Help Needed with LT1606

[europus2002]

Hi all ,
Does any one have experience playing with LT1606..16 bit ADC from Linear Tech... The question i have is for adjusting the reference so that my fullscale swing is +/- 2V. The data sheet does not affirmatively say anything about it. It just says that Full scale voltage is 4 x Vref... Can i load the band gap reference (internal and protected) so that it generates 0.5 or 1V reference so that my Full scale voltage can swing only upto 2 to 4V...Any other suggestions welcome ..I have to use 16 bit ADC capable of sampling above 200Ksps with BYTE interface capability......Thanks in advance..

 

[Skogsgurra]

Yes, you can force the reference voltage to a low 0.5 V, but I do not think that you will gain anything. Your reference voltage will probably be noisier than the internal reference. Do you want this to get all 16 bits in the result? Try using an external amplifier instead. That, of course needs a bipolar power supply.
Another way is to use averaging. That will give you any resolution you want.

Gunnar Englund

http://www.gke.org

 

[europus2002]

Hi skog, thanks for replying. You are absolutely right and the output of this ADC is to be taken for various analysis. I have designed an AGC (Automatic gain control)stage that ensures the signal amplitude peak just around the peak value taken by ADC for best resolution and this whole system is battery operated, so my voltages are down to +/- 5v to keep the consumption as low as possible.But it looks like i will have to increase the swing !! . Since every sample taken is equally important spectrum wise, i can't play much with signal.... Could you tell me more about this averaging ? Not sure if you are referring to averaging of signal in digital domain after conversion or average signal in analog domain?....

 

[Skogsgurra]

Average in the digital domain. You can either do a moving average over n samples or do an IIR filter by calculating k*x(0) + (1-k)*sum[x(-1)] and scale to suit. x(-1) means earlier samples. The k sets the degree of filtering. A small k - say 1/16 - will make a lot of filtering. The latter technique is very much the same as using a first order low pass filter in analogue domain.

Gunnar Englund

http://www.gke.org