In Reply to: Question on multimeters posted by Nathan Hopkins on 01/04/01 at 11:58 PM:
Multimeter accuracy specifications for analog meters and digital meters are done differently. You're more interested in DMMs, so I'll hit that first. The accuracy spec of a DMM is expressed as a percentage of the reading (not full scale)± so many counts of the least significant digit (LSD). Nearly every digital measurement has a minimum error of ±1 count. It's just the nature of the beast. If you ever look at a digital oscilloscope with the input grounded (no signal applied), the trace will jump up and down at least one bit, indicating this digital anomoly.
In the case of your RS meter's DCV ranges, the spec is ±1%, ±2 counts. This means that if the reading is 10.00 volts, the actual error could be as high as (10.00)(0.01) which is 0.1 volt plus two counts. Since the LSD represents 0.01 volts, we'd add 0.02 volts to that error giving 0.12 volts error. The actual voltage being read could be as high as 10.12 volts.
This same reading, since the error is plus-or-minus, could be as low as that same percentage + LSD bobble or 9.88 volts.
On your meter, it works that way on every range. Since a DMM has at the heart of it a dc voltmeter, everything else must be converted to dc volts for it to make a measurement, and DCV will be the most accurate measurement function. These conversions add error to the basic DCV measurement "engine", and that's why DCA is worse yet. The current shunts aren't that accurate in the RS meter. You'll find that the DCA ranges of a Tektronix, Fluke or Hewlett-Packard (now Agilent Technologies) are nearly as accurate as the DCV function because they spend extra money on good laser-trimmed shunts.
The ACA function has some of the worst accuracy because first a conversion must be made from ACA to ACV using those lousy shunts, and then the AC must be converted to DC using an AC converter, which throws in more inaccuracies. You didn't mention the accuracy of the resistance ranges, but I can assume that they're likely around ±3% or worse.
The accuracies of all DMMs are basically done in the same manner. I don't like to trash other folks equipment and that's not my intention here, but just for comparison, this particular meter isn't much more accurate than a mirror-scale Simpson 260-8M.
Mentioning analog meters, their accuracy specification are calculated against a FULL-SCALE reading. This is why one is cautioned with an analog VOM to make measurement as near to full scale as possible. If a VOM has an accuracy of ±3%, then the measuring 10 volts on the 10 volt range means that the accuracy of the measurement will be ±0.3 volts which translates to ±3% of the reading. If you measure 10 volts on the 100 volt range, the accuracy will be ±3 volts, which translates to a reading accuracy of ±30%!
Resistance measurement accuracies of a VOM, since the resistance scale is non-linear, is indicated as percent of arc of pointer movement and is a rather difficult specification to verify.