PI4CC

Contest Club
QRV since 1993.

PI4CC

Bandpass filters for PI4CC



Contesting is all about the ability to listen. Undesired strong signals, even out-of-band, can cause undesired QRM due to overloading receivers. Harmonics that end up in the next higher bands are even worse. With our current coax stub filters we can suppress small frequency-slices (notches), but the majority of the spectrum passes unattenuated. So, we embarked on a project to remove all signals that are outside the bands of interest.

This bandpass filter project aims at serveral goals:

  • Ability to handle high power (1 kW)
  • Low insertion loss to avoid heat build-up in order to avoid a cooling fan.
  • Adjacent band attenuation > 80 dB.
  • Easy reproduction, modular setup.
  • Rugged, stable housing.

Literature suggested a combination of Chebyshev and Cauer characteristics. The design was done by Elsie, followed by simulation with LT spice. We settled on a 4-pole induction-coupled Chebyshev band pass filter followed by a single Cauer stage to provide a notch for the adjacent band below. LT spice showed a next band rejection of 80 dB. Note that capacitor-coupled Chebyshev stages would have the "tail" on the "wrong" side for this purpose. The picture below is a combined LT-Spice simulation of 6 filters.


The attenuation values are summarised in this table.



As some 25A of HF current circulates in some stages, an initial test was done with an arbitrary tank circuit, powered by a transmitter.

For small C-values, typically a "C0G" dielectrum (losses of about 0.1%) is used. Heat build-up in this type of capacitors is limited. So far so good.



A test configuration with 4 Chebyshev stages with much bigger C's (which have dielectric losses of about 2%) showed too much heat generation. Therefore, the Russian C's were replaced by silver mica C's that have practically no loss (0.01%) and can handle about 12A each.

The test rig was loaded with 1 kW continuous power into a dummy load for 25 minutes. The silver-mica C's remained cold and only limited heat build-up was observed in the coils.



In order to get coils with better Q (less losses), an attempt with thicker wire was made. The disadvantage of bare copper is clearly that it oxydises rather quickly, which has a detrimental effect on surface conductivity.

The test rig was expanded with the Cauer notch filter (right hand coil and capacitor), which basically forms a ground path for the frequency (in this case ~1850 kHz) to be removed.


All measuring results for the 80m filter were captured in a single picture (horizontal 1-31 MHz, vertical 10 dB/div):
Purple line shows the 80m coax-stub notch filter.
Yellow line shows the 4-pole Chebyshev filter.
Blue shows the same filter including the Cauer stage (a notch on 160m).
The second harmonic (which sits in the 40m band) has about 38 db attenuation and 160M will be > 70 db suppressed as compared to the current coax stub notch filter.


Finding optimum Q experimentally by variation of length, diameter and wire thickness is next to impossible. A mathematical approach is unavoidable. As proper calculations are far more complicated than the formulas found in textbooks, software (incuding real-time animation) was written to arrive at the desired trade-offs between size, wire, material and heat dissipation. The picture shows diameter, L/D and surface temperature for a given self-induction, material, wire thickness and RMS current. This model is to be published in due course.


Suitable capacitors are being sourced from American Technical Ceramics www.atceramics.com. ATC provides component and custom integrated packaging solutions for the RF, microwave and telecommunications industries.


Most C's are received. Now it time to design the PCB.


The PCB.







A universal PCB is developed. This will held all combinations of all type of filters. You will need 5 PCB's per filter.

The PCB's has arrived and its time to build a filter.The first filter section assembled.


Five of these units will form the bandpass filter


A prototype RF wise housing with the 40m modules inside.


A sweep on the 40M filter. A nice > 80dB recejction of 80M band and above 40M 80dB on 20M slowly going up to 70db rejection on 10M.

Finally we found the correct way to make the housing. We are inspired by Hammond, but we make the cover of perforated aluminum. A few photo's with contruction details are here.


The same 40M bandpass filter but in the new housing and fine-tuned.
Look add those figures.
In band < 0,3 dB loss and > 77dB on any other contest band!

© Copyright 1996 - 2026 PI4CC All rights reserved.