Janos Konya
Member
Those who do not have the opportunity to create a custom filter, or do not feel like creating one, can choose one from the offerings of stores.
It is not always easy to select the right piece, as graphs and tables that accumulate technical parameters do not help the average person how to consider the data in their own circumstances.
I’m trying to simply explain the things you need to know minimally beyond trial luck, especially for the simple end user.
- Question of necessity
As the antenna is taller than nearby landmarks and buildings, the receiver’s input circuits are besieged by strong signals that don’t interest us. This phenomenon can also be a problem for professional receivers, but this is especially true when a low-cost, software-defined radio is in use. The problem is that tuning is not done in the "traditional" way, so preselection is not provided at the input. In exchange for almost unlimited tuning freedom, we get other possibilities for error. I’m not immersed in the technical details, but I just need to mention one example: Even the SDR receiver’s well-designed pre-circuits are unable to prevent overdrive and distortion on the 8 or 10-bit A/D converter, because of the given low dynamic range.
I read an entry in which there were occasionally detected airplanes when the antenna was placed high, even less than before in the window of the room. The owner explained the phenomenon by saying that the signal from the planes was too strong. Well, not like that. The airplane signal can only be too strong if your antenna is at the end of the runway and the airplane is just flying over it. I'm 2 miles away from an airport, but I haven't even experienced 70 percent of the maximum possible signal strength. Read the signal strenght of the closest airplanes to make sure that it is true.
Extremely strong stations in the commercial FM band, ubiquitous GSM towers and all kinds of radio frequency signals, even from industrial sources reach the antenna. (indoors, near the room window, this is not such a striking phenomenon)
-> Very often, if you need to turn off automatic gain control (AGC) and manually significantly reduce the input sensitivity for the signal decoding to work - you don't really need attenuation, but a good filtering instead. (!)
The small dynamic range mentioned above can be more easily preserved to decode aircraft messages by checking if additional interference signals "sit" on the device’s own noise. (Check the noise level with the antenna input shortened by a dummy load - 50 or 75 ohm, depending on the impedance of the receiver). Read the noise floor using SDRSharp or other receiver software. The electronics can handle the range above the noise level up to 0dB (or just a bit larger signals) without significant distortion - this is the usable dynamic range.
The interfering signal can also come from the power supply or the motherboard itself via the USB port, or the power cord. Where possible, use interference filter ferrite cores that can be clipped at a point near the ends of the wires. If you use a poor quality power supply (PSU) or AC adapter for the LNA -the result will be distorted or false signals. Make sure that shielding of the devices in the chain from the antenna to the receiver is OK and properly connected. They must have the same electric potential.
... now we can deal with filtering signals from the antenna.
- Selection criteria:
Without wishing to be exhaustive, let us review a few aspects
- Size
Either our filter have to be placed directly under the antenna in our box, or we have enough space on the desk. -> Size may limit options.
Cavity and comb filters are usually quite large,
Hairpin filters (Stripline or microstrip category) at this frequency are not really small and it is not easy to use a naked PCB filter (without housing), near other metal objects.
LC filters are small, but nothing is perfect - they also have drawbacks
- Tech specs
- The required bandwidth (It also depends on where we live.)
In US, you may be interested in the 978 Mhz UAT band as well. Both communication methods are used, depending on the flight level and the planned route. (above a certain altitude and/or destination on another continent, -> Mode S/ADS-B 1090 MHz is mandatory.)
You can find details here: https://www.aopa.org/go-fly/aircraft-and-ownership/ads-b/where-is-ads-b-out-required
In Europe, you can find really strong GSM signals at 960 MHz, so a bandpass filter tuned to 970-1100 MHz is unacceptable.
(Examples: The light blue Flightaware filter for those living in the United States, while the dark blue for European feeders)
- degree of slope and attenuation at out-of-band frequencies
Built-in filters for receivers and amplifiers are also LC filters like the FA filters mentioned above.
The attenuation of the LC filter is mostly below 40 dB - but plenty of brands tend to be much closer to 30, sometimes even less. It has a moderate slope, hairpin filters are really the worst in this field.
The ceramic filters' attenuation is ~50 dB (not all of them, so be careful), a 3-pole cavity is 60-70, and both have good definite separation between the designed passband and the external frequencies.
4-5 pole cavity or comb filters can achieve attenuation of 80-100 dB or better. Note that every 3 dB difference is a half-size signal (double when it is amplifying) 80-100 dB attenuation is huge but often necessary - since an RTLSDR dongle has only 40-50 dB dynamic range at 1090 MHz.
If you do not feel like playing with formulas of Decibels conversion into amplitude ratio: 80 dB attenuation is about 0.0001 times the original signal.
dB calculations for voltage and power are different, due to the aspect from where we inspect the values.
This strong attenuation is necessary to decrease the amplitude of huge interfering signals below the 0dB level at least, ideally close to the noise floor. A strong signal may cause problems by mirroring other signals, like a frequency mixer. The sum and difference values of two different frequencies are also displayed - mixing the modulations of the two original frequencies as well. The generated false signals can result in further intermodulation and cross-modulation distortions ... It is the beginning of a possible chain reaction.
A modern superheterodyne receiver has a tuned part for picking the wanted radio station, then uses multiple mixer and filter units in chain - thus have a great selectivity. Simple SDR receivers skip this process and directly tune the oscillator frequency (or its harmonic) to the target frequency - meanwhile a really wide range of frequencies are still alive on the same ADC unit. This is why a strong signal anywhere(!) can cause problems.
- Insertion loss:
(The amount of loss from the useful signal that we pay in exchange for filtering. It also depends on the Q-factor of the filter elements, the target bandwidth, and the coupling matrix technique. The smaller the better. Anyway, all type of units have insertion loss, at least due to connections .)
LC filters are quite small and inexpensive (to build), but have an insertion loss of 2.5 to 4 dB and less sharp boundary between the filtered band and unwanted signals.
An insertion loss of 3 dB means that you can only measure half of the original signal strength (!) after the filter. Combined with LNA, it’s still okay because the amplifier raises the weak signals well above the noise level - creating a good signal level to compensate the cable and filter losses. The ideal low-noise amplifier only compensates for the expected losses, plus has enough amplification to raise the weaker signals above the receiver's noise - but doesn't overdo it. In extreme cases, no lightning strike is required for receiver failure - a pulse-type interference signal at the amplifier input can easily damage the sensitive receiver input.
As we know the saying: Size matters! Added by me: "the right size"
After the antenna, an LNA has to come first and then you can insert the filter, otherwise the weak signals will disappear before they reach the amplifier. Using an inline amplifier designed for 75-ohm TV reception is not always the best choice, as most of them have a 5 dB noise level - wasting the already limited dynamic range. Always check this parameter before purchasing. It is fine below 1 dB.
Antenna -> LNA -> Filter -> Receiver: this is the case when there must be a power inserter between the filter and the amplifier. This is a simple circuit of a few components but pay attention to the impedance and the usable frequency range.
Hairpin and ceramic chip filters are also lossy. (- 4dB or worse) The specs of cheap hairpins on average PCB is poor. (The substrate material and the entire lamination process must be extremely precise to achieve the correct Q factor. )
By the way, ceramic-chip based filters are more selective and have much better attenuation than Hairpin and LC filters.
(In small metal boxes you can find usually an LC or a ceramic filter)
In this area, cavity and comb filters have won, mostly with a less than 1 dB loss.
- The price of filters recently available in stores
Hairpin filters are the cheapest and this price is valid for marking the "weak solution" in our application. Many of them consist the filter elements only, and there are no matching network and harmonic response handling on them. Not recommended.
Their advantage lies in the cheap and fast manufacturability, but most of the time the manufacturers save by not using the right substrate. Hidden antennas for mobile devices are made using this technology, although more and more often the right shape is punched from a solid conductor to create good quality communication in high-end devices.
We find LC and ceramic chip based filters in the low and medium price range. Both can be good enough if your receiver is not suffering from really huge GSM signals.
Ceramic filters have a bit more insertion loss, but also a significantly better filtering. With a good LNA after the antenna - I would prefer a ceramic filter.
Cavity and comb filters are the most expensive, but it worth checking prices and parameters not only on pages of feeders' webshop. Though, their size and the expenses are in focus when we make our decision, they might be better in some products, so keep an eye on them. All the other specifications, so the expected results are definetly the best with them, no doubt. This filter family can be your choice even if you do not really need a sharp and strong filtering but you prefer the low insertion loss - to keep more from the original signal.
--------------------------------------------
Let's see a pair of filters to compare what you get for the money ... (I won't mention brands, but you can see them by following the links)
This one is very expensive: 196 EUR and - for me - it seems to be something between DIY and amateur. It will definitely work, but a 3 pole comb filter should not be so expensive. OK, it is just money, but I have my own opinion about it... (Update: It is not available anymore)
The other one is not in a feeder webshop, it's only 41 EUR -> only one-fifth of the price of the above version. It is smaller than a 6 cm cube and will surely fit in the box under the antenna. (looks like an angry LEGO cube) This filter is a 4-pole (real) cavity filter and is "carved" from a nice piece of metal. Inside, it has a well-designed coupling matrix and is precisely tuned and optimized to provide the expected measurement curve. Interestingly, this 4-pole filter has 7 tuning screws. 4 above the rods and 3 between them which are actually extra tunable capacities between the filter units. A hole drilled in the longitudinal axis of the rods not only provides space for the tuning screw but also has a positive effect on the Q-factor of the element. (remember the relationship between skin effect and surface area from a high-frequency conductivity perspective) We also find a coupling between the high impedance points of the first and third bars, which is probably used to tune the filter symmetry or compensating of unwanted responses. This well engineered filter has about 20-30 dB better attenuation in the 960 MHz GSM band than the "feeder-friendly" web store offering above.
(Tip: Do not disassemble the filter, because even the slightest movement of the lid can ruin the factory tuning. You can't tune this back properly at home. Instead using your screwdriver, you can have a look at the internals on the picture below. )
Easy to pick the better one...
----------------------------------------------
Some pictures of the frequency spectrum:
Without filtering, with ceramic filter and then with cavity filter
The thin spikes come from nearby radar and its signals pass through any filter like a Jedi laser saber
Without filtering, I had to adjust the gain to 38.0-40.2, with the ceramic filter 43.9, and now it works best with a cavity filter and agc on.
Comparing the 1090MHz parts on the second and third pics, you can see that not just the GSM signals have disappeared but we got a double sized ADS-B signal thanks to the 3 dB less insertion loss - compared to the ceramic one. Two birds in one fell swoop
Using filters is not a magic that increases the possible distance of detected signals from the antenna, do not expect this. The theoretical maximum distance can already be achieved with a medium quality antenna if it is in a good position. Filters can prevent interfering signals from affecting useful signals already received by the antenna. Not else...
That is all for now. I hope this helps a little.
Regards,
Janos
PS: This article was inspired by a previous writing of ab cd, so I complement his respectable work in helping the community.
https://forum.planefinder.net/threads/do-i-need-a-filter.1027/
It is not always easy to select the right piece, as graphs and tables that accumulate technical parameters do not help the average person how to consider the data in their own circumstances.
I’m trying to simply explain the things you need to know minimally beyond trial luck, especially for the simple end user.
- Question of necessity
As the antenna is taller than nearby landmarks and buildings, the receiver’s input circuits are besieged by strong signals that don’t interest us. This phenomenon can also be a problem for professional receivers, but this is especially true when a low-cost, software-defined radio is in use. The problem is that tuning is not done in the "traditional" way, so preselection is not provided at the input. In exchange for almost unlimited tuning freedom, we get other possibilities for error. I’m not immersed in the technical details, but I just need to mention one example: Even the SDR receiver’s well-designed pre-circuits are unable to prevent overdrive and distortion on the 8 or 10-bit A/D converter, because of the given low dynamic range.
I read an entry in which there were occasionally detected airplanes when the antenna was placed high, even less than before in the window of the room. The owner explained the phenomenon by saying that the signal from the planes was too strong. Well, not like that. The airplane signal can only be too strong if your antenna is at the end of the runway and the airplane is just flying over it. I'm 2 miles away from an airport, but I haven't even experienced 70 percent of the maximum possible signal strength. Read the signal strenght of the closest airplanes to make sure that it is true.
Extremely strong stations in the commercial FM band, ubiquitous GSM towers and all kinds of radio frequency signals, even from industrial sources reach the antenna. (indoors, near the room window, this is not such a striking phenomenon)
-> Very often, if you need to turn off automatic gain control (AGC) and manually significantly reduce the input sensitivity for the signal decoding to work - you don't really need attenuation, but a good filtering instead. (!)
The small dynamic range mentioned above can be more easily preserved to decode aircraft messages by checking if additional interference signals "sit" on the device’s own noise. (Check the noise level with the antenna input shortened by a dummy load - 50 or 75 ohm, depending on the impedance of the receiver). Read the noise floor using SDRSharp or other receiver software. The electronics can handle the range above the noise level up to 0dB (or just a bit larger signals) without significant distortion - this is the usable dynamic range.
The interfering signal can also come from the power supply or the motherboard itself via the USB port, or the power cord. Where possible, use interference filter ferrite cores that can be clipped at a point near the ends of the wires. If you use a poor quality power supply (PSU) or AC adapter for the LNA -the result will be distorted or false signals. Make sure that shielding of the devices in the chain from the antenna to the receiver is OK and properly connected. They must have the same electric potential.
... now we can deal with filtering signals from the antenna.
- Selection criteria:
Without wishing to be exhaustive, let us review a few aspects
- Size
Either our filter have to be placed directly under the antenna in our box, or we have enough space on the desk. -> Size may limit options.
Cavity and comb filters are usually quite large,
Hairpin filters (Stripline or microstrip category) at this frequency are not really small and it is not easy to use a naked PCB filter (without housing), near other metal objects.
LC filters are small, but nothing is perfect - they also have drawbacks
- Tech specs
- The required bandwidth (It also depends on where we live.)
In US, you may be interested in the 978 Mhz UAT band as well. Both communication methods are used, depending on the flight level and the planned route. (above a certain altitude and/or destination on another continent, -> Mode S/ADS-B 1090 MHz is mandatory.)
You can find details here: https://www.aopa.org/go-fly/aircraft-and-ownership/ads-b/where-is-ads-b-out-required
In Europe, you can find really strong GSM signals at 960 MHz, so a bandpass filter tuned to 970-1100 MHz is unacceptable.
(Examples: The light blue Flightaware filter for those living in the United States, while the dark blue for European feeders)
- degree of slope and attenuation at out-of-band frequencies
Built-in filters for receivers and amplifiers are also LC filters like the FA filters mentioned above.
The attenuation of the LC filter is mostly below 40 dB - but plenty of brands tend to be much closer to 30, sometimes even less. It has a moderate slope, hairpin filters are really the worst in this field.
The ceramic filters' attenuation is ~50 dB (not all of them, so be careful), a 3-pole cavity is 60-70, and both have good definite separation between the designed passband and the external frequencies.
4-5 pole cavity or comb filters can achieve attenuation of 80-100 dB or better. Note that every 3 dB difference is a half-size signal (double when it is amplifying) 80-100 dB attenuation is huge but often necessary - since an RTLSDR dongle has only 40-50 dB dynamic range at 1090 MHz.
If you do not feel like playing with formulas of Decibels conversion into amplitude ratio: 80 dB attenuation is about 0.0001 times the original signal.
dB calculations for voltage and power are different, due to the aspect from where we inspect the values.
This strong attenuation is necessary to decrease the amplitude of huge interfering signals below the 0dB level at least, ideally close to the noise floor. A strong signal may cause problems by mirroring other signals, like a frequency mixer. The sum and difference values of two different frequencies are also displayed - mixing the modulations of the two original frequencies as well. The generated false signals can result in further intermodulation and cross-modulation distortions ... It is the beginning of a possible chain reaction.
A modern superheterodyne receiver has a tuned part for picking the wanted radio station, then uses multiple mixer and filter units in chain - thus have a great selectivity. Simple SDR receivers skip this process and directly tune the oscillator frequency (or its harmonic) to the target frequency - meanwhile a really wide range of frequencies are still alive on the same ADC unit. This is why a strong signal anywhere(!) can cause problems.
- Insertion loss:
(The amount of loss from the useful signal that we pay in exchange for filtering. It also depends on the Q-factor of the filter elements, the target bandwidth, and the coupling matrix technique. The smaller the better. Anyway, all type of units have insertion loss, at least due to connections .)
LC filters are quite small and inexpensive (to build), but have an insertion loss of 2.5 to 4 dB and less sharp boundary between the filtered band and unwanted signals.
An insertion loss of 3 dB means that you can only measure half of the original signal strength (!) after the filter. Combined with LNA, it’s still okay because the amplifier raises the weak signals well above the noise level - creating a good signal level to compensate the cable and filter losses. The ideal low-noise amplifier only compensates for the expected losses, plus has enough amplification to raise the weaker signals above the receiver's noise - but doesn't overdo it. In extreme cases, no lightning strike is required for receiver failure - a pulse-type interference signal at the amplifier input can easily damage the sensitive receiver input.
As we know the saying: Size matters! Added by me: "the right size"
After the antenna, an LNA has to come first and then you can insert the filter, otherwise the weak signals will disappear before they reach the amplifier. Using an inline amplifier designed for 75-ohm TV reception is not always the best choice, as most of them have a 5 dB noise level - wasting the already limited dynamic range. Always check this parameter before purchasing. It is fine below 1 dB.
Antenna -> LNA -> Filter -> Receiver: this is the case when there must be a power inserter between the filter and the amplifier. This is a simple circuit of a few components but pay attention to the impedance and the usable frequency range.
Hairpin and ceramic chip filters are also lossy. (- 4dB or worse) The specs of cheap hairpins on average PCB is poor. (The substrate material and the entire lamination process must be extremely precise to achieve the correct Q factor. )
By the way, ceramic-chip based filters are more selective and have much better attenuation than Hairpin and LC filters.
(In small metal boxes you can find usually an LC or a ceramic filter)
In this area, cavity and comb filters have won, mostly with a less than 1 dB loss.
- The price of filters recently available in stores
Hairpin filters are the cheapest and this price is valid for marking the "weak solution" in our application. Many of them consist the filter elements only, and there are no matching network and harmonic response handling on them. Not recommended.
Their advantage lies in the cheap and fast manufacturability, but most of the time the manufacturers save by not using the right substrate. Hidden antennas for mobile devices are made using this technology, although more and more often the right shape is punched from a solid conductor to create good quality communication in high-end devices.
We find LC and ceramic chip based filters in the low and medium price range. Both can be good enough if your receiver is not suffering from really huge GSM signals.
Ceramic filters have a bit more insertion loss, but also a significantly better filtering. With a good LNA after the antenna - I would prefer a ceramic filter.
Cavity and comb filters are the most expensive, but it worth checking prices and parameters not only on pages of feeders' webshop. Though, their size and the expenses are in focus when we make our decision, they might be better in some products, so keep an eye on them. All the other specifications, so the expected results are definetly the best with them, no doubt. This filter family can be your choice even if you do not really need a sharp and strong filtering but you prefer the low insertion loss - to keep more from the original signal.
--------------------------------------------
Let's see a pair of filters to compare what you get for the money ... (I won't mention brands, but you can see them by following the links)
This one is very expensive: 196 EUR and - for me - it seems to be something between DIY and amateur. It will definitely work, but a 3 pole comb filter should not be so expensive. OK, it is just money, but I have my own opinion about it... (Update: It is not available anymore)
The other one is not in a feeder webshop, it's only 41 EUR -> only one-fifth of the price of the above version. It is smaller than a 6 cm cube and will surely fit in the box under the antenna. (looks like an angry LEGO cube) This filter is a 4-pole (real) cavity filter and is "carved" from a nice piece of metal. Inside, it has a well-designed coupling matrix and is precisely tuned and optimized to provide the expected measurement curve. Interestingly, this 4-pole filter has 7 tuning screws. 4 above the rods and 3 between them which are actually extra tunable capacities between the filter units. A hole drilled in the longitudinal axis of the rods not only provides space for the tuning screw but also has a positive effect on the Q-factor of the element. (remember the relationship between skin effect and surface area from a high-frequency conductivity perspective) We also find a coupling between the high impedance points of the first and third bars, which is probably used to tune the filter symmetry or compensating of unwanted responses. This well engineered filter has about 20-30 dB better attenuation in the 960 MHz GSM band than the "feeder-friendly" web store offering above.
(Tip: Do not disassemble the filter, because even the slightest movement of the lid can ruin the factory tuning. You can't tune this back properly at home. Instead using your screwdriver, you can have a look at the internals on the picture below. )
Easy to pick the better one...
----------------------------------------------
Some pictures of the frequency spectrum:
Without filtering, with ceramic filter and then with cavity filter
The thin spikes come from nearby radar and its signals pass through any filter like a Jedi laser saber
Without filtering, I had to adjust the gain to 38.0-40.2, with the ceramic filter 43.9, and now it works best with a cavity filter and agc on.
Comparing the 1090MHz parts on the second and third pics, you can see that not just the GSM signals have disappeared but we got a double sized ADS-B signal thanks to the 3 dB less insertion loss - compared to the ceramic one. Two birds in one fell swoop
Using filters is not a magic that increases the possible distance of detected signals from the antenna, do not expect this. The theoretical maximum distance can already be achieved with a medium quality antenna if it is in a good position. Filters can prevent interfering signals from affecting useful signals already received by the antenna. Not else...
That is all for now. I hope this helps a little.
Regards,
Janos
PS: This article was inspired by a previous writing of ab cd, so I complement his respectable work in helping the community.
https://forum.planefinder.net/threads/do-i-need-a-filter.1027/
Last edited:
