This web pages describes a do it yourself 70 MHz high power low pass filter that is easy to make and from easy to find parts.
The low pass filter is a five pole Chebyshev filter as shown in Fig.1.
Fig. 1. Schematic of the low pass filter.
The filter is made so that the 70 MHz band is inside the last ripple of the pass band characteristic in order to make the attenuation slope as close to 70,5 MHz as possible providing maximum attenuation of the harmonics and other unwanted signals.
Unfortunately I am not very mechanically skilled so I chose a ready made box that is reasonably prised and may be soldered upon. Initially I tried to make C1 and C2 from FR4-PCB material but the loss was to high at 70 MHz. Instead I chose two door knob capacitors. The theoretical value is 56 pF that is close to the 57 pF value that is used in many W3DZZ traps thus easy to find. But the value is not critical ±2 pF should not cause any problems. The door knob capacitors and brass plates also serve as mechanical assembling and soldering points.
It is important that the coils cannot "see" each other. Therefore the coils are placed the way they are. Alternatively it might be necessary to install separation walls between the coils.
Wilko, PA1WBU, has found that the box may have a big impact on the performance of the filter. Separation walls and a small box will detune the filter and result in a poorer performance. Thus a big box is recommended unless you are willing to compensate for the change in stray capacitances etc.
The easiest way to adjust the filter is by connecting it to a network analyser or spectrum analyser with a tracking generator and monitoring the transfer characteristic while expanding or squeezing the coils. In case you do not have the right instruments you may tune the filter for minimum pass band attenuation.
Fig. 2. Low pass filter transfer characteristics. Pass band attenuation is 0,2 dB, and second harmonic is attenuated 42 dB and third harmonic is attenuated 57 dB. The FM Band is attenuated by more than 13 dB.
The pass band loss of 0,2 dB might with some additional effort be reduced to 0,1 dB by using a box that is not made from tinned iron and making the inductors from thicker wire, e.g. the centre conductor of a H-100 cable. Another idea is to wind ALL the coils on the same uncut wire. But this will probably take a couple of experiments to get it right.
|C1, C2||57 pF doorknob, value is not critical ±2 pF may also be used|
|L1, L3||3 turns, Ø16 mm, 12 mm long, 2 mm enamelled wire|
|L2||4 turns, Ø16 mm, 15 mm long, 2 mm enamelled wire|
|Metal sheet box||74 x 148 x 50 mm3. Schubert Weissblechgehäuse type|
|Connectors||N, BNC or other types|
|Nuts and bolts||Eight M3, 8 mm long|
|Brass plates||Two pieces 10 x 30 x 1 mm3|
I have used H-100 for the coils and 2 x 27 pF 3 kV disk ceramics in parallel for each of the capacitors, and flattened 10 mm cu pipe for the supports. I drilled the box for doorknob types by they failed mechanically when I screwed it all together so it looks a bit of a hash now...! Anyway - results are similar to yours:
The attenuation curve drops nicely to these levels once the lid goes on!
Filter made for the OE5QL beacon.
Bo, OZ2M, www.rudius.net/oz2m