Passive filter design software using preferred component values

[perdrix]

Hi folks

Looking to build a 10MHz bandpass filter to convert a square wave to a sine.

I would like to find some desing software for this that doesn't propose totally unobtainium component values - it's anot as iff I have any 229.76uH inductors lying around or even ordereable.

Hence the wish to use preferred value parts - I'll quite happily accept a wider bandpass for this trade off.

I tried "Elsie" but AFAICT that doesn't have any option to tell it to use only preferred values - but I could easily have missed it.

David

[Benta]

At that frequency you'll have so many parasitic effects that using calculated values is hopeless anyway.

I don't know your signal levels, but a possibility could be using 10.7 MHz adjustable IF coil sets. Cheap and easy to find. Or you could scavenge them from an old FM radio.


Edit: they're actually called "IF transformers".

[Mechatrommer]

any spice simulation will do.. tina-ti? ltspice?

[orbanp]

Quote from: perdrix on Today at 12:04:18 AM

...
I tried "Elsie" but AFAICT that doesn't have any option to tell it to use only preferred values - but I could easily have missed it.
...

Hi David,

Looks like you missed it!
On the "Design" page there is a "Nearst 5%" button, as well as a "Specify L" button, those let you tune the component values.

DISCLAIMER:
I have not updated "Elsie" for a while, my version is Student Edition 2.77, from 2016.
Newer versions might behave differently.
Do check the help pages!

Good luck, Peter

[perdrix]

Yes, you are right it was a separate tab, and I was looking for a tick box or similar.   

Thanks
David

[G0HZU]

At that frequency you'll have so many parasitic effects that using calculated values is hopeless anyway.

Are you sure about that? The centre frequency is only 10MHz.

[Benta]

At that frequency you'll have so many parasitic effects that using calculated values is hopeless anyway.

Are you sure about that? The centre frequency is only 10MHz.

It's been my experience, at least. For the caps you're down in the pF range, for the coils uH. Layout and wires start playing quite a large role.

[mc172]

Iowa Hills "RFFilter" http://iowahills.com/9RFFiltersPage.html

[G0HZU]

At that frequency you'll have so many parasitic effects that using calculated values is hopeless anyway.

Are you sure about that? The centre frequency is only 10MHz.

It's been my experience, at least. For the caps you're down in the pF range, for the coils uH. Layout and wires start playing quite a large role.

If you were correct, Coilcraft would have gone out of business years ago. It's possible to model and build decent BPFs up towards a GHz or so using fixed inductor values from Coilcraft

https://www.coilcraft.com/prod_smind.cfm

[Benta]

I don't question building filters up to 1 GHz with fixed inductors. I do question the modeling, though. But I've only done it a few times, and my calculations were always off compared to the measurements afterwards.

[RandallMcRee]

If you want to turn 10Mhz square into sine only a low-pass filter is needed--you just need to suppress the higher harmonics which are all in the 20MHz and up range. A bandpass filter takes more components than equivalent LPF.

[rhb]

I suggest trying this:

http://www.tonnesoftware.com/svcdownload.html

It designs filters using Standard Value Components.  Tonne Software have quite a few other programs besides Elsie.

[Kalvin]

A crystal-based passive bandpass filter is simple and requires very few components in addition to the crystal(s) itself. The Q-value is very high compared to equivalent LC-filter, so you will required only few stages.

https://hackaday.com/2016/03/13/designing-a-crystal-ladder-bandpass-filter/

[G0HZU]

I don't question building filters up to 1 GHz with fixed inductors. I do question the modeling, though. But I've only done it a few times, and my calculations were always off compared to the measurements afterwards.

At 1GHz it's best to use a VNA to take an s2p model of the inductor when it is mounted on the correct PCB material. The other alternatives are to use Coilcraft's own s2p files or try using the Modelithics range of models (expensive). Even then, at 1GHz you also have to model the PCB layout to get a sensible overall model of the complete filter.

But at 10MHz I think things are a lot easier.

[mc172]

I would agree - at 10 MHz the layout is hardly important. As long as the coils aren't touching each other, and you aren't looking for 50+ dB of rejection, it isn't going to be a problem whatever you do.

[David Chamberlain]

Iowa Hills "RFFilter" http://iowahills.com/9RFFiltersPage.html

Also came to say Iowa Hills one. On the download page 'RF Filter Designer Ver. 2.2'

It has an option for 'std LC Values'