Re: Current measuring resistor calibration

On Sat, 27 Jun 2009 08:28:03 -0700 (PDT), MooseFET
<kensmith@xxxxxxxxx> wrote:

On Jun 26, 8:36 am, John Larkin
<jjlar...@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
On Fri, 26 Jun 2009 07:30:05 -0700 (PDT), MooseFET

<kensm...@xxxxxxxxx> wrote:
On Jun 26, 7:12 am, John Larkin
<jjlar...@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
[... stuff then diffamp ...]

Get rid of the home-made diffamp and use a CMOS 24-bit delta-sigma
adc, AD7793 maybe. They have differential inputs that are essentially
perfect, programmable gain ranges, nanovolt drift, and are linear to a
few PPM. And add a signal-level mux to get the switch resistances out
of the measurement path. And get some good resistors. And you're done.

Better yet, run the whole analog and ADC process on an isolated supply
that has a point on the measurement circuit as its common.  This way,
the common mode is only equal to a little drop in some wiring.

Yup, it's not hard to float a serial adc. Analog Devices will sell you
a single part that will provided isolated power and isolate the data.


The OP will be working with fairly low voltages. He is also workin at
low frequencies. He may gain some advantage from a flying capacitor
voltage multiplier before the ADC.


In+ --o o-----+------+---
\ ! !
! ! !
=== ! !
! === !
/ ! !
In- --o o-----+ !
! ===
In+ --o o-----+ !
/ ! !
! === !
=== ! !
! ! !
\ ! !
In- --o o-----+------+---


If charge injection doesn't trash the accuracy. That becomes another
tricky design issue which I'm guessing he doesn't need.

A 24-bit delta-sigma is a huge amount of precision, already worked out
by experts, for sale cheap.

John



.