We Can Make Multimeter Reviews Better

[Richard_S]

I want to start by paying tribute to those that care about safety. Making sure that the meter has fuses and protection and will keep me safe when I probe around in that electrical substation is genuinely a good thing. Only I don't have an electrical substation nor three-phase at my house. I can't even remember the last time I measured mains voltage. I know that these things matter even when you are on the far side of the transformer, but even that is rare for me.

What I measure predominantly are small ac voltage signals and it is here that multimeters are pretty useless. What may surprise people more, is that even the expensive high-performance models are poor at small-signal ac measurements. Not all of them, but most.

Yet, I cannot remember reading or watching a review that took a look at small-signal ac measurements. I'm sure there are some, but they are far from the norm. We get loads of detail on the high-voltage, high-current side of things and that is great alongside lots of attention to dc accuracy, but ac precision and how useful a meter is for small signal work is regularly omitted altogether. What we all need to do is to balance our coverage of multimeters by mentioning the higher bandwidth and lower voltage signals alongside the bigger signals and dc performances we cover now.

Accuracy
Its not uncommon to find that a very expensive high-precision bench multimeter reading dc to better than 0.03% has an accuracy of worse than 1% for ac voltages. A much prized 6.5 digit meter may well perform significantly worse than another much cheaper device with only 4.5 digits. It is very hard to find meters even amongst the elite that prioritise ac accuracy. The poor ac accuracy performance is missed in so many reviews so we should make a point of discussing ac accuracy and especially for small signals..

Range
When it comes to small signals, the situation is made even worse because of the lack of low-level ranges. Meters that have a dc millivolt range or even stray into microvolts sometimes start 5, 10 or even 30 times this for ac signals. This disparity is rarely mentioned and we really should. The combination of poor accuracy and the absence of low measurement ranges interact to make many meters singularly useless for small signal measurement, yet we almost never say.

Bandwidth
It is rare to find a meter that will handle the full audio bandwidth which is what I am most interested in. Meters often retain their accuracy only in the tens of Hertz or hundreds of Hertz. There are meters that are good for this, and this is one area where paying more often helps, but you have to look for them and you can't take it for granted. Again, this shortcoming is rarely mentioned in our reviews. While audio is what matters to me, people working on servo systems and dc motor drives as well as all sorts of sensor and drive circuits need wider bandwidths, in some cases way beyond the audio range I'm looking for. I am amazed that ac accuracy, ac ranges and ac bandwidth are not standard talking points in every multimeter review.

RMS
I am often shocked at people talking about the RMS capability of a multimeter without mentioning bandwidth. People will recommend looking for RMS capability for use with audio for example. Why do they do this? In most cases the meter does not cover the audio range and so whether it measures true RMS or not is irrelevant. Secondly, true RMS is unnecessary when measuring things like gain and line-up levels using tone. Keep in mind that meters are always less accurate when measuring true RMS, so where the meter has a choice, you are generally better off turning this function off. Where RMS is at its most useful is when measuring noise, but yet few multimeters have the range to usefully measure noise in audio equipment (or servo systems for that matter). If they could, fewer still have A-weighting which you may well need at some point, for sound at least. We make far too much fuss about RMS and rarely explain how limited its use really is. How many multimeter reviews have you read where the author never draws attention to the limited bandwidth?

Decibels
The ability to present the reading in dB is often scored well in reviews when in practice, I find that it is rarely useful. If you want to line a piece of equipment up, then you can convert from dB first and simply measure in volts. Where direct reading of decibels does become useful is when tracking changing values and most multimeters are too slow for their dB reading displays to be relevant.

Please throw rocks at my thinking here. Tell me I'm wrong. Point out that all this stuff is regularly covered. Happy to be shown to be incorrect on this if I am.

[bdunham7]

Please throw rocks at my thinking here.

RTFM.   

And don't consider products that don't have reasonably complete specifications in that manual.  Reviews are no substitute for proper specifications.

A few things about specific points you made:

AC accuracy is an order of magnitude or so more difficult to be precise with than DC, even more so at high or low frequencies or complex signals.  But I can't think of any decent meters that have as wide of a disparity between them as you claim.  Any examples? 

As far as BW, meters intended for mains power applications may have reduced BW, but most decent meters with separate TRMS converter chips will cover at least to 20kHz, at least in voltage.  Where they almost all fall down is BW for AC current.  The otherwise modestly performing BK Precision 2831E was specified to do current with reasonable accuracy out to 20kHz, which exceeds most of its competitors.

Being able to switch between average-responding RMS reading and true RMS isn't a feature I've ever seen and A-weighting really wouldn't be appropriate for an RMS-reading DMM.  It is true that many use cases don't require TRMS, but I'm not sure why you think they are necessarily more accurate without it. 

Decibel readings may not be a big thing to you or most people, but the function is very easy and cheap to implement so there's no savings to be had by eliminating it.

Your issue with small-signal AC readings is probably residual counts and the lower specified threshold (in the manual, of course) that prevents you from getting any meaningful readings below 1-5% of the lowest range.  A review will only tell you how that particular unit does on that particular day, the actual performance will vary between units and is only guaranteed to meet the spec.  As an example, take the Fluke 289 on the 5VAC range reading a 1kHz test signal.  The meter is specified for accuracy from 2% to 100% of the specified range, meaning over 100mV.  One example displayed about 230 residual counts (with zero input) equivalent to 23mV, another showed 0 residual counts and doesn't start showing any reading at all until the signal reaches about 16mV.  Both meters are well within spec above 100mV.  But two reviewers with two different meters might come to different conclusions. 

b/t/w, to reinforce your point about TRMS, my average-responding Fluke 27 (3.2V range) will read that same 1kHz signal all the way down to 1mV within 1 count accuracy. 

[TheDefpom]

You obviously haven’t seen my reviews where I do things like check detection threshold’s and accuracy for the entire voltage range, including bandwidth vs accuracy.

I do this on a Datron 4700 calibrator.

[alm]

The problem with testing accuracy in reviews is that accuracy and bandwidth is generally based on probability distributions, and you won't learn much about a probability distribution from a single sample. Yet manufacturers are not likely to ship say a hundred meters for a review, never mind the amount of work in collecting data on 100 meters. If you are just talking about specifications, then surely you don't need a reviewer to read the specifications?

I can understand low voltage AC response, though I agree with bdunham7 that you really shouldn't be relying on the meter performing better than the guaranteed specs. And particularly should not expect because a single sample exceeded the specifications by a certain amount, that a meter you buy a year later will do the same.

Keep in mind that meters are always less accurate when measuring true RMS, so where the meter has a choice, you are generally better off turning this function off.

Do you have a reference for that? Any 5.5+ digit bench meter I'm familiar with is true RMS and has better accuracy than lower end meters. One meter I remember that can switch between true-RMS and average responding is the Keithley 2001-M, but the average responding mode has a lower accuracy than true RMS mode. I think the average-responding mode was just added for a military customer that had procedures based on avergage-responding meters, not for better accuracy. If anything, an average-responding meter makes more assumptions about signal shape and distortion. Any lab-grade ACV standard I'm familiar with, like thermal voltage converters, are true RMS.

[rsjsouza]

Maybe we are few and far between. Although I don't perform accuracy AC measurements in small signal scenarios as defpom does, in my reviews I usually measure bandwidth and try to educate the audience on its applicability to audio and how TRMS influences measurements of non-sinusoidal waveforms.

Regarding accuracy, alm and bdunham7 mentioned very important points, especially due to the variability of each unit and the wide variance of accuracy across the range.

[bdunham7]

Do you have a reference for that? Any 5.5+ digit bench meter I'm familiar with is true RMS and has better accuracy than lower end meters. One meter I remember that can switch between true-RMS and average responding is the Keithley 2001-M, but the average responding mode has a lower accuracy than true RMS mode. I think the average-responding mode was just added for a military customer that had procedures based on avergage-responding meters, not for better accuracy. If anything, an average-responding meter makes more assumptions about signal shape and distortion. Any lab-grade ACV standard I'm familiar with, like thermal voltage converters, are true RMS.

I think the superiority of the average responding meters is only evident at the very low end of the scale where the residual counts start to matter, at least if we're talking about comparing meters in roughly the same class.  But in most cases involving 5.5+ digits, average responding meters are older models that don't have the lower ranges on ACV.  The Fluke 8800A, for example, responds fairly accurately all the way down to the bottom of its 2VAC range, but competing 5.5+ digit TRMS meters like the 8842A have a 200mVAC range. 

On meters that are switchable or based on TRMS models, if they simply use the averaging feature of the TRMS chip along with different calibration constants, you may get an average-responding meter with all the disadvantages and none of the benefits of a dedicated average-responding converter. 

[J-R]

What minimum voltage are you actually targeting?

The features mentioned seem somewhat niche, so I can understand why reviewers might not be focused on them.  The DMM is a multifunction device so there are going to be strengths and weaknesses due to cost/design/market.  Plenty of forum members have some specific DMM due to an aspect it excels at.

For audio repair older analog AC millivolt meters still seem popular.  The "best" one I have currently is the Lodestar MV-3100A which has a 300uV range and 1MHz bandwidth.  My various bench DMMs were OK down to around 500uV at 100kHz, and in some cases to 200uV and 300kHz and beyond.  For regular measurements of these and smaller inputs, perhaps I'd get a nanovolt amplifier.

[2N3055]

I want to start by paying tribute to those that care about safety. Making sure that the meter has fuses and protection and will keep me safe when I probe around in that electrical substation is genuinely a good thing. Only I don't have an electrical substation nor three-phase at my house. I can't even remember the last time I measured mains voltage. I know that these things matter even when you are on the far side of the transformer, but even that is rare for me.

What I measure predominantly are small ac voltage signals and it is here that multimeters are pretty useless. What may surprise people more, is that even the expensive high-performance models are poor at small-signal ac measurements. Not all of them, but most.

Yet, I cannot remember reading or watching a review that took a look at small-signal ac measurements. I'm sure there are some, but they are far from the norm. We get loads of detail on the high-voltage, high-current side of things and that is great alongside lots of attention to dc accuracy, but ac precision and how useful a meter is for small signal work is regularly omitted altogether. What we all need to do is to balance our coverage of multimeters by mentioning the higher bandwidth and lower voltage signals alongside the bigger signals and dc performances we cover now.

Accuracy
Its not uncommon to find that a very expensive high-precision bench multimeter reading dc to better than 0.03% has an accuracy of worse than 1% for ac voltages. A much prized 6.5 digit meter may well perform significantly worse than another much cheaper device with only 4.5 digits. It is very hard to find meters even amongst the elite that prioritise ac accuracy. The poor ac accuracy performance is missed in so many reviews so we should make a point of discussing ac accuracy and especially for small signals..

Range
When it comes to small signals, the situation is made even worse because of the lack of low-level ranges. Meters that have a dc millivolt range or even stray into microvolts sometimes start 5, 10 or even 30 times this for ac signals. This disparity is rarely mentioned and we really should. The combination of poor accuracy and the absence of low measurement ranges interact to make many meters singularly useless for small signal measurement, yet we almost never say.

Bandwidth
It is rare to find a meter that will handle the full audio bandwidth which is what I am most interested in. Meters often retain their accuracy only in the tens of Hertz or hundreds of Hertz. There are meters that are good for this, and this is one area where paying more often helps, but you have to look for them and you can't take it for granted. Again, this shortcoming is rarely mentioned in our reviews. While audio is what matters to me, people working on servo systems and dc motor drives as well as all sorts of sensor and drive circuits need wider bandwidths, in some cases way beyond the audio range I'm looking for. I am amazed that ac accuracy, ac ranges and ac bandwidth are not standard talking points in every multimeter review.

RMS
I am often shocked at people talking about the RMS capability of a multimeter without mentioning bandwidth. People will recommend looking for RMS capability for use with audio for example. Why do they do this? In most cases the meter does not cover the audio range and so whether it measures true RMS or not is irrelevant. Secondly, true RMS is unnecessary when measuring things like gain and line-up levels using tone. Keep in mind that meters are always less accurate when measuring true RMS, so where the meter has a choice, you are generally better off turning this function off. Where RMS is at its most useful is when measuring noise, but yet few multimeters have the range to usefully measure noise in audio equipment (or servo systems for that matter). If they could, fewer still have A-weighting which you may well need at some point, for sound at least. We make far too much fuss about RMS and rarely explain how limited its use really is. How many multimeter reviews have you read where the author never draws attention to the limited bandwidth?

Decibels
The ability to present the reading in dB is often scored well in reviews when in practice, I find that it is rarely useful. If you want to line a piece of equipment up, then you can convert from dB first and simply measure in volts. Where direct reading of decibels does become useful is when tracking changing values and most multimeters are too slow for their dB reading displays to be relevant.

Please throw rocks at my thinking here. Tell me I'm wrong. Point out that all this stuff is regularly covered. Happy to be shown to be incorrect on this if I am.

From your writing, I would gather you need to measure voltage in audio range. Probably for audio amplifiers stuff.
What you need is probably specialty low voltage AC voltmeter.
There are many articles and reviews that mention small AC measurements.
I would be curious to know what application needs better than 0.06% accuracy in audio range at levels of 10 µV RMS.

Accuracy: It depends on your definition of "high-precision bench multimeter". Even my "not very fancy" Rigol DM3068 (6 1/2 digit meter) has 0,06% accuracy on AC range (10Hz-20kHz) on 200 mV range. Simillary from Siglent and oodles of others. Also 5 1/2 digit meters from Siglent and Rigol will do 0.2% in audio range... TTi 1908 will too.. and bunch of others..
Range: There are meters (like I said, those are specific instruments) that have sensitivity you need. Standard multimeters with their unshielded banana connectors and probes are not fit for this purpose. You need coaxial, shielded inputs and probes for this. Maybe differential, even...
BW: I have 3 multimeters that have way more than audio BW: DM3068 (300kHz), MTX3293 (200kHz) and BM869S (100kHz). There are very many meters that have very wide BW.
dB: dB measurements are simple recalculation. On well implemented meter you can specify load impedance and get direct dBm reading. Speed will be defined by meter AC measurement speed. On DM3068 (for instance) i can specify AC RMS filter and in fastest setting, I get maybe 10 measurements per second. Enable dBm and voila.... Realtime response.

But you are correct. Most multimeters are not audio meters. Audio meters are specialized instruments, which multimeters are not.

To add, I also have a 16 Bit picoscope that is best AC voltmeter you can buy. 5MHz BW, scope, meter, and several dozens of measurements, spectrum etc...
People also use audio analysers for accurate level measurements.

If you want a simple and inexpensive solution, a preamplifier and decent meter does a good job too.

Or shall we take this question and set it differently?
How about you lay out what problem you need fixed ( "I need to measure this and this within these and these parameters and constraints. What you guys can recommend?") and I'm sure some solution will come out of it...

[Kleinstein]

For the general accuracy, this something that can not be tested with a few meters in a short time. Here one can only check of the samples actually meet the claimed accuracy specs, which they should all do with some margin for drift.
A problem with just looking at the specs is that even for the manufacturer with 100s of meters to test it is hard to estimate the accuracy. Different companies may use different guard factors and even than they are sometimes off in either direction for some meters. Often only time tells which meters are very stable and which tend to drift or fail early.

How far one may go down and still get a sensible reading can vary between units - can't use 1 or 2 meters to really judge on this. It can make a difference how the RMS readings are done. There are now also some handheld meter that do RMS the digital way. This often comes with a limited BW, fast response (rarely specified - so need to actually test), reasonable good performance at low frequency and small fraction of the range (at least if done right). So one can usually tell from a test and sometimes the specs.

The analog RMS chips are tricky with small amplitudes: the smaller the voltage the smaller the bandwidth of the RMS chip. Another somewhat tricky point is the performance at lower frequency - that is less than some 40 Hz. Accuracy can suffer already quite a bit at 20 Hz. For the meters based on the analog RMS chips there is compromise between the low frequency performance and response time - it could be interesting to see on which side they are: relative fast response and poor LF accuracy or good accuracy also at 20-40 Hz and than a more sluggish response.

[armandine2]

we can make Multimeter reviews better

I've mentioned before that I don't believe there are Oscars for this line of work - but for content makers their viewing figures seem to suffice. 

[bdunham7]

For the meters based on the analog RMS chips there is compromise between the low frequency performance and response time - it could be interesting to see on which side they are: relative fast response and poor LF accuracy or good accuracy also at 20-40 Hz and than a more sluggish response.

That compromise is sort of baked into the problem and it is there even if you don't use an analog TRMS chip.  Better and newer meters (think HP 34401A and up/newer) have selectable AC filters and work pretty well down to 3Hz or so. 

[Richard_S]

Well, thanks everyone, what an interesting set of replies! I still think making reviews even better, even more effective, address a wider range of users, is a laudable aim though I am beginning to think that may be an unpopular view. Firstly, to those who say here that "I have got a great meter for..." Well done to you. I'm not sure that really speaks to the issue though. As for "reading the f-ing manual", is that really a response to reviews not being as well constructed as they could be?

Look at a bunch of multimeter videos and see how often they say "it's got a basic dc accuracy of..." but do not point out that its ac voltage accuracy is also very good, or perhaps more importantly when it is not. The dc accuracy is the banner headline figure but the ac accuracy is always worthy of discussion, always should get a mention. Always in the context of how the low ac ranges stack up against the dc too because having acceptable accuracy but no low ranges is just as much a problem.

Several times people mention that it is harder to achieve precision on ac. Is that a reason not to direct the viewer or reader to understand the performance of an individual meter? I would have thought if anything that it makes it more important.

I would just ask whether you think enough reviews make a big enough effort to explain the meter's basic ac accuracy and characterise it as good or bad?

[Kleinstein]

For many uses the accuracy of the AC measurements is not that important. Anything mains related is a bit variable anyway. There are a few exceptions, especially with audio. The more relevant part for AC are the limitations is the frequency and crest factor. Not all DMMs can measure 20 Hz or 10 kHz AC reasonably well - this could still be an acceptable limitation as most measurements tend to be main frequency and maybe the 2xmains frequency ripple after a rectifier.  A point not always clear in the specs is that some meters have AC mV with DC coupling only. This can be anoying when looking at ripple.

For the meters based on the analog RMS chips there is compromise between the low frequency performance and response time - it could be interesting to see on which side they are: relative fast response and poor LF accuracy or good accuracy also at 20-40 Hz and than a more sluggish response.

That compromise is sort of baked into the problem and it is there even if you don't use an analog TRMS chip.  Better and newer meters (think HP 34401A and up/newer) have selectable AC filters and work pretty well down to 3Hz or so. 

There is a principle limitation, but the analog RMS chips are quite a bit slower than the limit that can be reached in the digital solution. Ideally digital RMS the response takes the actual frequency into account and can react from one period to the next. Even with just a fixed filter it is the difference between a finite response time (FIR filter like running average) and an exponential decay (and sometime worse) for the analog RMS. Switching the filter speed with analog RMS is more like a work around to have the choice. The newer KS bench DMMs (3446x - not sure about the 34410) use digital RMS.

One can quite clearly see the difference when measuring an AC voltage and than remove the probes / turn off the source. With digital RMS one usually gets one reading in between and than a zero. Analog RMS takes several samples to get down and sometimes several seconds to really reach zero.

[2N3055]

For many uses the accuracy of the AC measurements is not that important. Anything mains related is a bit variable anyway. There are a few exceptions, especially with audio. The more relevant part for AC are the limitations is the frequency and crest factor. Not all DMMs can measure 20 Hz or 10 kHz AC reasonably well - this could still be an acceptable limitation as most measurements tend to be main frequency and maybe the 2xmains frequency ripple after a rectifier.  A point not always clear in the specs is that some meters have AC mV with DC coupling only. This can be anoying when looking at ripple.

For the meters based on the analog RMS chips there is compromise between the low frequency performance and response time - it could be interesting to see on which side they are: relative fast response and poor LF accuracy or good accuracy also at 20-40 Hz and than a more sluggish response.

That compromise is sort of baked into the problem and it is there even if you don't use an analog TRMS chip.  Better and newer meters (think HP 34401A and up/newer) have selectable AC filters and work pretty well down to 3Hz or so. 

There is a principle limitation, but the analog RMS chips are quite a bit slower than the limit that can be reached in the digital solution. Ideally digital RMS the response takes the actual frequency into account and can react from one period to the next. Even with just a fixed filter it is the difference between a finite response time (FIR filter like running average) and an exponential decay (and sometime worse) for the analog RMS. Switching the filter speed with analog RMS is more like a work around to have the choice. The newer KS bench DMMs (3446x - not sure about the 34410) use digital RMS.

One can quite clearly see the difference when measuring an AC voltage and than remove the probes / turn off the source. With digital RMS one usually gets one reading in between and than a zero. Analog RMS takes several samples to get down and sometimes several seconds to really reach zero.

KS 34410A KS 34401A uses RMS chip from AD. But 34410/34411A did have digital measurements and that architecture is probably the basis for 3446x  series....

EDITED error. Sorry.

[EEVblog]

Several times people mention that it is harder to achieve precision on ac. Is that a reason not to direct the viewer or reader to understand the performance of an individual meter? I would have thought if anything that it makes it more important.
I would just ask whether you think enough reviews make a big enough effort to explain the meter's basic ac accuracy and characterise it as good or bad?

Good thread topic.
One of hte problem is that a decent review is already long enough, and adding (and explaining the importance/lack of importance of low level AC signal measurmeents and bandwidth you could argue adds excess length to the video for what most would consider little benefit. Sure it might be very important, maybe the most important parameter for some people, the number fo people in the catgeory is arguably low.
I think most reviewers know this and is why they tends to skip or gloss over such things.

And I know some poeple will say " just use timestamps", but the fact is video length matters.

Every reviewer also has a set of their own personal interests, and so they tend to focus more on those things. Joe Smith for example is mostly focussed on electrical and physical robustness and comparison, and for some poeple that may be the most important factor, but for others it could be way down the list.

[Micke]

KS 34410A uses RMS chip from AD. But 34411A did have digital measurements and that architecture is probably the basis for 3446x  series....

Both HP/A/K 34410A and 34411A uses Digital sampling for ACV and ACA  . Crest factor can be up to 10:1

[2N3055]

KS 34410A uses RMS chip from AD. But 34411A did have digital measurements and that architecture is probably the basis for 3446x  series....

Both HP/A/K 34410A and 34411A uses Digital sampling for ACV and ACA  . Crest factor can be up to 10:1

 

You are right... Idiot dyslexic me confused 34401 and 34410...
I used 34411 so that one I know how it works... I completely forgot about 34410...

Post corrected. Thank you !

[bdunham7]

As for "reading the f-ing manual", is that really a response to reviews not being as well constructed as they could be?

I would just ask whether you think enough reviews make a big enough effort to explain the meter's basic ac accuracy and characterise it as good or bad?

I suppose it all depends on what the overall goal of a review is.  If it is to characterize a single example of a meter that either doesn't have detailed specifications or for viewers that don't read manuals or don't understand those specifications, then perhaps some attempt to discuss the subject might be worthwhile.  But there have been extended discussions here about the particulars issues with various meter's AC performance and those get quite complex and detailed.  IMO, there's not much use in trying to simplify these issues enough to cram them into a reasonable length YouTube review.  At some point, simple becomes wrong.

So here's what you'd need to include in order to have any reasonable claim of covering the subject, IMO:

1) Input impedance characteristics, including whether or not they change on each range.  This includes resistance, capacitance, AC or DC coupling and the lower cutoff frequency of that coupling.

2) The type of AC RMS conversion--separate digital, analog chip, on-chipset digital etc.  Then the bandwidth of the system, as well as residual counts and lower thresholds.  I'd want to know whether the meter goes to zero whenever it is below a threshold, if the BW is lesser at lower levels, what minimum % of scale you need to acheive reasonable accuracy, maximum crest factor and more things I can't think of at the moment.

3) The performance of the input divider/amplifer--at what point does it clip/protect, if it is DC coupled is it possible to saturate it with DC and get a wrong reading instead of OL, and again more things as they occur to the reviewer.

4) The performance of the DC system (ADC + conditioning circuitry) after the AC conversion, if applicable.  How often does the display update, settling time, accuracy on all the ranges, etc.

5) Perhaps comparison of the example with the specifications, if the specs are detailed enough.  However, I did give a concrete example above of exactly how this could not be all that helpful with only one sample.

Unless you cover all of that, IMO, you aren't covering the AC subsystem.  At best, what you seem to be proposing would be to review the performance of a particular meter in a specific use case that you have in mind, such as measuring some relatively low-impedance, low-voltage audio signal such as perhaps a line-out or something.  Characterizing a meter as "good" or "bad" would be an example of simplifying something until it is wrong.

[Richard_S]

Thank you to everyone that is contributing. I'm finding this a very rewarding discussion. I hope others are too.

I know that some think adding a bit of detail on ac signal measurements would be time consuming in reviews, but could we not just add "basic ac voltage accuracy" to compliment the "basic dc voltage accuracy" that is so beloved of the manufacturers themselves as well as the community? Add to this a clear statement about the lowest ac range and I think you have defined the small signal performance. A few tens of seconds at most.

I have found it interesting that people think low-level ac signals are very niche. They may well be. Because they are so vital for me, it's difficult for me to see how many, or how few, others also need it. I have to confess that I am surprised though. It would seem to me that many people have at some point built, refurbished, or repaired a piece of audio gear. Maybe that's just the people I know. As well as the sound, loads of servo systems use signals that are both vaguely in the same frequency range, and also share signal levels with sound equipment. Lots of the systems in cars and airplanes for example. While a lot of power supply ripple falls into the same frequencies and voltages. If there is one thing that I have found more educational than any other here, it is that small ac signals are a very specialist interest.

On the RMS topic, I completely get why power users need to know RMS values, to help them understand what real power is being delivered, expended, lost or whatever. For small audio signals I was making the point that RMS is useful for two things. Knowing the realtime operational level of an audio signal and noise. All too often I read or watch someone saying that a multimeter "has to have RMS" for audio applications. Well, for realtime measurements of complex signals a multimeter is totally unsuitable, so its only real benefit is for measuring noise. Here it is a great asset but I wanted to point out that often the meter:

• does not have the bandwidth to make audio measurements
• does not have a range low enough for noise
• is required to assess a weighted measurement (almost always A) and multimeters just don't have weighting circuits

Meaning that users can be driven to buy totally unsuitable RMS-reading meters without understanding their limitations. That it would be useful to educate people that an average reading unit may well have suited them better, if it had a lower ac range and greater ac accuracy. It seems to me, and please say if I am wrong on this, that away from the very high-class and consequently high-priced (when new) meters, RMS is a relatively costly option that often reduces accuracy.

[alm]

As for "reading the f-ing manual", is that really a response to reviews not being as well constructed as they could be?

The problem is that a review can not make any statements about performance like accuracy or bandwidth beyond performance beyond "the performance of this meter is consistent with the published specifications". Just like you can't conclude from measuring the height of one person from Australia and one person from the US if people in Australia are taller than people in the US. In both cases you need to describe the entire population based on a representative sample of sufficient size. Asking this from a reviewer is unreasonable. So the manual or data sheet is the best information you're going to find. If the data sheet is incomplete or unreliable, then don't buy a meter from that manufacturer if performance is important to you.

Well, for realtime measurements of complex signals a multimeter is totally unsuitable

Can you explain? Anything within the bandwidth and crest factor of the DMM should be represented reasonably well. There are plenty of DMMs with bandwidth well beyond the audio bandwidth.

Meaning that users can be driven to buy totally unsuitable RMS-reading meters without understanding their limitations. That it would be useful to educate people that an average reading unit may well have suited them better, if it had a lower ac range and greater ac accuracy. It seems to me, and please say if I am wrong on this, that away from the very high-class and consequently high-priced (when new) meters, RMS is a relatively costly option that often reduces accuracy.

Can you show examples of two DMMs in the same price class, one average responding with higher accuracy and lower ranges, and one true RMS with lower accuracy? Or is this just a hypothetical? This "RMS is a relatively costly option" was true up to maybe twenty years ago, but I think you'd be hard-pressed to find any decent average-responding DMM these days.

[Richard_S]

As for "reading the f-ing manual", is that really a response to reviews not being as well constructed as they could be?

I would just ask whether you think enough reviews make a big enough effort to explain the meter's basic ac accuracy and characterise it as good or bad?

I suppose it all depends on what the overall goal of a review is.  If it is to characterize a single example of a meter that either doesn't have detailed specifications or for viewers that don't read manuals or don't understand those specifications, then perhaps some attempt to discuss the subject might be worthwhile.  But there have been extended discussions here about the particulars issues with various meter's AC performance and those get quite complex and detailed.  IMO, there's not much use in trying to simplify these issues enough to cram them into a reasonable length YouTube review.  At some point, simple becomes wrong.

So here's what you'd need to include in order to have any reasonable claim of covering the subject, IMO:

1) Input impedance characteristics, including whether or not they change on each range.  This includes resistance, capacitance, AC or DC coupling and the lower cutoff frequency of that coupling.

2) The type of AC RMS conversion--separate digital, analog chip, on-chipset digital etc.  Then the bandwidth of the system, as well as residual counts and lower thresholds.  I'd want to know whether the meter goes to zero whenever it is below a threshold, if the BW is lesser at lower levels, what minimum % of scale you need to acheive reasonable accuracy, maximum crest factor and more things I can't think of at the moment.

3) The performance of the input divider/amplifer--at what point does it clip/protect, if it is DC coupled is it possible to saturate it with DC and get a wrong reading instead of OL, and again more things as they occur to the reviewer.

4) The performance of the DC system (ADC + conditioning circuitry) after the AC conversion, if applicable.  How often does the display update, settling time, accuracy on all the ranges, etc.

5) Perhaps comparison of the example with the specifications, if the specs are detailed enough.  However, I did give a concrete example above of exactly how this could not be all that helpful with only one sample.

Unless you cover all of that, IMO, you aren't covering the AC subsystem.  At best, what you seem to be proposing would be to review the performance of a particular meter in a specific use case that you have in mind, such as measuring some relatively low-impedance, low-voltage audio signal such as perhaps a line-out or something.  Characterizing a meter as "good" or "bad" would be an example of simplifying something until it is wrong.

Great thoughts and thanks for taking the time and sharing them bdunham7. Thankfully, I think for audio (but also to some degree for stuff like servo systems, sonar and other similar signals) life is a lot simpler.

1) Audio output impedances tend to vary from a few ohms to a few hundred ohms while inputs are a few thousand to a few tens of thousands. As best I know, this means that unless the meter is particularly incompetently designed, its resistive impedance is going to be sufficiently high for it not to matter. Then as for the reactive part; well audio is close enough to dc that it too just won't matter either. So very different to RF measurement thankfully.

2) The bandwidth matters, and absolutely needs to be discussed. Excluding the really high-priced examples, many meters just do not have the bandwidth required. I'm not sure the question of how the RMS is converted does matter though, its more the performance that needs addressing. I don't think audio test is that demanding. Thankfully for audio, crest factor is largely irrelevant. Either you are trying to measure operational levels - how loud the sound is - where crest factors cab be extreme in audio but also where a multimeter is no use (so we don't care), or you are measuring test tones (which when measuring level, will be sine waves), or you are measuring noise. If you are measuring noise then crest factor does not play a significant role (or if it is significant, it means that the noise is so high you no longer care about the exact level).

3) These things could theoretically be a problem I guess, but I have just never met these issues in practice.

4) Once again, we are lucky in audio. Settling time is largely irrelevant because operational measurement where it would matter can't be done with a multimeter anyway. For test tones and noise, contrary to some other signal types, longer is better. So more or less, whatever the meter can do, it will be sufficient.

[bdunham7]

There is a principle limitation, but the analog RMS chips are quite a bit slower than the limit that can be reached in the digital solution. Ideally digital RMS the response takes the actual frequency into account and can react from one period to the next. Even with just a fixed filter it is the difference between a finite response time (FIR filter like running average) and an exponential decay (and sometime worse) for the analog RMS. Switching the filter speed with analog RMS is more like a work around to have the choice. The newer KS bench DMMs (3446x - not sure about the 34410) use digital RMS.

One can quite clearly see the difference when measuring an AC voltage and than remove the probes / turn off the source. With digital RMS one usually gets one reading in between and than a zero. Analog RMS takes several samples to get down and sometimes several seconds to really reach zero.

Maybe this isn't the right thread, but I'd protest that things are not as simple as that.  The apparent settling time of a particular DMM using analog TRMS is not necessarily due to the response time of the analog TRMS system.  The DC system afterwards integrates over a period of time as well and often has a smoothing function.  Very few meters allow you to change the AC TRMS filter speed and the DC aperture independently (although I do have one that does) so this shows up as an apparent difference in settling times.  A quick experiment using two roughly equivalent meters, the HP 34401A and the Fluke 8846A, shows me that with the default 20Hz AC filter and reading a 5VAC 60Hz signal, the HP does take a few readings to settle but the 8846A will settle with one intermediate reading, whether from 5VAC to 1VAC or from 5VAC to residual counts (all using a fixed range).

One thing to realize is the the AC filter, whether expressed in Hz or ms, is not a high pass filter.  The TRMS converter will typically function down to DC if you can get past the blocking capacitor.  If you set the AC filter to a setting much higher than the input frequency but then use a long DC aperture (or averaging) you then get something closer to an average reading than RMS.  This is just a result of the nature of AC and the exact same thing happens in a digital system if you aren't careful.

Any AC measurement is necessarily a gated, computed result and one big difference between digital and analog is the shape of the gate.  However, there's a reason HP retained the option of analog TRMS in the 3458A and that is because a digital system with a rectangular gate (if you use one) has it's own issues.  You mentioned synchronizing the gate to the period of the signal for a fast response and when that works it's fine--but it doesn't always work.  The folks at HPAK and Keithley have put a lot of work into getting those digital subsystems to work appropriately and you still need the right gate time. 

My Fluke 8505A, which is not an integrating meter in the typical sense, gives me the option of manually setting a TRMS settling time of 100ms or 500ms and another separate option of how many samples to acquire and average for each displayed reading.  The default, IIRC, is 2⁷, but you can select any number from 1 to 2¹⁷ samples per reading (it takes 240 samples per second).  Using this, I can pretty readily demonstrate that the settling time of the TRMS converter is a non-issue for human-readable results.  If you were looking at logging faster changes in higher frequency signals, then perhaps you'd want a shorter gate/faster settling time, but that would be acheivable with either type of system.

[EEVblog]

Unless you cover all of that, IMO, you aren't covering the AC subsystem.  At best, what you seem to be proposing would be to review the performance of a particular meter in a specific use case that you have in mind, such as measuring some relatively low-impedance, low-voltage audio signal such as perhaps a line-out or something.  Characterizing a meter as "good" or "bad" would be an example of simplifying something until it is wrong.

That's the thing. If you spent say 10 minutes of a review fully charctering the AC performance and then compared it to say the best performing model on the market, may people without knowing better might think that's "bad", when in every other (likely more important) way the meter is better.
This is why to me, arguing over whether or not a meter survives Joe Smith's overload tests slighter better than some other meter is likely moot for most people. So it's likely better to simply say in a review that it's UL tested, tick, do a few basic tests like 1000V and 240V on the ohms range and that's that, tick, and they design and layout physically looks ok, tick.

I'd put AC performance of a meter so far down the list of things to consider for Joe Average that it's just not worth adding to a review.
That doesn't mean there isn't a niche market for that, just like Joe has found for his overload testing, many people enjoy those comparisons and find them useful.
So nothing to stop someone becoming the go-to channel for detailed AC performance test comparisons. But I just don't think it's something that needs to be in every review.

[EEVblog]

As for "reading the f-ing manual", is that really a response to reviews not being as well constructed as they could be?

The problem is that a review can not make any statements about performance like accuracy or bandwidth beyond performance beyond "the performance of this meter is consistent with the published specifications". Just like you can't conclude from measuring the height of one person from Australia and one person from the US if people in Australia are taller than people in the US. In both cases you need to describe the entire population based on a representative sample of sufficient size. Asking this from a reviewer is unreasonable. So the manual or data sheet is the best information you're going to find. If the data sheet is incomplete or unreliable, then don't buy a meter from that manufacturer if performance is important to you.

Yes, the spot accuracy test in a review is essentially just a "warm fuzzy" test to potentially spot an issue.
Any decent meter will be calibrated adjusted at the factory, so if you review test a meter with a 0.5% spec and you measure 0.4% out then you can certainly go "somethingsup here..."
But you can't take two reviews of different meters with one showing 5.000 and one showing 5.002 and say that one is "more accurate" than the other, that's not a thing.

Unfortunately the other issue with reviews is that can't keep having a detailed explanation in every review video of how accuracy works etc, the videos wouldn't become too long and convoluted. So you have to link to different videos on this topics and leave it up to the viewers to ensure they know what it all means.

One good example of this, even know not electronics, is Thunderf00t's style of debunking videos. The video will be about X, but his style is to include full detail on X, Y, and Z etc in the same video just in case that's the only video of his you'll ever watch. The result is a much longer video that repeats so much stuff from other videos. There is some merit to that of course, but if you are a regular viewer it's all redundant.

[Richard_S]

As for "reading the f-ing manual", is that really a response to reviews not being as well constructed as they could be?

The problem is that a review can not make any statements about performance like accuracy or bandwidth beyond performance beyond "the performance of this meter is consistent with the published specifications". Just like you can't conclude from measuring the height of one person from Australia and one person from the US if people in Australia are taller than people in the US. In both cases you need to describe the entire population based on a representative sample of sufficient size. Asking this from a reviewer is unreasonable. So the manual or data sheet is the best information you're going to find. If the data sheet is incomplete or unreliable, then don't buy a meter from that manufacturer if performance is important to you.

Well, for realtime measurements of complex signals a multimeter is totally unsuitable

Can you explain? Anything within the bandwidth and crest factor of the DMM should be represented reasonably well. There are plenty of DMMs with bandwidth well beyond the audio bandwidth.

Meaning that users can be driven to buy totally unsuitable RMS-reading meters without understanding their limitations. That it would be useful to educate people that an average reading unit may well have suited them better, if it had a lower ac range and greater ac accuracy. It seems to me, and please say if I am wrong on this, that away from the very high-class and consequently high-priced (when new) meters, RMS is a relatively costly option that often reduces accuracy.

Can you show examples of two DMMs in the same price class, one average responding with higher accuracy and lower ranges, and one true RMS with lower accuracy? Or is this just a hypothetical? This "RMS is a relatively costly option" was true up to maybe twenty years ago, but I think you'd be hard-pressed to find any decent average-responding DMM these days.

Hi Alm and thanks for taking the time to think about this and respond. I genuinely do not understand your first point and it is I am sure my fault.

Maybe I can explain what I find frustrating in meter reviews. I note that reviewers will almost always mention the maximum ac voltage that a meter's manual will claim it can handle. They will always mention the claimed basic dc accuracy. Not only that, they will use this figure to classify the meter. Finally, they will often then mention the minimum dc voltage range.

By contrast I note that the minimum ac voltage range gets mentioned less frequently. Even though its maximum almost always gets a look in. My overall point here is that while there are many people servicing electrical substations and three-phase motor drives and that kind of high-voltage gear, there are some that work on small signal stuff too and I think they get left out. It would seem from what people say here though that more are working on the big high-power systems like electrical substations or three-phase motors than would ever want to fix an audio amplifier, radio, or audio interface unit. A fact I genuinely found surprising.

My contention was that because it is frequently ignored, small-signal ac performance does not get put in perspective. I never hear the phrase "basic ac accuracy" for example. Why? In your reply here, you would seem to be saying, and I am sure I must have got this wrong, that mentioning the quoted basic dc accuracy is straightforward, but to mention the quoted ac accuracy is inappropriate without testing many examples of the same model of meter. If that is so, why? What is different between ac and dc in this regard?

As for operational audio measurement, I am happy to explain. If you want to do this you need a special meter, the dynamic response of which needs to be dedicated to the task. These dynamics were often called the ballistics of the meter. As well as a particular meter, the user also needs to have been trained to understand what the reading means and it is the movement of the needle (or other analogue indicator) that they will read, not just its position. So, no one can or would use the digital display of a DMM for this and any multimeter's analogue bar graph will not carry the correct ballistics. DMMs will also generally not update frequently enough, and rarely have the resolution required.

I'll search out some example meters. I note a feeling though that for a lot of contributors here, their own favourite meter is in the $1,000 to $2,000 (US) class and this is what they think of as a typical mutimeter. In my world, such meters are rare, and I think that the vast majority of people looking for help are buying $100 to $200 units.