Glad to see you are back up and running, and you are using adsb-receiver…Good Choice !
The signal graph looks good to me, the mean level is well inside the upper graph limit giving a nice set of green bars, peaking just over the bar.Depends though , sometimes setting less gain can result in better overall performance. @mgunther found this with his station and he discovered the optimum setting to be around gain 44-45 for his location and setup.
Probably best to try various gain settings for fixed periods of say 10 mins and observe the different results such as range and msg rate on the graphs in 1hr view option.
Okay, I think I understand now.
Lowering the gain until the blue peaks are below the red line is best.
Does anyone have a screenshot of what a good/textbook adsb signal graph looks like so that we can try to replicate it on our stations ?
A “track” is a series of associated messages/positions from the same aircraft.
A new track is created when a new aircraft is seen.
Tracks get removed after a period of inactivity.
Tracks that only ever got one message are usually noise: the aircraft wasn’t real, the aircraft address was just garbage and never got repeated.
If you get a lot of those, that means that you have a noisy system, perhaps overamplified.
How do you rate graphs below, for my 2 RPis?
Do i need to set my gain to a lower value? Right now gain is ‘max’, and NO amplifier.
Oliver (obj)'s expert comments will be very valuable.
Please note that Range is in nautical miles.
Set up 1 of 2:
Indoor Cantenna, near a large window >> 12 ft / 4 m RG6 coax >> 6 inch / 15 cm RG174 F-to-mcx pigtail >> DVB-T plugged directly into RPi Model-2 >> 30 ft / 10 m Network cable to router.
Set up 2 of 2:
Indoor 8 element Coco, near a large window, with Impedance Matching (0.85 x λ/2 coax + 2.5 pf cap) >> 12 ft / 4 m RG6 coax >> 12 inch / 30 cm RG316 F-to-mcx pigtail >> DVB-T plugged directly into RPi B+ >> 30 ft / 10 m Network cable to router.
Sorry, the graphs are only for few hours as I have formatted microSD card & done fresh install this morning (because I forgot password, and could not recover / reset it)
I see the dump1090 code doesn’t output tracks until they have 2 or more messages. That makes sense since I never see these junk tracks in the web view or on the output ports, even though my log says I should be seeing new garbage tracks every second.
Will be tweaking Gain settings more tomorrow (currently have set to 44.5).
Can we confirm what we’re aiming for…?
No signals “clipping” above -3db (no red lines / spikes)(reduce gain till peak is below the line/ less than -3db)
More positions / aircraft is good
Less “single positions” is good
Remind me, whats the reason not to use AGC, vs manual gain control?
(Also, afair, AGC has to be “tuned” to the particular dongle?)
Of course, the goal here is to remove spurious results from “single plots” - anything over say, 2 positions/aircraft is an accurate / true plot?
So in theory, pos/aircraft could as low as that? Mine’s currently 12’ish.
Is there any book / online resource anyone could point me at that explains
a)what we’re aiming for, and
b) why / what this represent?
Just so I can learn - save us all asking the same questions!
They’re all just statistics that tell you how the dongle / demodulator is behaving; the end goal is surely to improve your coverage (range, message rate).
The stats are there guide you towards what might be a better set of settings, and to let you monitor the results when you change things.
There is no definitive “right answer”
One reason it is quite difficult to get an ideal setting is that what works best will also depend on how many aircraft you can see and how far away they are from you. If you live somewhere where the majority of aircraft are overflying at high altitude, and are spread out across the whole area you can see, then your statistics will look quite different to if you live right next door to a busy airport.
I live about 15 miles from Heathrow, and one of the approach holding stacks is right next to me, so I see a very large number of aircraft at short range. These are also invariably very strong signals, so if I reduce the gain to account for them then the range suffers quite a bit. I tend to tune the gain to get the highest message rate per aircraft - this seems to give the best overall results, including range and message rate. It’s still a bit tricky, since you can only make comparisons between similar numbers of aircraft as the message rate per aircraft varies according to the number of planes visible.
I have been doing some experiments with capturing data and plotting the number of aircraft against the message rate, rather than doing it against time to get an idea of how the receiver behaves under different circumstances. Getting the data is a bit convoluted, but it looks promising. I’ll post some results when I have enough data.
Yes I’ve been using that for a while, but it is still plotting the rate against time. I want to plot the rate against the number of aircraft. At present I am doing it with sbsmeter, which can log the number of messages, but it uses the SBS basestation message format. It stores the message rate and the number of aircraft once per minute in csv format so you can import it into excel to manipulate.
Here are some of the graphs.
This one shows the total message rate on the y-axis against the number of aircraft on the x-axis. You can see that the message rate increases linearly until around 60 aircraft are visible, after which the rate of increase reduces. This shows that messages are being lost due to garbling - the more aircraft present, the greater the likelihood that they will transmit over each other. The close to linear the graph is, the more efficient the receiver is.
This graph shows the number of air-air messages received against the number of aircraft. These are the TCAS collision avoidance messages - the more aircraft within range of each other and the closer together they are, the more often they transmit. These messages make up the largest number when it is busy, so are likely the main cause of the loss of ADS-B positions.
Presenting data like this makes it easier to show differences between receiver settings - eg too high a gain setting will result in the receiver being over saturated and more messages being lost.
Ideally I’d like some kind of program that continuously logs the data and presents it like this, but there doesn’t seem to be anything available at the moment. The information is already present in the dump1090 json pages, so I don’t think it would be too hard to do in python using gnuplot or similar to produce the graphs. Unfortunately my coding knowledge is pretty basic, so I wouldn’t know where to start with it.
Just a heads up I am working on adding database support to the ADS-B Receiver project with long term data collection too be added soon. Admittedly it has been taking longer than it should due to work but it’s on it way. Like you said won’t be hard to do but might be hard on SD cards.