMS is difficult. No-one would willingly design a scaled averaging meter unless budgetary or design constraints force them to, which makes it a strange thing to actually want/go out of your way to get.

If that was his point, he wasn't very clear about it.

It looks to me like his point is "Nobody would sit down and try to develop an instrument specifically to output this value." BECAUSE "...scaling the rectified average to be like the RMS would be for a pure sine wave is simply a botch without any physical meaning."

Furthermore, TRMS is not difficult; you just have to use the right meter movement.  Moving iron, thermocouple and electrodynamometer movements inherently respond to TRMS AC+DC.

VOMs used a D'Arsonval movement because they wanted greater sensitivity to DC voltage and current, and ohms measurement.  Having made that choice of movement, it was necessary to use a rectifier type AC measurement.  Back then the waveforms they were measuring were usually sine waves since there were no highly distorted waveforms such as you get from a triac light dimmer, for example, so the rectifier type AC meter did the job.

The usual measurements of grid related voltages don't need a meter with a high input impedance, so there is no need to use a high sensitivity D'Arsonval movement.  Any movement is suitable for measurements of grid powered loads.  Notice that even the VOMs with D'Arsonval movements have fairly low input resistance for AC voltage measurement.  For example, the input resistance for AC measurements on the Simpson 260 was 1000 ohms/volt, versus 20,000 ohms/volt for DC voltage measurements.  This is not a problem when measuring grid related voltages where there is plenty of power to operate the meter.

The reason VOM's AC measurements were made with a rectifier type method is not because TRMS measurements were difficult.

When a meter was needed that was not a combination meter like a VOM, but was only a voltmeter, it was easier and cheaper to use a TRMS responding movement such as this, rather than the more complicated rectifier type:



Here's an example of a multipurpose analog meter that measures volts and amps, TRMS, with .5% accuracy:

[The Electrician]

Except, most "averaging" meters are actually scaled average, and will give a meaningless number.

It's not meaningless; it's the average multiplied by 1.1107.  If the applied voltage is a pure sine wave, then the value given is the true RMS value of the sine wave; this is not meaningless.

For other waveforms, it's also not meaningless.  Just divide the reading by 1.1107 and you have the true average of the absolute value.

[dacman]

The Fluke 8020A (circa 1979) is average responding and the Fluke 8060A (circa 1982) is RMS responding, but the 8060A has a noise floor that the 8020A does not have (5% of range originally then revised to 10% of range).  10% of range on the 200 mVac range is 20 mVac.

One difference between the Fluke 27 II and Fluke 28 II is that the 28 II is RMS responding and is stated to be specified from 3% to 100% of range, where the 27 II is (deduced to be) specified from 0% to 100% of range.  3% of range on the 600 mVac range is 18 mVac.

I have a test that I perform which calls out a specific handheld DMM that is average responding.  The test limit is below the noise floor of both of these RMS handheld DMMs.

[DAIRVINE]

 And measuring square-wave output inverters with a view to powering resistive toasters or incandescent light bulbs.

For this use case AVERAGE is more accurate than RMS. With a square wave the volts is in essence constant, so power is proportional to current. The load does not matter.

Wrong. Since the absolute voltage of a balanced square wave is essentially constant, the squared voltage is also constant. This means that RMS, average AND peak voltage will all give the same answer and be equally accurate (modulo noise sensitivity). Except, most "averaging" meters are actually scaled average, and will give a meaningless number.

RS20: I was perhaps unclear. For a pure resistive load then RMS is best, regardless of waveform. For a voltage square wave, then AVERAGE current is best, regardless of load, because power= volts*average current, and volts is constant. This also applies to DC.

The use case for RMS is a pure resistive load. For most other use cases peak or average may be more useful.
The other use case for RMS is that most professionals have RMS meters and we can all measure the same number, provided crest factor is not too high. For audio and RF I think you need specialist tools.