K5PC - Phil Clements is right on the button about our most popular
subject here. (see excerpt below) Amen!
I have found myself in the #3 category in my designs over the past
25yrs. Have spent months taming a 190 MHz parasite in a 100 MHz
cavity amplifier with 4CX3500A in it. The solution was built in
hundreds of commercially sold transmitters - some of which are still
running in FM stations worldwide. It used an LC trap, using Nichrome
wire in the L to deQ the circuit and allow it to work with a variety
of tubes without the customers needing a network analyzer to set it
up cold.
Have gone the extra mile to build a suppressor with resistors, copper
hairpin inductors and FERRITE tiles in it, for my recent (1998
design) 2 x TH555A tetrode push pull 100 kW + amplifier at 2.8 MHz.
On the network analyzer it worked well. Under high power, it wasn't
needed, so it sits in the spare parts crib now.
Sometimes a new amplifier circuit seems like it will probably talk
back and sing if you only look at scalar measurements such as grid
dipper, network analyzer lightly coupled, but it doesn't happen under
power. So you stop short of sticking a pile of crap/trap components
in there that will detract from reliability in the long term.
My most recent experience with high power oscillations was in taming
a pair of Marconi BW1643J2 (Amperex 8918) 400 kW triodes which were
configured in cathode follower mode, driving a resonant load of a few
kohnms Z. Based on the original papers on Cathode followers from
around WWII when they were proven to be excellent for video
amplifiers (with small tubes), the mode of oscillation was determined
to be when the cathode load capacitance configured the circuit to
look like a colpitts oscillator. By adding series R to each grid, it
was silenced. It took a year to determine empirically the correct R,
first it was 25 ohms with a pair of water cooled Bird loads, which
burned up regularly. Then I went to a high power 100 ohm R, which
didn't silence the amp but did hold up. Ultimately we have designed a
special load with a resistor from Ceswid/Carborundum, having internal
water flow for cooling. What was needed was to add positive R in
parallel with the negative R presented by the Colpitts model, enough
to make the net impedance positive at the grid. No one else in the
world has cathode followers working at this power level due to fear
of them oscillating. We love them because the output impedance is 11
Ohms up to the low HF frequencies.
Now, I am working to add HOM suppressors (higher order mode) for a 3
MW amplifier under construction, having a Thales TH628 Diacrode
(double ended tetrode) as the active device. Because of the geometry,
there are tendancies for such tubes to support TE21 and 31 circular
WG modes in the space between screen and plate. I have modelled them
using Superfish, a cylindrical EM solver, and they have been observed
at the tube factory in France in their test cavity. The tube is about
the size of a full keg of beer, so you can imagine what i mean as for
large geometry. Operating in class AB1, it will be necessary to add
suppressor components before it ever runs high power. These are most
likely to be of the "waveguide beyond cutoff" style, loaded with
water or air cooled absorber material at the ends. They must be cut
into the walls of the big cavity and be situated in the appropiate
locations for best suppression while not dissipating much fundamental
(200 MHz) power or they would be toast.
It is a costly endeavor, as each iteration requires metal fab work,
welding, brazing, machining.
Wish me luck on this one. I hope to report success at some juncture
in the coming year. Meanwhile i keep reading here hoping that the
miracle cure is found and sold by a ham. So far, nada.
Tnx Phil. I better shut up now.
73
John
K5PRO
>From: "Phil Clements" <philk5pc@tyler.net>
>Any "doubting Thomases" can do an easy experiment. Just breadboard up
>a circuit with a glass tube sans suppressors, and observe the results. I don't
>think anyone could possibly deny the existence of VHF parasitics. We have
>been living with them since at least the 1920's. The argument is how to
>identify them and prevent further oscillations. As ceramic tubes became
>popular, with their shorter leads and compact designs, it was found some
>could be tamed sans suppressor, or with only an inductor (3CX3000A7, etc.)
>or with any old suppressor laying in the junque box. In other words, with the
>advances in tube technology, suppressor design had room for error.
>
>There now seem to be four camps out there.
>1. Those who believe ALL tubes and layouts will have VHF parasitics.
>2. Those who believe that a suppressor of a certain makeup and design
> is better than anyone else's design.
>3. Those who start out from scratch on a new amp design, and tackle each
> problem as it comes up, installing only the preventative devices necessary
> (if any) then quitting when the beast is tamed.
>4. Those who will bet with you either way that the Sun will rise in the East
>tomorrow
> morning.
>
>In the early days of amp engineering I suspect a lot more trial-and-error went
>on than slide rule sliding. Those less experienced in this art need to give
>the math and computer programs a rest every once in a while, and get out
>the drill and soldering iron.
|