dump1090 mutability - best configuration


#1

Hi all!

I’m using dump1090 muta 1.15 dev installed via the adsb-feeder-script.

While trying to find out the best setting I’ve experienced unexpected behaviour:
Setting gain from “max” to “agc” made the message rate drop, but nb of planes remained stable.

I’m using an inline amp near the antenna (on the roof) and flightaware’s filter.

Can someone support me getting the best configuration?
Here are the results of the change (reverted after one day)

BR
Philipp


#2

“agc” is effectively “max + a bit” when receiving ADS-B for various obscure reasons. Looks like you’re just overloading the receiver input if you set it to agc.


#3

Ah, thanks.

Is there a positive effect when using a defined gain-value.
Can I see a good setting in the first chart? E.g. green surface close to red line but not crossing…


#4

“max” is something like 49.6dB. You might want to try a few dB lower (45?) and see if that gives better results, but max looks reasonable from your graphs.

(edit: and I should point out that the gain numbers don’t necessarily correspond to reality, the librtlsdr gain curve isn’t calibrated very well)


#5

I found that a setting of 45 works well in my installation. It also lowers the CPU consumption. You really have to experiment to find the best value in your situation.

phg2k, you may want to update JP’s script and re-run just the portal installation. The messages/aircraft/sec graph has been updated and scaled differently. You will see the message rate much clearer.


#6

Philipp,
As you have found out, more gain is not always better. With the graphing you have set up, you should be able to adjust the gain one step at a time and see what impact it has on your statistics. You may find that running a lower gain increases both planes and messages, or raises one while lowering the other. I have multiple locations with the same setup, 1 under tree cover requires higher gain than one in clear sky (44.5 vs 25.4). From the look of your graphs, you are close to the best setting for your current configuration. I have attached graphs from two of my feeders that are 20 miles apart with the gain dialed in.

Cheers!
LitterBug

Under tree cover 15’ above ground 44.5 Gain

Clear sky on tower 50’ above ground 24.5 Gain


#7

For my location, in NYC, I prefer some sort of agc.
I have commercial aircraft flying overhead at 2000-3000ft blasting out 200-500W on their transponders(heading into KLGA).
I also have GA aircraft flying up and down the Hudson at 300-1200 feet alt and 300-2000ft laterally pumping out 200W minimum.
NYPD also loves to fly over at 150ft four or five times a day.
There is just too much variation from busy times to late at night.


#8

Sure, but “agc” does not actually give you auto gain control for ADS-B signals.


#9

thanks for your ideas!

  • updated adsbportal in order to get clearer graphs

  • start with 45 as gain value and the perform some try and error cycles.
    will let u know what is the best result.

In planefinder’s overview, the installation of dump-muta is clearly visible (lost a few k planes with posistions per day) and got lots more of positionless planes.
is this a gain issue or something else? some other post (don’t remember where) said that there is a bug in 1.15dev.

thanks again!

Philipp


#10

There are no bugs in 1.15~dev! :wink: But more seriously, I don’t know of any outstanding bugs in position reporting. If you find one, please report it.

I don’t know what you’re comparing to, so no idea why it changed. Maybe the old version had bugs that were fixed.


#11

OK, is there an AGC system built into the dongles?
AGC has been around almost 100 years.


#12

The dongle can do AGC, but the problem is the very short duration of the transmissions for Mode S and ADS-B. They are of the order of 120 microseconds, which is far too short for any AGC system to react to. Since you are receiving many aircraft simultaneously, all at different ranges and with different signal strengths, which one is it supposed to be reacting to? The best option is to find a fixed gain level that is not so high that powerful signals overload the dongle, and not too low that the weaker ones are lost. Remember that ADS-B is a digital mode, so the aim is to get sufficient signal strength to decode, not the highest signal possible. In a receiver with greater dynamic range, this would not be such an issue since it could better cope with the large differences in signal strength.

AGC works well for receiving continuous or longer transmissions from a single transmitter, but it doesn’t work for this.


#13

OK, so should I aim for a peak of -3dbFS?


#14

found the thread:
http://discussions.flightaware.com/dump1090-mutability-1-14-or-1-15dev-performance-t36286.html
Thu Nov 12, 2015 3:13 pm

Seems to be the same effect…


#15

This is getting confusing to me too… :unamused:
I have set “–gain -10” for dump1090-muta v1.15~dev at 11:51 UTC today - just before I tried “–gain 45” and “–gain 40”.


root@rpi:~ # Sat Mar  5 11:36:32 2016 UTC  dump1090-mutability v1.15~dev starting up.
Using sample converter: UC8, integer/table path, with power measurement
Found 1 device(s):
0: Realtek, RTL2838UHIDIR, SN: 00000001 (currently selected)
Found Rafael Micro R820T tuner
Closest available gain: 40.20 dB
Setting gain to: 40.20 dB
Gain reported by device: 40.20 dB

root@rpi:~ # Sat Mar  5 11:51:24 2016 UTC  dump1090-mutability v1.15~dev starting up.
Using sample converter: UC8, integer/table path, with power measurement
Found 1 device(s):
0: Realtek, RTL2838UHIDIR, SN: 00000001 (currently selected)
Found Rafael Micro R820T tuner
Using automatic gain control.
Gain reported by device: 0.00 dB

This is the result after app. 30 minutes:

I can see more aircraft being tracked; the load on the dongle is rising; messages > -3dBFS is shown, but is this right?
This is my setting in “/etc/fr24feed.ini” at the moment, being used by fr24 / PiAware / pfclient all together:


procargs="--net --net-sbs-port 30003 --gain -10 --mlat --lat 53.x --lon 6.y --json-location-accuracy 2 --measure-noise --phase-enhance --oversample"

FYI I use a 4 legged spider antenna inside (that is not giving better range reception over a 52mm cut whip antenna), connected with cheap TV coax (75ohm - just like the DVB-T dongles have) and this pigtail: http://www.dx.com/p/dvb-t-rf-coaxial-to-mcx-tv-antenna-connector-black-22cm-cable-165089

R820T --> MCX --> PAL --> cable 3m --> PL259 --> SO239

https://farm2.staticflickr.com/1681/24763590789_4b5109fbbd.jpg
Any specific advice / explanation on the gain settings for basic usage (and the antenna) welcome!

Cheers, Henk


#16

I am using gain -10 for all antennas without an amplifier and it’s the best setting for these receivers. With an amplifier I’ve to lower the gain to reduce noise (-3dbfs messages) to not overload the dongle.

Changing settings just before high noon can be dangerous in Europe, because there is always a peak at around 12:00. :slight_smile:

Just look at the statistics for one day or better a week to know if it is the right setting for you.


#17

Try using the RTL1090 method I describe part way down in this thread. I have some pre-amps which require as low as 19.7dB setting at the dongle. :open_mouth:


#18

I did some more testing on gain settings with an inside cut-to-52mm whip antenne, with the base opened and copper shielding connected to a can.
The 4-legged spider just wasn’t performing and so as was a simple dipole.

From my experience after app. one week, not setting the gain at all (so leave it to default: max gain) gives the best S/N image.
Setting “–gain -10” brings in too much of > -3dBFS messages over here (noise / distortion, right?), since lots of planes and medical helicopters fly by.
My range won’t get passed 100NM very often, but I don’t have free view around, since I’m in the city centre.

So here are some gain settings I have tried with levels if noted (X means bad) :



19.7 X
28   X
36.4 X
40.2 X Noise: -11.6, Mean Level: -3
42 ... Noise: -42.1, Mean Level: -26
45 ... Noise: -41.9, Mean level: -22
48 ... Noise: -41.9, Mean Level: -24
49 ... Noise: -38.5, Mean Level: -18

max = 49.6


One needs to keep a “diary” of all changes and effects, because memorizing is too hard. :slight_smile:


#19

Question :
gain -10 means:

RTLSDR gain in dB.

If set to “max” (the default) the maximum supported gain is used.

If set to “agc”, the tuner AGC is used to set the gain.

GAIN=“38.6”

max= 49.6 -10 ?


#20

“-10” is equivalent to “agc”