> ? I am not trying to use Thevenin's. I am using the max power transfer
> theorem, which essentially states that when half of the total P is being
> dissipated in the generator, transfer of P to the load is max.
Nowhere in the maximum power transfer theorem does it consider
source efficiency.
The maximum power transfer theorem simply gives two
conclusions:
1.) When the magnitude and angle of an impedance connected to
the terminals of a network is adjustable, the largest power results
when Z load is the conjugate of the output impedance of the
network when the load is removed.
2.) When only the magnitude is adjustable, maximum power is
transferred when the absolute magnitude of the load impedance is
equal to the network or source impedance magnitude.
As a matter of fact, the Maximum Power Transfer Theorem only
refers back to Thevenin and Norton models, both of which rule out
using the model in any application for describing source efficiency.
In all cases the source is a black box, and the matching process
does not define anything in that black box including energy
conversion efficiency.
> >If you do the interchange you will find your 60 percent amplifier
> >model made by wrongly applying Thevenin's theorem to a PA
> >suddenly becomes less than 50% efficient with the very same
> >parameters of operation.
> >
> ? I don't see how doing the Thevenin/Norton interchange is going to solve
> the discrepancy. If the T/N interchange is valid, why not do it twice and
> we are back to square one?.
The rules say the models are fully interchangeable. If they do not
interchange, you are misapplying the model. It is not your "right" to
pick the particular model that gives the results you want to use at
any given moment, unless you are invent a 300 mpg carburetor for
a 66 Caddy or a CFA antenna.
> ? True . It has everything to with dissipative resistances -- i.e., the
> dissipative ESR in the generator and the dissipative resistance of the
> load. When power in the load is the maximum possible, an equal amount of
> power will be dissipated in the ESR in the generator. Thus, the
> efficiency is 50% at max. power transfer to the load.
Georgia Power's generators just went up in smoke.
Seriously, let me use a generator as an example. Assume we have
a generator with a fixed shaft speed and unlimited torque compared
to the power we are consuming. If we fix the field winding
magnitude by opening the feedback loop to the regulator and do a
load pull on the generator, we will find maximum power is being
delivered at that fixed field amplitude for an impedance almost
exactly equal to the load impedance before the regulation was
removed.
If we increase ZL, voltage rises but current drops faster so net
power decreases. If we decrease ZL, current increases but voltage
drops faster. A load pull shown optimum load Z occurs at a fixed
impedance, which happens (in a well designed system) to be equal
to the load impedance set by the regulator at any given load
condition. The alternator or generator efficiency at that point is far
more than 50%, yet maximum or optimum power is being
transferred.
Same in an RF system.
> > and only a person who never read the theorems would
> >think it did or that it could be used to describe efficiency in a PA.
> >
> ? Yours truly used to teach Thevenin's Theorem and Norton's Theorm to
> technicians who worked at the Pacific Misslie Range. I think I see what
> your trying to get at, and some of Maxwell's stuff makes sense, however,
> this apparently is not good enough for the boys in Newington.
That's because they think the source Z must always be dissipative.
An easy mistake, if you look at the pictures and don't read the text.
> ? ok, if by self-capacitance you mean the grid's total capacitance to
> other tube elements.
>
> > That self-capacitance is in
> >the order of a few pico-farads.
>
> ? For an 8877, grid/cathode C is 40pf and grid/anode C is 11pF. Is 51pF
> a few pF in your opinion?
Did we suddenly change tubes? If so that's fine.
The self resonant frequency of the 8877 grid is up over 600 MHz
with a proper grounding scheme and no series C.
You are welcome to calculate the change in frequency when 600
pF or so is added from the grid to ground, especially when that 600
pF is in series with two inch long number 18 leads.
> >If you add an EQUAL amount of
> >capacitance in series with the grid you shift the resonant frequency LESS
> >than 50% higher. That would require about 10 pF or so in an 811A.
> >
> ? close enough
>
> >Adding a few hundred pF barely changes the self-neutralizing
> >frequency of the tube, when that parameter is properly measured.
> >
> ? I agree that the resonance shift is relatively small, however, perhaps
> it was far enough to decrease feedthough a bit at the anode-resonance.
No, they plainly claim it adds negative feedback.
> >Second, adding series reactance INCREASES feedthrough on
> >every other frequency.
>
> ? Maximal VHF feedthrough/feedback occurs primarily at grid-resonance.
> How could moving the grid resonance a bit higher make much difference at
> HF?
We could look at the 30L1, but I don't have the values handy
without digging. Let's look at the SB220 mistake that Kenwood
also copied.
The SB-220 uses three 220 pF caps in parallel. The reactance of
those caps is 65 ohms on 80 meters. Those caps are driven by two
voltage sources, one from the control grid to ground and one from
the anode to ground.
The control grid to cathode capacitance is 8.3 pF in parallel with
the grid to cathode resistance sourced from perhaps 70 volts of RF.
The anode to grid capacitance driving that divider is 4.7 pF sourced
from about 2500 volts of RF.
Please tell me if this circuit works as claimed, and adds negative
feedback from the cathode to grid by the capacitive divider formed
by the G-K capacitance.
By my calculations, the bulk of the feedback voltage appearing on
the grid is from anode to grid. It also varies greatly from band to
band. Phase of that feedback will vary with anode tuning and with
G-K resistance, since a tube that draws grid current has a fairly
low G-K resistance ONLY when the grid is driven positive with
respect to the cathode.
Do you think feedback that varies over the entire RF cycle because
it is shunted by a diode is a good idea? I don't.
Now that circuit was a great idea in the 30S1, where the screen of
the tube prevented the high anode RF voltage from driving the grid.
The tetrode was also class AB1, where the grid never drew current
in normal operation. In that case the grid-cathode impedance was
almost entirely the G-K capacitance, and never varied with drive
level or output tuning unless the grid was driven accidentally into
current. At that point, the grid to cathode impedance would sharply
drop and the grid would follow the cathode reducing risk of damage
to the tube. A very good idea if you want to protect a tube from
poor "tuners" or heavy footed operators.
But when that same circuit is applied to a triode, especially one
operated in sub-class 2 with grid current, all hell breaks loose. The
main feedback source is now the anode, and the phase and level
changes with drive, operating frequency, and tuning.
Regardless of what goes on at 60 MHz and higher, adding
feedback from the anode to cathode by raising the grid impedance
is bad engineering.... unless you are trying to build a TPTG
oscillator or a distortion box.
> > Collins (and Bill Orr when he "borrowed" the
> >idea from Collins) claim this is to add "negative feedback".
> >
> >That's pure nonsense, as anyone who calculates the grid voltage
> >produced by the capacitive divider formed by the cathode-grid
> >impedance and grid-to-chassis impedance would be able to see.
> >
> ? Nonsense indeed, Tom. Everybody makes mistakes.
Very true. I believe in this case the mistake was seeing a good
idea, not understanding it, and copying that good idea into a circuit
where it just doesn't work the same.
It doesn't mean they didn't know what they were doing, it just
means they didn't understand what they were copying.
For a tiny improvement over one frequency range (that could be
achieved some other way), it is not wise to make everything else
worse.
That circuit certainly belongs in AB1 cathode driven tetrodes, which
is where it was originally patented. It has no business in multiband
triodes or AB2 triodes.
73, Tom W8JI
w8ji@contesting.com
--
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
|