At 07:44 PM 11/23/97 EST, you wrote:
>No need to repeat the long stuff from Steve but I feel that he made a few
>points that I think are wrong.
Carl, I believe that you are talking about a totally different failure
mechanism, that caused by an incorrect load, and which is something that
occurs over some finite, measurable period of time, sometimes a few seconds
or more? I and Rich were talking about PARASITIC OSCILLATIONS, those things
that cause your amp to go BANG!! in the night, then it's all over before
you've even stopped blinking.
>1. Once a sustained arc occurs in the Tune cap or across the bandswitch
>it becomes a very wideband signal source and of course it will be heard
>all over the spectrum. It is called a spark gap.
Not true; "sustained" arcs are a very low resistance path because they form
a "plasma" envelope which is highly conductive, just like a metallic shield.
As a result, solid arcs radiate very little RF energy.
OTOH, arcs that *do* radiate broadband energy are those that are constantly
starting and stopping so that there is a very sharp rise and fall time,
which is what generates wideband energy extending into the microwaves. This
could occur as material is burned off the capacitor or bandswitch plates.
But the energy dissipated within such arcs is generally only a fraction of
the total amplifier output power. In any case, it's quite easy to tell the
difference between such sustained arcs and the infamous parasitic oscillations.
Recall that when the space shuttle is re-entering the atmosphere at very
high speed, a plasma "shield" is created around the shuttle through which no
radio wave communication is possible. This plasma "shield" persists until
the shuttle has been sufficiently slowed by air friction that the plasma
dissipates.
> However that arc is not
>a parasitic. It MAY have been started by the parasitic or something
>entirely else.
According to Mr. Rauch, there is NO SUCH THING as a "parasitic oscillation".
Since he cannot be speaking of a "parasitic oscillation", I guess I
misunderstood the thrust of his earlier comments, too: he was *really* only
addressing the problem caused by a severe mismatch at the output of the
amplifier; a bad antenna, a loose antenna connection, or a T/R relay not
fully closed. *These* kinds of failure-inducing mechanisms are known and
understood by nearly everybody, I think. Take your antenna off your radio,
and your radio's likely to go up in smoke. Might take a few seconds, maybe
even a minute. But the radio usually survives for at least a *few* seconds,
and the damage is often not even catastrophic. In any case, when it results
in arcing across tank capacitors or even bandswitch contacts, the damage
usually takes a few seconds to become a complete disaster, before which you
can always lift the key, attach the antenna, and sometimes continue on
without even knowing that you arced your amplifier.
Not so with most parasitic oscillations; they're extremely quick, usually
pretty loud, meters peg both ways, and the amplifier usually doesn't work no
mo. And often, the tube doesn't survive more than a couple of them, either.
>2. A true parasitic will take place with the amp terminated in a perfect
>50 Ohm load...DC to light...since it never gets past the tank circuit in
>the first place. Try it yourself; drive the pi-net from either direction
>with a 120-200MHz signal and see what comes out the other end.
It's absolutely correct that a "true parasitic" can occur with ANY load (or
absence thereof), and that was part of my point. But it's not necessarily
true that the VHF energy generated at the tube anode cannot pass through the
tank circuit. Whether it can or not depends upon what impedance that tank
presents to that particular frequency.
For example, it's quite easy to design a Pi-net tank that acts as a low-pass
filter below some HF cutoff frequency, but which also passes energy with
less attenuation at higher frequencies; an elliptical LPF is a good example.
The parameter that you specify to order filters which are designed to avoid
this higher-frequency energy transfer is called "ultimate attenuation". It's
quite often a problem with "store-bought" microwave bandpass filters; at
some much higher frequency, the filters pass RF energy virtually
unattenuated. In microwave filters, this is usually caused by what's called
"moding" whereby the filter enclosure itself passes energy unattenuated from
one end to another.
A closer-to-home example is the crystal filter in your radio. Specifying a
filter to have 80 dB of ultimate rejection will do no good if the board on
which the filter is mounted cannot, by itself, provide that much rejection.
It doesn't take very much coupling between PC board traces, even at 9 MHz,
to provide -60 dB coupling; imagine how much coupling you might see between
turns of that pi-net output inductor in your amplifier at a hundred
megahertz; or between bandswitch contacts or wafers.
In the case of a HF amplifier, it's quite possible for the pi-net inductor
itself to have sufficient stray capacitance, either from the coil turns to
ground or from turn to turn, that the inductor passes energy at some V/UHF
frequency virtually unattenuated. A very similar design principle allows the
construction of distributed amplifiers operating over a decade frequency
range (for example, from 20 to 200 MHz). Extremely careful consideration and
compensation of all strays is necessary.
>3. What you see in the SB-220 and many other amps is what I call a
>secondary reasonant circuit formed by the shorted turn effect of the tank
>coil - bandswitch combination.
[snip]
This sounds more like you're really disturbing one resonant circuit by
creating another while using the same components which cause the first. It's
an idea I haven't heard of before but which should prove useful.
>I may not have all the terminology correct since "I aint no durn
>injuneer" but I do feel that the basic idea is correct.
>I have run SB-220's with the wide spaced tune cap and an experimental
>plate xfmr at power levels of 1500-1800W with nary a trace of arcs once
>the snubber is installed. This also assumes that the parasitic
>suppressors are replaced with improved versions since those original 2W
>carbons just cant take the 10-15M RF at those power levels.
Makes sense to me: destroy (or at least detune) the original VHF resonance
in the plate tank circuit, and away goes the original parasitic. Of course,
*another* one might pop up...but change the cap or where it's connected, and
you can displace that resonance, too.
Thanks for the tip, Carl; I'll have to remember that lil trick! 73, Steve Ko0U/1
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|