Is there a maximum throughput of the RTL-SDR/Pi combination?


#1

It’s busy out there. At the moment I’ve got over 240 aircraft on my screen and am seeing more traffic than I’ve seen before in my short time of running this.

Is there a point at which this setup will get swamped and start dropping traffic?

Keith.


#2

The per-message CPU cost is not very much, most of the cost is in the scanning-for-preambles part which is mostly independent of message rate, on a Pi 2 or newer I doubt dump1090 will ever get close to using a full CPU core which would be the limit. At 2k+/sec messages you are starting to get a lot of message collisions that can’t be decoded anyway, so it’s somewhat self-limiting.


#3

That’s good to know, thank you. I think this is probably the busiest time of day so I don’t expect that I’m likely to see much more. I can squeeze another dB or so of signal out of this system and then the law of diminishing returns will be in effect.


#4

That explains why there are less planes and a lot of incomplete reports at “Prime Time” - may i need to reduce gain below 20db…

Regards

Volker


#5

I regularly get 1800 messages/sec and have seen 2000 (200-300 aurcraft). The RPi is not apparently struggling. I have tried many ways to improve on this and suspect I have hit the natural limit for my location. The messages are simply overlaping/destroying each other.


#6

That’s why i move to Mode-S Beast next time - looks like some FA feeders are using Mode-S beast with RPi.


#7

Mode-S beast / Airspy / PlaneFinder showed no improvement over FA orange RTL stick for my location. hence my conclusion that messages were overlapping/destroying each other.


#8

OK - what is the solution? Reducing gain did not help - it will decrease range, but not the message rate…


#9

There isn’t really a solution, it is a result of how Mode S works.

You could use sectorized antennas to reduce the collisions that each sees.


#10

Beast ordered - will report later what happen. If there is no success i will ad a second sector - antenna any suggestions ?


#11

I use these:-
moonraker.eu/avionics-and-sh … al-antenna
I haven’t worked out how to tune them yet. My problem is that I can get a vertical higher, hence a greater range(I am only about 60’/20m AMSL).

If you are located in an electrically noisy area then I highly recommend a cavity filter.
shop.jetvision.de/epages/6480790 … ucts/71010

I have one on my radarcape(with hab/nevis amp), airspy and beast. I took it off the radarcape yesterday(to test) and dropped 400-600 packets per second.
There are so many aircraft (hence TCAS), radar (@1030Mhz) and cell towers in my area that I am probably getting swamped with noise.

You can sectorise and still use mlat with the beast, but only using two inputs(The doc says 4 but is old. I confirmed with the company that it is only two). Don’t enable Mode A/C, it will swamp your device and FA doesn’t use it much.
You could use two beasts (with two inputs) and combine the display using VRS (virtualradarserver.co.uk/). I consolidate all my feeds this way. It is great for range maps.


#12

Thanks …


#13

I left out that all of my coax is LMR400, even the 1’/30cm jumpers.
This is to keep loss and noise low. I also use Ferrite toroids on my USB and power cables.
I also have the RPI’s in metal cases.

My next step is to move the cavity filter and Amp to the mast head. I just need to find a way to power the amp by a USB Bias-t injector.


#14

Ick - ick - ick!
(that’s for powering the amp from a USB port! LMR400 is good stuff, and you know putting the amp close to the antenna is the right way to go.)
If you insist on using USB port power, since you have to make a funky connector/adapter anyway, I’d do a pretty stiff multi-section LC filter as USB power can carry significant noise! Good ferrite cores and multiple caps, such as 1, 10, 100nF, 10uF, all very low ESR ceramics to provide that low ESR across a broad band.

bob k6rtm


#15

Bob,
The problem is I only have POE in the attic. I currently have the amp next to the Odriod and Radarcape so can power it by USB from either. I suppose I could try to run another cat5e to the attic to power a STD bias-t injector.
I am looking at something like this
5209fafb-a-62cb3a1a-s-sites.goo … edirects=0

I power two other HAB amps using Bias-t (RTL-SDR and airspy), however, the radarcape doesn’t provide bias-t for the 1090 antenna, as far as I am aware.

Jon


#16

Why not keep the filter snugged up next to the dongle to attenuate whatever junk the cable run picked up on the way down? (running from the premise you’ll be running the LNA at the antenna) Rather take the insertion loss post amplification than before - so long as your amp has a high enough IP3 so it doesn’t overload that is.

EDIT: Sorry, just noticed you are using HAB - so it’s all built-in together. :frowning:
OR - you also mentioned cavity, so that would be stand-alone…


#17

OK - looks like the BEAST was the solution for me - dump1090 shows me a message rate around 1400 with 204 planes- (Dongle was around 2100 with 120 planes)


#18

Out of interest, how straightforward is it to replace an RTL-SDR dongle with a Beast on a Raspberry Pi? I assume it’s not the same as swapping dongles where they’re all compatible.


#19

You need to install beast-splitter first -> ads-b-flight-tracking-f21/mode-s-beast-to-dump1090-t37135.html