M0KGK logo
MØKGK Amateur Radio Pages
Softrock40, last updated 2016-11-30
[ Home ] [ Keyer ] [ PSK31 ] [ Softrock40 ]

Softrock40

The Softrock 40 is a Software Defined Radio kit covering a portion of the 40m band and available from AmQRP (at the time of writing all kits are sold out).

The kit consists of a number of discrete components and a number of ICs, which are of the surface mount variety.  Care needs to be taken with soldering to ensure that there are no solder tracks or spikes between pins.

softrock40-2
SMD side of board

The other side of the board contains the discrete components and three connections which are (1) the USB plug which only serves to provide power, (2) the RF coax input, and (3) the audio output carrying the I and Q signals.

softrock40-1
Discrete component side of board

The I/Q signals are fed into the soundcard and this is where the clever processing of the PC comes in, courtesy of the Flexradio software.

flexradio
Flexradio screen shot (click pic for larger version)

Initial findings

Despite taking great care, there was a solder bridge between two pins on the quad analogue switch.  The software config causing a cyclic thump but both problems were quickly fixed after a run through the FAQ.

Initial results looked good with the only observation that the design centre frequency of 7056.000kHz was actually measured at 7054.430kHz - this offset can be quickly fixed in the software to give accurate readings.  Sound quality and sensitivity were good, here's a short sample of a received signal.  7047kHz LSB .wav file (196k bytes)

40m at night yielded many signals - in fact, too many.  Back to back testing against another receiver showed signals that simply should not have been there.  While trying to listen to the Russian CW beacons around 7039kHz, there was a strong AM station from the Indian subcontinent obliterating any CW activity.  A quick search showed that this station was transmitting on 7135kHz at 40dB over S9, the reception at 7039 was of an image 96kHz away.

I'm guessing the cause of this is decimation caused by the 48kHz sample rate and insufficient anti-alias filtering on the sound card input causing signals above the Nyquist frequency to end up in the baseband I/Q signal.  Some tests were run to establish the extent of the imaging:

spurs
Image problems caused by inadequate anti-alias filtering on soundcard

Shown above is a graph of the response to image frequencies.  The Softrock40 was tuned to 7042kHz with the AGC set to Fixed and a signal generator swept up and down to check the level of the spurs.

The dBm values are read from the Flexradio screen and have not been calibrated although given the severity of the problem, this is arguably academic.  The lop-sided response is caused by the input bandpass filter centre frequency working out a little higher than the intended range of the Softrock40.

23-Oct-2005: As a potential fix, a Delta-44 96kHz/24bit board has been ordered which I hope will yield better performance than the low-rent soundcard built into the motherboard on which the tests were performed.  If this doesn't bring about an acceptable performance improvement, a high order outboard LPF will be needed to constrain the I/Q feed from the Softrock40 to within the Nyquist frequencies.

Delta-44 Soundcard

06-Nov-2005: The Delta card has arrived, and the difference in performance is absolutely astounding.  Click the pic on the right to see the substantial difference in noise performance between the AC97 onboard chipset and the Delta-44.

I was going to repeat the test above with the spurs and overlay them on the graph, but this exercise proved pointless.  At the same 48kHz sampling rate all unwanted images were down in the noise and could not be measured.

Quality of received audio using Softrock 40 / Delta-44 / PowerSDR is excellent.  I did some back to back listening tests on CW and SSB during busy contests between the Softrock and the Yaesu FT-857D which happened to be plugged in at the time.  In all cases, the readability of weak DX signals from the SDR was better.  The signal to noise seemed to be noticeably better too.

The 857 is not graced with the best receiver in the world, so it's my intention to do back-to-back tests when time permits with the FT-990 and FT-101E, both of which are (IMHO) better performers.  Next steps are to get the Softrock into an enclosure and feed with an external supply.

 

These pages and contents are Copyright ©2004-2017 Duncan Munro M0KGK.  All trademarks used herein are acknowledged.  Email any queries about these pages to postmaster@duncanamps.com