Tony:

I can toss in some thoughts on the subject.

At all costs, remain linear for powering the system.  Switchers are the
harbinger of death in this sort of application.

Strain gauges like to be driven with a current source, such that you
regulate the current through the bridge.  I read in an AD data book that
this compensates for wiring loss and noise (?!).  A bridge can be driven
by an op amp, NPN, and sense resistor (IE 10 ohms) in the bottom.  I
would do ascii art, but it pisses me off.

Investigate the ratiometric method.  A/D's these days are more and more
likely to be capable of rigging the reference to an external voltage.
If you skip the precision reference and link the reference to the bridge
exite voltage, you get better tracking (really).

You must use an instrumentation amp on the return pair.  The old standby
tripple op amp 'electrometer amplifier' configuration requires hand
matched resistors for CMRR and you DONT want to spend the rest of your
life with an ohm meter in your hand and a fistful of resistors.  And
even at that the performance still cannot beat today's cheap IA's in
plastic, like the AD620/621/622.  You might need another IA on the
supply to the bridge to drive the reference on the A/D to be ratiometric
IF you opt for the bridge current regulation in the low side.  Unless
your A/D has ref+ and ref-.  Get all of that?

What kind of performance / resolution are you looking for?  I saw the
new AD7730 sigma delta family in the Analog Devices book, and it blew me
away.  Without pulling off a genuine miracle in layout teqcniques, you
could probably do 16 bit resolution and skip all of the above.  See the
data sheet and you will see what I mean.  BIG bucks, though.

In answer to your original question on supplies, you can simply use
linear regulators, and filter very well with a resistor, electrolytic,
and ceramic on each power feed.  That will clean up to, oh, 5mV or
less.  MORE IMPORTANT THAN THE REGULATORS, though, is to use either a
superb starred ground layout, or even better a genuine ground plane.
When you do, you will begin to see databook performance from your
circuit, and it will make you beam with pride.

G'luck.
Chris Eddy, PE
Pioneer Microsystems, Inc.

TONY NIXON 54964 wrote:

> Has anyone got any pointers to a low noise 5V power supply with good
> temp stability. Mains operated, 1A max O/P.
>
> I need it to power 24 strain guages for data aquisition.
>



Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Christopher Eddy, PE
Content-Disposition: attachment; filename="vcard.vcf"

Attachment converted: wonderland:vcard.vcf (TEXT/CSOm) (00007F96)