This is my first time posting so I would like to start by saying thanks to the community since this forum has helped me a lot with setting up my ADSB receiver location.
I am currently running the following setup: FA antenna → N to F → 3m quality coax → F to SMA → FA ProStick → USB pigtail → RPi 3 (PiAware, dump1090-fa with VRS server installed). It is working very well but I was wondering about a few things which I can’t find an explanation for myself. Following are some of my performance graphs:
As you can see, I’m averaging about 1300-1400 messages/s with 200-250 planes during the day. My question is: why is the percentage of aircraft without positions much higher during the day than during the night? The message rate per aircraft seems to be relatively low during the day when compared to nighttime, which also might be a contributing factor. I wouldn’t expect that aircraft broadcast more messages during nighttime for some reason.
I’ve been thinking about this and have a few ideas:
a) Hardware limitations - Either the ProStick or RPi can’t keep up with the amount of incoming messages. If it were indeed a hardware ‘issue’ (it really isn’t an issue since performance is pretty great), I’d suspect it to be the ProStick since the Pi runs at about 20% CPU.
b) Software limitations - I’ve read somewhere that the practical limit for dump1090 seems to be around 1600 messages per second, but can’t verify that claim.
I’d like to improve my setup if that’s feasible and am curious to hear your thoughts! Thanks in advance,
that’s true - the message rate self-limits anywhere around 1500 messages per second. this is limited by the protocol itself and the message collisions there. at my site even when i see 300 aircrafts i get 1500 m/s - in the late evening about 100 aicrafts will result in 1400 m/s. you can get a small boost when using radarcapes and/or sectorized directional antennas. but the result is in no ratio to money/effort.
a second thing at least at my site is - on sunny days there are many small private aircrafts without position data (no mlat because low altitude and no mode-s position transponder). as soon as it gets dark outside they are away and with them lots of flights without position data
You have numbers to die for. I live nearby our main airport and I’m not even close to your results. You’re doing pretty well IMHO
In Holland F connectors are primarily used for 75 Ohm coax cable. The connector usually doesn’t have a typical resistance rating, but if you use 75 Ohm coax, you might introduce avoidable cable/insertion losses as the antenna and Prostick are optimized for 50 Ohm
One question:
What is your definition of a message? Does it include partial, incomplete or test messages as well or is it limited to the main type codes as 11/17/18? This document helped me to understand what you’re receiving and which signals are useful. If you optimize for high numbers instead of high quality data, then it’s easy to split one feed, have it processed by different applications and merge it again as an example.
Lies, damned lies and statistics:
To illustrate the previous remark: I have one antenna on a high pole mounted to the side of the house connected to a 3 pole filter and a radarcape. A second similar antenna is connected to 5 meter coax, an Uputronics ceramic LNA and an Airspy. I’m aware of the less-than-optimal location of the LNA. The Airspy is connected to a PC running ADSB_Spy which feeds Planeplotter and Modesmixer2 which feeds socat on a Piaware. Basically: an optimized and isolated Radatcape, and two configurations fed by the same antenna and ADSB_Spy. The Airspy connected to modesmixer2 and PiAware produces the highest number of positions and aircraft; Planeplotter connected to the same source has a much lower message rate but a tenfold higher number of planes in the > 400km zone and Radarcape is lagging a bit.
hmmm - astonishing! my experience was different - the radarcape harvested the most messages, the nooelec-dongle/habamp slightly lower messages but slightly higher number of aircrafts and the airspy was less with both params while eating excessive cpu-power …
I just got the message that the German Wizard has prepared the package with the best antenna and the best signal splitter there is. I’ll add another two meters to the antenna mast to see if it improves my results.
The bottom line of my reply was: It’s easy to manipulate data that msg/sec and aircraft numbers are not the best indicators to judge the quality of a configuration.
who’s the german wizzard? guenther? what antenna did you buy - and not really one of the 300$ splitters
btw. your stats are not too bad - where you should look for improvement in my opinion is to extend range. and it would be easier for others to find your stats-page if you set up as your website in user settings of fa-forum. click my user-icon and you see what i mean …
Thanks for your replies! The ADSB decoding document looks promising. I was looking for something like this but couldn’t find what I was looking for until now. @Akubra, the message count I referred to is the one specified on the dump1090-fa webpage (and my performance graphs) - I am not familiar with the internals (yet) but will try to learn something about it using the document you specified.
I agree with you that further improvements aren’t that efficient (cost-wise) so I’ll probably focus on the little things to improve for now, haha. I originally ordered an Airspy Mini as my tuner solution, but unfortunately it never arrived and I’m now stuck in a PayPal dispute situation… I would’ve liked to test its performance with my setup.
i made the experience two years ago - improvement from 96% possible to 98% possiible is about 1.800$ while the setup for 96% is 200$ all inclusive. save your money and have some good dinners. that’s why i unmounted all this stuff and now this is my whole setup:
p.s. maybe soon we all need our money in the pocket to run away from europe if the awesome looking lady in berlin and all the other european airheads continue with their activities …
Got it, thanks! I added the link. Off course I’m talking about Guenter. The A3 ADSB is my choice, I’ve build an antenna mast that allows multiple antennas. If you mean the $412 splitter, than the answer is: ehhh… yes. I live in a city with lots of antennas and high buildings. I’m doing the best I can.
that’s with the splitter crazy! anyways that is the perfect solution for 1by1 tests. but again - if you spend so much money i’d give the yagi a try. i choose the cheaper one for about 150$ they have the same (more heavy duty) for 350$. easy to guess which one you will order
p.s. don’t know if you can upgrade your radarcape to the two antenna type - because the you could feed two antennas to one flightaware-site account …
Concerning core elements, I prefer to buy the best product because I know that the irritation of not making progress due to opting for the lower quality is greater than the pain of the price difference. It wouldn’t surprise me if you made similar choices in the past after seeing pictures of your setup.
The Yagi might be the next step - I asked you in the other topic if some of your unused equipment was for sale. It’s not a game of spending the most money, but trial and error for me. Unfortunately I was a little intoxicated when I was fed up with the poor performance of the $8 China dongle which led to purchasing the Radarcape. I have the wrong one with one port and no video out (I don’t want video out BTW)
OT: I attached a Raspberry Pi to my scope and generated a pulse wave on the GPIO and compared it with a function generator. It is amazing what a difference the “kill all rf”, disconnecting the HDMI cable, keyboard and mouse make. Putting ferrite clamps on the ethernet cable and USB power cable contributed big time. I know that a pulse wave via the GPIO is different from processing antenna signals, but these external connectors/wifi/BT impact the Raspberry.
so - finally all parts of the script seem to work as intended.
i feel a little bit guilty - when you were asking back then for the ‘optimal setup’ and i posted all the jetvision stuff i did not realize that you really would buy every single pice :)))
and yes - often i choose the higher quality too - but with the yagi i used it only in my attic and saw no reason for the more robust/professional variant. as far as i know there is a radarcape without video but with secondary receiver. i’d go and ask guenther how to upgrade. he is a nice guy and will find a solution. btw. he told me that the radarcapes are manufactured just 5 miles from my house.
haha - there is another forum member ‘jonhawkes’ who also has a radarcape and is very into all this ferrite and noise thing. while i believe that your measurings were right - my personal findings were that all this has nearly zero influence in the number of aircrafts and messages you decode. in ads-b the problems with message collisions etc. is such a higher level that all these other factors nearly are neglectable. for use as a traditional sdr-receiver this is of course different …
edit: yes found the one without video but second receiver
edit2: btw i use the raspberry where gui is installed simply via vnc from other computers in my network - this way no mouse/keyboard/hdmi is needed nor makes noise
I see no reason for guilt, you definitely inspired me to explore unknown options for me back then, but I didn’t use it as a shopping list. I definitely bought the wrong radarcape version, but climbing the ranks is not my main goal. I have different setups and professional grade test equipment, but I can’t explain the differences. That’s the reason for the splitter: to feed every configuration with the same signal.
I’m just a nerd, our house is on all three phases connected to the grid due to the amount of projects I’ve worked on in the past. When you showed the pic with the Yagi, my first thought is: “I should experiment with that”, followed by: “could a directional antenna work in my area”. Five seconds later I’m convinced it would be fun to play with one
Buying that splitter is a textbook example of poor decision making. It requires a DC injection of at least 12V which none of the common dongles can provide. An undocumented feature is that a DC feed connected to the external power connector is also passed on to the antenna output instead of just powering the splitter. 12V is much more than all common LNAs can handle.
what is/was jetvisions advice to power the splitter in an ads-b environment?
p.s. i can fully understand that you are annoyed spending so much money and now it does not work at all. but the dc passing is a common behavior and on jetvisions website you find this well documented. i’m sure together with guenther you will find a good solution for this problem!
p.p.s. another idea were to disble the dc pass - with inserting a dc-block between splitter and lna …
Don’t get me wrong, I’m not blaming anyone other than myself. I should have asked more questions before buying especially because the documentation is pretty limited. When I first fired it up powered with a V3 RTLSDR dongle via Bias Tee, the results were below par and within seconds the Pi casing started to melt - that’s how hot the dongle got. I hooked it up to a function generator and an oscilloscope and discovered what was happening. I used a lab DC power supply and found out that 12V is required to get acceptable signal losses. The easiest solution is to solder a LDO on the LNA. Let’s see what that does first.
nobody is to blame for - it was just a mistake that always can happen before soldering at my lna i’d just use one of these things (nearly zero cost and insertion loss): https://ww2.minicircuits.com/pdfs/BLK-89+.pdf
btw. why did you wanted an active splitter - the loss of 3,2db a passive has shouldn’t be a problem when used directly after the lna with about 15db gain?
It was meant to be used for different configurations. I’ve attached an antenna mast to the side of my house with the intention to attach multiple different antennas with and without LNA. The splitter would allow me to feed the signal to multiple dongles and a spectrum analyzer. To determine the differences between antennas it would merely be feeding the splitter with a different cable. As I want to have test equipement simultaneously attached, I needed at least a three port splitter. Secondly, I want to test whether your statement is true or not. The effects of saturation or the impact on the signal to noise ratio tests are on the agenda, just as the quality of different filters.
Where can you buy those DC blocks? I also have a hard time finding a 12V Bias T. The splitter is currently connected with crocodile clamps to the alternative power input. Fine for the time being but not for the long run.
ok - understand if you have more than two output ports then no question an active splitter is the better choice. i just did a short google search (dc block sma 50 ohm) when i read your 12v problem and thought this were the easiest way as it’s probably the same that is built into the global invacom splitters with dc-blocking. there are different options - but as the uputronics lna and my splitter are mini-circuits i thought their dc-blocks are fine too. just call mini-circuits germany Mini-Circuits Bauteile beim offiziellen Distributor in Deutschland!
