Question about DMM sampling rate

[pfm]

Typically I have seen DMMs refresh rate is about 3/sec. Is that the display rate or the actual signal sampling rate ? I suppose there must be some kind of an ADC in DMMs ? and so it must have a certain sampling rate ? Is that sampling rate much higher than the display rate or both are same ?
Thanks.

[zaoka]

It differs from model to model. For example Gossen Metrawatt sames and 10/s and refreshes display 2/s.

I found that regular DMM that does refresh 5 times/s works the best for me.

[David Hess]

Digital multimeters usually integrate over some fraction of the greatest common factor of 50 Hz and 60 Hz like 10, 5, or 2 Hz so that they can reject 50 and 60 Hz common mode noise.  Faster ones will integrate over one power line cycle but must be set which gives them a maximum sample rate of 25 or 30 Hz.

[pfm]

David,
so display rate would always be less than sampling rate, am I understanding this right ?
And if that is correct then what do the functions MIN/MAX on the dmm display - the max/min of sampled values or the max/min of display values ?
Same question for logging - If the dmm has logging capability the values you see on the PC graph are those raw sampled values or the values that the dmm chose to display ?
If brand or mode of dmm matters then I am only talking about general consumer mid-level like Tenma, Amprobe, Extech, Mastech....

[Maxlor]

Display rate is less or equal to the sampling rate. Hopefully  .

how fast min/max are depends on the multimeter and the mode. Cheaper ones will just choose the min/max values from the displayed values, others will pick choose from raw ADC values that are sampled at a much higher rate. Some meters (like the BM257) even let you choose which mode you want (they call it "Crest" and "Peak"). Generally, you'll have the study the multimeter manual to figure out how exactly the manufacturer has implemented the feature.

All the multimeters whose logging or serial interface features I've used have been sampling at either the display update rate, or something slower like 1Hz.

[David Hess]

David,
so display rate would always be less than sampling rate, am I understanding this right ?

It will be because even the slow sample rate provided by most integrating ADCs used in multimeters would be too fast to follow anyway.  I have a number of frequency counters which do the same thing; they have an update rate control which is used to slow the display rate down below the sample rate.

And if that is correct then what do the functions MIN/MAX on the dmm display - the max/min of sampled values or the max/min of display values ?

I have never looked into this in detail but Maxlor covered it.  A couple of my meters support min/max and peak hold but I have never tested them to see how they work other than "fast".

Same question for logging - If the dmm has logging capability the values you see on the PC graph are those raw sampled values or the values that the dmm chose to display ?
If brand or mode of dmm matters then I am only talking about general consumer mid-level like Tenma, Amprobe, Extech, Mastech....

The DMM specifications or user manual should say.

[pfm]

This whole thing came up when I was talking with a friend about measuring the highest voltage and current(and thereby power) for music signals going to a loudspeaker. Music is not a simple sine wave of one frequency but a complex wave of a mix of low(20hz) to high(10khz+) frequencies and lots of their harmonics. See the attached pictures. The first picture(music_wave1.png) is about 1 second of music. The leftmost waves are about 100hz and the right most ones are about 4Khz, both with lots of harmonics. The other two are zoomed in areas of the high frequencies in the first picture so they can be seen clearly.

Assuming its a TrueRms meter, my concerns/questinos were - would the meter miss out on some peaks because it would be sampling at a much lower rate than a typical audio/music frequency ? It could possibly randomly catch the peaks but that would be all by chance. Is my understanding right ?

Now to catch these fast peaks what would work better - Peak hold(crest) or Max option ?

[DanielS]

Now to catch these fast peaks what would work better - Peak hold(crest) or Max option ?

Neither since the RMS value of a peak depends on how long that peak is actually held during the sampling interval: if you have a 10V supply with a 1ns 1kV pulse, the RMS voltage if you do mathematical integration will still be very close to 10V.

RMS is a form of averaging: the equivalent DC voltage or current that would yield the same energy transfer over a given observation interval. Not a peak/max thing.

If you want to calculate the RMS value of a signal faster than your ADC, you can do non-linear analog integration to approximate integrating the square of your input signal, sample the output, apply corrections and calculate RMS from that. You can get better than 10% accuracy with a relatively simple circuit this way, which is far more accurate than using peak and crest factor on non-sinusoidal input.

[dom0]

When measuring complex AC signals with a DMM there are a few steps to obey in order to get a good measurement

- True-RMS is a must. Fake RMS ("rectify and multiply with x") won't work with much else than a 50/60 Hz sine wave.
- A proper DMM specifies a maximum crest factor. Your signal may not have a crest factor higher than that or the measurement will be off
- The AC bandwidth of the DMM must be sufficient to "capture" all important parts of the signals' spectrum. The bandwidth might be as low as a few kHz or as high as 100-1000 kHz.
- These criteria are independent of the method used by the DMM. Some models may use a fast ADC, some may use analog circuitry to get the job done.

Now you are asking for something slightly different. You want peak values. Peak values don't really make sense with RMS as it is a special averaging method (more or less).

You are likely way better of to use a two channel oscilloscope for this job. One ch for current, one for voltage. Mind the potentials.

I might remark that you usually don't do any serious measurements on amplifiers using highly complex signals like music. Depending on what you want to measure continuous or specifically crafted, often modulated or interfered signals are much more useful, even for really simple tests. (Like testing whether the cooling is sufficient or needs adjustment.)