Multimeters at 500.000 counts ... Do we need more than that ?

[Kiriakos-GR]

Multimeters at 500.000 counts ... Do we need more than that in a handhold DMM ?

The question sounds simple, but, is it out there any easy answers ?
Should we say to the manufactures .... stop we do not need more counts, use your R&D for making better other aspects than that.

[BravoV]

Its easy answer, depends who "we" are talking about, and which market segment is those manufacturers are targeting or planning for new product.

If "we" are electricians that work mostly on high voltage power mains or power installations like yourself, then 500.000 counts is way overkill, c'mon, you have to admit that too.

Have few friends that are experienced electricians, mostly they claimed that that those high counts meters especially hand held usually is not for the real job, but merely for bragging factor.

But say if "we" are bunch of calibration operators that are working of ppm degrees everyday when doing high precision calibration job, I would say 500.000 counts still not enough, hell, even with 1.000.000 count / 1 ppm resolution will not make these guys completely happy.

[Kiriakos-GR]

Well my own wish list about the 500,000 counts portable solutions, will be an sample rate of 6 times per second.

Until now and probably the start of 2012, the max that we have see is up to 1.25/S nominal .. not bad, but still away from my liking. 

[Conrad Hoffman]

I like my 6 1/2 digit bench meters, but in a handheld it would seem like a distraction. My old Fluke 77 does everything I need in a handheld, but I'm old school. Now, if a meter also has a frequency counter function, I can see where having more digits would be nice. Even the counters built into digital scopes leave a lot to be desired in terms of resolution and update. IMO, fast update is a hugely desirable feature.

[alm]

Fast sampling rates have always been the domain of the system/bench meters. Even a 1981 HP 3456A will do 120 readings per second with auto zero or 210 without at 5.5 digits of resolution. A modern Agilent 34410A will do 10k readings per second at 5.5 digits. I guess the main reason is that these specs are especially important in system applications, where reading rate is directly proportional to throughput. For example if you want to characterize a part at all possible combinations of parameters. Not much point in hundreds or thousands readings per second if the purpose is just to display them for the user, since the human brain is unable to process information that fast.

Not sure why the handhelds are unable to sample at something like 10 readings per second, I find a meter with a faster update rate much nicer to use. So I use $100 bench meters instead of paying an order of magnitude more for a handheld with similar specs. Maybe the power budget and parts count of handheld meters makes it hard to increase the reading rate? Or maybe they just don't care.

Resolution is not the important spec, I care more about accuracy. What's the point of a 5.75 digit meter if the accuracy is the same as a 4.5 digit one? But for most applications, usability issues like display quality (LED or VFD is much nicer than LCD) and reading rate are much more important in my opinion.

[McMonster]

Until now and probably the start of 2012, the max that we have see is up to 1.25/S nominal .. not bad, but still away from my liking.

For the little I could check where I am now for my SANWA PC5000a the refresh rate is actually much higher than 1.25 times per second. On V DC and mV DC it's almost identical on both settings while simply holding the probes in hands, about 5 times per second. I'll do more test when I'll be back at my lab tomorrow.

[Balaur]

Personally, I'm perfectly happy with 4000 or 6000 counts with a good 0.1%+1d.
I consider other features far more important, but it's difficult to find a multimeter with all these:
- fast autoranging
- a rotary switch with all the possible measurement choices, or at least with some reasonable defaults
- decent AC bandwidth (at least audio, come on, it's not this hard)

Also, I would really appreciate an extended diode range (up to at least 4.5-5V) with a selectable current up to 10mA (let's say) that could test all those white and UV LEDs. Most of the multimeters are pathetic in this field.

[Tony R]

I guess it depends on what you want to do with it. If I was an electrician I may not be concerned in anything past a decimal in my everyday work. My major is computer engineering, and many times I'm only concerned with if it is a high or low. But sometimes when I'm doing analog work knowing it as accurately as I can down. It also comes in handy when you are doing high precision stuff, like working with 0.01% tolerance resistors.

You wonder at which point does the resistance/inductance/etc. in the probes and internals of the DMM make the least significance decimal meaningless...

The point is perhaps the high counts is not for everyone, but they are useful if you need it.

Sometimes the most significant data is in the most insignificant details.

[alm]

You wonder at which point does the resistance/inductance/etc. in the probes and internals of the DMM make the least significance decimal meaningless...

Four wire measurements help with this. But if you check the accuracy specs, the last digit is rarely accurate (how many 4.5 digit meters have a basic accuracy of 100ppm?), although it may help with relative measurements (resolution versus accuracy).

Sometimes the most significant data is in the most insignificant details.

There are definitely cases where everything but the last digits is boring and predictable. For example when measuring the temperature coefficient of resistors. Even a crappy resistor won't change from 1 kohm to 2 kohm due to small temperature changes.

[EEVblog]

Not sure why the handhelds are unable to sample at something like 10 readings per second

Power consumption may have a fair bit to do with that?

Resolution is not the important spec, I care more about accuracy. What's the point of a 5.75 digit meter if the accuracy is the same as a 4.5 digit one?

Resolution can be very useful, even in a 1% accuracy grade instrument, to allow you to match components and do relative difference measurements etc.
Generally, even a basic 1% class instrument will be just as short term stable and linear as a high class instrument, so it's possible to use it for such work.

But for most applications, usability issues like display quality (LED or VFD is much nicer than LCD) and reading rate are much more important in my opinion.

LED and FVD also rule out practical everyday use handheld instruments of course.
LCD is the only solution for handheld meters.

Dave.

[Kiriakos-GR]

Until now and probably the start of 2012, the max that we have see is up to 1.25/S nominal .. not bad, but still away from my liking.

For the little I could check where I am now for my SANWA PC5000a the refresh rate is actually much higher than 1.25 times per second. On V DC and mV DC it's almost identical on both settings while simply holding the probes in hands, about 5 times per second. I'll do more test when I'll be back at my lab tomorrow.

I will use some terminology of car mechanics ...   

At 50.000 counts it works with a faster gear box 5 samples nominal, that is good enough.
And at 500.000  uses a slower gear box, check yous Users manual. ( I have read it all ready ) 

[alm]

Resolution can be very useful, even in a 1% accuracy grade instrument, to allow you to match components and do relative difference measurements etc.
Generally, even a basic 1% class instrument will be just as short term stable and linear as a high class instrument, so it's possible to use it for such work.

This assumes good linearity and low noise, which are rarely specified for affordable meters. Is the difference between 100.01 mV and 100.06 mV significant or is it just noise? Hard to tell without manufacturer specs or gathering statistics.

LED and FVD also rule out practical everyday use handheld instruments of course.
LCD is the only solution for handheld meters.

What about OLED?   Even in LCD there are varying degrees of suckiness, some have better viewing angle and contrast than others. The speed and better display does lead me to prefer bench meters for some tasks, though.

[amspire]

Resolution can be very useful, even in a 1% accuracy grade instrument, to allow you to match components and do relative difference measurements etc.
Generally, even a basic 1% class instrument will be just as short term stable and linear as a high class instrument, so it's possible to use it for such work.

This assumes good linearity and low noise, which are rarely specified for affordable meters. Is the difference between 100.01 mV and 100.06 mV significant or is it just noise? Hard to tell without manufacturer specs or gathering statistics.

Well not really. If the last digit  or two is unstable, then yes that is not much use. But if it is fairly stable, then the extra digits are very useful.  You can use the resolution for matching resistors, comparing voltage references.

Also if the meter auto zero works well, then the extra resolution is great for measuring the null point on a bridge circuit like a wheatsone bridge.

Richard

[Conrad Hoffman]

Balaur- +1 on all your points. I hadn't even thought of them, but I'd sure buy a meter with those features.

Concerning absolute accuracy, somebody tell me an application where 100.0 VDC vs 100.5 VDC actually makes any difference at all. Confirming some spec, sure, but I mean where circuit function is affected. I don't think anybody is making analog computers anymore. Come on, I'm waiting!

[IanB]

if the meter auto zero works well

Tell me about auto zero. Every previous meter I have owned reads 0.000 on a voltage range if I short the probes together. By my new Protek 6500 is not following my expectation in this regard. On the mV range it has an offset of -0.009 mV with shorted probes, and on the V range it has an offset of -0.0005 V. This offends my sense of tidiness. What do other 50,000 count meters do with the same test? Do I have an oddball, or is this par for the course with a high resolution meter?

[EEVblog]

Concerning absolute accuracy, somebody tell me an application where 100.0 VDC vs 100.5 VDC actually makes any difference at all. Confirming some spec, sure, but I mean where circuit function is affected. I don't think anybody is making analog computers anymore. Come on, I'm waiting!

I could start listing some reasons, but better to ask yourself:
- Why do meters have an accuracy spec at all?
- Why do various accuracy/price ranges exist if absolute accuracy doesn't matter in practice? (all the way from 1%+ to 0.0001%)
- Why do people get their meters calibrated? (which is mostly checking it's absolute accuracy)

Dave.

[amspire]

Concerning absolute accuracy, somebody tell me an application where 100.0 VDC vs 100.5 VDC actually makes any difference at all. Confirming some spec, sure, but I mean where circuit function is affected. I don't think anybody is making analog computers anymore. Come on, I'm waiting!

The most obvious case is when you are relying on two different meters to read a voltage drop. Say you are in one building and your mate is at the other end of the wires in another building.

Say it is 100V at your end, and 99V at the other end.

Ideally you would like to be able to trust the readings without having to construct a meter calibration chart to match up the two meters.

The other much more important case is when you have just realized that your new $400 multimeter is out by 0.5% and that 100.5V reading is sending you into an extreme state of depression.

[ToBeFrank]

- Why do meters have an accuracy spec at all?

Because if you don't give an engineer a spiffy spec to meet, he won't know what to do?

[BravoV]

Assuming we're talking "handheld" multimeter, I think we must agree to certain extend that no single multimeter will make all parties happy.

Field electrician work meter is definitely different vs multimeter for EE or designer working in a nicely & comfy bench table in the lab environment.

Although I'm not an electrician, I would imagine if I was a high voltage or heavy industry electrician that spend most of my times in harsh or even danger environment like in antenna tower or room with noisy filled with heavy machinery, rather than it can displays gazillion digits accuracy,  I rather preferred if my multimeter's probe somehow can be used as a safety harness, or even better the meter itself can be used safely as a hammer block lol .. like to bang out out a rusty metal enclosure door of the power distribution box that is hanging hundreds meters above the ground, thats stubbornly refused to open while I was still hanging up there (yeah, I'm exaggerating) .

Just a reminder, the thread starter is an industrial electrician, imo it is not realistic to expect an industrial electrician at the field task frequently to do resistors sorting to get a 0.0001% pair for the accurate voltage divider.

Regarding display refresh speed, even it is capable to display at human eyes rates, c'mon, again realistically, do you trust your own eyes for a single event of a quick miliseconds reading spike rather than looking at the min or max value captured by the meter ?

[alm]

About the importance of refresh speed, I've had a fluctuating signal that appeared constant with a meter only updating a few times per second, but was clearly unstable on a faster meter. A fake-analog bar graph would also have worked in this case. Min/max might be used to quantify the fluctuations, but you have to be aware of them first.

Another issue is adjusting trimmers so the voltage is 10.00 V. Humans need fast feedback since we don't have PID controllers. It's very annoying to start turning and have to wait a while before you even know if you're going in the right direction. A bar graph is not very useful if you're down to the least significant digits.

[Lightages]

Concerning absolute accuracy, somebody tell me an application where 100.0 VDC vs 100.5 VDC actually makes any difference at all. Confirming some spec, sure, but I mean where circuit function is affected.

That is easy. Solar power system batteries. Their life depends entirely on proper voltages while being charged, and the balance of the voltages and current involved within a bank of batteries.
While not usually in the range of 100 VDC , voltages around 50 and 60 are quite common and a difference of 0.2 VDC between different batteries in a bank can mean many amperes imbalance between batteries causing some to not charge properly or discharge faster than the overall average.

[Kiriakos-GR]

Just a reminder, the thread starter is an industrial electrician, imo it is not realistic to expect an industrial electrician at the field task frequently to do resistors sorting to get a 0.0001% pair for the accurate voltage divider.

What is this ? another one " warning warning an electrician plays with our tools !!  " 

Well I am from the modern ones.   
Equipped with the education of 90s so to handle the future.
 

[Conrad Hoffman]

Lightages, as long as only one meter is involved, absolute accuracy doesn't matter (within reason). It's only if you want to measure a bank, and somebody else wants to measure a bank with their meter, that you'd have a problem. Or, if you're recording data for future use. Even then the percentage accuracy you're talking at lower voltages is pretty undemanding. We all like resolution and many applications need as much as you can get, but in 35 years of doing this I've come across very few (none) situations where absolute accuracy (as in agrees with NIST or other international standards) amounts to a hill of beans. Most of the applications I can think of have been obsolete for years. Don't get me wrong, I'm a metrology nut and maintain multiple voltage standards plus a bunch of other stuff, and don't even get excited until an instrument has at least 7 knobs, but it's basically a high and low world today.

[Kiriakos-GR]

Accuracy in measurements its a must have for example in the industry (makers) or repairs (technicians) of the electronic scales.
Electronic scales of any size and capable for light weight items , or really huge.

Every single gram or kilo stands for a pile of money, yes this tasks need accuracy and resolution.
     

[Conrad Hoffman]

But don't you calibrate scales with weights?