I have the same problem.
1500msg/sec seems the limit for me with FA antenna mounted externally and expensive cabling to FA dongle. Pretty sure I have the gain optimized at this point. I would like to get the antenna higher up on the peak of the roof - but I doubt this would improve numbers during the busy hours - though I might get some further out reports at night and some more contacts on the ground. I’m getting 900k-1m msg/day so I really have nothing to complain about.
Helmholtz equation.
Signals at this frequency don’t really penetrate very far into solids, so basically, most of what you’re receiving with an internal antenna is sort of coming in through a window and bouncing around. In an attic, since roofing is fairly thin, you will get coverage through the roof albeit with some attenuation- but if you’re like me, in a building with brick walls, you’re not going to get coverage through those.
Here’s how this equation affects wifi coverage based on router location, presented visually:
Here’s an article about the helmholtz equation and how it relates to wifi propogation - wifi has even less solid penetration than 1090mhz, but many of the underlying principles here are relatable.
It’s likely that plays a part. I get noticeably different coverage based on the position in the roof space the aerial is placed as the gable ends are brick. What is less apparent is why I sometimes get drastic differences with the antenna in the same place. There must be something not quite right with the feed line, as I’m getting about 40 miles less range than I know is possible with the antenna in its current position.
If the sensible upper decode limit for a R-Pi + RTL dongle is around 1600/second, what is the feasibility of running 2 (or more) instances of dump1090 on a newer Pi and ‘pinning’ each instance to a core and ‘internally’ combining their outputs to produce one single output (feed)?
I have a picture in my mind of 2 dongles with either separate beam (or sector) antennas, one pointing North and one pointing South, on a R-PI3 quad core with each instance pinned to a core.
Sound do-able or Pi in the sky? 
Reason I ask is I am considering putting a system on an 300 metre ASL hilltop site near me and I have tested an RTL dongle into my laptop with a simple quarter wave antenna held over my head and can easily ‘see’ out to 250nm and over 200 aircraft on screen and wonder if a Pi would get overwhelmed at peak times, especially with an antenna having some gain?
Nigel.
My testing so far suggests the RTL dongle works out best for aircraft counts at around 1600 messages/per second. I live in a busy area and display shows 200 to 300 aircraft most of the day. The Pi3 handles this with ease. Swapping from a Cantenna to the FAntenna increased number of aircraft by about 15% at the expense of gaps in logging of overhead/local flights - which I found worthwhile.
This can be done but you will need good power and you will need to run two separate copies of piaware if you want mlat (you can combine the feeds for internal use, just don’t give the combined feed to piaware, it needs a separate feed per receiver for mlat to work).
Probably no need to pin dump1090 to separate cores, a Pi 3 has tons of CPU to spare for this.
Is it OK to run two copies of piaware on one R-Pi and feed those two to FA with just the one MAC address?
Nigel.
Inside also suffers more from equipment interference.
My Bedroom setup is behind a tall dresser we keep our TV on. The FA Antennas are sitting in the corner behind and below the TV.
I noticed that my starts dropped significantly at 8PM last night (Site 17065). This is the time my wife went to bed watching TV.
I still have the plastic cases on these Dongles(1090 and 978). I have metal case and will swap them over when I get a chance. It may help a little with interference. The Antennas are large so it may not help at all.
So… just as a quick follow up… the takeaway for me is that I stopped worrying too much about the message rate. I have seen big changes in range and report rate while the message rate just nudged a little up or down. I guess the amount of messages sent by aircraft makes this system work pretty decent even when a significant amount of collisions occur.
Still, not everything fine with my site. I’ll open up a new thread for that.
My tests on RPi3 suggest FA stick easily runs up tp high 1600’s without aircraft loss and max’s at 1800 with losses. Airspy runs up to 2200’s and max’s at 2400. With airspy running dump1090 etc on same Pi makes no real difference to messages but using web pages upsets MLAT.
Very busy air day (even given Covid) today, and was messing with tweaking gain settings to maximize coverage from some of my sites today. Top performer has managed to get to 1800. I’m pretty sure that is the max I’ve ever seen from any of my sites.
Site is a mountain top at 11,700’ in south central Colorado. It’s got a DPD productions antenna (I think that is about 9 dBi gain), with a RTL-SDR 1090 Amp at the antenna, LDF-1 1/4" hardline running into the building (about 35’) Polyphaser lightning protector, and another 10’ or so of LDF 1. Goes through a splitter, and then to an RTL-SDR on a Raspberry Pi 3 B.
I might have to think about playing with some directional antennas and multiple feeds from there next summer.
Chuck
Those two are your weakest links. You will saturate the receiver input ADC and, if you fix that (Airspy) the processing power is not there.
However there is a limit of ADS-B messages, based on conflicts of packets. Probably around 1900?
airspy mini to replace the rtl-sdr v3 might be cheaper than adding directional antennas.
(external bias-t as well as the at least nominally the airspy mini bias-t isn’t beefy enough to feed the rtl-sdr filtered LNA)
1 Like
AirSpy Mini sold at Aliexpress has a “Hi-Res Audio” chip also, and one can enjoy Hi Fi music while receiving ADS-B 
2 Likes
Well … then don’t get a knockoff …
1 Like