> >That's correct George. Orr (the person who coined the name "semi
> >and super cathode driven") applies description to ANY PA (
> >including triodes ) using a grid tied back to the input directly
> >or through a divider (the divider also includes stay capacitance
> >in the tube) and illustrates and describes the use of triodes in his
> >July 1967 QST article.
> George: So is 'semi' where K is less than unity, and 'super' where K is
> unity?
Rich, I can fax you a copy of the article. That would be easier. But
simply, Orr labels the system where the grid is fed out-of phase
signal with respect to the cathode SEMI. He labels the system where
the cathode and grid have in-phase voltage of any ratio SUPER.
> I conceed that I overlooked the effect of stay capacitance, Tom.
I'm sorry Rich, we all overlook things from time to time. That's why
I like to use peer review in designs.
Look carefully at the effect of the time-varying grid to cathode
impedance. That is something Orr and Sayer overlooked in their
analysis.10 pF of grid cathode capacitance does NOT make a
good linear divider in a PA that draws grid current, it's nearly as
bad as the G2DAF system.
I received calls and letters trying to pressure me into using this
circuit even though it decreased gain flatness and IMD reliability in
a PA. The system fails a complete mathematical analysis in a
triode or any amplifier drawing grid current, as well as an A-B test
in any single PA unless it uses a well shielded tetrode or pentode
that never draws grid current.
> >The important point I was making is the grids
belong grounded in a
> >grounded grid PA. Any time any grid is RF coupled directly or
> >through a divider to the input or output circuit, stability is decreased.
> Apparently, Collins didn't think so. I have never heard of an
> instability problem with the Collins 30S-1 amplifier, and the grid is RF
> coupled directly to the cathode.
No, the grid is coupled to ground through a small mica cap, the
screen is directly grounded. As I said before, the system works in
the 30S1. It only works properly in multi-grid tubes that do not
consume control grid current.
> >A second flaw is any internal tube arc is easily passed on to the
> >cathode and other circuits when the grid floats.
> Three out of the five 3-500Zs with bent filament helices were removed
> from an amplifier (LK-550) wherein the grids are directly-grounded.
I have never looked carefully at how "well" the LK-550 is designed,
since I refuse to look at them. Re-engineering a Yugo costs more than
starting from scratch, a PA is likely the same.
> >... ... A simple
voltage division
> >analysis that includes grid/cathode dynamic impedance will prove
> >this.
> Please show us the analysis, Tom.
Using the capacitive divider values obtained in Orr's article or used
in the 30L1, and plugging in the time-varying grid impedance across
the upper capacitor (about 6 pF per tube in the 811A).
Here's what we get.......
At 3.8 MHz, when the grid is NOT drawing current (over most of the RF
cycle), the RF grid voltage is determined by 6 pF (7k ohms) over
220 pF (190 ohms). The grid voltage is about 2.6% of the drive
voltage.
When the grid reaches full conduction, the grid cathode impedance is
a few hundred ohms. I don't have an exact value, but let's assume
it goes to 200 ohms. The grid is now at nearly 50% of the RF drive
voltage. Since the resistance slope is non-linear with grid current,
the RF feedback is also non-linear with grid current. Maximum gain
reduction occurs during grid current peaks.
Not only that, the gain reducing feedback from the
capacitive divider increases on ten meters, while the gain reducing
feedback from the grid impedance decreases.
Do you think non-linear RF feedback, providing maximum gain
compression when peak levels are reached, is a good idea?
It falls way down in the "not good" class to me, no matter who
suggests it. When a system fails both theory and real tests, I
consider it flawed.
I could tune a SKD amplifier to "look OK" with a steady
two tone signal, especially on bands where the feedback is a little
more correct, but all PA's I've tried it in became more critical to
tune and less stable. The only exception was in a AB1 4CX5000. It
improved that PA.
Collins had stability mods for the triode AB2 30L1, but none I'm
aware of for the tetrode AB1 30S1.
73, Tom W8JI
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|