The Airspy receivers are heavy users of the USB interfaces. Are there any tests or evaluations on using Pi4’s USB2 or USB3 connections for the best performance with the Airspy receivers?
I suspect we would have to assume the Airspy’s were the only devices connected via USB. Looking at the spec’s on the receivers, no mention if they are set up for USB2, 3.0 or 3.1 Has anyone have some information or recommendations? I don’t recall any posts about the preferred connection. Not sure just what criteria to use to get a feel on which works better. Thoughts?
As a test, swapped my R2 from USB2 over to USB3, on an Pi4/4G. Should be interesting.
Not really a test. But i have put my Airspy mini randomly either to the left or the middle USB-Port of the Pi4 with no difference in result.
Even a Pi3 delivers same results while using the same config settings.
Maybe it changes if you select 20MHz sampling instead of 12MHz. But this is overall not reliable regarding MLAT AFAIK
Swapped Airspy R2 to USB3 around 1:40 GMT last night. CPU usage increases by about 6%. This implies more messages. With 6:15 hours until the end of day collection period, I currently show 2454 planes and 447,269 messages. This seems to be somewhat higher than the usual Monday traffic at this time. It could be a random thing, but the cpu usage chart took a that 6% jump at the same time as the cable was moved to USB3. Also seeing higher 250+ m message counts, usually around 400 and today at 1637 so far.
Something is happening, just not enough info yet to see a definite trend. I will run a similar test with a Pi4 / FlightAware Blue receiver on USB2. Laying down a baseline now for a day or so and then changing to USB3. Might be interesting, might be nothing.
Now I need to add that pizza pie plate to the bottom of the antenna… See message #114 of Confused with ADS-B range
Dang, always some other variable tossed into the mix to mess up experiments. Well good thing one of my experiments Pi4 / FA Blue just started up again on USB2 about an hour or so ago. Then again, it changes every day, so bit of the “hit and a miss” kind of testing.
Been running an air spy mini at 20MHz on a pi 4 for about 4 1/2 years with no mlat issues. It has no additional cooling and despite being in a loft which gets up to over 40⁰C ambient in summer it hasn’t missed a beat.
Been doing some testing on a Pi4 / FA Blue dongle. And yes, using a 6 inch extension cable. The test question: Does it matter which USB2 or USB3 connection matter for the FL Blue dongle? NOTE: you do not need to power cycle when the dongle is moved, dump1090 will recognize the removal and quickly find and initialize the dongle. Reception will immediately start up with out power cycle or reboots.
I did my testing at hourly intervals, and over a 4 hour period it did make a difference, about 10 - 20% more messages on USB2 vs USB3.
EDIT:
Changed my test setup to 20 MHz for the last 90 Minutes. No changes except the CPU load (overall +20%) and the Airspy SNR (peak increased from 25 to 33 avg)
Tested the Airspy R2 on both USB2 and USB3. Host Pi is a Pi4, Feeder Type: PiAware (SD Card) 9.0.1. with airspy_adsb v2.2-RC30-test to support the Airspy Receiver at 20MHz sampling rate. Preamble Filter set to 60.
OPTIONS= -v -t 300 -f 1 -w 5 -P 8 -e 60
I consistently receive higher message rates when connected via USB3 sockets, and additionally, significantly more 250+ mi positions. This implies to me, more planes too, though because traffic varies so much, cannot prove one way or the other.
Result: Airspy Likes USB3. Kudo’s to @prog and @wiedehopf for their coding skills to make this work so well.
I see noticeably more messages with 20MHz and slightly more positions, but that doesn’t always translate to more visible aircraft. I’ve never seen 12MHz give better results than 20MHz though, so if you have the CPU to run 20 then it’s advisable to do so. You aren’t going to lose a huge amount of performance by running 12MHz however.
I should adapt my airspy script to run over a few days and get some more definitive measurements.
I have activated 20 MHz this morning at 11:00 without changing any other parameters
-E is at 60, -C at 90
Not really an impact. I assume it’s because of the terrain here. My reception is perfect and does not improve, independent from the device. The Airspy is only giving a higher message rate than Airsquitter or FA-Stick, but no improvements on range or seen aircraft.
Edit:
What i also have seen is a change in the FA stats. With the Airsquitter i have a similar amount of ADS-B and MLAT compared to the Airspy. But the Airspy reports a lot more “other” positions.
@foxhunter: Did you restart your airspy_adsb software? I usually see a short drop in the graphs and a spike at the restart of the software. Airspy_adsb doesn’t monitor changes to the configuration file in /etc/default, thus the need to restart it.
My traffic is a generally a bit higher than yours, and it’s when it’s busy you get the most benefit. If you are already receiving the vast majority of messages with 12MHz sample rate then you won’t see any benefit by running 20MHz, because there’s nothing else to receive. These things can be very dependent on your location and traffic and individual installation.
Your maximum range doesn’t look particularly limited - it’s higher than mine is. For me though I have a lot of local traffic which brings the average range down.
I was running several attempts to find out. So far it really looks that my range is only limited by the terrain (e.g far distance mountains) and there is nothing i can improve by myself (without relocation, but this will be a topic in two years on retirement)
Peak number of aircraft in summer is around 160 to 180 (ADS-B)
All devices i tested so far were showing a very similar distance. And the traffic depends on the flight plans.
I will change back to 12 MHz for another 24 hours and check the impact
