Well i haven't explained it quite correctly i guess.
Its actually a 20mv signal from a transducer which has to be brought at a 5v level so i will amplify it 250 times.
Now this 5 volt signal has to have a measurable voltage as low as 100µv.
100uV resolution on a 5V signal is a BIG project, layout and design will be absolutely critical - noise and layout problems can easily swamp the lower bits.
Yes it is a big project and thats why i need some inputs.That was a good starting point Nigel but i really hope that you may post a link or something of this sort.If someone has already been through then it will be that much easier for me.Anyways comments like these are welcome
Thanks and regards
100uV resolution on a 5V signal is a BIG project, layout and design will be absolutely critical - noise and layout problems can easily swamp the lower bits.
You could use programmable gain amplifier and increase the gain if the signal level is too low.
This will give you a lower number of volts/bit at lower signal levels and a higher voltage/bit at higher signal levels.
That is a practical, cost effective, solution that I have seen in the real world. Often, only two gain levels are necessary and you can switch between the outputs of two cheap, fixed gain op amps with a cheap cmos analog switch. In that way, you can avoid compensating for output offset voltage changes, and other issues that come with changing the gain of a single amplifier in precision applications. Of course, in high speed applications that method is problematic due to the switching time and recovery time for the higher gain amplifier to come out of saturation.