Measurement accuracy question

[taydin]

When a multimeter's measurement accuracy is stated to be 0.5% for the 200 mV range, what is the test voltage used? Is it the middle of the range (100 mV) or as close as possible to the range upper limit? (199mV)

With high end multimeters, I see that an exact voltage is specified with the tolerance numbers, but with most multimeters I don't see a voltage specified. Just a tolerance for a range.

[Kleinstein]

The accuracies are usually given as percentage of the measured value plus a part proportional to the range, which may be specified as counts.
With low resolution meters the absolute error part may be +- 1 count and in this case not extra noted. With analog meters the usual specs is only to give the part proportional to the range, as this is often the dominant part.

The rating is valid for all values up the the nominal full scale, at least for DC. For AC there can be additional limitations for small readings (e.g. only for > 1 or 5% of the range).

[Fungus]

When a multimeter's measurement accuracy is stated to be 0.5% for the 200 mV range, what is the test voltage used? Is it the middle of the range (100 mV) or as close as possible to the range upper limit? (199mV)

Usually a bit less than maximum.

eg. I've seen a lot of 2000 count meters where the voltage is 180mV.

OTOH the scale is perfectly linear so the 0.5% is applicable across the whole range.

[taydin]

The rating is valid for all values up the the nominal full scale, at least for DC. For AC there can be additional limitations for small readings (e.g. only for > 1 or 5% of the range).

There also should be a minimum voltage for the lowest DCV range, which I don't generally see (Keysight specifies it in some meters, but I haven't seen it specified in many others). As the voltage approaches the resolution of the DMM, the error would get larger and larger, eventually making the measurement completely uncertain. So I was wondering if there is an unwritten convention here.

[Fungus]

As the voltage approaches the resolution of the DMM, the error would get larger and larger, eventually making the measurement completely uncertain.

If you measure 1Mv on the 200mV range of a 2000 count meter then you obviously don't have enough digits to measure with 0.5% precision.

An autoranging meter might be able to drop down to a 20mV range when that happens but eventually you'll hit the bottom, yes.

[taydin]

OTOH the scale is perfectly linear so the 0.5% is applicable across the whole range.

I think this is the case when we are talking about one of the intermediary ranges (2V, 20V etc). But there needs to be a lower limit for the lowest range.

[Kleinstein]

The error would be 0.5% plus 1 count (could be more if noted, especially the AC ranges, current and low voltages may have more).  At large voltages the 0.5% are the important part, but at low voltage the +-1 count part dominates and takes into account the limited resolution. So there usually is no need for an extra lower limit. Analog AC is because of the RMS converter.