Low Pass Filter for 20 meters - determine the best cutoff frequency

[Jeff Weinmann]

Hello,

I'm looking to 3 design a 3 pole low pass filter with the aid of the following site:

FYI its Dale Weatherington's site, one of the co-founders of Hayes Modems FYI:

http://www.wa4dsy.net/filter/filterdesign.html

I'm looking into filtering for a 20 meter frequency of 14.090 Mhz and this is what I came up with:

http://www.wa4dsy.net/cgi-bin/lc_filter4?FilterResponse=Lowpass&poles=3&cutoff=25&funits=MHZ&Z=50

There are tons of sites showing HOW to calculate the -3db cut off frequency, but I'm looking to minimize the db loss at the working frequency of 14.090 Mhz.

I picked 25 Mhz because its below the 28 Mhz harmonic, but I'm also looking to filter out as much between 14.300 Mhz and above. I would like to do this for other bands but I would like to understand more why I should pick a particular cut off frequency instead of guessing.

The purpose of the filter is to clean up the square wave signal of an SI5351a clock.

Any help would be appreciated!  73 W8ZLW

[bd139]

Start with how many dB you need to knock off the first harmonic, work out how many toroids you can be assed to wind, convert that to the filter order and work back to the cut off frequency. It will probably be well in the flat low insertion loss part of the filter.

If it’s a low pass you probably need more than a third order filter though. 5th at the very least.

Edit: also dont assume the Si5351A is a 50 ohm match for the filter as well so measuring insertion loss will be difficult.

[UR5FFR]

Try RFSimm99 software and use real Q for inductors

[fourfathom]

What will you be driving with the output of your filter?  This really determines the level of filtering needed.  And what filter loss can you tolerate?  Sometimes accepting a higher loss can give you a sharper or more tolerant filter.  BTW, with the Si5351 output set for "high" drive, the output Z is supposed to be close to 50 Ohms.

You might try something like this filter.  It's not flat, but broadly peaked around 14 MHz.  The 2nd harmonic attenuation is around -50dB, and the third harmonic attenuation is close to -65dB.  I am assuming 1 Ohm DC resistance for the inductors, which are T37-6 Iron Powder, 14 turns.  This doesn't have the return loss of a nice Butterworth Pi filter, but it's decent.  The loaded Q for this circuit is low enough that reasonable component losses and tolerances aren't very critical.  Note the tapped capacitor input and output -- this allows the Q to be independently set, and also provides the necessary AC coupling for the Si5351 output. 

I use this topology a lot, as I work with synthesized signals that may have significant sampling artifacts in addition to the textbook harmonics, both above and below the desired frequency.  The Si5351 is pretty clean in this regard though, so a flatter LPF is probably fine. 

[radiolistener]

I picked 25 Mhz because its below the 28 Mhz harmonic, but I'm also looking to filter out as much between 14.300 Mhz and above. I would like to do this for other bands but I would like to understand more why I should pick a particular cut off frequency instead of guessing.

You're needs to take into account, how much harmonics should be filtered. When your harmonic is more far from cut-off frequency you get better result. If you're using 3-pole Butterworth LPF with cut-off 25 MHz, your 28 MHz harmonic will be attenuated for just -4.73 dB. I think you will not be happy with that

[radiolistener]

The purpose of the filter is to clean up the square wave signal of an SI5351a clock.

I think it's better to use cut-off 14 MHz. In that case your filter will have -18.13 dB attenuation for 28 MHz and about -3.09 dB insertion loss. So, the second harmonic will be -18.13 - -3.09 = -15.09 dBc, which is much better, but still is not enough to comply with transmitter requirements. If you want to use it as transmitter, the second harmonic should be at least -40 dBc or even -60 dBc for some countries.

Another way is to use 7-pole LPF, with cut-off at about 14.5 MHz, it will give you -40 dB attenuation for 28 MHz and insertion loss about -2.07 dB. So the second harmonic will be -40 - -2.07 = -37.93 dBc. Which is almost fits with transmitter requirements. You can add additional filter after power amplifier in order to improve that result.

[Jeff Weinmann]

Thanks for the input,

I'm attempting a WSPR beacon station with only 10 milliwatts max,  So I will be using surface mount inductors like the project here:

http://pe1ryy.blogspot.com/2015/04/low-pass-filter-added.html

So at 10mw,  I'm trying to save as many db as I can.  I'm not exactly blasting the airwaves with noise at 10mw, but its still considered illegal.  The article above as a good start for me though.

[fourfathom]

In that case, look at this design, a 5th order Chebychev low-pass filter, 0.5dB passband ripple, Corner freq = 15 MHz:  Just add input and output shunt capacitors to the filter in that blog post.
14.096 MHz fundamental attenuation = 0.2dB
2nd harmonic attenuation = -40dB
3rd harmonic attenuation = -59dB
Since the 2nd harmonic from the Si5351 is already quite low, the output will easily meet the ham transmitter spectral purity requirements.  You can also push the cutoff frequency a bit higher and still meet purity specs.
There will be some additional loss, mostly depending on the inductor Q (as with any similar filter).
Here's a link to that filter calculator: https://rf-tools.com/lc-filter/

For several years I was running a 20m WSPR transmitter based on a RPi (just the RPi, the signal was generated by the internal counters).  This put out about 10mW and with a filter like my previous post and a simple dipole, was heard from Japan to Alaska to Europe to New Zealand.  Here's a link to a ham-club presentation I did about this: http://wb6cxc.com/wp-content/uploads/2016/03/Ham-Presentation1.pdf

By the way, the Si5351 waveforms shown in that blog you referenced look pretty bad -- they are typically much cleaner squarewaves than shown.  I suspect poor layout or measuring technique.

[Jeff Weinmann]

Thank you for a most informative reply!

The antenna I will be using is a shark mini 20 hamstick, connected to a cb mirror mount and grounded with a ground spike and radials if needed. I've done this setup before (with no filter) and 'illegally' had my WSPR signal reach California from a QTH of Orlando. 

Thank you for taking the time and I will definitely be checking out your link!

73 W8ZLW

[Jeff Weinmann]

One more question,

You mentioned I should use 'shunt' capacitors along with the filter,  where would they go in the attached filter schematic?

What value should the shunt capacitors typically be?

thanks again!

[mag_therm]

USA Hams have spurious  TX limits at  43dB below the in-band average power.

I have the Coilcraft chip inductors Designer Kit for 1812CS inductors.
Kit has a good range of L, from 1 uH to 33 uH. and 10 pcs of each value  There are other kit ranges too. Reasonable price.

Also have the Johanson 0805 Ceramic Chip Capacitor Kit. Quite expensive , but also  a good range of 5% npo

I have designed receiver downconverter input filters bandpass for 20 metre and 30 meter. Artworked but not built yet.

I might be on here for advice  if they don't work !

I am using qucs Smith chart and functions for m-derived filters.

[fourfathom]

The filter above is complete, essentially a two-section Pi, AKA a 5th-order Pi network.  When I said "add shunt capacitors" I was referring to the L/C/L "Tee" filter seen in the pe1ryy blog.  Take that filter, add shunt capacitors at the input and output and you end up with the two-section Pi (which has much sharper roll off and better harmonic attenuation.)   The new capacitors won't add much if any noticeable attenuation, it's generally the inductors that are lossy, at least in practice.

What measuring equipment do you have?  Building a filter can involve adjustments to make sure it is set to the right frequency and shape.  The filters I have shown are fairly "low-Q" and should perform well with normal care and component tolerances, but it's probably a good idea to at least do a frequency sweep to check the cutoff region.  Since you have the Si5351 driving this thing you should be able to easily test across a wide frequency range.  Get 5% (or better) components, COG/NPO capacitors, and it should be close enough.

That first filter I showed (with the LtSpice simulation) is essentially the same as the 2-section Pi we are discussing, just with tapped capacitor in/out sections, which allows more flexibility in component selection.

[Jeff Weinmann]

I have a 80Mhz function generator and Oscilloscope so testing is not an issue   I do plan on making my own pcbs with 0805 inductors and capacitors.  should be fun!   My Agilent O-scope can do Bode Plots,  I just have to read up again how.

thanks again I will try and source the parts for this design and see what kind of response I get.

[fourfathom]

While you can make your own SMT filter boards using an xacto knife, using KiCad and having the boards fabbed is so cheap and easy that I usually go that way.  Here's a board I use, with a flexible front-end filter (with optional notch-filter components), and a 20dB preamp, for driving a cheap SDR receiver.  (the 3D model for the SMA connector is off a bit, the actual one fits perfectly.)