I was just saying once you hear the noise floor very clearly it’s loud enough that you don’t need to turn up the volume any more, as you would only hear the noise floor louder and couldn’t possible hear any more signals that you couldn’t hear before.
I have rephrased my last post as well as it was a little rough around the edges
But with a decent antenna and some not too shabby coax i’m sure you are more limited by terrain than anything.
Anyway the rtl-sdr blog LNA is not that expensive so if you want to try it, go for it
Having the LNA not too far from the antenna connected with a good cable is of course important because that influences the relative strength of signals and internal LNA noise.
But i’m pretty sure you are aware.
You could always see if you have equally good reception with a gain of 15.
Then you can be sure you are in the optimal gain band.
While with a typical dongle that band might be relatively narrow and you might need to chose between airplanes close by and far away, with the airspy you should have a fairly wide range of gains that work equally well.
I just did another test leaving the uputronics filtered preamp away. My message rate dropped from 750/s to 570/s and visible planes from 63 to 40. The difference was much smaller with the cheaper Nooelec SDR. So I may benefit from an additional cavity filter right below the antenna and an additional LNA. Or modifying the Uputronics LNA for higher gain…
That’s a strange inference you are making here. Checking with a gain of 15 makes more sense to me. Would be 6dB less.
Could you check how your reception is with gain of 15? I’m curious now
Removing the complete preamp will reduce the signal by about 17 dB right? Also it removes filtering.
So i don’t think that’s a good comparison.
Anyway if you have more than 3 m of coax between the antenna and the preamp, moving the preamp directly below the roof might sure make sense.
I did that test already before and I redid it just now. The rate drops at gain 15. It’s difficult to determine how much exactly because the radio needs time to adjust and in the meanwhile the number of planes may have changed. Right now I had both ~12-14% less aircraft and messages/s at gain 15.
I live in the approach corridor of a small airport. This evening, many planes flew over my antenna at low altitude. I noticed that I lost track of them when they were closer than 1-3 km. Then I reduced gain from 21 to 18 and now I can see the planes all the time.
So I think the best algorithm for right gain is to wait for some aircraft to fly in low altitude over the antenna. Increase gain until you loose track of them (dotted line in the map) but you can see them again once they are sufficiently far away. Then reduce gain by 3 to 4.
Yeah it really all depends on what compromise you want to make.
Many seem to prefer range over local low altitude planes and then it makes sense to use the strong messages as a guideline.
While i get more messages with a gain of 25 i might reduce to 23 again when i get more low flying planes in the summer.
Still a little surprised, i would have thought the dynamic range of the airspy would be enough to use a gain of 21.
Maybe the reduction in range/messages you saw using gain 15 was just a fluctuation in air traffic and not indicative of actually losing anything.
I completely agree. I did not say there is no drop of message rate going from 21 to 18. There clearly is one from 21 to 15. But there are hundreds of receivers catching those high altitude planes anyway. Being able to catch a low flying signal inside my valley nobody else is able to catch is more valuable to me. Even the higher dynamic range of an Airspy R2 is limited at some point. I would need something like a compressor in audio recording studios
Now I have stopped using gain optimization scripts. Instead I do it by trial & error.
On Skyview map’s table, click RSSI column heading to display highest to lowest, and assess how many messages > -3 dBfs
Reduce grain by 3 dB, again check >-3 dBfs. If the number reduced, further reduce gain by 3 dB and repeat. Keep an eye on max range and plane count. It is a trade-off situation, plane count and max range may vary in opposite directions e.g. plane count increasing, max range decreasing. Decide which setting I like the best.
The command only works for dump1090 using a dongle directly.
In the case of airspy feeding dump1090 the command just won’t work because dump1090 won’t count strong_messages. Also the accepted messages will be under a different key (remote instead of local) in the json file.
Can you explain this a bit further? Sure the Airspy has a USB receptable (female) whereas other SDR have a USB plug (male). But this can’t be the reason why the command works or fails. You can connect an SDR like Nooelec using an extension USB cord too.
Anyway I am just curious. I still think the optimal gain setting algorighm should look like this:
Start with a safe medium gain that shows at least some planes.
Log RSSI and distance to antenna for each plane for 24 h.
Search the log for planes that get lost while approaching and reappear afterwards. Calculate the average distance to antenna for each such flight between disappearance and reappearance.
If there is a common minimal radius around the antenna where most (let’s say 75%) of the aircraft disappear then gain is too strong. Reduce gain by 1 (3 dB) and step back to 2. for 24 h.
If there is no such common minimal radius, means 95 % of the closest passing aircraft are visible all the time then gain is probably too low. Increase gain by 1 and run the algorithm again for 24 h.
Don’t change gain if not more than 25 % but not less than 5 % of the aircraft are lost when passing closeby.
The percentages 75% and 5% should be adjustable to the users taste. Who can program such an algorithm?
This is my airspy in my attic. It runs full gain with an external bias-T Amplifier(hab/nevis) with no problems.
I do run a cavity filter before the amp because I have a lot of RF noise.
I find that it is deaf without an amp.
Note that the older airpsy adsb code used a lot of CPU(more than an RPI3 could handle). My setup above uses an Odroid XU4. The new code runs fine on an RPI3.
You can check only for something to parse json files so it’s a little bit easier to read.
Basically my command looks at the last section (total) of that file.
When dump1090 connects itself to a dongle via USB it needs to do all that work. Messages derived like that are considered “local”.
Now if it gets the messages from airspy_adsb for example the messages are considered “remote” and it’s programmed not to count strong_messages the same way.
All the signal diagnostics were not done for “remote” messages because dump1090 for example doesn’t know the signal level between messages. A count of strong messages would still make sense when fed with beast data as those contain a signal strength indication. But it’s just not programmed that way.
Does that make sense?
Anyway have you tried 19 or 20? As you probably have read those gain steps are actually 3dB per step.
Thanks for the json advice. All common SDR are connected via USB, they just have different vendor/product IDs. It looks like the different receiver-type in /boot/piaware-config.txt creates the different behaviour. Anyway,
I think my algorithm is better because it actually counts the missed flights due to too strong signal. Strong, but not too strong signals are favorable and should not be limited. In a perfect system all planes would generate strong signals.
Right now I work with a gain of 17 since with 18 I still had some missed close flights.
You own a skyscraper directly in front of a busy airport? You must have a beautiful view up there. Enjoy!
I would not call the Airspy deaf without the preamp. I still see more planes than with the Nooelec. But better is the enemy of good.
With an rtl-sdr type dongle dump1090 talks directly to the dongle via USB.
In the case of an airspy device on the other hand airspy_adsb talks to the dongle via USB.
dump1090 then talks to airspy_adsb via the network interface (directly looping back to the pi of course but it’s still technically network traffic)
The receiver-type for piaware makes dump1090 not try and use an rtl-sdr dongle. Because it wouldn’t find one in your case. The USB communication is different and a different driver provided by airspy_adsb is needed.
If you don’t want to lose very close by aircraft and you know they are in sight of the antenna your approach is of course very good.
Many users of flightaware optimize for the numbers, that’s why i made this thread. Also it is a very good starting point to maybe further reduce gain if you notice the planes very close aren’t being received.
So while your approach might be perfect the numerical approach of the percentage of strong messages is still better compared to the gain optimization script that is floating around several forums. That script can create really random results when the traffic changes while the script compares message and plane counts with different gain settings.
Comparing last Friday with this Friday on his FA stats (https://de.flightaware.com/adsb/stats/user/burgdorfer#stats-28783)
Last Friday the gain was still set higher i presume?
Anyway it probably doesn’t change much because the mountains limit his range anyway.
