I am working with FMCOMMS4 and using No-OS drivers (of course with my modifications in both HDL and Software). I am also using an Rx FIR filter with decimation factor of 2.The filter and sampling frequency are configured so that the usable bandwidth is nearly 53 MHz with IQ sample rate of 60 MHz. I have the following two questions :
1) Although the entire 53 MHz is usable, the noise floor is not perfectly flat over the entire bandwidth. When I use my custom average detector, I see that the noise floor is uneven, as in it is 10 dB more near the edges as compared to the centre. This is not acceptable for my application as the spectrum looks quite weird (see attached figures). So, can someone tell me what's the source of this and how can this be addressed ?
2) There is also a considerable DC spur even when there is no signal in that specific band/region. So, this is something introduced by the receiver stage. The DC spur in most bands is 4-5 dB more than noise floor, however, sometimes it could be as high as 10-12 dB (see attached figure 1). How can I solve this problem ?
Any help would be greatly appreciated ! Thanks !