RF filter design tool

[enemra]

Hello friends,

I am looking for a source to design the RF filters (bandpass filters) and simulate. For instance, MATLAB/PSpice or web based tools such https://markimicrowave.com/technical-resources/tools/lc-filter-design-tool/ might do the job but practical results will always be different to theoretical. I would like to know from you experts, based on your experience, can you suggest me the best option to go with?

I am asking this because I have a radio board that is tuned for 868MHz. However, the RF IC can support 435MHz too. I want to replace the filter circuit. The ideal case for me would be to simulate the existing circuit and then change the components in the software accordingly to support 435MHz.

Thanks

[ftg]

If you want free then QUCS Studio is among the better out there.
Do also grab the examples to see how to use it.

https://qucsstudio.de/

[Karel]

https://qucs.sourceforge.net/screenshots.html

[Solder_Junkie]

I have found Elsie to give very good results but it needs realistic values of Q for the inductors and capacitors. In the case of UHF, component lead lengths, track lengths/widths and the effect of nearby shielding, will probably make it necessary to fine tune the filter on a trial and error basis. My experience with Elsie is limited to HF, but the measured response was very close to the calculated.

https://tonnesoftware.com/elsie.html

SJ

[rfclown]

I use QUCS. I've never tried the Studio version. It is very important to include the parasitics for the response to match what you build. For the components I use the manufacturer's S parameters, and minimize any interconnect length between components (put the parts right next to each other). If you don't minimize the board traces, you'll have to model those.

[enemra]

Thanks friends, I will try QUCS and Elsie..

[rfclown]

For low frequency (audio etc) where parasitics aren't important and I can use ideal LC values, I also use LTspice. With QUCS I use S parameter simulation and it can model PCB microstrip lines.

[enemra]

Well, if you are trying to place the passives next to each other, won't it be an issue with phase mismatching?

For instance, considering a random circuit as attached, the manufacturer says certain passives should have specific distances.... if I understood this correctly, it matters only for phase matching isnt it and it wont affect impedance. Ain't I right?

[ftg]

At those frequencies the distance matters, as the microstrip is part of the matching circuit.
In QUCS one can just put in the appropriate microstrips between components to simulate that.

[rfclown]

Like ftg said, in that circuit the length of line between the IC and lumped component is part of an impedance matching circuit. You can use a length of transmission line as an impedance transformer. That microstrip line appears to be 50 ohms. If the impedance at the IC pin isn't 50 ohms, then a length of line will transform it to something else which is a function of the line length. If the impedance at the IC pin is 50 ohms, then the length doesn't matter (expect for resistive and dielectric losses), it will stay 50 ohms independent of line length. Again, like ftg said, you can play with that in QUCS simulation to get a feel for what happens. Find the transmission line component that lets you specify impedance and length. Place a S-parameter source (Power Source), transmission line, and resistor to ground. If the transmission line is 50 ohms and resistor is 50 ohms, then the length won't matter. S11 will be matched always. If the resistor isn't 50 ohms, or the transmission line isn't 50 ohms, S11 will be a function of the line length.

[radiolistener]

at frequencies more than 100 MHz it's very hard to tune such filters because it become very sensitive to any drop of solder, it's hard to predict Q-factor of components, parasite capacitance and inductance, also it's hard to measure it. At 400+ MHz it's become even harder

[MarkT]

I've had resonable success making VHF surface-mount filter networks just from LTSpice simulation and using quality components like Coilcraft SMT inductors and branded C0G SMT caps.  Using a small 0603 size helps in taming parasitics and I found stripboard can be used as microstrip if you glue a layer of copper foil to the underside - and edge-on SMA connectors fit nicely.