LimeSDR XTRX: narrow internal spurs in RX spectrum — can they be removed?

[Ben_Gant]

Hi all,

I’m seeing narrow spurs in the RX spectrum of a LimeSDR XTRX (FW 4, HW 2, PCIe /dev/limepcie0) with the stock gr-limesuiteng source block. Settings as in the attached screenshot: LO 1585e6, sample rate 122.88e6, LPF 20e6, NCO 0, antenna “auto”.

The attached spectrum (FFT 32768, Blackman-Harris, 200 averages) shows a noise floor around −110 dB with a DC spike and narrow spurs at roughly −46, −31, −15, +12.5, +31, +46 and +61 MHz offsets, around −90…−95 dB. They are present with nothing connected to the RX input, their offsets don’t change with the sample rate (tested 50 / 90 / 122.88 MSPS), and RX DC/IQ calibration doesn’t affect them. I have an IQ recording of this and can share it.

Is there a way to get rid of these spurs, or is this just how the board is? Any known settings or tricks would be appreciated.

Thanks!

[andrewback]

Tagging @Karolis.

[cryo]

+1 for this fix

[N0YKG]

The fact that the spurs are not symmetrical around the center frequency makes me wonder if they are a sign of IQ Imbalance - have your checked the calibration?

[aardric]

I ran the example using the wide LNAW for the Antenna port of the source and observe much the same asymmetrical spectrum as presented above. Setting the input shorting switch of the LMS7002M 0x010D[1]EN_INSHSW_W_RFE results in the symmetrical spectrum attached suggesting that some of those spurs are actually RFI picked up by the board which would not surprise me given that the board is mounted in a slot within a noisy PC enclosure.