The 5m USB cable could well be the problem if you haven’t tried swapping that. Especially if it’s a cheap one. I wonder if just the voltage drop across it is enough to make the dongle a bit flaky. 5m of passive cable is right at the limit of what USB is theoretically meant to do, and with cheap components you probably don’t even get that.
Where is your dongle? On the mast? They’re not exactly weatherproof.
You can be (and in your case, are) connected to more than one mlat server behind the scenes; it’s possible for one to be accepting data while another isn’t. The control panel status shows the worst case. I restarted one to turn on some more debugging which is probably what reinstated your feed for a little while, though it has disabled itself again now. (Basically, you are getting enough sync problems that once it gets inhibited, it never uninhibits - you need a long term average of no more than one sync problem per 120 seconds to uninhibit, and you’re running at more like one per 90 seconds)
I only have the one 5mtr USB cable at present in which is being powered by the Pi. Would it make any difference if I applied an additional 5v power adapter between the Pi and Dongle?
Yes, the dongle is attached to the mast. However, it is in a waterproof casing with air vents for air to circulate and extract heat from the dongle. Outside air temp is 11c degrees and dongle is currently at 14c degrees, as I’ve included a wired temp sensor inside the casing taped to dongle
OK mate. So, as asked, if I use 10mtrs of RG6 cable from antenna into dongle and put dongle directly into Pi. What loss would I get if I’m using a 5db Collinear Antenna - no filters or amps?
That’s just it buddy. I don’t have a USB hub. I have 5mtrs of USB Lead from Dongle to Pi. Gotta a feeling, as it’s a cheapo USB Lead from eBay. It’s not good enough for what I need it for.
Just out of interest, now that a/c are counted not flights, how does a non-mlated a/c fit into the count if it IS mlated later in a 24hr period or vice-versa?
Which category would it fall under? Would it stay in what ever it was seen as first and not change the other count later(like switch category)?
It’s counted as mlat for a particular hour or day bucket if (a) any mlat positions were seen from it by the receiver within the bucket’s timespan and (b) no ADS-B positions were seen from it by the receiver within the bucket’s timespan.
So yes, over the course of an hour or a day a particular aircraft can potentially start off being “no position”, move to being mlat, then move to being ADS-B as different position types arrive.
(the mlat/ADS-B case is mostly just a double-check, usually there are no mlat positions generated for an ADS-B aircraft)
Awesome, makes sense.
Thanks for the clarification of that and the quick answer.
To follow up;
So that could potentially been seen in the daily stats as the “Other” category going from say 5 a/c down to 1 but mlat or ads-b(or split somehow between the 2) increasing by the 4 that were “Other”, right?
Yes, that’s right, if you’re looking at the current stats bucket while it’s still being generated. Once the hour or day is over they won’t change further.
A question for obj: I was wondering if you are doing any accuracy characterisation on mlat positions? For example, finding an mlat solution for an aircraft that is also transmitting ADS-B for comparison. This would give some indication of the quality of the coverage of a particular area other than just, are there sufficient receivers to perform an mlat calculation. Are you able to determine the likely accuracy purely from the geometric distribution of the receivers, or are there too many other factors involved? Do you know the best/worse case scenarios possible with this mlat system or is it sufficiently good that any mlat fix is acceptable?
It would be better to use WF100, which has largely replaced RG-6 for satellite TV installations. They have lower attenuation and DC resistance than RG-6 since that has a steel core and aluminium shield. If you plan on putting an amplifier at the antenna end of the cable some time in the future that is something that can make quite a noticeable difference. If you want even lower attenuation, you could use WF125 which is a bit thicker, but costs more.
Don’t use cheap TV coax, or cheap satellite WF65 coax as they will have much higher attenuation.
I suspect that might be circular logic. As I understand it the ADS-B aircraft are already used to calibrate the mlat system. We don’t have information on actual time of arrival of signals between different mlat stations, just the delta time between signal arrival. Using a fully positioned ADS-B aircraft as a reference point allows you to calibrate that.
I crosschecked against ADS-B aircraft very early on in development when I was shaking the bugs out of the solver, but that did suffer from circular reasoning, it was really just testing whether the code was working, not the overall system. It might be nice to do ADS-B crosschecks again at some point but it will need a little thought about how it should work as you really need to select some aircraft to look at (how?) and exclude them from the sync process. That also needs a few client/infrastructure changes too as currently there’s no way to ask for all messages for an ADS-B equipped aircraft (the server just asks for an aircraft, and the client works out whether that should be ADS-B or mlat data based on the messages it hears)
The solver does maintain estimates of the measurement error in the individual pseudoranges based on the observed errors when updating clock sync, and those then feed through the Kalman filter etc and end up producing an error estimate for the final position. That’s used to decide when to actually emit positions. That error estimate bundles up both the measurement errors and the dilution-of-precision from the receiver geometry.
You can actually calculate the accuracy due to receiver geometry directly if you wanted to, it only depends on the receiver positions, but it’s not so useful unless you have good data on receiver coverage too, which is tricky.
(speaking of coverage, there are some new coverage maps up at flightaware.com/adsb/coverage that are worth checking out)
Hi Caius, thanks for your reply. Obj suggested WF100 to me as well. I was going to try Maplin but they are currently Out Of Stock nationally. However, I have found a WF100 equivalent, called CB10R which has the same 75ohms as the WF100. Would this be suitable or is it like the RG6 which shouldn’t go outside?
Update 13:00GMT. I have now found some Webro WF100. However, I do have some concerns with this 6.4mm dia cable fitting into an SMA Plug at one end and SMA Female at the other. Correct me if I’m wrong but aren’t these plugs suitable for thinner cable such as RG174 which is 2.8mm dia, as I don’t really wish to buy different plugs and have step-down adapters?
RG6 and WF100 are both used with F-type connectors, which are a little easer to find and install than SMA because they’re widely used for satellite installs. You can pick up screwfit F-type connectors pretty easily which don’t require tools to install (well, maybe a pair of pliers to twist them on).
Picking good cable probably outweighs the connector loss. Anything much thinner than WF100 is going to start picking up a fair amount of loss even over just a 10m run.
As it happens, I’ve bought 10mtrs of WF100 which comes with 2 unconnected F-Type plugs free - so I can cut the required length. I was originally going to buy 5mtrs and create an extension but decided to buy an N-Type Plug and run a new continuous feed from antenna straight into dongle without any breaks - hence my SMA female into my SMA/MCX pigtail.
Update: 14:30GMT. So, this will be my new configuration.
Antenna
N Type Male/F Type Female Adapter
F Type Male onto 10mtrs WF100 ending with F Type Male
F Type Female/SMA Female Adapter
SMA/MCX Pigtail into Dongle
Dongle into Raspi
Hopefully, shouldn’t lose too much but still far better than my cheap USB Lead
As I understand it the ADS-B aircraft are already used to calibrate the mlat system. We don’t have information on actual time of arrival of signals between different mlat stations, just the delta time between signal arrival. Using a fully positioned ADS-B aircraft as a reference point allows you to calibrate that.