Hi Dave,
My english is not good, and i'm not sure what You mean ,when
write "LF gain instabilty - >0.1Hz ".
In this area (weighing, micro_nano_volts_DC) i feel myself still as beginner , after 5 years work.
First scale i saw was with AD7715 , first(3000 divisions) we made was with AD7730 and 8051. I have played about year with that drift and problem was "solved" in zero-tracking when low changes are observed . This was incorrect, i know. After first 1000 pcs we change ADC to Texas ADS1240 (price reason). In same time i developed weighing indicator with AD7730 and MSP430.
Then i saw Flitec indicator with AC excitation. I repeated this
and we pass certification for platform scale with 2 intervals -
in that time here wasn't procedure for indicator certifiyng.
Here helps digital filtering too. But my filtering is not good
enough as Flitec's. I'm using digital Bessel low-pass filter.
In last time i'm playing with "cosine" filter , it has some advantages , but is not what Flintec uses. (What filtering
You use , Dave?)

2..3 years after this AnDev's distributor gave to us some
Micrel bridge to use with AD7730. He says this will improve
performance( i think - behavior) of AD7730. I was not tryed, because i allready have AC bridge on transistors. But fact is
AD says about problem 3..4 years after first samples of AD7730,
and as far i know, such info is not published.
We still use AD7730(AC excitation , without software "antidrifting")for platforms and truck scales.
I'm planning to check AD7714 in same place - lower power consumption,lower price.
For retail scales we use last 4..5 years ADS1240 with 5VDC excitation without drift :-) . Last 5 monts we are with success trying retail scales with MSP430F425 - micro wit 16 bit SD, with "autozero" opamp preamplifier.

From long time i know about Crystal, but have no luck to try,
here was not distributor and so on. With AD7730 we get
200..400 nV "legal-for-trade" divison . From datasheets i expect
You have 100nV division. Do You?



regards,
Stefan