Filter hacking for 2395 ISS TV

[g0mjw]

Colin G4EML

Something is not quite right with the tuning. Has one of the tuners reached its limit perhaps? Maybe the spacing needs to be adjusted? Incidentally, why are all the tuning screws on the same side? They usually alternate in this type of filter.

Mike

[g4eml]

Interdigital filters have the resonators on alternating sides. Comb style filters like this have them all on the same side. Apparently comb style can have the resonators closer together for the same coupling, so the end result is a shorter filter.

Nothing is maxed out on the tuning, I could get a partially flat response like this or a more symmetrical one with a lot of ripple. I suspect adding the two extra resonators at the same spacing as the 3 pole version is not the correct thing to do, the coupling between the resonators is almost certainly not optimum.

Colin G4EML

[g4saq]

I can't contribute any technical insights to this thread but I just want to say how inspiring it is to see this collaboration on the forum. Great stuff. I look forward to copying your final solution! 73, Russ, G4SAQ

[g0mjw]

Interdigital filters have the resonators on alternating sides. Comb style filters like this have them all on the same side. Apparently comb style can have the resonators closer together for the same coupling, so the end result is a shorter filter.

Nothing is maxed out on the tuning, I could get a partially flat response like this or a more symmetrical one with a lot of ripple. I suspect adding the two extra resonators at the same spacing as the 3 pole version is not the correct thing to do, the coupling between the resonators is almost certainly not optimum.

Colin G4EML

Yes, that's probably it. A lot of ripple might be over-coupled. Can you perhaps reduce the spacing a bit? I found this on line https://www.changpuak.ch/electronics/co ... signer.php that suggests a for a 246MHz filter a spacing of 24mm for a 5 resonator filter vs 22mm for a 3 resonator filter. That's not 24 GHz but perhaps the principle remains. They are also not evenly spaced, the outer ones closer than the inner ones, which is interesting and might also matter.

Then I came across this https://www.rfshop.co.uk/C&IDES.DOC

Which goes into the maths with a worked example for a 5 GHz 5 pole. It might be worth calculating for the box section you have. I have not tried it but the algebra doesn't look to complicated.

Mike

[g4eml]

Adjusting the spacing is not easy as it would probably also need moving the input and ouput connectors to match. That's a lot of holes that need to be moved.

I may try adding additional screws in between the posts. That is supposed to adjust the coupling. Might be worth a try.

Colin.

[g0mjw]

I would start again with the rectangular tube Gareth found. I think square probably increases these unexpected modes.

Mike

[g4eml]

Rectangular tube will behave slightly differently but I don't think that is the problem I am seeing and I think adding poles will pose the same issues.

I have ordered some of the 40 x 30 tube to test later this week.

I have just asked ChatGPT a few questions and it is suggesting that three poles spaced 70 70 could be replaced with 5 poles spaced 70 74 74 70.

Whilst I don't always agree with what it says that does sound reasonable.

Maybe some more drilling tomorrow.

Colin.

[g0mjw]

That does sound reasonable.

The articles seem to imply there are calculations that get you close and some element of magic involved in reaching the final solution, due to reality being more complicated. I was thinking the rectangular would mean fewer evanescent modes to worry about, but it is also less material so lighter and perhaps cheaper. I didn't see any thicker walled stuff that would be better for taping into.

Here is the sadly missed Doug, VK3UM programme for calculating interdigital filters https://www.vk5dj.com/InterDigitalFilte ... nstall.zip - it still works but I don't think it helps here.

Mike

[g4eml]

That's Interesting, when set up for 2395 MHz, 36mm box depth, 10mm rods and +- 5MHz bandwidth that program comes up with spacings of 69.8 74.8 69.8 for a 4 pole filter. Those are surprisingly close!

Colin.