Page 1 of 1

20dB nonlinear response on RSP2 ?

Posted: Mon Jan 09, 2017 5:13 am
by razvan

I am puzzled by the very non-linear response that I observe on my new RSP2 unit.
While it varies with the sampling rate, the extreme case is the 4MHz sampling rate, where it displays a 20dB (!) variation between the central 1600kHz and the outer parts of the band. It looks pretty horrible to me and in practical terms it prevents the simultaneous monitoring of let's say, 80m and 40m bands, as I'll describe below. (Attached are the snapshots of the screen including all the relevant controls)

Is this normal ?!?

sdrplay4mhzNoAntenna_1of2.jpg (185.53 KiB) Viewed 2056 times

sdrplay4mhzNoAntenna_2of2.jpg (101.95 KiB) Viewed 2056 times

I received the unit last week and I connected my 40m OCF type antenna to the high-Z port. I've seen various levels of noise as I scan from 1MHz to 30MHz, as my antenna resonates, but for what I describe below the antenna is disconnected to eliminate its influence from the picture.

With nothing connected to any port(!) -- I observed this unevenness of the received (noise) spectrum at pretty much all settings of the RSP2: from 2MHz to 10MHz and various decimations, with various degrees of nonlinearity. With the exception of the 10MHz case, all other settings show more or less abrupt cuts at the edge of the spectrum, shortening the advertised total bandwidth by 500khz-1mhz. In some cases (5Mhz and above, I'd say) the flatness of the spectrum is reasonable, but for 3mhz and especially for 4mhz sampling rates, the device response can differ by -20dB outside the center. Observe the 4mhz case, in the figure, where outside the center 1600khz the response drops by 20dB, to progressively drop to nil 1mhz farther.
Signals will be modulated with this response function, impairing substantially the reception outside the central 1600khz.

As a practical case, I tried to simultaneously monitor the 40m and 80m bands. In order to keep the cpu load under control and more importantly to obtain the best spectrum resolution, one needs to keep the sampled band to the narrowest setting -- in this case 4mhz. At this setting the two VRX will sit at the extremes of the sampled band, where the response is severely suppressed and consequently one will receive nothing, as I observed directly. Move one VRX into the center 1600khz portion by shifting the LO and you're in business -- unfortunately that eliminates the possibility to watch 2 bands simult. at this SR setting. (my CPU won't bear more than 4mhz SR with two VRX at 32k FFT resol. without skipping, so... )

So, back to the 4mhz and 3mhz SR response nonlinearity - what is this all about?! Do I have a defective unit?
(I can post more figures, with the response at other sampling rates, if necessary)



Re: 20dB nonlinear response on RSP2 ?

Posted: Mon Jan 09, 2017 9:14 am
by Tech_Support
Hello Razavan,
This is entirely normal and not a fault. If you look at the top left hand corner, the IF tuner bandwidth is selected as 1.536 MHz, so the roll off in the noise response, is entirely due to the tuner IF filter. The software selects the nearest IF bandwidth that is lower than or equal to the sample rate selected. It is important that the IF bandwidth should be less or equal to the sample rate to prevent spurious signals due to aliasing. The tuner used in the RSP2 as selectable IF bandwidths of 200 kHz, 300 kHz, 600 kHz, 1.536 MHz, 5 MHz, 6 MHz, 7 MHz and 8 MHz. If you increase the sample rate to 5 MHz, you will see that the spectral response will look completely flat.


SDRplay Tech_Support

Re: 20dB nonlinear response on RSP2 ?

Posted: Fri Jan 13, 2017 3:22 pm
by razvan
Thanks for the prompt and clear response.