Six Section Remote Controlled Low Pass Filter

Updated 20160731
I designed a companion low-pass filter to the 20W amplifier. These boards are now sold out, and my new amplifier has the filter on a single PCB. A photo of the bare PCB:
A bullet point design spec:
  • Size 100x80mm, half euro-card
  • Covers all HF range, 160m-10m (note some banks cover 2 bands)
  • Through hole parts used where they have to be, toroids + relays
  • Laid out to optimise loss - thru path on higher bands is shorter
  • Optimal grounding around toroids, learning from commercial layouts
  • Economically priced relays are used
  • Will handle up to 100W with T68 toroids and appropriately rated capacitors
  • Ceramic SMD capacitors to reduce size and cost
  • Can use SMD or thru-hole filter capacitors, your choice!
  • SMD capacitor pads will accept 0805 or 1206 size parts
  • Filter bank selection by direct logic (switch), or;
  • By binary code, 3/8 decoder (decodes brought out on connector), or;
  • By I2C bus control, you choose which way!
  • Gerber files released under public licence, get your own made

Designing this board took a long time with several restarts. Most commercial rigs use 6 banks to cover 160-10m. Realistically only 6 filters with 7 poles will fit on a board this size. The toroid dimensions dictate that. Having the decoding section takes less PCB space than another filter bank, but adds a lot more optional functionality. Omron G5V-1 relays are used in many 100W Japanese radios. They are easily available, small and cheap.

The hardest part of building this board is winding the toroids. Its best to start with the higher frequency bands as they have fewer turns, and get the hang of winding them evenly. The silkscreen markings on the board will show you which way the toroids have to go. You start with a short tail under the winding, and proceed clockwise with the turns.

Choosing the filter capacitors determines the power the board will handle. For up to 40W PEP SSB power, 100V rated parts will just about cope. SMD capacitors of 0805 or 1206 from reliable manufacturers will perform in this application equally well as expensive and bulky silver mica caps. The thing to avoid is unbranded 50V Chinese capacitors, they will burn up! Modern 200V RF rated “1111” capacitors will also fit the pad design, are ideal for handling 100W CW, and far smaller than silver mica caps. Thru-hole pads are provided for anyone with a junk-box stock of suitable RF rated capacitors - I leave the choice to you.

For up to 100W you will need larger toroids, ideally T68. The board is designed to accept T68 as a maximum size. I put on T68 for the top band bank in the prototype because the number of turns with the wire thickness available was easier. The circuit diagram lists numbers of turns. Be careful over the different permeability of the toroids (which is indicated by their colour).

*WARNING* Some iron dust toroids sold on eBay are Chinese copies and perform badly. They can often be recognised as lighter in colour than the genuine parts.

Production data:
Gerber file release Iss.A dated 20141228
Circuit diagram Iss.A dated 20150211
Items list (Excel format) dated 20151026
Component positions for placement

The prototype board fully populated and undergoing test:

Filter Responses
Measured with an SDR-Kits VNWA, the low pass responses of the prototype are below. Any preceding amplifier should be of class A or Class AB push-pull. Class C amplifiers will need higher stop band attenuation to get enough harmonic rejection. Typically modern balanced amplifiers using class AB MOSFETs will give a clean signal with low second harmonic to avoid EMC issues, and conform to the conditions of your ham licence.

Starting with 160m, here are S21 plots of the filter banks on the prototype:






Filter values are originally from the Y2000 ARRL handbook and given here. You are welcome to design your filters and the positions for thru-hole capacitors may help. Good software for designing filters is ELSIE. It is possible to switch out all the filter banks and have an external one for 6m, or whatever else you want.

Optimising capacitor values
If used with the matching amplifier, the input capacitors can be adjusted to improve matching (to 50 ohm output) and therefore the gain/output power. I have only done this with STP16NF06 devices at the moment. Mitsubishi RD16HHF1 will be different. You will need a reasonably accurate, and >30W rated, 50Ω dummy load, and at least a good quality SWR/power meter to do these tests. This tweak is entirely optional, you will find the percentage improvement may not be very much.

All I do is place a parallel capacitor at the input while driving the amplifier module to about 20W. Monitoring the power, I place a capacitor (about 20% of the value) across the input capacitor of the filter board. If the power goes up, try the next value up in series. If the power goes down, try the same value on the output. Almost always an increase of one of these capacitors above the stock values (given on circuit diagram) will bring a small improvement. In some cases there will be no improvement possible.

Any increase in filter capacitance will increase the low-pass effect. So it is best to perform this optimisation when transmitting at the top end of the range, e.g. 1999kHz for top band, 14349kHz for 20m. Also be careful with "dual band" banks to check the matching is improved on both. If one goes up and the other down, it is best to leave well alone.

By this cut & try method, a few extra watts can be pulled from the amplifier and board combination. The values I came up with were:
160m - C7/C8(output): 150pF
80m - C44/C51(output): 100pF
40m - C9/C10 (input): 68pF
30/20m - C34/C35(input): 47pF
18/15m - C16/C18(input): 27pF
12/10m - C27/C28(input): 15pF

These part numbers are the input when the amplifier and filter are stacked. In other words with J2 as the input and J1 as the output. If using coax, the left pads as input and right pads as output. Of course if you wire the board the other way round the component values listed will "reverse".

The filter board is designed to generally accept common sized capacitors if you need some special filter response. I recommend the
ELSIE program for designing your own filters.