[volt-nuts] 3458A calibration - linearity
Dr. Frank Stellmach
drfrank.stellmach at freenet.de
Tue Aug 11 06:55:50 UTC 2009
Predrag,
Please download the HP Journal Vol 40, #2 (April 1989) from here:
http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1989-04.pdf
The multislope rundown conversion is explained in detail (p.8 ff), also
why its linearity is nearly perfect.
The main negative effect of multislope A/D conversion, which is
cancelled in the 3458A, is the charge injection by the analogue switches.
A strictly symmetrical switching pattern is used for every possible
situation. So the charge injection effects are cancelld out.
The external passive components (R,C) are relatively uncritical compared
to this effect.
But besides the linearity of the A/D converter, there is another
nonlinearity in the 3458A.
On the 1000V range, a quadratic nonlinearity over increasing input
voltage is introduced by dissipation effect in the 100:1 divider.
Therefore, up to 12ppm additional error has to be taken into account.
That's a big disadvantga of this AUTOCAL technique.
Other longscale DVMs use external 1000V / 500V reference points to
calculate and cancel this effect.
Frank Stellmach
More information about the volt-nuts
mailing list