System Blue - Early Integration
#1
Hi all,

A few notes on the integration approach I used for System Blue - congrats to the team for a great solution with some powerful/easy to use built-in tools to get up and running quickly (particularly the Signals page with noise floor checkbox/noise floor p-p measurements)

H-field:

Started with manual control and E-field input disabled - gain at minimum on 1A and 1B (no 1C connected). Gradually increased the gain while finding optimum antenna placement, rotating the antenna to identify directions of noise sources. After some time I found a location where the noise spikes on the red channel were the only real interference - based on other postings I understand these are generated by the controller and were significantly improved by switching off the HP filter. During early testing I was plagued by significant interference at 120/s (related to TV/dimmer switch 60Hz) but only in the relatively near field.

E-field:

Same approach - manual control with H-field inputs disabled - gain at minimum on 2A. Gradually increased gain while finding a relatively quiet place for the antenna. Grounding the controller board to electrical ground (copper water pipe) improved performance, but the biggest gain was realized by using a dedicated copper grounding rod pushed into the ground.

A couple of questions:

1. I notice that running in Automatic mode dials back the gains significantly from what I can accomplish in manual mode with minimal interference. This is the opposite of what I found during early testing, where the system would get caught in a cycle of automatically increasing gain until it entered interference mode, and then significantly back off the gain (repeats). I assumed that the goal would be to get the max sensitivity from all stations without entering interference mode, but this does not appear to be the case (perhaps based on the increased density of stations)? What is the goal of the algorithm?

2. During testing I had a number of occasions where the controller would stop responding - seemingly related to burst mode interference. External pings would fail, the GPS LED would extinguish and the network LED would remain on despite the network cable being unplugged. Although the problem does not appear to occur during normal (non-interference) operation, I would have expected the watchdog to restart the controller in this scenario. Any ideas for further troubleshooting?

Thanks,
Peter
Stations: 1588
Reply
#2
(2016-09-13, 17:22)pjb01 Wrote: 1. I notice that running in Automatic mode dials back the gains significantly from what I can accomplish in manual mode with minimal interference. This is the opposite of what I found during early testing, where the system would get caught in a cycle of automatically increasing gain until it entered interference mode, and then significantly back off the gain (repeats). I assumed that the goal would be to get the max sensitivity from all stations without entering interference mode, but this does not appear to be the case (perhaps based on the increased density of stations)? What is the goal of the algorithm?
The on-board automatic adjustment tries to adjust the gains, so that there's a visible noise floor so that the thresholds can be set to something around or below 100mV. If the noise floor is not at a constant level within some time period and/or there is interference independent from the noise floor, then the automatic adjustments will fail. In that case I suggest a manual upper gain limit. There's also an automatic amplitude filter which can help in specific problems with interference.

This all is independently from automatic adjustments controlled by the server, which is currently very very basic.


(2016-09-13, 17:22)pjb01 Wrote: 2. During testing I had a number of occasions where the controller would stop responding - seemingly related to burst mode interference. External pings would fail, the GPS LED would extinguish and the network LED would remain on despite the network cable being unplugged. Although the problem does not appear to occur during normal (non-interference) operation, I would have expected the watchdog to restart the controller in this scenario. Any ideas for further troubleshooting?
Is there any sign of activity, like blinking LED or sounds? If not, then push button BOOT1 for ~10 seconds. If you hear a beep, then the system is still running. If not, then the CPU stopped working. A hardware watchdog should always reset the system after a few seconds. The watchdog can only be disabled by pressing BOOT1 during boot. If you have a TTL-serial-USB-converter you can check the serial log in case this happens again.

What's your station id? You should log in on Blitzortung.org and then switch back to the forum tab so that your forum user will be connected to that. :-)
Stations: 538, 1534, 1712, 2034, 2219
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)