Need advice on bandpass filter design (uni project).

[Refrigerator]

I've been assigned to make a bandpass filter as one of my semester projects for university.
I think designing it is no big deal but some questions have popped up along the way. I'm using https://rf-tools.com/lc-filter/ to help me design this filter and ADS 2020 to test the design.
Basically i've noticed that stopband attenuation plays a big role in the filter design but in my case the stopband attenuation wasn't given, meaning that i have to choose it myself.
I don't have much experience in RF design so i don't know what level of attenuation is a good ballpark figure to base my filter off.
For example, the calculator linked above defaults to -50 dB, but i don't know whether it's normal or some pie in the sky figure. What's a common real-world number i should be aiming for?

Parameters are: center frequency 145 MHz, -3 dB at 144-146 MHz (passband), 5-th order, discrete components, FR-4 substrate, SMA plugs.
Impedances are not given but they'll be the standard 50 .

Also i'm not sure which type of filter to choose. At first i was leaning more towards elliptic or Chebyshev, but now i'm thinking more along the lines of Butterworth.
Problem with Butterworth is that the calculator gives me capacitor values in the femto Farrads and i don't think i can get my hands on those.
Some freedom is given in the filter design as long as the initial criteria is met so i think some microstrip action wouldn't hurt.
How hard would it be to design a discrete 490 fF or 151 fF capacitor on a PCB? I don't think stray inductance would be much of a problem given the µH inductors next to the fF caps.

[FenTiger]

I think those picohenry valued inductors are going to give you bigger problems.

I've only done a little bit of VHF filter design, but I've never had much luck at getting realistic component values out of online calculators. I've had far better results from Elsie: http://tonnesoftware.com/elsie.html

[TheMG]

The required attenuation outside of the passband (how far out from the passband?) really does depend on the intended purpose of the filter, and is not entirely easy to predict as real-world components, the PCB layout and the physical construction all have parasitic inductance and capacitance that is going to affect the stopband characteristic at certain frequencies, especially if the aim is to have a large amount of attenuation.

Since no specific requirement was given, just design the filter for the given requirements, following best practices for component choice and layout, and what you end up with for attenuation outside of the passband, is whatever you end up with. Find a calculator that doesn't require entering an attenuation requirement, or do the calculations manually to determine the required component values.

A practical VHF filter is normally going to have inductor values in the tens or hundreds of nH. Discrete inductors in the pH and capacitors in fF values do not exist, as parasitic inductance and capacitance of leads and traces would exceed that. PCB inductors and capacitors of such values would not typically be used for anything lower than microwave frequencies, at which point you'd be using stripline filters which is a completely different animal.

[T3sl4co1l]

Not to mention the other capacitors that are fractional-pF, in series with 2.5uH inductors that need a hell of a lot less than 0.5pF stray between their ends and to ground.

Yeah, this ladder topology needs to be transformed to a coupled-resonators topology.  This is better at high Q like this.  Roughly speaking, tune up a bunch of resonators (helical resonators are okay at this frequency) to 145MHz, then couple them together just enough to achieve the desired bandwidth.  Couple say a magnetic loop to the input and output and you're done.

Probably better off buying one, the 2m HAM band is a standard thing after all.  Not to say it's not a good learning experience to design and construct a filter, but it will take a while and if you have other priorities in this project, don't let it hold that up.

Tim

[Refrigerator]

Probably better off buying one, the 2m HAM band is a standard thing after all.  Not to say it's not a good learning experience to design and construct a filter, but it will take a while and if you have other priorities in this project, don't let it hold that up.

Tim

Buying is not an option because the point of this project is to apply the knowledge we gained in this course by designing, making and testing an RF filter. 

[M0HZH]

I don't think it's possible to meet the requirements (2MHz -3dB bandwidth at 145MHz from a 5-pole filter) with discrete components. My best attempt would be coupled resonators, designing for a high working impedance (1-2kohm) and using impedance transformers at the input & output.

For these requirements, the most common practical solution would be some sort of helical resonator.

[cdev]

Two helical resonators only lightly coupled (with a short piece of mini coax) or allowed to influence one another some ideally varying amount with a hole in the shielding with short taps near the shorted ends at the appropriate point for your low impedance ports. (but they are interchangeable) So each one is an LC resonant circuit at the desired frequency. There are lots of designs online.  You dont really need test equipment if you have a receiver and signals you want to receive are audible and constantly available. But it would be very helpful if you do have it. Thicker wire works best by far but you can use thinner stiff wire make sure you use a metal box and that it is shut because any movement or deformation of the coil will change the frequency it resonantes at and you will need to retune it. Solder on thin bare copper wire makes it stiffer. You can use this for higher frequencies too. (but it will be smaller) Capacitor quality matters more as you go higher in frequency, Ive read. Some trimmer capacitors are too lossy to work well.

[cdev]

Did they give you any other requirements like size?

I've been assigned to make a bandpass filter as one of my semester projects for university.
I think designing it is no big deal but some questions have popped up along the way. I'm using https://rf-tools.com/lc-filter/ to help me design this filter and ADS 2020 to test the design.
Basically i've noticed that stopband attenuation plays a big role in the filter design but in my case the stopband attenuation wasn't given, meaning that i have to choose it myself.
I don't have much experience in RF design so i don't know what level of attenuation is a good ballpark figure to base my filter off.
For example, the calculator linked above defaults to -50 dB, but i don't know whether it's normal or some pie in the sky figure. What's a common real-world number i should be aiming for?

Parameters are: center frequency 145 MHz, -3 dB at 144-146 MHz (passband), 5-th order, discrete components, FR-4 substrate, SMA plugs.
Impedances are not given but they'll be the standard 50 .

Also i'm not sure which type of filter to choose. At first i was leaning more towards elliptic or Chebyshev, but now i'm thinking more along the lines of Butterworth.
Problem with Butterworth is that the calculator gives me capacitor values in the femto Farrads and i don't think i can get my hands on those.
Some freedom is given in the filter design as long as the initial criteria is met so i think some microstrip action wouldn't hurt.
How hard would it be to design a discrete 490 fF or 151 fF capacitor on a PCB? I don't think stray inductance would be much of a problem given the µH inductors next to the fF caps.

[David Hess]

The first thing I thought of when reading the requirements is helical resonator.  I will not say that a discrete implementation is impossible but tuning will be a problem.  Below is a photograph of a 2 meter helical resonator which I designed and made by hand several years ago.

In the design process, I started with the wire diameter which I had available to adjust the other dimensions.  This was actually the second helical resonator I designed and built; the first one did not have overlapping seams for greatly increased strength and used plastic film trimmers which severely limited performance compared to the piston trimmers.

If I built another one, I would implement variable coupling between sections using sliding windows.  This one is over-coupled although that ended up working out well in practice.

[WPXS472]

As someone with some experience in building filters on a commercial scale, let me just say that good filter people are born, not made. I agree with the above statements that a helical resonator is probably your best bet. Filter design is all mathmatics, but filter building is a whole different matter. I would try to get an old, VHF high band radio, something like a GE MASTR2 and rip the preselector out. Then study that and build your filter in a similar manner. Don't copy, but use what you learn from studying that old preselector. The company I worked for in a former life made a series of helical filters that worked from about 125 MHz to 500 MHz all in the same housing, and meeting the same specs. They were pilot selection filters for an older CATV manufacturer, There were challenges in doing this, but we managed to overcome them and made filters for several years. They finally settled on a standard pilot frequency and had ceramic resonator filters produced. Something they should have done from the beginning, in my opinion.

[cdev]

David, That looks very nice!

I notice you have what appear to be threaded caps over the adjusthent trimmers. Thats to protect the values from being accidentally detuned, if the filter gets knocked around, right? 

[David Hess]

I notice you have what appear to be threaded caps over the adjusthent trimmers. Thats to protect the values from being accidentally detuned, if the filter gets knocked around, right?

The caps came with the piston trimmers and include a plastic gasket on the inside.  The piston trimmer adjustments are pretty stiff so I doubt routine shock would change them so the caps are more to prevent contamination and idle maladjustment.  If you look carefully, you can see the hex nuts which secure the adjustment side of the piston trimmers through the copper clad board but I think I soldered them on the inside once mounted.

The piston trimmers are 1 to 10 picofarads and provide a very smooth adjustment over a couple of MHz.

I opted for direct coupling as shown on the open end and you can see where I adjusted the coupling for best matching and never cleaned it up.  Because the open end is the low impedance side, attaching the cover has little effect.  A better implementation would silver plate the inside and seal the cover.

[cdev]

What is a good way to silver plate copper items without breaking the bank? I suppose, electolysis might work, if you have some silver. I remember seeing this discussed here, I should just search for it.

Thats pretty cool if you can actually brush on or paint on or rub on silver.

[T3sl4co1l]

Don't worry about plating.  A few more percent in component Q isn't important at a system Q of only ~100.  Plain copper will give a Q around 500 to 1000 at this frequency, I think??

Tim

[grouchobyte]

I've been assigned to make a bandpass filter as one of my semester projects for university.
I think designing it is no big deal but some questions have popped up along the way. I'm using https://rf-tools.com/lc-filter/ to help me design this filter and ADS 2020 to test the design.
Basically i've noticed that stopband attenuation plays a big role in the filter design but in my case the stopband attenuation wasn't given, meaning that i have to choose it myself.
I don't have much experience in RF design so i don't know what level of attenuation is a good ballpark figure to base my filter off.
For example, the calculator linked above defaults to -50 dB, but i don't know whether it's normal or some pie in the sky figure. What's a common real-world number i should be aiming for?

Parameters are: center frequency 145 MHz, -3 dB at 144-146 MHz (passband), 5-th order, discrete components, FR-4 substrate, SMA plugs.
Impedances are not given but they'll be the standard 50 .

Also i'm not sure which type of filter to choose. At first i was leaning more towards elliptic or Chebyshev, but now i'm thinking more along the lines of Butterworth.

Problem with Butterworth is that the calculator gives me capacitor values in the femto Farrads and i don't think i can get my hands on those.
Some freedom is given in the filter design as long as the initial criteria is met so i think some microstrip action wouldn't hurt.
How hard would it be to design a discrete 490 fF or 151 fF capacitor on a PCB? I don't think stray inductance would be much of a problem given the µH inductors next to the fF caps.

This may help.....PM me if you need some additional details. I haven't been on the forum lately so it might be hit or miss reaching me. Hope I got the details of your requirement more or less correct

-Bob/ Grouchobyte

[Refrigerator]

Thanks for the input but i don't think i could get away with all microstrip design when my requirements call for discrete components.
I think i might have painted myself into a corner with this one.   

Too bad the whole corona thing messed up my plans and i couldn't submit my own parameters for a custom filter in time.
I'll try playing around with various filters and parameters until i get something usable.

[Refrigerator]

I think those picohenry valued inductors are going to give you bigger problems.

I've only done a little bit of VHF filter design, but I've never had much luck at getting realistic component values out of online calculators. I've had far better results from Elsie: http://tonnesoftware.com/elsie.html

I tried running the program but it didn't run. Says it's 32-bit is there a way to run it on win10 64 bit?

Edit: got it running, turns out the installer just takes ages to start running, in my case at least.

[Refrigerator]

Here's an idea i had, can the circled components be made into microstrips?

[grouchobyte]

Thanks for the input but i don't think i could get away with all microstrip design when my requirements call for discrete components.
I think i might have painted myself into a corner with this one.   

Too bad the whole corona thing messed up my plans and i couldn't submit my own parameters for a custom filter in time.
I'll try playing around with various filters and parameters until i get something usable.

You can try and submit the microstrip design. It does have some discrete capacitors and there is really nothing wrong with the critical structures being microstrip....so its kind of a hybrid design. Your prof or TA might like the simplicity or elegance of an less standard approach. During my teaching years, I would have allowed it.

Anyways, Good luck finding suitable discretes

Bob/grouchobyte

[grouchobyte]

Here's an idea i had, can the circled components be made into microstrips?

What you circled are the microstrip structures in my proposed filter design
Are you finally coming around to my way of thinking?

Bob

[Refrigerator]

Here's an idea i had, can the circled components be made into microstrips?

What you circled are the microstrip structures in my proposed filter design
Are you finally coming around to my way of thinking?

Bob

Oh i think i didn't quite understand your first post then, i thought it was purely a microstrip design.
But i agree on the hybrid idea, it says discrete components but not how many (or to what %) so i think it would be ok.

[radioactive]

The first thing I thought of when reading the requirements is helical resonator.  I will not say that a discrete implementation is impossible but tuning will be a problem.  Below is a photograph of a 2 meter helical resonator which I designed and made by hand several years ago.
...

That looks pretty cool!  Here is some more VHF bandpass filter porn.  This one is tunable.  I didn't make it though.  Something I bought on ebay in order to measure really low level 2nd order harmonics from a high power amplifier.  (block the fundamental while passing the harmonics).

[radioactive]

Here is a Qucs simulation that might be a start.  I modified a 150-174 MHz BPF I designed in the past (made it a bit narrower and centered on 145 MHz).  I've included the simple inductor and capacitor models along with the simulation in the zip file.  I think if you construct on FR4 and use air-core wire-wound inductors / ceramic capacitors,  it should match pretty close to the analytical simulation.  It will not meet your requirements for passband BW.  This is more like 10 MHz vs 2 MHz as is.  Still might be useful in some way.

[David Hess]

Don't worry about plating.  A few more percent in component Q isn't important at a system Q of only ~100.  Plain copper will give a Q around 500 to 1000 at this frequency, I think??

I seem to recall that I got a Q of about 900 but measurement was difficult.  The references I worked from suggested building it 10% oversized to make up for the loss of Q if silver plating was not used.

[Refrigerator]

I received the go-ahead to make the helical resonator so that's what's i'm going to do 
I've found an online calculator for helical filters at https://www.changpuak.ch/electronics/Helical_Bandpass_Filter_Designer.php
Are there any good ways to simulate such filter?