Roger Parsons wrote:
>I will probably be sorry, but I am going to stick my
>nose into this one:
>
>
I thought I would be sorry but without getting emotional about the
issues we have good technical discussions and I always learn from them.
>1. It is certainly true that a Class B grounded grid
>amplifier will have a varying input impedance over the
>driving cycle.
>
>2. If a matching network is placed at the input, it
>can be designed to transform that (instantaneous)
>impedance into some other value.
>
>3. If the matching network is lossless, the
>transformed value will vary in just the same way as
>the drive impedance, but the values will have been
>transformed.
>
>4. If a lossless length of coax is connected between
>the amplifier input and the matching network, then the
>impedance will still vary over the drive cycle, but
>the values will have been transformed again.
>
>5. Therefore, with perfect components, there is no
>difference in principle between the matching network
>being at the amplifier or separated from it.
>
>6. Pefect components do not exist, so the matching
>network/coax will introduce some loss. This loss will
>depend on the network configuration and type/length of
>coax. The variation in the drive impedance will
>therefore be marginally less, but good quality real
>components should not be dramatically different from
>perfect ones.
>
>
Here is a point I think I differ with you. It's not only about loss but
characteristic impedance of coax when operated without the proper source
and termination impedance. Things get real weird when we do that and,
as other posters have said, we can see transformation effects that can't
really be predicted given the variations at the cathode. And, as you
said, we don't have any perfect components.
>7. If a transceiver with an automatic ATU is used and
>it hunts without an input matching network, it will
>hunt with one, assuming the circuit losses are low. It
>is also conceivable that it may not be able to match
>the impedance seen at the end of a coax connected to
>the amplfier input.
>
>
It's been shown that that isn't true. I don't know the math nor the
high level theory but I have "seen" it.
>8. Although grounded grid amplifiers are often
>considered to be unconditionally stable, that is not
>always true (believe me, I know!), and a capacitor
>across the input may help stability. This is
>automatically achieved with a pi network and may be a
>good reason to use one close to the tube.
>
>
I agree. Stability is not automatic in any amplifier. There are too
many so called grounded grid amps that have so much stray inductance in
the grid circuit that I'm surprised they don't oscillate all the time.
>9. The only way that I know to make the drive
>impedance appear effectively constant over the driving
>cycle is to swamp it with a dominating admittance.
>This increases the drive power required, probably by a
>significant amount.
>
>
Others have suggested using resistance too. It obviously works in some
form but I don't think that it does anything to improve the IMD of the
amp. That's probably where I'm headed with the discussion. We can all
get power out of an amp by driving it with "something" but when that
signal is unleashed on the bands is it friendly to our neighbors up and
down the band or is it one of those things that (I'm sure) some run that
way to clear themselves a hole (listen on 75 meters any night or on all
the bands on any contest weekend)? I suppose one might argue that it
serves a purpose but I feel that we are obligated to make the signal as
clean as we can.
>10. If the feed impedance of the amplifier is
>significantly different from the drive impedance of
>the exciter, then a matching network can reduce the
>driving power required by providing better matching.
>
>11. Any flywheel effect from a matching network at the
>input to the amplifier will be no different from a
>flywheel effect from a matching network some distance
>from the amplifier. The commonly recommmended Q for
>input matching pi networks is so low as to produce a
>negligible flywheel effect.
>
>
We differ in opinion on two counts. True the Q is lower than you'll
find in the plate, but with Eimac's recommendation of an input Q of 4
and by using a compromise Q of 2, things appear to work well for many
people. As Rich says in his paper on the SB-220, that amp's original
input circuit had a Q lower than 1 and simply didn't work. Raising the
Q of that input circuit to 2 yielded enough Q to improve amp performance
without destroying the SWR across the bands. Putting a length of coax
between the cathode and the matching network goes back to the point in
your #6. You end up with transformation and loss that at best hides the
matching network from the cathode or at least disguises it as something
else.
>12. It is a lot more convenient if all impedances
>within a station are kept constant. No retuning is
>then necessary when the amplifier is/is not in
>circuit.
>
>
Amen! Will said the same thing to me and no greater truth has been
spoken. I believe that if you can achieve that goal, you will also get
a cleaner signal assuming you're not going to over drive the amp.
>I can see no mechanism by which the IP performance of
>the amplifier can be improved by the use of a matching
>network physically at its input, rather than slightly
>separated, always provided that the amplifier/system
>is stable.
>
>
Here the term "slightly" may have to be defined. Even Collins Radio
separated them "slightly" inside some well known amps but not by several
feet as is the jumper between an amp and most rigs driving them. I've
read where people have reported problems with short coax jumpers between
the cathode and tuned input of only one foot. This may take us to
consider the capacitance of the coax too, I don't know.
>Hard hat firmly in place.
>
>73 Roger
>VE3ZI
>
>
Thank you for expressing your opinions Roger! No hard hat needed :~)
73, Tony W4ZT
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|