Calibration of Software/Measure activity

Hal Murray hmurray at megapathdsl.net
Sun Oct 21 22:02:13 EDT 2007


it.daniher at gmail.com said:
> As work on a USB ADC continues, somewhere along the lines the problem
> of calibration came up both with regards to the sound card, and the
> in-development peripheral. 

There are 2 dimensions to calibration: time and voltage.

How accurate do you need?  (or want?)

Calibrating the time should be simple.  All you need to know is the frequency 
that the A/D is collecting samples, so time how long that takes...  A simple 
user program could grab the time, collect a large number of samples, then 
grab the time again and do the calculations.  You need a lot of data so that 
the time to get started/stopped doesn't introduce much of an error.  Try a 
second or so, and compare that with twice as long.

That assumes your CPU clock is accurate.  (You are comparing the CPU clock 
with the A/D clock.)  That's probably a reasonable assumption.

The sound card runs off the CPU crystal.  If you are using it, the nominal 
value should be very close.  (It would be worth checking.)

The schematic that you posted the URL for a few days ago used an 8051 without an external crystal.  I assume it's running on an internal R/C oscillator.  See Mitch's comments.  The data sheet probably has some specs for accuracy.

The 8051 has a couple of pins labeld XTAL.  t might be worth leaving space on the board for a crystal connected to those pins.  Using it will probably require tweaking a mode bit in the 8051.


I don't know a great way to calibrate the voltage.

The obvious quick and easy way is to compare the A/C with a good meter.  That requires a good meter.  Handheld DVMs are usually good for around 1%.

You can get chips that make a reference voltage.  Higher accuracy or better temperature stability cost more.  Google for LM4132 if you want a straw man.  There are probably cheaper ones.  That was just the first one I hit.

I remember an old tale of using fresh batteries for a reference.  It woiuld be fun to test that.


-- 
These are my opinions, not necessarily my employer's.  I hate spam.






More information about the Devel mailing list