Average measuring multimeters more precise than TrueRMS counterparts?

[nightfire]

Just browsed through some Fluke manuals, and stumbled upon the 27II/28II series of multimeters, that, to my understanding are basically 83V/87V in a modified housing and some additional certifications (like mining sites).
The 27 II is average-measuring AC like the Fluke 77 IV, and in the manual https://dam-assets.fluke.com/s3fs-public/2x_2____umger0100.pdf  on page 16 it is stated, that with TrueRMS multimeters it is normal, that when shorting the test leads in ACV mode, some remaining values might be displayed.
Also I noted that the ACV accuracy of the 27II is noted with 0.5% + 3 digits, the 28 II is noted with 0.7% + 4 digits for frequencies around 50 Hz.

Question here: Are those Average-measuring meters more precise due to the fact they do not have to deal with the whole TrueRMS stuff, or the compensation techniques to get rid of induction energy via the test leads?
Might this be a reason why a company like Fluke also provides such meters to the customers (aside from the rumors that the military is a driving force or buyer for the 77 IV)?

[alm]

Just browsed through some Fluke manuals, and stumbled upon the 27II/28II series of multimeters, that, to my understanding are basically 83V/87V in a modified housing and some additional certifications (like mining sites).
The 27 II is average-measuring AC like the Fluke 77 IV, and in the manual https://dam-assets.fluke.com/s3fs-public/2x_2____umger0100.pdf  on page 16 it is stated, that with TrueRMS multimeters it is normal, that when shorting the test leads in ACV mode, some remaining values might be displayed.
Also I noted that the ACV accuracy of the 27II is noted with 0.5% + 3 digits, the 28 II is noted with 0.7% + 4 digits for frequencies around 50 Hz.

The manual will probably also say that this offset does not affect accuracy above 5% of full scale or something like that. And the specifications will say the AC accuracy is only guaranteed from X% to 100% of full scale. True RMS meters don't perform well at the very bottom end of the scale.

Question here: Are those Average-measuring meters more precise due to the fact they do not have to deal with the whole TrueRMS stuff, or the compensation techniques to get rid of induction energy via the test leads?
Might this be a reason why a company like Fluke also provides such meters to the customers (aside from the rumors that the military is a driving force or buyer for the 77 IV)?

Average responding and True-RMS meters can both be made accurate. The reason Fluke is still selling average-responding meters is that these meters give different readings for non-sinusoidal signals, and some procedures will have been written with this in mind. If there's an F-16 manual saying "measure the signal between pins 1 and 16 on connector A with an average-responding meter and ensure it's between 1.1 V and 1.2 V", then the cheapest thing to do is buy an average-responding meter.

The accuracy of an average-responding meter is of somewhat limited value unless you can ensure the signal is sinusoidal with low distortion. Otherwise the crest factor will likely swamp any uncertainty of the meter.

[MadTux]

Perhaps not use an "electrician DMM" when you actually care about ppm RMS???

0.5% is plenty enough for any electrical work, for precicion electronics I'd rather get something like a 12bit DSO, where I can measure RMS signal a a waveform and do the math afterwards....

[bdunham7]

Question here: Are those Average-measuring meters more precise due to the fact they do not have to deal with the whole TrueRMS stuff

One major difference between an analog TRMS converter like the AD737 and a typical averaging converter is the internal dynamic range.  The TRMS converter's first operation is to square the input, then it integrates and then there is a logarithmic amp.  The dynamic range of that middle part will be higher than the range of signals at the input.  The TRMS converter may be quite limited by noise and offset at the low end and by crest factor at the top end.  So if you put a similar amount of effort into an averaging converter as you do a TRMS version, you'll probably get better results, especially at the low end of the scale.  The Fluke models you mentioned actually use the same AD737 for the averaging and TRMS models as the chip has both modes.  I don't know if using it in the averaging mode solves the dynamic range issue and I suspect it doesn't.

If you move away from the analog TRMS systems to something like a sampling system (like Keysight's 'TrueVolt' ) this distinction may all but disappear.

[HKJ]

Mains voltage is not a perfect sinus anymore, i.e. a average meter will not show correct voltage, but a true rms may.

[Fungus]

Mains voltage is not a perfect sinus anymore, i.e. a average meter will not show correct voltage, but a true rms may.

Mains voltage can easily vary by a few volts over a single day so why does it matter?

[HKJ]

Mains voltage is not a perfect sinus anymore, i.e. a average meter will not show correct voltage, but a true rms may.

Mains voltage can easily vary by a few volts over a single day so why does it matter?

Seen from that point of view it do not matter, but if you want to measure it precisely it is a argument for true rms, even if the mean has slightly better specifications.

[Kleinstein]

Average responding and true RMS AC to DC conversion give a different result, like looking at the empty and max. weight of a car. There is not that much sense in comparing the precision as they are different.

AFAIK the hardware inside some of the Fluke meters is actually very similar. Still the same TrueRMS chip, but in the average responding version a filtering capacitor is left out. So in this case I would not expect a much different precision between the 2 versions.  Dave has a video about converting the average to true RMS version by just adding the cap. It still does not make much sense because it requires a new calibration / adjustment.

A average responding converter can be build with less expensive parts (thus found in most cheap meters) and may be also more stable (if one really cares) than the usual analog RMS chips. Still it depends on the implementation and the analog chips like AD636 are not the only option for RMS. An extra accurate average responding meter is rare - maybe if one of the better ones with digital RMS would offer that option  (could be, but I don't know), but those are also quite good with true RMS.

[nightfire]

Thank you all for the input regarding inner workings of the ADC!
For me i was curious about the driving force behind those different notations regarding the same multimeter family, and why quite similar multimeters a) exist other than for some marketing reasons and b) if it has some technical benefit if the user could guarantee that the measured voltage was a sine wave.

I am aware that at least in the 77 IV, a simple cap is the difference- and other multimeters like my Agilent U1272A have similar ACV spec than the Fluke 28II, so in this regard there is also nothing distinctively better.
The only handheld DMM I can remember I have actively looked at the specs that would be significantly better that 0.5% ACV tolerance would have been the Hioki DT4282 with its I think 0.2%

[radiolistener]

Mains voltage can easily vary by a few volts over a single day so why does it matter?

Sometimes it can easily vary for 50 Volts within one minute and even more, just about a month ago I was seen Voltage rise to 250 Volt and then immediate drop down below 190 Volts. The difference between TRMS and average measurement of mains can be up to 100 Volts. So, it have sense to use TRMS DMM for proper Voltage check, because it shows real effective Voltage.

[bdunham7]

The difference between TRMS and average measurement of mains can be up to 100 Volts.

That would be pretty extreme to be off 43%.   An average-responding, sine-RMS-adjusted (multiplied by ~1.11 as all of them do) will be off by 11% for a square wave and 4% for a triangle wave.  Even if you put rectifier in the line for a half-wave sine, that would only be a ~28% error.  I doubt your mains are that bad!

[Fungus]

why quite similar multimeters a) exist other than for some marketing reasons

People buy them, so ... Fluke makes them

Like many things "Fluke" I suspect it's just old procedures/clients who don't want to change their system.

and b) if it has some technical benefit

No.

[radiolistener]

I doubt your mains are that bad!

When rocket or a bomb hits some building in the city it often leads to a high Voltage change in the mains, later - after a couple of seconds you can hear a strong air shock wave from the blast, if it happens too close shock wave can break glass in home windows or even knock out windows entirely, but first you can see it as a high Voltage change in the mains. It often leads to burn out home electronics, such as induction stove on kitchen. Some people was seen over 320 Volts in the mains and its duration was more than minute.

It's a good idea to buy overvoltage/undervoltage protection switch in order to protect your home electronics.

[Electro Detective]

320 volts reading on an averaging meter or RMS?

Analogue or digital meter?

Peak readings perhaps?

Is it 320 volts in a 120v area

or 220v ?

or 230/240v ?


Either way only correctly rated fast blow mains fuses have a chance to save anything at 320 volts RMS for 'more than a minute'

and they need to do that in nanoseconds, not under or within a minute

or it's game over for home electronics.


Quote radiolisterner:
"It's a good idea to buy overvoltage/undervoltage protection switch in order to protect your home electronics."

It better be  FAST sensing and acting switch, with repetitive performance,
not just 'one or two surges and BANG!..'

[HKJ]

320 volts reading on an averaging meter or RMS?

Analogue or digital meter?

Peak readings perhaps?

Is it 320 volts in a 120v area

or 220v ?

or 230/240v ?


Either way only correctly rated fast blow mains fuses have a chance to save anything at 320 volts RMS for 'more than a minute'

and they need to do that in nanoseconds, not under or within a minute

or it's game over for home electronics.

You can easily get 320V or more (in 230V countries) if the neutral wire is broken.

[Electro Detective]

You can easily get 320V or more (in 230V countries) if the neutral wire is broken.

Good point although I can't see that happening easily in a single phase domestic zone,

unless there is a 3 phase supply out in the street feeding it,

and anything goes with a neutral loss and or earthing/ground fault   

[HKJ]

Good point although I can't see that happening easily in a single phase domestic zone,

unless there is a 3 phase supply out in the street feeding it,

and anything goes with a neutral loss and or earthing/ground fault   

In Europe a 3 phase distribution to each building is fairly common (I do not know the system in Ukraine).

And you do not need a war to get these faults, just a electrician that is a bit sloppy.

[Kleinstein]

A neutral loss can also happen with a "single" phase system. The 110 V is in many areas 2x110 V - so that 220 V is also available if needed or after a fault.

[radiolistener]

320 volts reading on an averaging meter or RMS?

Analogue or digital meter?

Peak readings perhaps?

It was shown on protection switch in electrical switch box, which probably using some kind of averaging meter. With normal Voltage it shows about 220 Volts.

Is it 320 volts in a 120v area

or 220v ?

or 230/240v ?

This is for mains with standard 220 Volts 50 Hz.
Usually, when all is ok, Voltage varying from 220 to 240 Volts during the day, sometimes up to 250 Volts.
But when something goes wrong, it can easily go below 190 Volts or above 250 Volts.

As other mentioned above, when neutral line is broken you can easily get 380 Volts in your home mains socket (difference between two phase lines which is connected to a load with shared neutral line).

Good point although I can't see that happening easily in a single phase domestic zone,

Multi-storey buildings have 3 phase power supply, phases are distributed between apartments. And broken neutral line happens pretty often under heavy loads.

[robert.rozee]

here in new zealand, 230v 3-phase is distributed to residential areas, with each house fed with a single phase so that every 3rd house down the street shares the same phase. the neutral wire is then considerably lighter gauge, as along the street the neutral current sums towards zero. this saves considerably on the quantity of copper in the network.

as a result, a short between neutral and one phase may cause a (hopefully brief) drop in voltage to 1/3rd of the houses in the vicinity, and a rise in voltage to the remaining 2/3rd. in a war zone, where multiple dwellings are impacted in quick succession, i could see this creating severe problems.


cheers,
rob.

[David Hess]

For what it is worth, the RMS meters that I have which also support average AC return identical results with calibration waveforms.  In cases where I have a choice, I usually select average reading mode because it settles so much faster even though I then have to apply a correction for the waveform type.

0.5% is plenty enough for any electrical work, for precicion electronics I'd rather get something like a 12bit DSO, where I can measure RMS signal a a waveform and do the math afterwards....

A DSO, even a 12 bit one, is unlikely to deliver better results.  Input offset, gain, and linearity will usually be worse, and the RMS conversion will include broadband noise, although averaging will remove that error term.  Older DSOs removed excess significant digits but I think unscrupulous companies stopped doing that to make their DSOs seem more accurate and precise than they really are.

Some digital voltmeters, even 20 years ago, used high resolution sampling and then DSP for RMS measurements like Keysight's "Truevolt" does.

The Fluke models you mentioned actually use the same AD737 for the averaging and TRMS models as the chip has both modes.  I don't know if using it in the averaging mode solves the dynamic range issue and I suspect it doesn't.

From the looks of the application notes, average computation mode on the AD737 disables the RMS averaging part so the average of the square root of the square is calculated, instead of the average of the square root of the average of the square, resulting in the same dynamic range.  This would also preserve the same offset and gain errors, and the same frequency response.

In my opinion switching between RMS and average reading mode to produce inconsistent measurements would be worse than limiting the performance of average reading mode, although I have the same complaint about multimeters which produce inconsistent results because their input resistance changes with the range selected; that wasted hours to days of my time before I got into the habit of looking for that particular malady.

[mawyatt]

This gets into an interesting area regarding Analog RMS computation vs. Digital.

With Analog you have the limited dynamic range due to the non-linear log type conversions for the square and square-root functions, as well as the scaling and offsets of these conversions.

With Digital you have a sampling system will all the effects and such, but also the benefits of directly doing the computations in the digital domain.

The net result IMO is the Digital solution is superior in most cases, and lends itself to the constant CMOS improvements with feature size reductions. Also, the better DMMs from Keysight and Keithley use Digital computation methods for RMS and those folks certainly know what they are doing, which alone would tilt towards the Digital solution.

Best,

[Kleinstein]

Digital RMS is not just used in higher end DMMs. There are also lower end handheld DMMs that use digital RMS as part of some of the chip sets. These cheap solutions usually have a rather limited bandwidth (e.g. 1 - 5 kHz), but otherwise they work quite well. Especially the essentially instant response (similar to the DC part) is very conventient. By nature the analog RMS part is slow from the averaing filter. The filter also limits the accuracy below some 20 Hz.

I see a chance that the DMM chips get better (faster) and replace the old analog RMS solution at least for most of the hand helds. The power metering chips show that it is possible at a low price and good performance.

[mawyatt]

Certainly makes sense wrt the CMOS trend of more & faster computational power with smaller chips, which yields lower recurring chip cost!!

Works with the high resolution SD ADCs also, perfect example of how the "analog" world has benefited from the CMOS digital revolution.

Best,

[David Hess]

The digital computation might as well be error free, but it is at the mercy of the high resolution sampling, which is no trivial task.  We now have inexpensive integrated sampling ADCs with incredible performance.

[Kleinstein]

The performance of the analog RMS chips like AD736 or AD8436 is not that great: The accuracy and BW is only good for not too small an amplitude and the frequency range is somewhat limited to the low side (e.g. already no longer really RMS below some 20-40 Hz, depending on the filter).
It does not need that much resolution (e.g. 12-14 bit ) to get comparable or better performance, with different weak points for both systems.
The ADC can be sampling, maybe a little better than those found in µCs, but also a SD ADCs. AFAIk the DMM chip-sets with integrated RMS use SD ADCs - so the same technique as the main ADC for DC, just faster.

The digital energy / power meter chips use something like 16+ bit 100 kHz range SD-ADCs, somewhat similar to the audio ADCs, though with better gain stability. They need some of the resolution to get good accuracy over a large current range (like still 1% accuracy with 0.1% of the FS current). Chances are some of the BW limiting is intentional because of the standards for power metering.
A more normal handheld meter should get way with less than 5 digits for AC and thus less ADC resolution.

[mawyatt]

The digital computation might as well be error free, but it is at the mercy of the high resolution sampling, which is no trivial task.  We now have inexpensive integrated sampling ADCs with incredible performance.

Yes the newer IC ADCs are amazing in performance vs. cost, a direct tribute to the digital influenced CMOS scaling

Best,

[radiolistener]

The digital computation might as well be error free, but it is at the mercy of the high resolution sampling, which is no trivial task.  We now have inexpensive integrated sampling ADCs with incredible performance.

it is available for a long time, but they have high power consumption which is not acceptable for handheld DMM which is powered from a battery.

[David Hess]

The digital computation might as well be error free, but it is at the mercy of the high resolution sampling, which is no trivial task.  We now have inexpensive integrated sampling ADCs with incredible performance.

it is available for a long time, but they have high power consumption which is not acceptable for handheld DMM which is powered from a battery.

Analog Devices has suitable ADCs now like the AD7988-1 which draw only a couple hundred microamps, so the power draw of the ADC itself would be insignificant.  Of course that is not the sort of part which would be found in a device intended for the lowest cost market.

[Kleinstein]

The chips for power metering use digital calculation, mainly for the true power, but many do also calculate the RMS values for the current and voltage. This market is somewhat price sensitive and not that small in numbers. However the chips indeed use a bit much power (e.g. 2-10 mA range) for bettery operation.  Just a DMM function would be 1/2 or 1/3 of this with only 1 channel instead of 2 ADCs and power, current and voltage and maybe a few more results.  Some of the new SAR chips are indeed rather low power and would be OK for battery operation.

Not sure how much worse a system using a µC internal 12 ADC would be. The computation part is not that bad and would not need a very fast µC  (e.g. some 4 MHz clock for a ARM based µC could be enough). Chances are a µC with a slightly improved (e.g. 14 bit) ADC may be all it takes to get digital RMS with good performance. Even with only a 10 Bit ADC digital RMS is already at about 3 digit level and useful for many tasks.