[CQ-Contest] Bandpass filter
w8ji at w8ji.com
Mon Jul 9 05:09:30 PDT 2012
> I can relate to the problem of capacitors with lower than sufficient
> rating. I use the ICE 419 banpass filters while on DXpeditions. I have
> A models and two B models. Both models do a good job rejecting
> crosstalk. However, the caps in the 20 meter and up tuned circuits have
> heated up, resulting in high VSWR. This usually happenes when the
> switches the transmitter to a different band but does not remember to
> the filter until AFTER transmitting and seeing high VSWR. Ten meters is
> far the most vulnerable section. I have a large supply of 270pf 500v
> mica caps on hand to make repairs. Two of these caps are wired in
> to get 540pf for the tank circuit. If I could find caps rated at 1000v, I
> would use them.
I added a SPICE model of two low pass 3.6 MHz filters to my webpage. You'll
see the stress on parts there for a 100 volt peak source through a 50 ohm
resistor, which is about like a 50 watt transmitter when matched.
K6LL Dave is correct.
Manufacturers traditionally use capacitors that have too much ESR, or not
enough dissipation to handle the ESR. It isn't very often there is a voltage
failure. Sometimes a higher voltage part has increased physical size and
ability to get rid of heat faster than ESR increases from the thicker
dielectric, so it handles current better.
This isn't exactly like resistors. We have to be careful splitting values
into smaller caps in parallel to increase overall current rating. Smaller
values often have lower safe current ratings, even when not voltage limited
by X*I voltage. It is easy to get into a situation where two caps in
parallel of half value barely increase total safe current through both caps,
and I've even seen it go backwards thermally for the same net current!!
Another factor is thermal stability of capacitance.
The snubber micas Dave linked to are good capacitors, that's why you see
them in some amplifier filters and tank circuits (the ALS600 filter and in
some parts of other less than 1 kilowatt tank circuits). They have high
voltage breakdown, much higher than normal current rating, and are
temperature stable (like most micas).
The very best capacitors today are high voltage ceramic chip capacitors (the
network in the AL1200 and 1500 went from doorknobs to ceramic chips many
years ago, and the ALS1300 has chips). These chips are far superior to any
leaded part I'm aware of, in particular those manufactured by ATC.
Here's something else important. Some filters have long leads on capacitors.
This can do two things at higher frequencies. It can increase circulating
current in the capacitor, and it can decrease filter rejection at higher
Sometimes higher harmonics can be suppressed over 20 dB more, for the very
same filter design, using chip caps and good layout compared to a leaded
part. One company years ago sold a filter for lower frequencies that was
wide open near 30 MHz because of lead length. Chips not only have much lower
ESR and heat, they are much more temperature stable than doorknobs and have
higher current rating, and act much more like theoretical capacitors do at
higher frequencies. Snubbers are the second choice.
More information about the CQ-Contest