Low cost RF filter modules

[rhb]

As part of my Pixie 2 redesign effort I prototyped the new filter so I could properly sweep it with my 8560A.  After changing the filter, I put the kit parts on a similar PCB to sweep it.

Afterwards I realized I now had a couple of SMA-F filter modules.  So I covered them with electrical tape followed by copper foil tape and soldered the foil to the PCB to provide good shielding.  For under $1 each, pretty handy.

Now I just need to find the instructions for my label maker to figure out how to make multiline labels for the PCB foil side.

Have Fun!
Reg

[radiolistener]

Some notes:
1) cooking foil is too thin and has bad performance for shielding. It's better to use more thick copper foil.
2) keep shielding away from filter components as far as possible. At least keep not less than several tens of millimeters.
3) these yellow Chinese capacitors works very-very bad and have extreme high loss on RF. They are not intended to use on RF. This is very important for RF filter to use low loss capacitors and inductors. Use something like silver mica capacitors.
4) these EC24 green inductors have Q about 30-50. This is better than cheap inductors with high loss, but this is not much. So, it's better to replace it with air core inductor. But first replace capacitors, because these yellow capacitors are piece of shit.

ATC porcelain capactiors are good, they promises Q about 10000, but they are too expensive.

[rhb]

I was asked what they looked like at VHF, so here are sweeps to 100 MHz and 200 MHz.  The attenuation drops to almost zero by 1 GHz.

FYI The capacitors are Vishay and the inductors are Bourne JW Miller.  The  >60 dB suppression of the  Pixie 2nd & 3rd harmonics testifies to the effectiveness of the filter.  And there is no discernible insertion loss at the  7 MHz Cauer peak.

The copper foil pressed in place did shift the Cauer peak frequency of the prototype filter a bit, but for a junk box filter that's not really an issue.  I'm not likely to anticipate what filter I need for a future project.  These will get labeled and transfer functions saved in a notebook.   These are just a way of recycling my testing prototypes.

Have Fun!
Reg

[rhb]

Some notes:
1) cooking foil is too thin and has bad performance for shielding. It's better to use more thick copper foil.
2) keep shielding away from filter components as far as possible. At least keep not less than several tens of millimeters.
3) these yellow Chinese capacitors works very-very bad and have extreme high loss on RF. They are not intended to use on RF. This is very important for RF filter to use low loss capacitors and inductors. Use something like silver mica capacitors.
4) these EC24 green inductors have Q about 30-50. This is better than cheap inductors with high loss, but this is not much. So, it's better to replace it with air core inductor. But first replace capacitors, because these yellow capacitors are piece of shit.

ATC porcelain capactiors are good, they promises Q about 10000, but they are too expensive.

I designed an output LP filter to replace the stock one in an $18 kit that came with 10 xtals.  The kits are $7-8 with only one xtal. The stock filter does not meet current FCC spurious emission requirements.  The kit filter design  dates from the mid 90's when FCC exempted QRP  transmitters from the spurious emission regulations.  Since then they have removed that exemption.

So after making a prototype filter to test, removing the kit filter and replacing it, I mounted the kit filter so I could sweep it to show the difference.  I demonstrated with the transmitter that the kit as received did not meet the FCC requirements but with a filter which added a *single* part it exceeded requirements by almost 20 dB.  I happened to use 2 parts because I did not have a 360 pF cap and had to parallel a 330 and a 27.

Looking at the 2 test boards I decided that they might prove handy one day and decided to shield them with copper foil as being a cheap and very simple method.  I thought the cost and effort made it worth mentioning.  I expect I shall be doing it quite often with prototypes.  Left over from projects.

I posted SA sweeps showing the performance of the filters before (in another thread) and after shielding.  They clearly work as expected.

Based on the *color* of the capacitors you tell me they are no good, my shielding is no good and I should use air core inductors.  And do all this despite clear evidence that the new filter performs the task better than required and that the Butterworth filter supplied with the kit does exactly what it is expected to do and by analysis with Elsie can't meet FCC regulations.

Please excuse my thinking that your post  absurd.

Have Fun!
Reg

[bd139]

It's all fine. I've built far worse looking low pass filters for amateur radio use



As for capacitor choice, any half decent NP0/C0G ceramic MLCC is fine for LPFs at HF (and mostly VHF). Only important factor is the voltage and the dielectric. The inductors however are likely to be an issue. Mostly because the wire is so fine they will melt if you put too much current through them. They also aren't toroidal so there is no field confinement which can be an issue.

I usually build mine like this now

[rhb]

I just got $350 of toroids from Diz.  I used 7 cent solenoid inductors because I want to produce a design which can be sold for <$10 delivered from China and has substantially better performance than the current designs.

With 5-6 transistors and a quad op amp I'll use toroids and possibly shielded sections.  Realistically I can't compete with Wayne Burdick at the high end any time soon.  Though with luck I might live long enough to compete.  I know a lot about DSP.   But high performance multiband radios are expensive.  So I want to focus on low cost, high performance single band radios.  And I share Jim Williams' passion for analog.

I've always wanted an RF "erector set".  I've bought a lot of stuff, but this is my initial foray into building modules since I was in grad school 35 years ago.

I'd been looking at the prototypes and unhappy about the amount of work to shield them when I hit on the copper tape scheme.  Eventually I hope to do SMD stripline by milling pockets in a 2nd piece of single sided PCB.  That should make for some really high performance filters and other modules.

Have Fun!
Reg

[radiolistener]

Here is my 29 MHz LPF build from unknown Chinese SMD capacitors and EC24 KLS inductors. The total cost of all components, exclude PCB and SMA connectors is about 0.31 USD.

Technically it works, but I don't like it's performance.

[radiolistener]

Based on the *color* of the capacitors you tell me they are no good

I have capacitors bought on ali which looks exactly the same, even marking looks the same.
They are terrible. They have very high loss and almost unusable above 1-2 MHz.
I hear that these are clay capacitors from cheap clay produced for cheap electronics working at audio frequencies.

[radiolistener]

I usually build mine like this now

Looks nice, where did you bought core rings?

[rhb]

Here is my 29 MHz LPF build from unknown Chinese SMD capacitors and EC24 KLS inductors. The total cost of all components, exclude PCB and SMA connectors is about 0.31 USD.

Technically it works, but I don't like it's performance.

What are you expecting?  Looks like a Butterworth filter.  Is that measured or modeled?  If measured, how did you measure it?  What are you trying to do?

My 8560A photos are unambiguously actual performance on a good instrument which is working properly.  And my goal was quite explicit, suppress harmonics to comply with  the FCC <-43 dB requirement.  At <-60 dB I did that.

The point of my work was to demonstrate that if you use the wrong filter, the results don't meet FCC rules and that a filter which *does* meet FCC rules is only 4 parts instead of 3.  At 5-6 cents for the cap in 100 unit quantity, not a major cost increase.

I am not an expert on analog filter design.  I've spent about $300 on books on the subject and am stumbling though them.  I'm rather pleased with what I was able to do.

This thread was just about shielding filters for pennies by using copper foil tape.  The filters were detritus left over from curing the Pixie kit harmonics.  So I thought I'd make them more generally useful by adding shielding.

Reg

[radiolistener]

What are you expecting?  Looks like a Butterworth filter.  Is that measured or modeled?  If measured, how did you measure it?  What are you trying to do?

I'm expecting less losses in pass band and more sharp slope. This is measured results. It is measured with NanoVNA. I'm just share my experience with similar inductors and capacitors.

I also tried to test aluminum foil for food storage as a shielding for the filter. It gives no advantage.

The point of my work was to demonstrate that if you use the wrong filter, the results don't meet FCC rules and that a filter which *does* meet FCC rules is only 4 parts instead of 3.  At 5-6 cents for the cap in 100 unit quantity, not a major cost increase.

Sellers know why it doesn't comply with FCC requirements. But these 5-6 cents give them a lot of money, just because these things sold in a large quantity.

[bd139]

I usually build mine like this now

Looks nice, where did you bought core rings?

GQRP club

[rhb]

I just bought a huge bunch of toroids  from Diz at https://kitsandparts.com for very nominal prices.  GQRP only sells to members, though I've joined that.

Butterworth filters are maximally flat passband.  The skirt response is very poor for my application.  A member of the QRPTech list got a better response using Elsie by designing a 5 element Cauer and then deleting the 3rd harmonic notch to make it a 3 element Cauer.  However, it's only better if the output impedance is precisely 50 ohms.  At present I don't know the actual output impedance of the Pixie PA.

At present I'm more concerned about the +10 dBm LO leakage.  But I think I may be able to fix that by switching the LO and RF feeds when changing from RX to TX.

I neglected to measure the filters before I applied the copper foil. I'll do that next time.  Copper has the advantage it can be soldered to the PCB ground plane all around the edge.   The next time I make one I'll put a 50 ohm termination on the end and use an E field probe to measure the radiation before and after applying the copper foil.  I would not expect the shielding to have a significant effect on the thru response of the filter unless it was a very high (e.g. > -100 dB) reject filter.

Have Fun!
Reg