SDR-1000 Latency
While I was recording the audio stream from several radio time sources for the leap second last evening (see my Leap Second 2005 page), I realized that the same equipment setup would make it easy to measure the processing delay in the SDR-1000 compared to a traditional HF receiver. All I had to do was record the same signal on both receivers, and then measure the delay in the audio output. The WWV time ticks make a perfect signal source.
I tuned both the SDR-1000 and a Yaesu FT-817 to WWV on 15MHz (fed from the same antenna) and recorded the audio stream. By zooming in, it's easy to measure the delay between the output of the 1-second tick on the two receivers.
I was running PowerSDR version 1.4.5 Beta Preview 4. The sound card buffer size was 2048, and the DSP buffer was 1024. The computer was a 2.7GHz Pentium IV with 512MB RAM running Windows 2000. One quirk is that I was using VNC to control the PC, and that adds somewhat to the CPU load. PowerSDR reported the CPU usage running around 40%.
The capture PC was an Athlon 2000+ with a Delta 44 sound card. It was running Linux and the Audacity sound editing software. Data was captured at 96ksamples/second/ and 24 bits.
Here's a zoomed-in version of the tick on the FT-817.
And here's the same zoom of the tick on the SDR-1000.
The FT-817 tick starts at 3.1133 seconds. The SDR-1000 tick starts at 3.3009 seconds. The difference, which is the SDR-1000 latency compared to the FT-817, is 187.6 milliseconds.