RedPitaya comparison 125-14 bit vs 125-10 bit

There are two version for the RedPitaya on the market:

  1. STEM 125-10 (10-bit ADC/DAC)
  2. STEM 125-14 (14-bit ADC/DAC)

The 14-bit version is roughly 20% more expensive compared to the lower resolution version. In case one wants to operate a  dedicated RedPitaya 24h/7d with SDR-WSPR, thinking about the difference in performance with repsect to receiving more wspr messages, is probably a good idea.

The WSPR protocol is capable of decoding messages up to a signal to noise ration (SNR) of -28 dB (Wikipedia) with respect to a 2500 Hz reference bandwidth. Nevertheless I regularly get decodes till -31 dB.

This raises the question, if a 14-bit SDR has an advantage compared to a 10-bit SDR. The strong signal should not be influenced, but the weak signals might be lost, by not being able to resolve the antenna signal sufficiently. To my understanding, even a 10-bit ADC should should have sufficient dynamic range, but environmental conditions i.e. noise / strong other signals can limit the usable range.

To figure this out experimentally, I used a T-splitter after my DC2PD pre-amplifier and connected a 10-bit and a 14-bit RedPitaya. Both run the SDR-WSPR app and submit the decoded messages to the WSPRlive database. The only difference is the comment field, which hold on one side ‘Wdm-40dB’ and ‘Wdm-40dB-10bit’ on the other side. The ’40dB’ indicate that the pre-amplifier was set to 40dB gain and ‘Wdm’ that I use a 42m Windom antenna. This configuration was kept for the period 2017-05-26 to 2017-06-05.

My first impression was: no difference 🙂

One way of quantitatively comparing both receivers is to plot the histogram of the reported SNR during that period:

Comparing the reported signal-to-noise ratio of decoded WSPR meassges for a 10-bit and a 14-bit RedPitaya

As one can see, the 10-bit version performs pretty well. Also the -31 dB messages are decoded equally well by the lower dynamic range receiver. The full dashboard is called WSPR SNR and can also be used to compare the SNR statistics of different reporters.

In my opinion, using the 10-bit version in my setup works without significant difference, because I use a pre-amplifier that shift the signal into the upper bits of the ADC. Without it there might be a noticeable difference.

73 de Clemens / DL2ZZ / PA7T