Fourier image frequencies

[zappedagain]

I have a sensor signal where the Fourier transform is showing 'images' of the baseband spectrum repeated at higher frequecies. In the 1024 point frequency plot I'll see the baseband 'image' repeat around the 102, 204, 306... points. What are possible causes of this?

I know it isn't aliasing of the actual signal because several lowpass filters have been tried with no improvement, except for the noise floor going down as the bandwidth narrows. I've even replaced my original signal with a signal from an RF source and the same thing happens.

I suspect I may have distortion along my time axis; Due to the constraints of the system the points are not collected at a constant sample rate. Each sample gets logged with a timestamp and then the time data is processed into evenly spaced 'bins'. Is residual error in the binning process causing this phenomenon?

I also noticed that the images have varying amplitude, for example the image centered at bin 102 might be -30 dB down, the bin at 204 is -20 dB down, and the bin at 408 is -35 dB down. I expect this amplitude variation is showing the harmonic content of the distortion wave.

Does anyone have any other suggestions? Any insights into possible causes for this are appreciated.

 

[melone]

Sounds like harmonics of a baseband frequency (102Mhz). Try placing a decoupling cap that is tuned to that funadamental as close to the load(s) as possible.

 

[zappedagain]

Ah, it was introduced by the 'binning' process. Narrowing the bins shifts the images to higher frequencies.