Why do multimeters have "counts"?

[Fungus]

What's the thinking behind having "counts" on multimeters? Meters are 2000 count, 6000 count, etc.

Why aren't they all just nice decimal numbers - "1000 count", "10,000 count", "100,000 count"?

I've been scratching my head trying to come up with a reason they would do that. 

[MrSlack]

I built a DMM from scratch once. Not commercial grade but it worked. This used a simple integrating converter and some CMOS logic.

Basically what tends to happen in these is there is a crystal controlled clock pulsing at a constant known rate. The input circuitry of the DMM first scales the voltage down to something manageable, 200mV in my case. Then you charge up a small capacitor with a constant current until the voltage reaches the same voltage as the input (this is basically integration). The amount of time this takes is proportional to voltage and is measured by the count of pulses that go past between when the counting starts and when the voltage is equal to the input and this is popped on the display giving you the "count".

Higher counts count faster and therefore have more resolution.

[Muttley Snickers]

I don't know  , it might have something to do with the particular processor or chipset used, I have a Uni-T meter    that is either 8000 or 80000 count depending on the user selection, an entirely separate subject is the specifications which state that the meters basic dc accuracy should be +/- 0.05% plus a gazillion counts.   

What gets me is the number of digits with 3~3/4, 3~5/8, 4~87/64ths and all that rubbish  , mind you I was never much good with fractions.   

[teksturi]

If meter have 10000 counts then 1V range maximum reading is 9.999V. Meter is then 4 digit meter. If meter have 12000 counts maximum reading with 1V range is 11.999V.  Meter can now call that it is 4 1?2 digit meter (at least Keysight do that. This is not so black and white).  This is because it cannot show full range of 5 digit. Why anyone want to get meter with nice count number if that meter is not that great.

There is of course technical aspect behind all of this and little bit marketing.

[coppice]

Why aren't they all just nice decimal numbers - "1000 count", "10,000 count", "100,000 count"?

Isn't separating models by a factor of 10 in resolution a pretty coarse approach?

[The Soulman]

It is easy, the count number gives the maximum value that the display can display.
So a 2000 count meter could display maximum 2.000 volt on the 2 volt range, to measure
3 volts on this meter you would have to move up to the 20 volt range to have the meter display 03.00 volt.
So you're losing resolution.
More counts the better, I do a lot with 24 volt battery systems and have to use the 200volt range on my
2000 count meters but that gives a poor resolution of just 24.0 volt.
To be more accurate I've bought a 30.000 count meter that has a 30 volt range, it can display 24.000 volt gaining not one but two digits of resolution.
This also applies on every other range on the meter.
Am I making sense? Actually I think Dave has done a video on counts.
That 3 and so much quarter is just a different way of saying the same but in a more awkward way.

[The Soulman]

[Wytnucls]

Traditionally, multimeters were characterized by their number of digits (i.e. 3 1/2), as the first digit was usually limited to 1. This became confusing when labels like 3 3/4 or 4 4/5 digits came about, when the first digit could be any number. Specifying a number of counts for the display became easier to understand, although the number of digits method is retained for bench meters with high resolution.

Interesting Tektronix article:

 http://www.tek.com/document/whitepaper/understanding-handheld-dmm-specifications

[Fungus]

It is easy, the count number gives the maximum value that the display can display.
So a 2000 count meter could display maximum 2.000 volt on the 2 volt range, to measure
3 volts on this meter you would have to move up to the 20 volt range ...(snip)

Yes, I know what counts are, I want to know why.

It seems a weird way of thinking, I don't understand why it's universally done that way. Where did it start?

[Delta]

As a DMM is essentially a glorified DAC, so why aren't the number of counts nice 2^N numbers? ie a 13bit DAC would have 8192 counts, but meters have 6000 etc...

[nctnico]

What's the thinking behind having "counts" on multimeters? Meters are 2000 count, 6000 count, etc.

Why aren't they all just nice decimal numbers - "1000 count", "10,000 count", "100,000 count"?

I've been scratching my head trying to come up with a reason they would do that. 

Probably marketing and common sense to have meaningful measurement ranges for commonly used voltages.

[joeqsmith]

I have no clue about how it started.  My guess is a combination of marketing, advertising and sales groups came up with it as a way to avoid confusion or to set them apart from other brands?   If a big player started the ball rolling, everyone would follow to keep up.... 

http://www.tek.com/document/whitepaper/understanding-handheld-dmm-specifications

Good link.

If meter have 10000 counts then 1V range maximum reading is 9.999V. Meter is then 4 digit meter. If meter have 12000 counts maximum reading with 1V range is 11.999V.  Meter can now call that it is 4 1?2 digit meter (at least Keysight do that. This is not so black and white).  This is because it cannot show full range of 5 digit. Why anyone want to get meter with nice count number if that meter is not that great.

There is of course technical aspect behind all of this and little bit marketing.

So a Keysight meter with a 1KV range will actually read 10KV?  I wonder why the one I tested failed so early then.      

[Delta]

I have no clue about how it started.  My guess is a combination of marketing, advertising and sales groups came up with it as a way to avoid confusion or to set them apart from other brands?   If a big player started the ball rolling, everyone would follow to keep up.... 

http://www.tek.com/document/whitepaper/understanding-handheld-dmm-specifications

Good link.

If meter have 10000 counts then 1V range maximum reading is 9.999V. Meter is then 4 digit meter. If meter have 12000 counts maximum reading with 1V range is 11.999V.  Meter can now call that it is 4 1?2 digit meter (at least Keysight do that. This is not so black and white).  This is because it cannot show full range of 5 digit. Why anyone want to get meter with nice count number if that meter is not that great.

There is of course technical aspect behind all of this and little bit marketing.

So a Keysight meter with a 1KV range will actually read 10KV?  I wonder why the one I tested failed so early then.   

No.   

[EEVblog]

If meter have 10000 counts then 1V range maximum reading is 9.999V.

IME, meters will go slight over on such 100,000 count type ranges.

[Kleinstein]

Traditionally the ADCs in DMMs directly used the digits displayed and often directly used them decimal. So the 3 1/2 Digit = 2000 count meters are usually not binary 12 bit converters that might give +2047 to -2048. The 3 1/2 digit resolution was a long time favorite because there where cheap chips (ICL7106/7) available for this.

The digits number is also more a marketing language not a good technical specifications, there the counts are much better as they can deal with other ranges as well without a problem.

For confusion some meters alow some overrange: e.g. a nominal 100 V range might read up to about 120 V, though with possibly higher errors. Sometime they call such a meter 12000 counts and sometime 10000 counts.

[sync]

As a DMM is essentially a glorified DAC, so why aren't the number of counts nice 2^N numbers? ie a 13bit DAC would have 8192 counts, but meters have 6000 etc...

Because they usually use dual- or multi-slope ADCs. These are voltage to time converters. The time interval is measured by a counter. A lot of DMMs using BCD counters for that. It's not like a SAR ADC which has a native resolution of 2^n.

[Wytnucls]

Intermediate counts (for instance 4,000 or 6,000) make sense to get the most out of the chosen multimeter ADC, especially for low count multimeters.

http://zone.ni.com/reference/en-XX/help/370384N-01/dmm/resolution_bits_digits/

[newbrain]

What's the thinking behind having "counts" on multimeters? Meters are 2000 count, 6000 count, etc.

Why aren't they all just nice decimal numbers - "1000 count", "10,000 count", "100,000 count"?

I've been scratching my head trying to come up with a reason they would do that. 

Once you have made the effort to add that 1 (on the display and the circuitry) it's good to exploit it (giving you the common 0-1999 ranges).
Also: https://en.wikipedia.org/wiki/Benford%27s_law

[Fungus]

Once you have made the effort to add that 1 (on the display and the circuitry) it's good to exploit it (giving you the common 0-1999 ranges).
Also: https://en.wikipedia.org/wiki/Benford%27s_law

It's all so clear to me now... how could I have missed that?

Benford's law explains perfectly why all those cheap multimeters use 2000 counts.

When Fluke, et al., had to distinguish themselves so they made 4000 count meters (obviously twice as good as the cheap ones).

Then along came the marketing guy with his vision for a 6000 count meter ("Just imagine it...!") and it all went downhill from there.



PS: Not a lot of people know it but Benford's law is the main reason why the numbers on a slide rule are distributed unevenly.

[barry14]

The numbers on a slide rule form a logarithmic scale because a slide rule exploits the property of logarithms that convert multiplication of numbers into addition of the logarithms of those numbers. 
This is a universal property of logarithms.  Benford's law, which is not universal, has no connection to slide rules.

[zapta]

Why aren't they all just nice decimal numbers - "1000 count", "10,000 count", "100,000 count"?

Because "it gives you the extra push over the cliff"

[Fungus]

The numbers on a slide rule form a logarithmic scale because a slide rule exploits the property of logarithms that convert multiplication of numbers into addition of the logarithms of those numbers. 
This is a universal property of logarithms.  Benford's law, which is not universal, has no connection to slide rules.

You've been trolled... 

[pelule]

Yes, I know what counts are, I want to know why.

I hope, I understood correctly the basic question.
To "precisely" convert a voltage to a digital value, several methods are in use today.
In the "good" old days (until ~198x, even today for very high precision) this was done by a Voltage-To-Time conversion (Integrating Conversion, Single-Slope, Dual-Slope, Multi-Slope, Voltage-to-Frequency). This time/frequeny is digitally measure with a counter circuit - thus the result is named the COUNT.

BR
PeLuLe