I am trying to get the AD-FMcomms4 running with Nvidia Jetson TK1. For now the hardware setup consists of Nvidia Jetson board, Zedboard, and AD-Fmcomms4 such that the SPI lines from AD-Fcomms can be connected to either Jetson or Zedboard (can be swapped by a simple switch) and all other pins coming from the FMC connector of AD-Fcomms4 are connected to Zedboard (Including the lines for AUX ADC). The Nvidia Jetson is using the no-OS driver on a Linux Platform to set the parameters of AD-Fmcomms4 via SPI commands. The initialization command sequence is completely unaltered as given in the no-OS driver. The other parameters I am configuring using the API functions are - Center Frequency (618 Mhz) , ENSM mode (FDD), Sampling Rate (8 MHz), Bandwidth (8 MHz), Gain control (Manual - 40 db). The parameters configured by Tegra are verified in following way :
1) Have SPI lines from AD-FMcomms4 connected to Zedboard and open the GUI (IIO Oscilloscope).
2) Observe the FFT waverforms when HDTV receiver antenna is connected to FMcomms4 board.
3) Swap the SPI lines to Tegra, and configure the parameters.
As the ADC lines are still connected to Zedboard, we can view the FFT waveforms with these new parameters. I have attached the waveforms after setting above mentioned parameters using Tegra as well as Zedboard GUI. The waveforms obtained after setting parameters from Jetson seem to have unintended frequency components mirrored around center frequency, from the original components. Example : for LO Frequency 618 Mhz, the signals at 614.311 MHz and 620.311 MHz are the desired ones. However in case of Jetson, we also have significant signal components at 621.688 MHz, and 615.688 Mhz. From calculations, it is clear that 621.688 MHz component is mirror of 614.311 MHz about the center frequency and similar issue for 615.688 Mhz - a mirror of 620.311 Mhz.
I have been looking through the set of parameters that might effect this, but haven't got any success yet. Any suggestions or did someone come across similar issue ?