RTLSDR calibration - which reading to believe?

Bermuda1

New Member
I have been trying to receive ADS-B traffic using a "cantenna" and an RTLSDR. I am currently getting around 60 miles coverage from the upstairs window. My setup is just the cantenna, RTLSDR and a Raspberry Pi. No pre-amp or special co-ax at this time, just a short length (about 70cm) of RG58U + a short length of wire making up the BNC to SMA adapter.

During my research I found a reference suggesting that calibrating the RTLSDR migth bring some improvement, so for this purpose I downloaded and compiled kalibrate-rtl. The version cloned directly from the Github respository did not run on the Pi, which is another story, but this was ultimately resolved by downloading the ZIP package from this Github page:


This compiled and ran fine. However, I tried to calibrate using tow options:

Option 1) use kalibrate to scan for GSM900 signals and use the strongest one to calibrate the ppm offset.
Option 2)use rtl_test -p

The first option returned a value of -38.299. The second one after a few minutes settled at around 78. I re-ran the first option again and got -38.287 so a very similar reading again. So why are these two readings so significantly different, and which one do I trust/use?

I did play with both, but I'm not sure that I am seeing much difference to be fair, perhaps only a subtle one. The same can be said for using --enable-agc or --gain -10 on the dump1090 command line. The difference seems minimal. I have yet to experiment with using positive values for gain.
 

wiedehopf

Member
What is the make of your dongle?

With good dongles like the rtl-sdr v3 dongle you don't need any correction.

For the others the correction is also not too relevant according to tests done by some people.

Maybe try making a quick spider and see if it performs better than your cantenna:
https://discussions.flightaware.com/t/quick-spider-no-soldering-no-connector/19181

Having no other electronic devices in the same horizontal plane as the antenna can help as well.
But if the window is all you have to work with then you will be stuck with it.
 

Bermuda1

New Member
The one I am currently using is a v2. Not sure about the two that I have just ordered, but they are these:


Thanks for suggesting the quick spider antenna. I had a look and found a few bits that I could use to build a quick spider. I had a short 30cm approx piece of co-ax and an odd metre of twin and earth mains cable out of which I pinched the earth wire to make the radials. I also found a salvaged BNC socket. I reckon it took me about 30 mins to put together. I have attached a photo. The foil near be bottom is to shield the joint with the inverted BNC socket. I placed some insulating material around the core.

Performance seems to be at least as good as, but possibly slightly better than the cantenna. The difference is subjective but I think I am picking up aircraft from slightly further away. The difference, if any, is quite small, probably no more than 5-10 miles and I seem to be getting perhaps a couple more entries on the list than previously. Still, any improvement is welcome and one advantage of this antenna is that I don't have to come into close contact with the sharp edges of a metal can every time I attach or disconnect the co-ax! I did file down of course, but they do remain pretty sharp nonetheless and great care is needed if one is to avoid cut fingers.

I guess this kind of range is probably the best achievable without a pre-amp, so I guess that will be the next step although I will probably have a go at building a bow-tie antenna for airband first.
 

Attachments

Last edited:

wiedehopf

Member
Yeah the rtl-sdr blog produced dongles are better than the one you linked, they have a metal case and reduced noise.
You can the the rtl-sdr v3 dongles here:
https://www.rtl-sdr.com/buy-rtl-sdr-dvb-t-dongles/

Those also have a bias-t that can be activated via software, which is necessary for example for this LNA (https://rtl-sdr.com/new-product-rtl-sdr-blog-1090-mhz-ads-b-lna/)

Works very well together and due to filtering and amplification normally greatly increase range.
(Limited by terrain of course: https://discussions.flightaware.com/t/what-is-the-maximum-range-i-can-get/17248)

The antenna looks very nice :)

As you have probably already read, with a can and by reducing the whip length the DVB-T whip antennas can be made into acceptable cantennas :)
 

Bermuda1

New Member
I am experimenting with this one at the moment:


However, I have seen those from RTLSDR.COM on eBay and also these which seem to be similar:



The UZEUNER one comes with the necessary co-ax adapters which would be useful as I don't have any here. Also the warehouse is in the UK and the wait time is 4 days. Would this be OK, or is the RTLSDR.COM one better?
 
Last edited:

wiedehopf

Member
I wouldn't buy those. If you want to buy via ebay, use their store:
https://www.ebay.com/str/rtlsdrblog

The rest of the offerings pretending to be rtl-sdr.com blog v3 dongles are fakes.
(Or one of the offerings doesn't even pretend that it is the rtl-sdr blog v3 design.
It's of course still rtl-sdr compatible but not the design from the blog)

If you have to buy adapters separately that's unfortunate but unavoidable.
And you'll have to wait for shipping of course but you really don't want the fake ones.
 

Stealth

New Member
Anyone know much about this one:


It claims to have both a pre-amp and filtering built-in.
It would be ideal and reduce the amount of clutter?
It does have a pre-amp and a filter, although at the receiver end of the cable is not the best place for a pre-amp. I used one for about a year and it was fine. One thing to be aware of is that an additional filter may almost certainly be required if you are in a strong cell phone signal environment.

wiedehopf is very supportive of the rtl-sdr.com dongles and he will have done a fair bit of testing with them. The aluminium case for screening is a strong plus.
 

wiedehopf

Member
The external LNAs are better but the Flightaware Prostick does work.
Depending on the electrical noise though you may need the FA filter in addition.
(The LNA may get overloaded. The external rtl-sdr LNA has a high-pass filter on the input and lower gain on each stage to avoid overloading problems)

FA pro stick plus + filter is almost as much as just getting the rtl-sdr dongle + LNA, that's why i recommend the package that has the better performance.

You could also get just the rtl-sdr v3 dongle now and maybe buy an LNA later if you are short on cash or just want to experiment first :)
 

Bermuda1

New Member
FA pro stick plus + filter is almost as much as just getting the rtl-sdr dongle + LNA, that's why i recommend the package that has the better performance.
Having considered all of the helpful advice, I have decided to order the RTLSDR.COM dongle + their ADS-B LNA as this does seem to be the best option. It will be interesting to compare their performance with just an ordinary RTL dongle.
 
Last edited:

Bermuda1

New Member
Have to say that having tried out the RTLSDR dongle from RTLSDR.COM, performance is much improved and I can get planes up to a radius of about 100 miles now. This is still well short of prediction, but much better. The next step is to figure out how I get a proper map in Planefinder....
 

wiedehopf

Member
What do you mean a proper map?

Which dump1090 are you using?
I would highly recommend using dump1090-fa.
(I've written short instructions on how to install it and remove the old dump1090: https://github.com/wiedehopf/combine1090#installing-dump1090-fa)
If you are using dump1090-mutability 1.15 that is also a good option.

Other dump1090 options have worse reception because the demodulation isn't as good.

If you don't have the LNA yet, you should use gain -10 to get the best reception.
Having it in the window it's unlikely to get the full theoretical range, for that you need a position overlooking most stuff in the area.
 

Bermuda1

New Member
What do you mean a proper map?
Sorry, confusion on my part between PlaneFinder and PlanePlotter. It was getting a proper map in the latter that I was referring to. When you start you get a very pixellated map which you can change to an outline, but no indication of cities or major roads etc. My trial period has now ran out, so I can't do anything further with it now anyway and I don't have enough information to base a purchasing decision on. Its probably OT here anyway.

If you don't have the LNA yet, you should use gain -10 to get the best reception.
I experimented with that using the standard dongle and there seemed little difference between gain -10 (auto) and gain 36.4 which seemed about the optimum manually set gain value. Using gain -10 on the new RTLSDR.COM dongle with the LNA attached gives very poor results, while setting gain 36.4 appears to work quite well.

Which dump1090 are you using?
I would highly recommend using dump1090-fa.
(I've written short instructions on how to install it and remove the old dump1090: https://github.com/wiedehopf/combine1090#installing-dump1090-fa)
Thanks for the recommendation. I tried dump1090-fa but got just one plane! I think this is because of 'gain -10'. I tried to follow the config instructions here:
However, I seem to have no piaware-config or /boot/piaconfig.txt on my system. Instead, I edited /etc/default/dump1090-fa and restarted the service, which does seem to have applied the new setting, but this time no planes at all. I now think the problem may be that the biasT is not getting enabled and the LNA has no power, but I am unsure where/how to enable it so that it starts with or before the service? In any case, I found that installing dump1090-fa broke rtl_biast. This is what I get when I now try to enable biast:

Code:
Found Rafael Micro R820T tuner
rtl_biast: symbol lookup error: rtl_biast: undefined symbol: rtlsdr_set_bias_tee
Even though I have now removed dump1090-fa I still got the same error. I had to remove the librtlsdr0 library that was installed with dump1090-fa (but not auto removed) and re-install the rtl_sdr and rtl_biast versions from compiled sources to resolve the issue.

I don't mind giving dump1090-fa a try, but I would need to find a version of rtl_biast that works with it.

Having it in the window it's unlikely to get the full theoretical range, for that you need a position overlooking most stuff in the area.
Point well taken but the best I can do is put it in the loft, which I intend to do, so this will probably be as good as it gets.
 
Last edited:

Stealth

New Member
Regarding Planeplotter, I use it myself, but unless you plan on doing more than just looking at ADS traffic and uploading to a couple of sites it's probably not worth the effort.
 

Bermuda1

New Member
I am making some progress.

I removed the libraries that rtl_biast installed when I did 'sudo make install' and then re-installed dump1090-fa and re-compiled rtl_biast. This time rtl_biast worked without error.

That solved , I then moved on to getting bias-t enabled before the dump1090-fa service starts. I came cross some instructions on how that can be done here:


I can report that following this worked OK except that I had to do one more thing at the end of the process. After saving the file /etc/systemd/system/dump1090-fa.service.d/bias-t.conf in the nano editor, I also had to run this:

Code:
sudo systemctl daemon-reload
Otherwise when the service tried to start it generated an error indicating that the service had "changed on disk". After running that command, the service starts normally.

It is also worth noting that the path provided to the 'sudo nano' command must be the actual path where the rtl_biast binary is located on YOUR system. I place all of my sources in a ~/src direftory so I had to amend my path accordingly. For the final configuration I actually copied rtl_biast to /usr/local/bin where all of the other rtl_ binaries are found and supplied that path to the 'sudo nano' command instead.

So now bias-t gets enabled at startup when the dump1090-fa service starts. The receiving range does seem to have increased a bit over the generic dump1090 as I appear to be catching planes as far north as Preston/Leeds, as far down as the south coast and even one or two planes to the East over the North sea. Auto gain (-10) seems to work about the same as setting a gain of 36.4.

Regarding Planeplotter, I use it myself, but unless you plan on doing more than just looking at ADS traffic and uploading to a couple of sites it's probably not worth the effort.
Thanks. I will bear that in mind! I have a second dongle and managed to get ACARS reception to work but that data doesn't neccessarily have to go into Planeplotter.
 
Last edited:
Top