Reply by Rob February 3, 20082008-02-03
Hi,

I've spent some time reading about oversampling and I think I understand 
the fundamentals.  Oversample by some factor > desired bandwidth, 
decimate and increase ENOB (with certain assumptions about dithering, 
etc).

My signal of interest is slowly varying (1-2 Hz), and I'm using a 10-bit 
ADC, oversampling at 10 ksamples/sec.

Two questions:

1.  For oversampling, some documents talk about accumulate and decimate 
("dump") where after accumulation, LSBs are chopped off and the new 
sample has increased number of bits.

Do I still need to low-pass filter after the decimation?  Some documents 
mention the low-pass and some do not.

2.  Is a moving-average filter an acceptable low-pass if the answer to 
above is "yes"?

I understand the MAF will only help reduce noise, but will not help me 
with ENOB.

Thanks!
Rob.