The DSP filtering and detection were revamped quite significantly. Glad to see a sensible effect in real world crowded conditions.
Can you make a few sample dumps in this wonderful period so we can improve the algo further?
3 Likes
Traffic is starting to tail off today, so Iāll grab a couple next time itās busy.
I have an Airspy R2 and a FlightFeeder fed from the same antenna (splitter).
Airspy: SoNic67 ADS-B Feeder Statistics - FlightAware 61996
FlightFeeder (yellow):
SoNic67 ADS-B Feeder Statistics - FlightAware 56483
Are we allowed to talk about the elephant in the room?
airspy_adsb v2.2-RC30
1 Like
I think the more people ask, the more prog wants to stay on RC30.
Also i havenāt done a release in forever, iād have to set up all kinds of stuff to compile all the versions again. Not looking forward to that unless there are actual code changes.
So ā¦ expect RC30 / RC30-test to remain āthe shizzleā!
2 Likes
I see no significant changes for the time being. We have already optimized the current approach to the point the changes only result in statistical noise. Maybe low level code tricks, like SIMD, are the next step? But Iām not eager to go there. Aggressive code optimizations are the best way to stop the algorithmic exploration.
Yeah i think the question was about having a nice round v2.2 name instead of v2.2-RC30 ā¦
If iād get really bored iād brute force 3 sets of filters so you can have narrow, normal, wide.
There are two versions available right now, using either RC30 or RC30 test, but the differences are minimal.
@keithma iād recommend you only use airspy_adsb | sed -e 's/-RC30//'
Thats should fix the version 
5 Likes
Aye, thatās basically it. I wasnāt expecting any more changes because what we have is working incredibly well.
Cheeky
As for my memories, the named wide option worked nicely that time.
After a while, filtering changed a lot and multi-core processing was bornā¦
ā¦now, Iād have a try on the famous wide option - packing the good things together, if possible. (old and new ones too)
RC 30 Test2 (?)
Frankly, itās not worth the hassle. The data-fitted filters will behave randomly with different gain settings and traffic profiles.
You two can decide on reasonable development directions. We all others, who just play with the āend productā just want an āeven better toyā. 
Iām very happy with what weāve got.
2 Likes
mee too
ā¦as a forever kid, I want an even better. (always and right now) LOL
Wow, what a thread. Spent the last few days (on an off) reading through it. What a lot of information, and great work by Prog and wiedehopf in developing this software (and scripts), and to all of you for testing it and giving feedback.
I currently run 2 feeders off the same aerial, one with the blue Flight aware + stick, and the other with the orange stick. I started running 2 when I got the blue stick, and was using the orange stick as reference, the idea was to then keep the orange as test / reference, but life being life (and lockdowns looking after kids etc) I mostly just left both as is to do there stuff.
Set up is FlightAware aerial, 15m coax (H155 ā about 4.4db loss I think), uputronics LNA, Minicircuits splitter, then into either the blue or orange stick each plugged into a Pi 3B (and in the last few days fitted the light blue Flightaware filter in front of the LNA to see if that increased stats, but couldnāt see it made any difference). I live in Greenwich (London) almost under the flight path into Heathrow (due west), and also with a view down into City airport (North East). Aerial is about 10m high (just) above the roof line so generally clear view to North and West, but because I leave on a hill (slope) the roof of the houses to the south are slightly higher, and I have shooters hill to the east which blocks range.
My Flightaware stats are here:
iandrews ADS-B Feeder Statistics - FlightAware - Orange dongle
iandrews ADS-B Feeder Statistics - FlightAware - Blue dongle
Though I am (much) better than most around me, it seems I could do better, eg caius has about 500 more planes then me (about 50% more positions), and reidy0951 has maybe 100-200 more planes, (but about 30% less positions).
Looking at (one of) Keithās sites it seems he gets loads more planes further out than me. EG over north France, when these screen shots were taken he had about 326 planes (304 with positions), and me 197 planes (186 with positions). Now, I think he has a very good aerial position, and very clear view to the horizon, (so I would never get near to his), but I am hoping I can get some more out of mine.
Keith.
Me.
Iāve just ordered a Airspy mini (hence reading this thread), and by the sounds of it Iām hopeful it may up my plane count (and positions). Though I think it would work ok on the Pi 3B I am also planning on using a new Pi 4.
Any thoughts on whether the Airspy will help increase planes / positions, or any other advice. I think it would help if I could get the LNA near the aerial, I canāt get up to it though as it was put up when we had scaffolding up (just hope nothing goes wrong with it!!!), so if I was to cut existing wire and fit SMA connectors I could maybe get the LNA either at the bottom of pole (about 3 meters of wire from aerial) or maybe near the bottom of the wall (about 7 meters of wire from the aerial, would that be worth it?), but itās taking a big risk that if anything goes wrong I probably wont be able to replace wire to the aerial!?!?!? Think I will just wait until the Airspy is here and see what improvements (if any) it makes.
To add to this. Looking at my range plot on the stats page shows no positions at the outer ring between East and South which is probably the hill / roof lines, but also nothing for both the outer and next one in in the South West direction. Was just thinking that Crystal Palace TV transmitter was in that direction and wondering if that was causing lost signals, and then just saw this in another post:
However - if you have an Uputronics LNA, be prepared for the merest sniff of RF on VHF/UHF to totally flatten it.
So, as I have the Uputronics LNA is there maybe anything I can do if the TV signals are interfering.
In my experience of the Uputronics LNA, when I say it gets flattened, I really mean that the received signals drop down to complete zero in a fraction of a second. I suppose itās possible youāre being desensed from Crystal Palace but itās not the same symptoms I experienced at multiple sites with that LNA.
Aye, Iām basically very lucky being where I am, itās a fantastic spot for air traffic.
All signals are merged at the antenna, interference canāt block reception only in a certain direction, it will block reception regardless of direction.
2 Likes
Yes you will almost certainly see a dramatic increase in messages and aircraft. The airspy is far more capable than the rtl dongles.
Iāve never see more than 1800 msgs/s with 1600/s being more typical. The airspy regularly gets around 3000/s. Iāve seen over 400 aircraft simultaneously with the airspy.
I think waiting to see what performance you get before cutting the coax is probably a good plan. You might find itās not too much of a problem, though getting the amp as close to the antenna as possible is ideal.
Sell it and get the rtl-sdr.com triple filtered one instead. The filtering on the current version is not as tight as the original, but it should still be good enough. Iāve been using one for a couple of years and itās performed very well. It also has higher gain than the uptronics amp.
I suppose you could put a narrow band filter on the input of the amp, but it would probably cost you as much as the triple filtered LNA and wouldnāt be guaranteed to help.
You might find this thread interesting for seeing where your signal is getting blocked if you havenāt seen it already: Signal Strength Heatmap
@caius @keithma
If adding the filter in front of the uputronics doesnt improve stats, the uputronics is likely doing just fine.
With the coax already attenuating things, overload is less likely.
Oh, I agree. This is not the case of the Uputronics slamming the door closed at the whiff of RF that Iāve experienced.