Do I Need A Filter?

Seriously i am running both RPi with the Pro Plus stick connected directly without any issues. Seem to be an urban legend.

1 Like

My airpsy is connected directly to the pi. When I was using the rtl-sdr.com v2 dongle with built in bias-t to power the amp, I did have it connected to a powered USB hub, but that was mainly to reduce the power load on the USB port. Since switching to the pi 4 and airspy, I switched to an external bias-t powered from the 5v pins on the pi and am using the USB 3 port which has a higher current rating.

I’m not convinced of the value of separating the receiver from the pi for this application - ads-b isn’t really about pulling signals from near the noise floor since with a decent antenna you are almost always going to be limited by terrain rather than signal.

I did try using a USB extension with some ferrites on it, but didn’t see any change in reception so I took it out again. I think as long as you are using a decent quality power supply it should be good enough.

There are two main problem with longer cables (and some active repeaters, and even some hubs):

  1. voltage drop over poor quality cables; these dongles are power hungry;
  2. the RTL2832 has essentially zero buffer space, so any errors at all on the USB bus (e.g. due to EMI) result in sample data being dropped; the data is long gone by the time the host does a retry. This doesn’t matter much for general use where you can tolerate losing a few milliseconds of samples, but it fundamentally breaks multilateration which is relying on a predictable sample rate.

A good quality short cable is fine, and anecdotally with some ferrites it might help reduce interference from the Pi (though I haven’t systematically tested it myself). Plus it makes physically dealing with the dongle much easier if you don’t need to worry about a rigid connection and blocking other ports.

On one Pi, I am running both 1090 and 978. Two dongles are physically impossible to plug directly into Pi. I use one directly plugged in, and other through a short USB cable.

On other PI I run two instances of dump1090-fa, two instances of piaware, two instances of planefider feeder, two instances of flightradar24 feeder, and two instances of adsbexchange feeder. For two instances of dump1090-fa, I plugged one dongle directly into PI, and 2nd dongle (being physically impossible to plugin directly), is plugged through a short USB extender cable.

Both Pis and all 4 dongles are running perfectly well. All the 4 dongles used are listed below. All these have plastic cases, no metallic shielding:
-Flightaware ProStick Orange
-Flightaware ProStick Blue
-RadarBox FlightStick Green
-Generic DVB-T Black

 

For me all these worked very well :slight_smile:

Do you realize that you are explaining this to the guy who wrote much of the software that we are using? I am going to chance a guess that he has a pretty good idea what he is talking about.

1 Like

Unfortunately, for those using an Airspy (R2), forget about using anything much longer than the stock cable & quality is king. You can however get away with using active extensions with R820Tx based radios since they don’t pass near as much bandwidth through the pipe.

Rule of thumb when it comes to RF/noise is to keep the cables short and use high quality with ample shielding/grounding - as OBJ mentioned, voltage drop can and will be an issue, especially on units that have built-in LNA’s.

USB 2.0 ports should be fine to plug in direct as they are shielded by design (some better than others), USB 3.0+ is a different story - even though they ring at higher frequencies than we’re messing with (in specific ~2.4GHz range), noise is noise. That said, many use POE and/or either use or don’t disable wifi/bluetooth anyhow (Edit: OBJ addressed this on the stock image by implementing RFKill), so there isn’t much they can do to keep things super quiet to begin with, so no use losing much sleep over…

In the past i had two different USB-cables. One 1m and one 2.5m
The longer one did not show any difference to the directly plugged in stick, the 1m cable ended up in a 30% lower performance. Maybe simply a bad one.

Once the crisis is over i will go buying another USB extension and test it. I wanted to do this anyhow as the cable is drawing hard on the stick and i want to get it away from RPi

However i don’t have issues so far with noise (beside the noise my wife creates regarding my useless hobby :slight_smile: )

1 Like

ProStick wrapped in aluminum foil - RF shield

1 Like

That’s really crap advice.
While it’s great if you can find the cable + ferrites that work, you certainly can’t “expect" it to perform well unless you have tested and confirmed that it has not crippled your site

I had my whole device including the USB stick (connected directly) in a metal christmas cake box as an outdoor solution

The only impact i had was a significant drop in WiFi range due to the isolated box.
No impact to the ADSB environment.

How can i measure an impact by the interferences you’re talking about?

And I said it a couple of yeas ago.

Anecdotal evidence from users on this site show mixed benefits when adding a USB cable between the Pi and dongle.

If yours works - keep using it, but did you actually do a side-by-side comparison? (I did).

Did you read what you linked?

Can you see the bit about 2.4GHz?
ADS-B operates at 1.09 GHz

Also, 2.4GHz is well above what an R820T/T2 can receive.

Unfortunately, the white paper you’ve linked has only compared “USB-3 Not Connected” to “USB-3 Connected / operating”, but not “USB-2 operating on USB-3 port”. It’s probably reasonable to assume that a USB-3 port operating at USB-2 is not worse than a native USB-2.

Additionally, the white paper shows only one broard band spectral plot (fig. 2-2), but that was made by “directly probing one of the USB 3.0 transmit-pair signals”. ie. it is the power spectrum of the data stream, not what is radiated.

In summery, I have done what you suggest, and with the components I used, I could observe a very significant drop in performance.

2 Likes

RF is touchy enough (and the 1090MHz demodulator and the rtlsdr dongles themselves have enough quirks) that It’s quite reasonable to test and not just assume that the received wisdom is correct.

What’s the process you’re using to test changes to see if they improve your reception?

5 Likes

Doesn’t that require an USB 3.0 device to be connected to the Pi?
I wouldn’t expect USB 3.0 interference if you have only the SDR plugged in.

UNNECESSARY!!!

scan-without with-filter-850-900

 

Yes if the aluminum foil is wrepped loosly leaving air gap between plastic case and foil, as the air in gap will act as heat insulator, retarding flow of heat.

However I have wrapped the aluminum foil tightly around the case leaving no air gaps, and now alumunum foil will conduct heat without any heat insulation.

Hello,
I setup an ads-b receiver with a RTL-SDR dongle and external antenna.
All is running with pretty good results (FA’s statistics page reporting few positions in the range of 250+ mi).
I’ve done a scan and found signals at 800 and 950MHz (GSM I think).
Would a filter or preamp improve reception?

I tried to change gain settings of dump1090-fa and the best value is auto-gain (with max messages fall drastically).

Thank you

1 Like

Please specify which make SDR you are using.

That’s normal when using an uamplified SDR.

A filtered preamp would likely improve reception, yes.
(uputronics 1090 MHz preamp or rtl-sdr triple filtered LNA)

You can also try going to the blue FA prostick+ or the green AirNav Flightstick (those two have a filter / preamp built in).

But such things aren’t trivial and there is no guarantees it will improve reception.

It really doesn’t matter - if we want to get down to brass tacks as it concerns noise, USB 2.0 devices spit noise in 480Mhz multiples (don’t forget sinc at 1/2 clock freq’s as well), so there is a perceptible spike at 960Mhz from USB 2.0 by design - we’re behind the 8-ball out of the gate. I don’t think these devices have chokes?

Canada does not have GSM900, but I still get noise at 940 Mhz. The source seems to be the USB noise.

scan-without with-filter-900-950

I’d expect 940Mhz to be trunked radio or something like that, but it would look more intermittent I’d think… Guess it depends what the timeframe of the capture is as well. Shouldn’t see anything USB related at 940.