I was interested in Stefan's remark about needing AC 'as the 7730 has some unpredictable drift'. I've been in weighing far too long, and have long ago dismissed AC excitation in favour of ensuring all 'pairs' of thermocouples in the signal circuit are in close thermal proximity. I.e. Every change of conductor material in the chain is identical and mechanically close. I even go to trouble of mapping 'isotherms' across the PCB from the heat generating components (power regulator, etc) and ensuring the ADC signal input components are equally 'heated'. This frees me from the problems (manifold) of AC excitation.

I don't use the 7730. I was already on the 'Crystal Semi' route when the 7730 became available. I used the CS5520, and lately the CS5532 (last 10 years), but it does have problems, and I keep wondering if I should change horses. Major problem is LF gain instabilty - >0.1Hz - which of course walks through my filters. It is in the order of 1:20,000, so it is fine with my 1:10,000 certified (OIML 76 approved) weight indicator designs, but it is very annoying, nonetheless. It varies from device to device, and batch to batch, so I've long put it down to an on-chip effect - and it often DOES exceed the datasheet spec. Just not enough to make a big deal of it with the manufacturer. Therefore, I'm very interested when someone tells me the AD device may have a similar problem (perhaps it's inherent in delta-sigma conversion?).

Stefan - tell me more.

And Mohamed, you mustn't add a pre-amp - it'll just magnify your problem and add some of its own. Also - try to avoid a resistor chain in the excitation - why add further, very expensive, gain-error-inducing components? Although, I don't know what the 7730's ref input limits are... The CS5532 has 5v, which is great - but I discovered that it goes non-linear immediately you go north of 5v - so I give it 4.9v ;-)

Dave