we are battling noise issues of the WT subsystem. we do have several noise generators on the same board which is the source but I can not disable them during a measurement so we are trying to mitigate the noise at the AFE.
In order to track down the noise, I've shorted channels 3 and 4 and set the gain of the first stage to 81 with a 2.5K R. the gain of the second stage is 4 so the total gain of the first test is 324. a near by resistor divider sourced from VLDO bias the short to near mid scale
both channels are sampled sequentially in single conversion mode with the DAC = 0. I then sample and store 30 seconds of 4 channel data into memory until the test is complete and I offload the results. with this gain, I see about 10 counts peak to peak of "fir" on both channels 3 & 4 (-1360 base line).
When I increase the gain by 2.5 (1KΩ) and no other changes, I get a base line value in the -1310 range and fir in the 25 count range. this implies the fir is something internal to the AFE. we have the recommended 0.47uF caps on both Vref and VLDO . we need to cut this noise down to and/or up the gain to reach our target resolution of 10counts with a total gain around 1000.
can you recommend anything to improve the AFE's shorted input performance?
Thanks. marty
