[volt-nuts] Agilent calibration

[Charles Steinmetz]

Joe wrote:

>The way I read this is that if I send them a DMM that is within spec, they
>won't adjust it or provide pre/post data. Is this the case? If I spend over
>$200 sending a DMM to them, I want it adjusted to the best possible specs
>and I want the data. I do not want someone just saying that it is good
>enough and send it back to me. I can get that for $50 in El Paso.

The big difference is not between adjusting and not adjusting -- it 
is between getting a calibration "with full data" and getting one 
without data.  /The true value of calibration is not the adjustment 
-- it is the data./

Agilent doesn't just say it is good enough -- they tell you 
specifically how far off it is and quantify the statistical 
uncertainty of their measurement.  That is everything you need (i) to 
correct readings you make with the instrument and (ii) to be 
confident of the potential uncertainty of those measurements.

Let's say your meter has an uncertainty spec of +/- 15 uV (1.5 ppm) 
total at 10 V.  If your calibration certificate says the meter reads 
dead on at 10.000000 V, the reading shown on the display is your 
measurement result (with a certainty of 1.5 ppm, or +/- 15 counts 
from the reading) when you measure a 10 V source.  But the cal 
certificate could just as well say that the meter reads 10.000008 V 
when measuring a 10.000000 V source.  In that case, you know to 
subtract 0.000008 V from whatever the meter reads when you measure a 
10 V source to get your measurement result (again, with a certainty 
of 1.5 ppm, or +/- 15 counts from the corrected reading).  Of course, 
in the real world a voltage standard will have its own calibration 
offset, so you will make two corrections when you measure your house 
"10 V" standard to verify that your meter is still in calibration.

So, why wouldn't they adjust every instrument to be "spot 
on"?  Because metrologists have determined that, as a general matter, 
not messing with the adjustments results in overall better stability 
of instruments.

Adjusting instruments inevitably causes a new drift and settling 
cycle, so if you adjust everything as close to perfect as possible 
every time you calibrate, you will always be on the steepest portion 
of the settling curve.  On the other hand, you can benefit from the 
long, ever-decreasing tail of the settling cycle by not adjusting as 
long as the instrument is within the manufacturer's 
specifications.  Further, seeing the change from one calibration 
interval to the next, and the next, etc., increases the confidence 
you have in readings you make when the last calibration is not so 
fresh any more.

Best regards,

Charles