+- Blitzortung.org Forum (https://forum.blitzortung.org/mybb)
+-- Forum: Public Forums (https://forum.blitzortung.org/mybb/forumdisplay.php?fid=29)
+--- Forum: General Discussion (https://forum.blitzortung.org/mybb/forumdisplay.php?fid=31)
+--- Thread: Efficiency (/showthread.php?tid=1261)
RE: Efficiency - cutty - 2017-10-31
(2017-10-31, 10:02)pasense Wrote: Cutty, would it not be reasonable to discuss which tools are missing to enable the improvement of the station's perrformance?
Why of course it's reasonable! More than reasonable, in fact. Advanced tools may be necessary in many situations! I'm sorry if there was an inference NOT to!... I was merely suggesting the 'similar analog' of the compression artifacts, and suggesting that the basic 'computation' possibilities not be over analyzed for a specific isolated situations... that'll always be the case. I just like to keep in mind that not every body will understand 'waterfalls', or know how to use a scope or spectrum analyzer, etc, and the controller doesn't have a whole lot of memory available to 'install' certain 'tools' on board...
RE: Efficiency - micha.d - 2017-10-31
(2017-10-31, 07:39)pasense Wrote: ...
Currently, only a few percent of the signals my station sends are used in strokes by the network, but looking at the signals I am quite sure that almost all of them are from real sferics. ...
I have a similar feeling but I'm not so sure what would be the necessary measure. I see many signals that are very sharp and I monitored the live view when a lightning was close by. While having the audio beep to announce a signal and the map with the lightning circles in parallel, I think that a lot of signals should have been from the lightning. Nevertheless the map did not draw a line or dot for my station and the statistic told it was not used.
So my problem was not only to reduce the number but to find the reason why obviously some good signals were not used. So, I repeat my idea that restarted all the discussion after a long time. I would like to see WHICH signal was used and WHICH NOT. For signals that were not used, it would be great to know how far away they were from a good signal. Was it some µs e.g. due to filters timeshift or reflections or was it at least some ms because it was not releated to a sferic at all. Even for signals associated to a sferic the accuracy in µs or meters would be helpful.
And if the data is available additionally a histogram would be a good statistical method to sum up the behavior of one station. That should show the time deviation in µs and the amount of signals used or not used. An ideal station would then have a very sharp peak of signals used and nothing around it. A station with reflections e.g. from hils or buildings close by would have a broader curve, where only some signals were used. And a station with a slow filter would have something like a shift by some µs leading to many not used signals.
RE: Efficiency - mwaters - 2017-10-31
In order to register a stroke, both H-field antennas must sense it above the thresholds.
I see many signals here --that are obviously lightning-- that are not registered as a stroke for that reason. Only red, or only green, (and not both) pick it up the signal.
And it does not matter what any third H antenna (mine is a loop parallel with the earth) OR what the E-field antenna receives.
BOTH H-field antennas must sense it! Otherwise, it will only be a signal, and not a stroke. Some of the most successful stations in region 3 only have two H-field antennas.
(Note that stations with one H and one E antenna all have negative efficiencies! And there are many of them.)
RE: Efficiency - pasense - 2017-11-01
@mwaters,
I tend to disagree with your statement. If the two H antennas are oriented at 90 degrees and one sferic arrives that gives a maximum signal in one channel, then the signal in the other, perpendicular channel would be zero, but a stroke should be registered. Just at the moment, my station is receiving a lot of signals from a direction such that only one H channel gets a signal, and looking in the list of registered strokes a lot turn up with only one H channel. Just have a look at the image below.
Of course there are others (second image), where all channels are involved, but both H channels are not required.
RE: Efficiency - pasense - 2017-11-01
@micha.d,
I just checked the strokes where the signals from your station were involved, and it seem that it is working well. I did this by looking at the list of strokes in the statistics page. I do not know of any other way to check what strokes used the signals of a particular station. As I said above, some more information would be helpful, but it is also possible with what its available, but it takes some time.
The only comment I have regarding your station: it seems you have connected the two H field antennas to channel B and C; it would be better to shift to A and B. You can see that from the color of the traces in the signal plots below. I copied these from the list of strokes where your station was involved and they show typical signals.
RE: Efficiency - readbueno - 2017-11-01
Hi Pasense, I agree, I had my station, whilst I was in the initial set up, with only the "E" field functioning. I also set up the "H" channels individually.
I was still registering strokes and was able to optimise each channel, without the confusing array of squiggly lines that are present, when all the channels are activated at once.
Having said that, whilst in your first example, the Green signal is very clean and well defined, in your second example you have far too much gain!
The Red channel is flat topped and chopped and even the Green and Magenta channels have signs of distortion products.
Even in low noise situations, I believe that gains over 10 are very rarely needed and for signals within 2,000 km range, gains of 8*5 or even 5*5 will produce excellent results.
Reducing the gain actually makes the system more sensitive, because you are not amplifying the noise floor as much and therefore have a better signal to noise ratio.
I have made this suggestion to several stations now and they will confirm that their stroke rates and their locating rates are now much more efficient and their effectivity is more constant.
I know that you are in a special situation due to your location, but I would like to suggest that you try this for a few days and compare the results between various levels of lower gain and higher threshold levels.
You may be surprised and if not, it is only the matter of a few seconds to return to your present settings.
Respectfully yours,
Brian.
RE: Efficiency - pasense - 2017-11-01
Brian,
you are right, the signals go in saturation for these strokes, but these were actually the strongest signals I have seen so far, so I copied the signal traces. I am just experimenting in order to see if and how I can get also polarity information from strokes over the ocean (which is supposed to be difficult or impossible), but yes, I could reduced the gain of the H channels. Coming back to useful plots we would like to have: a histogram of maximum signal strength, both for signals sent and for signals used in strokes would make it easy to spot such situations: these would just go into the "overflow" bin. So far, I have to look at the list of strokes in order to find out if the amplification and thresholds are OK. Such histograms are standard in all situations where one has to adjust such parameters and I am sorely missing these here!
RE: Efficiency - cutty - 2017-11-01
Heh
.... and here's what happens when you have 33 strokes >500 km but a sporadic noise maker active,which does NOT trigger interference mode,....
83335 0 0% 0 0% 33 0%
...sigh.. sta 1439 1300UTC
RE: Efficiency - readbueno - 2017-11-01
Hi All, Stupid question time?
https://www.lightningmaps.org/blitzortung/europe/index.php?bo_page=archive&bo_show=strikes&bo_station_id=14853&bo_datetime_start=2017-11-01+11%3A50&lang=en&bo_strike_id=101230862&bo_other_graphs
Looking at the signals on this page there is obviously a lot of variability in signals even for the same stroke. If you look at the stations chosen for location nine of them are greens, six are reds, and five are blues. Some of the signals seem vastly over-driven, being cropped flat topped and distorted, some with various amounts of noise and some that are barely recognisable as lightning at all?
After looking at several pages like this I came to the conclusion that the criteria for being chosen for the location of strokes has nothing to do with quality of signals, but is just based on the first to arrive at the server. Greens send less information, so they predominate. Blues with H.O. or E.O. Send less info but more than greens or reds so they bring up the rear. there are very few 3 or 4 channel blues represented in this lottery?
Yes, this is all just for one stroke! I have not edited it in any way. The chosen stations are underlined.
My question involves the timing in micro seconds at the top of each graph at a quick scan this varies from a few + microseconds up to several + hundred microseconds, the average being +50-+60 microseconds! Only the fastest are chosen!
Does this timing have anything to do with the decision about whether your signal is chosen for the location calculations? Based on my short observation it seem to.
Secondary question. Is there anything that an individual station could improve on this time delay? Reduce channels and get a better result in the lists
Are there things that we should avoid doing that might increase this delay? Send good quality signals on all channels?
Last question, at least for the moment. Is this timing of great importance or is it the time that has the seconds to nine decimal places that is the important time?
I wish I had studied statistics more when I was young, and I wish there was a quicker way to get all this data from the archives.
Regards,
Brian.
@ Cutty mutual vigilance is the key!
RE: Efficiency - micha.d - 2017-11-01
(2017-11-01, 07:52)pasense Wrote: @micha.d,
I just checked the strokes where the signals from your station were involved, and it seem that it is working well. I did this by looking at the list of strokes in the statistics page. I do not know of any other way to check what strokes used the signals of a particular station. As I said above, some more information would be helpful, but it is also possible with what its available, but it takes some time.
The only comment I have regarding your station: it seems you have connected the two H field antennas to channel B and C; it would be better to shift to A and B. You can see that from the color of the traces in the signal plots below. I copied these from the list of strokes where your station was involved and they show typical signals.
Hi pasense,
I have read that channel A has more noise due to the GPS being close by. So I switched to B and C for testing. I now feel that it does not make much difference, at least with my antenna getting some interference from outside and NOT over the PCB itself. Even soldering a shield into the device did not make a difference:
But regarding the evaluation of separate channels I have some doubts. I'm also sure that single channels are used for strokes but signals coming in the direction of one of the two antennas seem to be prefered because there is a better chance to exceed the threshold. I think I can see that effect in one of the diagrams (https://www.lightningmaps.org/blitzortung/europe/index.php?bo_page=statistics&bo_show=longtime&lang=en&bo_station_id=14731), but it's most probably still to few data. My antennas are oriented in direction north-east <-> south-west and south-east <-> north-west and I think the green ratio is better in those two directions. To remove this effect we would need some info on the direction of the single antennas to calculate the direction of the signal source and effective signal of that direction. In my next live I'll go for 3 antennas to reduce this side effect.
EDIT: Before someone complains. I mixed up some numbers. The link above is not my station. But I stick to 3 antennas in the next live. Back to efficiency ...