On 21 February 2014 04:40, Shobhit Jain <[email protected]> wrote: > Hello all, > > I am designing a 1.25 GHz bandwidth 4096 channel spectrum analyzer using > Roach-2 (Rev-2) and ASIAA Sinica 5GSPS ADC kit. My design is based on tut-3 > and I am using ADC 5g yellow block in my design (1:1 demux, A&C Mode, clock > freq: 1250 MHz). Problem is that I am getting a constant spectrum (which I > don't understand) only when quant gain is very high (0Xffffffff) and no > spectrum for lower quant gain, even if I don't give any input to ADC. >
Do you have any idea what's coming out of the adc in the time domain? > > Also I have some questions, some may be quite trivial, but sorry for that: > > > First, what should be the clock frequency for Roach-2 (312.5 or 156.25 MHz)? > For two inputs, each sampled at 2500MSa/s, the FPGA clock should be 312.5 MHz. The FPGA clock rate (for 1:1 demux) will end up being 1/8 of the value you set in the adc yellow blocks. > > Second, is it necessary to run ADC-tests before running and compiling any > design using that ADC? Because if yes then I have got a result that > calibration test failed. > You need to at least run the adc mmcm phase calibration routine. From an email Rurik sent me about this, you can calibrate with: <> from adc5g import * set_test_mode(roach, 0) set_test_mode(roach, 1) sync_adc(roach) opt0, glitches0 = calibrate_mmcm_phase(roach, 0, ['scope_raw_0_snap',]) opt1, glitches1 = calibrate_mmcm_phase(roach, 1, ['scope_raw_1_snap',]) unset_test_mode(roach, 0) unset_test_mode(roach, 1) Where "roach" is a FpgaClient object and the "scope_raw...." are snapshot blocks hooked up to the raw (but signed binary converted) output of the ADC. The sync_adc here is a good idea since it will align the ramps of all cores of the chip. If either opt0 or opt1 are None then that calibration failed (which can happen for many reasons); in such a case you can look at the glitchesN lists to see the number of glitches at each MMCM phase. <> Also, as Dan says, I've got a single-input 4Ghz spectrometer (or 5ghz with 4-bit requantization, which i'm still working on) with this adc that you're welcome to use/hack if you want. Cheers, Jack > > Can someone please help me with this? > > > Shobhit > > > P.S.- I have attached an image of constant spectrum which I am getting.
