Janos Konya
Member
Summary of construction experience and some conclusions:
(You can find the plan and the link of the online calculator in the attached file - ID BPF plan)
!!! DC shorted filter - Do not forget switching the Bias-T in your dongle off !!!
- You have to insert the power after the filter, from where there is no more device towards the LNA and the antenna
Given that the filter had to be "manufactured" on my desk and I have a limited set of tools - I decided to use thin materials. This decision only brought short-term benefits. In terms of the end result I would have been better off working with more massive materials.
I bought the necessary components from a modeling shop. I chose a quarter-millimeter “thick” copper plate and a tube with a similar wall thickness - thinking that cutting and bending operations would be easy. That was the case ... but I paid the price elsewhere. The mechanical stability of the finished filter is questionable despite my benevolence and bias.
Due to the strong coupling, the resulting filter pattern is slightly pointed - instead of the ideal flat characteristic.
On the left side of the filtered range, the drawing widens in the direction of the lower frequencies - neither the slope nor the degree of filtering shows the expected parameters. The measured attenuation of the filter is about 35-38 dB at the bottom and better than 60 dB at the top. I can't accept the widening nature of the pattern - even if the attenuation is significant at more distant frequencies. The insertion loss is slightly below 1 dB, with a bandwidth of 25 MHz between -3 dB points.
Shortening the rods might help, but I don’t feel like fiddling with this instance anymore. Call it the number one experimental specimen and let it "rest in peace." The unsatisfactory result here is not the fault of the filter, I certainly should have paid more attention to the details of the implementation.
Interdigital filters are characterized by a slightly stronger coupling (due to the alternating bars), good frequency tunability, and relatively easier sizing...
...the next test subject will still be a Comb Filter.
A Comb Filter can be made in a slightly more compact size, maybe it’s easier to achieve narrower bandwidth - but that won’t be my goal as it would increase insertion loss. Reducing the latter is a primary consideration. I will also be able to control the amount of coupling with the length of the input and output probes. They are standalone tiny rods - not soldered points to the first and last fingers.
I will use a square aluminum tube or similar. The only challenge is thread drilling and sizing. For rods, this is not a negligible problem - I don’t have a fixed drill with a clamping device. We'll see...
In my next post, I will hopefully report more successes.
Regards,
Janos
Update: Comb filter project is done!
(You can find the plan and the link of the online calculator in the attached file - ID BPF plan)
!!! DC shorted filter - Do not forget switching the Bias-T in your dongle off !!!
- You have to insert the power after the filter, from where there is no more device towards the LNA and the antenna
Given that the filter had to be "manufactured" on my desk and I have a limited set of tools - I decided to use thin materials. This decision only brought short-term benefits. In terms of the end result I would have been better off working with more massive materials.
I bought the necessary components from a modeling shop. I chose a quarter-millimeter “thick” copper plate and a tube with a similar wall thickness - thinking that cutting and bending operations would be easy. That was the case ... but I paid the price elsewhere. The mechanical stability of the finished filter is questionable despite my benevolence and bias.
Due to the strong coupling, the resulting filter pattern is slightly pointed - instead of the ideal flat characteristic.
On the left side of the filtered range, the drawing widens in the direction of the lower frequencies - neither the slope nor the degree of filtering shows the expected parameters. The measured attenuation of the filter is about 35-38 dB at the bottom and better than 60 dB at the top. I can't accept the widening nature of the pattern - even if the attenuation is significant at more distant frequencies. The insertion loss is slightly below 1 dB, with a bandwidth of 25 MHz between -3 dB points.
Shortening the rods might help, but I don’t feel like fiddling with this instance anymore. Call it the number one experimental specimen and let it "rest in peace." The unsatisfactory result here is not the fault of the filter, I certainly should have paid more attention to the details of the implementation.
Interdigital filters are characterized by a slightly stronger coupling (due to the alternating bars), good frequency tunability, and relatively easier sizing...
...the next test subject will still be a Comb Filter.
A Comb Filter can be made in a slightly more compact size, maybe it’s easier to achieve narrower bandwidth - but that won’t be my goal as it would increase insertion loss. Reducing the latter is a primary consideration. I will also be able to control the amount of coupling with the length of the input and output probes. They are standalone tiny rods - not soldered points to the first and last fingers.
I will use a square aluminum tube or similar. The only challenge is thread drilling and sizing. For rods, this is not a negligible problem - I don’t have a fixed drill with a clamping device. We'll see...
In my next post, I will hopefully report more successes.
Regards,
Janos
Update: Comb filter project is done!
Last edited:
