[CQ-Contest] Bandpass filter

Jukka Klemola jpklemola at gmail.com
Sun Jul 8 03:02:24 PDT 2012

Tom has many things correct.

As a summary:
there is current flowing through filter components when there is
stopband energy trying to enter the filter.
It creates heat and thus components failures.
..equalling a filter failure.

- -   then to my long writing - -

Sorry this post became long.
Those interested maybe want to read the novel.

As in the Mythbusters, a warning there is some close-to-scientific content.

2012/7/7 Tom W8JI <w8ji at w8ji.com>:
>> I've had a C31 for years.  The separate feedline "option" for the C31
>> is QUITE the stretch. The numbers some previous posters have quoted
>> are conservative to my mind.  IMHO, the only station-safe
>> configuration to operate triple feedline C31XR at QRO is to also
>> spring for the high power NQN bandpass filters for 10/15/20.  That's
>> because, as others have posted, the typical exciter level bandpass
>> filter is not designed for 200+ watts out of band, and WILL go bad on
>> heat buildup in a contest. Only a matter of time before some cap dies.
> Here is the problem, although appearently it is difficult to get across. :-)
> Filters do not dissipate out-of-band power, they reflect it.

Filter does not pass the stopband power through.
That is, there is kind of no power loss through the filter as such.

Taking a more detailed look inside the filter complicates things.

>Stress on the
> filter will vary with filter placement on the line, and distance (because of
> standing waves) from the source. The antenna is the source, the receiver end
> the load, in the case of split feedline antennas.

Largely correct.
Also 'the other' TX frequency changes affect happenings in the filter
that is stopping that unwanted-on-this-cable signal.

This is a minor detail at this level we look at these things.

> This is probably esoteric, but there will be bad spots and sweet spots in
> the system where a capacitor's voltage rating or current rating might be
> overdone.

Now we come to the tasteful part of engineering.

This voltage/current happening to a set of components or maybe just
one single component in the filter causes filter failure.
The voltage/current induced phenomena are practically always due to
heat in ham graded commercial applications.

For passband this is clear as the passband calculus is normal
engineering practice.
When going over limits, something overheats.

For stopband analysis, Tom has the right idea but I am not sure if he
follows it through to actual cause of the component failure.

The root of this is when there is current going through a capacitor in
the filter, there is heat dissipated.
Inside the usually tiny and lightweight component that does not have
even a square inch of dissipating surface area.

To add to this, most of the components we have/see in the ham-grade
commercial filters are not completely lossless.

In reality, the first (usually) parallel components take the energy ..
they take the heat.

As a result, we agree there is stopband energy trying to enter the
filter and the first components are directly subjected to all of it.
Heat is one result.

I have not seen a broken filter that would have been broken clearly by
over voltage.
Coils are usually critical only on the passband - - usually.
Many have seen melted coils. Failure root cause usually coming from
something on the passband.

We need to understand the current breaks also capacitors .. that is
heat breaks the capacitors.

Those capacitors are components of a filter, thus we see a filter
failure due to power dissipation overload.
And - - this heat really adds up during the contest.

And yes, the capacitors age when they are heated up.
If marginal, they eventually show a failure.

To help this, the TX stubs reduce the stopband currents in the lines
(stubs attenuate), helping the BPFs survive as there is less current
flowing through the filter first components in the direction where the
unwanted energy comes from.

> The REAL problem is, based on the cheap filters I have seen, is the
> capacitors are just sized far too small. In filters I have seen, capacitors
> are barely large enough to last into 50 ohm loads at rated power. They have
> no current headroom, or voltage, and often are types with low Q in the
> capacitor.

Correct for the passband.
Stopband performance is not rated .. thus performance is based on some
experience in the best cases.
More typically it is based on luck.

> I added this section to a web page on coupled power:
> http://www.w8ji.com/antenna_coupling.htm

Suggested reading yes.

> Some say stub (or filter) location does not matter!

It is better to have a stub and/or a filter or a set of them than not
to have anything.

> I would NOT put a reject stub or filter right at the output port of an
> amplifier, I'd put it just under 1/4 wave away **on the reject band** for
> most amplifiers. I'd put it at the antenna, and/or plan the impedance the
> antenna sees at the stub, if I wanted to protect gear.

Exactly correct.
The output stage (exciter or amp) has a filter at the output, the
additional filter should be at least some percents of wavelength away
to really give an effect.

Look for K3NA stub article in the net. Eric has illustrated a dual
stub system where the individual stubs are about 1/4 waves from each
other .. I can verify the results by measurements at OH4A.

I had to build such a beast on 40 to try to reject the 20 harmonic. It
was not enough.
Then I added a separate trap using coils and capacitors for the 20m
harmonic (passing 40) and behold .. the harmonic level starts to be
bearable. Yagi separation is about 7 or 8 ft.

Yes .. I know I was digging for trouble but I have only two towers
here and decided to maximize the station performance.
There was also a bunch of engineering to learn (= read: some heated up
components burning fingers).

> We can brute force things, or we can plan and understand better. Either one
> solves the problem, but understanding how it really works and what really
> causes the issue is probably better for the community as a whole. Even if we
> can't have a universal recipe, we could at least stop propagating the silly
> ideas that all sources and loads are 50 j0 out-of-band, and filters absorb
> and/or dissipate a significant part of out-of-band energy.

We are fortunate to have a world that is not perfect.
It means there is some more Fun left.

> 73 Tom

Jukka OH6LI

More information about the CQ-Contest mailing list