Thank you for your replies. I see my mistake with the amazon order. This time, I’ll buy the Airspy mini on the Airspy website.
Like the colours scheme of your graphs.
Hi,
I guess I need one of our excellent developers to give me a proper answer, but I welcome anyone’s ideas.
Connected to the existing stream, I need a raw port where I want to extract the mode-s weather data and dump it to an external file for further analysis.
I am trying to use this project to implement it:
I created a virtual machine on a Win11 host and installed pyModeS on it. I connected the source to my airspy ads-b station via network.
The result is depressing. Only some possible hex icaos and callsigns are shown in the table.
-
Does our decoder filter out meteo data?
I didn’t notice any changes in airspy_adsb with the “–D none” argument either. -
…or is the project I installed unable to sort the data?
This is the main project for getting output: - containing the one above.
I tried this command line on the guest system:
“modeslive --source net --connect 192.168.0.106 30005 raw”
(raw or beast format is accepted in the project)
PS: My rtlsdr v3 dongle died a few months ago, so I can not involve other hardware than the existing airspy r2.
thx
Bought an Airspy Mini during the black Friday sale.
Read the post on configuring and have it up and running but have a few config questions.
Setup is Vinnant 5/8 antenna → 6ft RG58 → Uputronics 1090 LAN → AirSpy Mini, Pi 5.
airspy_adsb config is:
GAIN= 21
OPTIONS= -v -t 90 -f 1 -w 5 -P 8 -e 60 -b
SAMPLE_RATE= 20
Seems to be working fine and I’m not seeing anything in the logs beyond what looks like normal startup
Dec 01 05:30:38 tracker1090 systemd[1]: Started airspy_adsb.service - Airspy ADS-B receiver.
Dec 01 05:30:38 tracker1090 airspy_adsb[959763]: airspy_adsb v2.2-RC30
Dec 01 05:30:38 tracker1090 airspy_adsb[959763]: Listening for beast clients on port 47787
Dec 01 05:30:38 tracker1090 airspy_adsb[959763]: Acquired Airspy device with serial 258C62DC304CAB8F
Dec 01 05:30:38 tracker1090 airspy_adsb[959763]: Decoding started at 20 MSPS (Gain: 21; Preamble Filter: 60.0)
Dec 01 05:30:39 tracker1090 airspy_adsb[959763]: Push client connected to 127.0.0.1:30004 (beast)
I did notice the CPU frequency bounces between 2.4Ghz and 1.5Ghz.
CPU runs about 25% (when the clock is at 2.4Ghz).
It is not thermal or power thotteling as vcgencmd get_throttled returns 0x0 and vcgencmd measure_temp remains below 50C
On the quickstart guide there are some performance tweaks mentioned for the Pi3
changing the scaling governor from on demand to performance
forcing turbo
Disabling FIQ Fix
Are these settings applicable to a Pi 5 ?
I haven’t set up my pi5 for adsb yet so not sure about the cpu behavior you are experiencing.
Changing your gain to auto (gain= auto) would be my first suggestion. Using the target CPU setting, for example -C 75, would be my second. -e is not necessary as -C will increase the preamble filter to achieve the target CPU %.
Those settings aren’t really relevant. The pi 4, let alone 5 is powerful enough that you don’t need to mess with the CPU speed or governor.
The pi 5 is changing frequency because it doesn’t need the extra speed and shouldn’t cause any problem. Just check the journal for signs of lost samples.
Definitely use auto gain - from the testing I did when it was implemented, it selects an appropriate value the majority of the time. The results of that are earlier in this thread if you are interested. The airspy is quite tolerant of too much gain compared to an rtlsdr dongle but there is a slight performance advantage in having it set correctly.
You have so much CPU spare that any highish value of -C will result in the maximum preamble filter being used.
Thanks, didn’t think the governor was an issue bouth thought I would check.
You have so much CPU spare that any highish value of -C will result in the maximum preamble filter being used.
OPTIONS= -v -t 90 -f 1 -w 5 -P 8 -e 60 -b
I just forced the preamble filter to 60 with -e 60.
Is that not the correct thing to do?
It doesn’t matter too much as you aren’t likely to run out of CPU. -C sets a target CPU usage, but if it gets to the maximum without reaching it’s not a problem and functionally similar to just setting -e to 60 anyway.
Any benchmarks on this new Pi?
1 Like
What exactly are you looking for?
Pi5 is nice… I’m running a pretty full docker stack for adsb/uat. Just added an Airspy R2 and I’m running 24MSPS with about 35% total cpu according to graphs. The jump on Sunday morning is when I went from 20MSPS to 24MSPS on the Airspy.
Up to Sunday 6:00am it was forced to 2.4Ghz (force_turbo=1), after Sunday 6:00am removed that and actually set to clock down to 2.0Ghz (arm_freq=2000) to see if that stopped the ondemand governor from downclocking to 1.5Ghz (it didn’t) and to also see if it ran any cooler (it didn’t). Runs in near 50C and the Pi active cooler fan cycles on/off frequently. It’s winter here so max temp has been 10C. Running 20MSPS, no lost buffers.
I think summer temps will be an issue. A Pi4B in the same location with an RTL-SDR only running 8% CPU with a heat sync and fan is 20C today and hit 60C in July.
I’m running force_turbo=1 and have it overclocked to 2.6ghz. Active cooler is installed and the device is located inside an indoor closet with ventilation. I have not had any governor issues with downclocking.
Mostly CPU usage, MLAT stability, tolerence for other I/O intensive tasks.
I just received my Airspy Mini and wanted to thank everyone for this thread. I was up and running in mere minutes and everything appears to work fine. I was worried as I am not a Linux person, but slowly learning. I added to my PiAware image and it works good, including the graphs. I will read this thread … for more tweaks to my configuration.I was going in about 20 minutes including a precautionary SD card backup before hand. I am running a Raspberry Pi 4B 2 GB.
2 Likes
I was thinking of getting an Airspy Mini also. Interesting that the CPU utilization went from about 8% to 30% with the Airspy Mini. On the other hand, the CPU temperature is no higher. Or maybe it hasn’t reached steady state temperature. Can you tell yet if you are getting more aircraft and more positions reported with the Airspy Mini?
I did not get it for anything other than to be able to cope with close and far away aircraft better. The most planes I have seen at once is 32 and my messages never hit 100 / sec. So some would likely say the Airspy is overkill. I am concerned about some of my graph stats though, particularly snr, noise. I will do testing with / without the blue FA filter. It is too early to answer your questions right now.
Here are my current graphs …
Now that I have had it running for a day or so. I can say that it is better in that I have 2 systems and one always did better than the other. Now with the Airspy they are much closer in the results seen.
One has a typical antenna higher up and about 15 ft away with a better view of the surroundings but I do have terrain issues in all directions. The other system (the one with the Airspy) is on a spider antenna and it is lower down and it has more obstructions with the building, as it is mounted on the balcony.
As I live in an area with few planes, I am not stressing the Airspy. However I am just going to try to tweak things to tweak my performance. Really … that is probably only adjusting the gain as I cannot see that anything else is really going to be a problem.
I get about 300 planes per day and about 60,000 positions a day.
But in summary this is about my 2 systems. I am always looking at ways I can improve.
System 1 … roof top antenna about 8-9 dB gain, LMR400 coax (25 ft), Radarbox Green dongle, Raspberry Pi 3B.
System 2 … Spider antenna on balcony about 15 ft above ground (obstructed in about half of the coverage area), LMR 400 coax, 8 ft. Uputronics LNA, Airspy Mini, Raspberry Pi 4B.
Well if I got an Airspy Mini it would be to try to improve the number of positions. I notice I am lower than the best nearby site by about 15% in terms of aircraft number but lower by about 33% in terms of number of positions. I attribute that to drop outs of nearby aircraft due to too strong a signal. The Airspy Mini should do better because of increased dynamic range compared to the current RTL-SDR dongle.
The downside is the Airspy Mini is another $100 or so and its more power consumption, more heat and it doesn’t provide as much current through the bias-t as the RTL-SDR dongle (possibly requiring a separate outboard bias-t to power the LNA). A lot to pay for a marginal improvement in stats.
2 Likes
It really depends on your location and circumstances. In terms of power consumption, my pi 4 with airspy mini and external LNA consistently draws about 8-9W. The LNA is powered by the 5V rail from the pi via an external bias-T. Heat has never been an issue - the pi 4 has a small fan on it but the airspy doesn’t, and it’s mounted in a loft space which frequently reaches over 40°C in summer. It’s been running 4 years without issue.
The performance increase over the rtl dongle is significant however, but I live in a very busy area with lots of messages (over 3k/s during the day). I probably receive about 50% more messages with the airspy than I would with an rtl dongle, though that doesn’t translate to 50% more aircraft, just more complete coverage for those visible.
If you are in a less busy location you may not see such a large difference.
4 Likes