I am running a receiver for years now and put my antenna outside now.
Yesterday I had 278 total aircraft but only a message rate of 1250 msg/s.
I reach this rate with an aircraft count of 150 and up it is constant.
I tried different sticks (orange, blue and airnav) and different filters (first fa filter and a cavity filter) , all with nearly same result. Also tried different gain settings and looked at strong signals.
So I think it is maximum rate here because there are also mode ac signals around. If I activate --modeac I will get around 2200-2300 msg/s
I do not feed mlat results back to my Dump1090 client.
Had same experience with my indoor raspberry station.
So that could be too much traffic on that frequency or is it too much for a raspberry? I am located in north west part of Germany.
There indeed seems to be a limit to the message rate you can get using rtl-sdr compatible dongles.
Probably messages are overlapping but i’m not exactly sure about what the problem is.
But with an external LNA and a good dongle (the rtl-sdr blog v3 seems pretty good) at least I could reach message rates up to 1450.
To recover even more messages you’ll need an airspy or a ModeS Beast.
I’m not quite sure why they are able to pick up more messages though.
Another solution would be to sectorize your receiver.
2 antennas which cover 180 degress field of view each connected to their own USB receiver. (not really sure how to construct something like this)
If you want to run FA mlat both receivers will have to be on different stations.
Looking at your feed, the number of positions at <40 nm seems very low.
Maybe you should reduce your gain (Thoughts on optimizing gain)
Some of it may be dynamic range; some of it may be that the rtlsdr dongles just can’t get very many samples per bit. dump1090 runs at 2.4Msps which is about as fast as you can reliably run a rtlsdr at (yes, you can tell them to do 3.2Msps, but they’ll drop a ton of data and/or get stuck randomly).
At 2.4Msps dump1090 only has 1.2 samples per Mode S pulse (halfbit) to work with, so trying to disentangle two signals that are partially overlapping is not practical.
If you’re running at a higher sampling rate like the airspy / beast, then you can handle some types of overlap so long as they’re somewhat offset from each other.
