Double aircraft tracking on one system in identical setup, seeking root cause

FlightAware ADS-B Flight Tracking
1

I’m seeing 65/45 (81 msgs/sec) versus 122/105 (645 msgs/sec) on an identical ads-b setup on my roof. The only difference is a collinear antenna 4 feet down the mast on the 122/105 system. Let me describe the setup and what made me choose that tentative root cause.

I have 2 ads-b systems setup on my roof separated by about 30 feet, both are around 45 feet above ground level with each other). Each uses a 20ft run on RG8x down to a RPI powered by POE.
The antennas are 2 spider antenna’s that I built (8 radial groundplanes) and tuned with my VNA.
The dongle’s used on both are: Flightaware Pro Stick Plus (gain setting -10 on both).

This morning I noted both were sitting around the 80-100 positions per second value, which was odd since before I went on vacation the higher performing one was typically over 700 messages per second. I double-checked the dump1090-fa settings between both and then went out to inspect the antenna’s.

Here’s where it gets weird: I noticed the RG-174 lead on the collinear antenna that I had built and installed 4 feet down the mast of the higher performing system had come unscrewed and was dangling down the mast.
I hooked it back up and immediately noticed the messages per second jumped back up to over 700 per/sec on the OTHER antenna + dongle.

This collinear lead is connected to a different dongle on the same higher performing RPI setup but is not being used. That dongle is a RTL blog + RTL blog LNA.

To me this is very strange that all I have to do is unscrew a lead to an essentially unused collinear antenna and get much lower positions per second. How could a 1090mhz collinear 4’ft down the mast from the active spider antenna + dongle effect the active one?

EDIT: changed positions/sec to messages/sec

2

Best guess: Your higher performing setup is actually the RTL dongle connected to the colinear.

How do you configure the dongles so the correct one is chosen?
Easy way to find out: Unplug the RTL dongle :wink:

Also i would be skeptical of two basically identical setups performing that differently.

P.S.: It is messages not positions per second or are you really talking positions?

3

Here’s my higher performing site:

I select any of the 3 dongles via –

root@odc2lora:~# rtl_test Found 3 device(s): 0: Realtek, RTL2838UHIDIR, SN: 00000001 1: Realtek, RTL2832U, SN: 00001000 2: Realtek, RTL2838UHIDIR, SN: 00000002

and thus use: --device-index 1 (to select the Flightaware dongle)
Note: I’ve written serial numbers into eeprom on all my dongles

I know which one is which because I can do interactive on the ones that dump1090 is not using and it fails to open them on the one it is using…

it’s messages per / second, sorry for the mistake.

4

It would still be a good test to pull out the RTL dongle and see what happens :slight_smile:

If the messages per seconds drop it could be that somehow the amplified signal is in mysterious ways feeding over into the other dongle (signal from the LNA)

5

Wouldn’t you use --device-index 00001000 to select the dongle?

The other index can change on reboot i believe, can it not?

6

Well, perhaps infrastructure issues versus magical amplification…
Pulled the collinear dongle (easy cause its RG-174 not RG8x):
778.2 messages / sec before
677 messages / sec after

Well within 2 minute change over variance.
So now thinking cable issues on the RG8x since I had to lower the mast to reconnect the RG-174 and I re-tightened all.

7

Indeed it is.

No idea what’s going on then and why the two dongles differ so much in their performance.

Anyway if you have the 1090MHz filtered LNA it should provide the best performance in my experience. Gain -10 won’t give you good performance on that one though. A gain of 23 is a good starting point for that setup :slight_smile:

On the other setups lower gains might also give you better message numbers without reducing the range.

8

Oh another idea: Analyze the complete thing of the cable + antenna with your VNA.

Could that tell you about cable problems or not?

9

absolutely it will tell me, one antenna I built I tuned with a short pigtail, another with the final cable that I used on the mast. The tuned with cable worked much better than tuned with pigtail.

I ran piaware-gain-optimizer and tested all the high values… I think it was 22.7 that gave the best results with my home-made antenna.

10

Have you ever tested the rtl setup with one of the spiders? :slight_smile:

Or is your highrange station already the rtl dongle? I get a little confused with the 2 stations 4 dongles 2 pis :wink:

11

Not with a tuned one as yet. I tested it with the Flightaware stick antenna. That antenna had a different radiation pattern versus the spider ones I built. I would mostly pick up high altitude planes, while the spider did both high AND low altitude planes exceptionally well. So I removed the Flightaware stick from the line-up for now.

At some point I’ll test a spider + RTL + RTL LNA. At this point I was more concerned with getting the two setups working the same to give me a production one and another for testing various antenna designs.

I’m running the 2 active ads-b stations with Flightaware dongle’s. The built-in filter + LNA on the Pro stick simplified my outdoor case wiring.

12

Are you sure the gain wasn’t just too high (low altitude planes generally have the strongest signal due to them being close by and close to the horizon)? The FA antennas are regarded quite highly :wink:

You can also just use the LNA indoors. No really necessary to put it outdoors.

Yeah i can understand. But obviously the 2 cable runs with antenna are very different right now :slight_smile:
Or it’s the dongle but that can be easily tested.

13

Adjust manually the gain on both of them.

14

If the cable attenuation is high enough then -10 is the best setting as it provides more gain than 49.

But if you need -10 with the prostick you definitely need to improve your antenna / cable setup :slight_smile:

15

Are both your masts grounded?
Are both your cables grounded?
A ground on one vs the other could make that much difference, I’d bet.

16

The high performing one has a heavy copper lead grounding it with 30ft to a ground rod and a splice to the metal roof, the other has 2 strands of cat5e grounding it to the metal roof (leftover from a bad cat5e cable on the roof).

I suspect the metal roof while mostly at ground level is probably not as good a ground by the time it makes it the 40ft thru the roof to the weaker performing mast/antenna. This one is in the middle of the metal roof (instead of on the side like the other) and is hard to reach with a ground cable that reaches directly to an earth rod. Both antennas on the masts are at least 15 feet above the metal roof highest spot.

17

Just noting something because you seem to have quite a bit of a setup: Don’t know the data on your particular cables, but the types are generally not too good for 1090 MHz.
RG-8X might be ok but RG174 will kill the performance even if your colinear is well made.

18

I’ve only been doing these sites since Dec 5th, the RG174 was the initial coax for the first site. As I learned more I replaced the ads-b cables with RG8x, and moved the RG174 to other antenna’s (8xxMhz stuff). I now have LMR400 to install on both of the ads-b sites (replacing the RG8x). So it has been an iterative process.

Adding a 2 foot pigtail of rg174 to the rg8x end in one experiment cut around 10% of my messages / sec. I removed that one fast!
I’ve also noticed a large difference in quality of the cables connectors. One RG8x I tried if you wiggled it the number of planes would drop off by half.

19

I found root cause for the difference between the 2 sites.
Let me describe the site hardware first:
(high performing) Odroid-C2
(moderate performing) Raspberry PI 3+

I disabled both bluetooth and WIFI (its powered by POE so using ethernet) on the RPI3+ and now have the same number of planes and messages / second on both systems.
So apparently there is some interference with those enabled on the RPI3+.
Anyway I now have two high performing systems…

20

Hmm that’s interesting. Don’t think i ever had any problems with that but i guess i wouldn’t know always using WiFi. (RPI 3+)

Right now i’m using the RTL-blog LNA so i don’t think WiFi intereference is a factor as the signal into the dongle is much stronger.
But maybe it interferes even in a later stage. Maybe i will try an ethernet cable some time :slight_smile:

I’m pretty limited by terrain anyway so it’s hard for me to tell much difference.
You seem to have a pretty good “view”.
Curious how the RTL combo will do at your location.