[hpsdr] SDR Received Noise

[PETER BUCK]

Hi All,
A good topic for all the experts to get out the flame-throwers and fire-hoses! Just some points for discussion here which relate particularly to SDR. Have long been concerned with rising man-made noise on HF - its a favourite topic of discussion as we all well know. I decided to review the shack layout, and at the same time check that there were no obvious noise sources in the shack. The tests below were relatively crude, but I have not seen the issues clearly answered in Ham radio press.
1. Most of my DC powered equipment is run from a battery pack, charged via a transformer fed constant voltage PSU. Equipment fed via switch mode PSU (displays and one laptop) have been exhaustively checked to ensure they don't raise the noise floor in the shack or as detected by an external antenna.
2. Two computers, either of which can drive the radio, were both connected via ethernet switch and Cat 6 cables to the router. Some ethernet switches are great noise generators - mine wasn't but a pick-up coil placed near the switch handling heavy 100T traffic saw the noise background increase by 1-2dB in a 2.4kHz bandwidth. The system was re-wired direct to the router with shorter cables so the switch was eliminated. So no need for Cat 7 cabling?
3. The pick-up coil was then located at the Hermes ethernet port. It was noted that when random band noise up to about 80dBm (common here) was fed via the antenna socket, the pick-up coil collected RF noise from the ethernet port running at 100T of about 3-4dBm above ambient. A similar result was obtained with Metis. Interestingly, a pure tone of same strength fed to the antenna socket gave no detectable increase in ethernet noise. Can/does any of this ethernet related noise get re-cycled in the Hermes circuits? Were test results published related to this potential problem with ethernet? I did similar tests with Phil's QS1R but could not find any similar issues with the USB2 cables.
4. On-air reports of band noise vary very widely. Even if all S meters were calibrated to the same standard (utopia?), there remains the problem of whether peak or average signal strength should be reported, and what is a standard time period for averaging the noise? I have noted that even though Hermes has a much lower noise floor than the QS1R receiver, when operating on bands 7MHz and below using the same mode/bandwidth/agc/no preamp or attenuator/no NR, the QS1R nearly always displays less band noise, but the wanted signals are similar. As expected, with these levels of incoming noise, whether dither and random are on or off makes no significant difference. 
Let the fun begin!
Pete, G3LWT
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20141129/34ef3d2f/attachment-0001.htm>