The SDR32, a STM32 based software defined radio

[danfo098]

So I started working on a STM32 based software defined radio and so far I've come up with a block diagram and some very rough calculations for the specifications. The idea is to make a fairly simple but still quite competent SDR for experimentation with various digital modulation techniques.

Some basic specs:
0.1 to 6 GHz transmit frequency range
1 to 600 kHz programmable Tx/Rx bandwidth (although the DAC will probably limit the max Tx BW somewhat)
0.1 to 4 GHz receive frequency range (couldn't find a 0.1 to 6 GHz IQ demod)
USB 2.0 interface to PC
More specs in the attached image.

Check out the block diagram. Questions, comments and suggestions are welcome 

[ve7xen]

Nice looking design. Are you planning on implementing the baseband processing on the STM32 itself, or via the PC interface with typical SDR software? How is the real-life performance of the STM32 ADCs/DACs? There's certainly enough DSP power on the STM32 to make a fully self contained SDR at least for simple modes, which could make the basis for an interesting QRP rig.

[danfo098]

I'm mostly interested in and planning on doing all the processing on the STM32 but the USB2.0 chip is there for the possibility to connect it to GNU Radio.

The transmit will be fairly simple (for simple modulation types) just lookup tables in RAM with the waveforms and using the DMA controller to send them to the DAC, a minimum of CPU cycles will be needed for Tx. The limiting factor will be the DAC settling time for higher bitrates I think. Generating a full swing ~100 kHz signal with 180 degree phase shifts is probably the upper limit for the DAC which gives a maximum Tx bandwidth of 200 kHz.

The receive will be quite a bit more CPU-intensive. It will be interesting to see what performance (bitrate) is acheivable for various modulations (BPSK, QPSK, FSK etc) in realtime with the STM32, probably not very high... The SDRAM could be used for buffering bursts (8 seconds of sampling memory) and doing the processing in non realtime.

The ADC/DAC is probably not great performance wise for this application but I wanted to keep the component count to a minimum for a simpler design so that's why I didn't go for external DAC/ADC's. The ADC max sampling rate is 2 MSa/s and so decimation could be used to improve the ADC specs below.

ADC typical specs:
ENOB 10.8 bits
SINAD 67 dB
SNR 68 dB
THD -72 dB

[ve7xen]

Came across this while poking around other microcontroller-based SDR designs, seems they have PSK31 working at least in a similar MCU: http://www.stm32-sdr.com/

Cortex-M4 at 100+ MHz should be plenty powerful enough to implement most modes I would think, especially if you make use of the DSP functions. I'm not a signal processing guy though so maybe it's harder than it looks. Codec2/FDMDV developer VK5DGR is also working on an STM32F4-based modem for that digital voice protocol, would be neat to see that directly integrated in an SDR.

Subscribed, this project is up my alley and I hope it goes well.

[mathias]

Any particular reason why all of the RF components are from Hittite?

[danfo098]

I was not aware of PSK31 (I'm not a ham) but after checking the specification I think the STM32 could handle it easily as is already proved, however when the bitrate is 1000 times higher I think the STM32 will start sweating. The STM32-SDR seems interesting, might be able to borrow some code from that project. Codec2/FDMDV also seems interesting.

Regarding the Hittite components I've used them before and I like them and for this case they had a nice selection of wideband components that fitted this project. Suggestions for better components from other manufacturers are always welcome though.