hi tim, i suggest you phase lock your 20 MHz generator to the frequency synthesizer that's generating the adc clock, and setting the input frequency to the exact center of a spectral bin. (that insures that every spectra will have the same exact and integer number of sine waves in it).
for example, if you are sampling at 1 Gsps, and you have 1024 spectral bins over a 500 MHz band, then each spectral bin is 488,281 Hz wide, so you should set your input frequency to 19.53125 MHz, which will be the center of bin 40. best wishes, dan On Tue, Jun 4, 2013 at 3:43 PM, Timothy L. Madden <mad...@umail.ucsb.edu>wrote: > Hi All, > > I am working with the Phil Lubin's Deepspace group at UCSB on a ROACH > based spectrometer. I have been testing tutorial 3 using a 1 Volt > peak to peak sine wave with a frequency of 20 MHz. When set the gain > to a reasonable level it outputs a sharp peak in a consistent > location, but the amplitude of the peak changes quite a bit (on the > order of 10%) between accumulations. I just wanted to know if this > was normal, as it makes calibration irritating to say the least. I am > using a revision A ADC2x1000-8. > > > Thank you for your time, > > Luke Madden >