FlightAware Discussions

Do I Need A Filter?

See my graph above.
Seriously i am not such an enthusiast. For me it’s a hobby and my setup is working to my satisfaction. So i am fine with it.

1 Like

Where do you get your data from?

No one else (with any RF experience) agrees with you.


No. An external filter is beneficial in many installations where a receiver with an in built filter such as the ProStick plus is used. It depends very much on the RF environment at the receiving location. At my location, an additional filter increases the number of aircraft and messages received by a ProStick plus by about 30%. This is because I am close to FM and DAB broadcast transmitters as well as a large number of PMR and GSM transmitters. Without any additional filtering these transmissions cause intermodulation products to be generated by the amplifier in the ProStick before the in built filter. Filtering out these signals significantly improves reception at my location.

1 Like

The problem of the blue stick is that the filter is designed after the LNA

I wouldn’t really describe it as a problem, all design decisions are a compromise. If you put the filter first you will inevitably increase the noise figure of the receiver. Having the amplifier first means you have to decide such things as how far you need to go with its linearity, 3rd order intercept point etc. Each of these decisions have implications on cost, power consumption and so on. Personally I think the ProStick plus is an excellent example of a design where these compromises have been made correctly. It works fine as it is for most people. Making one that would work perfectly out of the box here would be too costly and I think would be overkill for 90% of installations.


Agreed. Let’s take it as a kind of limitation based of the design

Agreed. I started with it my journey of flight tracking and was never disappointed.
It’s way better than lot of other sticks

I would like to see a side-by-side comparison of two identical devices, same antennas same cabling.
One direclty, the other with the improvements of hdtvspace


A parallel study of the AirSpy Mini vs ProStick
Parallel Receivers
A similar setup could test with various filters as the variable.

1 Like

I am not a technician on this and have almost no knowledge, but from logical perspective i would assume that filtering out the unwanted frequencies and then amplify the “good” ones makes most sense, or?
Otherwise i would also amplify the noise which makes it harder to filter the good ones, or am I wromg?

Normally a filter before the LNA is not good.

With some LNAs, like the uputronics models, it works a whole lot better to have a filter outside the of the LNA. Otherwise, in some locations, it gets overloaded. I tested mine with a radarcape, Airspy, FA dongles and RTL-SDR dongles. For my location, the filter on the outside of the LNA always produced better results.

I do need to re-test with the RTL-SDR LNA. Many people have seen better results from strong interference when using this LNA.

It is common to have a filter on the input of a receiver in the professional wireless broadcast world.

For a specific example, here in the UK horse races can be covered by 3 or 4 broadcast companies, sometimes more, all running in the same 2GHz band as WiFi etc.

If broadcasters didn’t use filters on the inputs of their receivers, they can become overloaded by adjacent channel interference when another mobile camera Txs is closer than the one they are trying to receive.

Yes, they do literally run in adjacent channels due to the limited bandwidth and number of wireless camera TXs.

Most racecourses now have strong WiFi access points, just to make the available band even narrower.


As a general rule of thumb - perhaps. BUT (There is always that in the RF world), it depends on location just like this entire thread. Some environments may call for filtering prior to the LNA due to excess noise/interference which will overload (think of sound clipping) the amp. Yes, there will be insertion loss prior to amplification which is usually a detriment, but in some cases it may be required. Garbage in, extra distorted garbage out is not always preferential, so there is always a compromise.


I have learned it both from academic study of RF engineering and having worked as a professional RF engineer for many years. I design radio systems for a living and sorry, you are just plain wrong. Have you ever seen a professional RF installation? If so, have you ever wondered why all those really expensive cavity filters are installed immediately in front of the receivers? They are not there to compromise the performance as you seem to think.


I asked this earlier, but I didn’t get an answer. What is your methodology for testing this? You need to provide some evidence if you are going to make broad claims. Some installs definitely do benefit from an additional filter, regardless of what you believe the rule to be, though I’m not going to claim that it’s necessarily the best solution if you were designing the whole receive chain from scratch (maybe they just have too much gain in the first place).

(one interesting comparison would be an external filter vs. a broadband attenuator with similar loss)


I see what you are doing here. :joy: This one is not intuitive.
In general, you need the lowest achievable noise figure in the band of interest, but this could cause IMD problems if the front-end is wide open to out-of-band blockers. Why? The linearity and sensitivity are duals. Improving one will degrade the other.
One way of evading this duality is using filters. All filters will exhibit some insertion loss that adds up to the noise figure of the receiver, which reduces the sensitivity, but the reward is that the out-of-band blockers will be removed from the equation. The utility of the filter only shows up when its insertion loss is lower than the de-sensing that happens without filtering. In some situations, reducing the gain a notch (ie. using an attenuator) will be just as good as adding a filter.
On the other hand, even when IMD is not the problem, having the lowest possible NF might be overkill because of some local noise masking the weak signals, or even when the antenna has enough gain to overcome the system noise floor. In this specific case, an attenuator will improve the linearity without degrading the sensitivity and will turn out just as good as a filter, and probably even better if there are some in-band blockers.
Commercial receivers are optimized to cover the majority of use cases, but not all. The RTL-SDR ADSB LNA has a sloppy but low loss band pass filter in front of the MMIC. This filter is implemented using a PCB printed inductor, which is basically a wide band resonator around 1090MHz. The attenuation of out-of-band blockers is relatively low, but it is sufficient for the following amplification stage to operate with a decent dynamic range margin. When combined with the front-end stage MMIC (half of it, actually), the combination defines the noise figure of the system. The subsequent filtering gets increasingly sharper until the band is clear-cut from the rest of the spectrum, without changing the noise figure by much.
How to make the best use of this filtered preamp? Use an antenna with enough gain that gives a noise floor that is barely stronger than the noise floor of the preamp. But again, one must pay attention to the radiation pattern of the antenna.
So, back to the question “Do I Ned A Filter”, until something non-electronic is used for radio communications, ultimate answer will remain: It depends.


What a load of drivel.
A decent cavity or inter-digital filter has less loss than most peoples feeder.
If you are in a high-RF environment, the amp (be it an LNA or not) will still amplify all the unwanted signals by 20~30dB (which you are then expecting the filter to remove), this can easily saturate the amp.
In this situation, putting the filter first is the obvious solution.


I did make something very like that test, with a goal of learning if the Mode S signal range is Noise Figure limited.

First I ran my system for a few weeks and plotted maximum ranges. (It’s a pretty good system with >200 nmi range at all azimuths)

Then I added a 3dB RF attenuator between the antenna and the front end, and I raised the SDR gain by 3 dB. This should have kept the RF system gain the same but worsened the NF by 3 dB.

The system max range was no different on average.

To me that implied the system performance is not NF limited, but mainly is radio horizon limited.

And yes, my cavity filter is placed between the antenna and the LNA.

The transmit power can be several hundred watts (at least for airliners), which is easily enough to reach the horizon in clear air, so I’m not surprised that’s the result you got. For an antenna in clear air, there shouldn’t be much difficulty receiving aircraft out to the terrain limit.

Having a lower noise floor should be more significant in less than perfect conditions - where there are trees in the way, or inside a loft space or something like that, or for receiving light aircraft with much lower powered transmitters at low altitude.


In my dense urban location the spectrum is far noisier than kTB, I see the external noise floor at least 40 dB higher than thermal near 1 GHz. So as you say the receiver NF is not very significant.

You’re also right that the LOS link margin is high, especially since the intended receiver is a ground interrogator with a high gain antenna.

That makes me suspect the standards were based on IFF and jammer margin is included. The aircraft transmitter power specification may be based on what could be obtained from a 2C39 triode circa 1945. That’s only my speculation.


Can i run airspy mini with heatmap.py
No supported devices found ?

Negative, give SpectrumSpy a whirl - it’s packaged with SDR#. :slight_smile: