[AMPS] Grid capacitors
w8jitom@postoffice.worldnet.att.net
w8jitom@postoffice.worldnet.att.net
Thu, 10 Jul 1997 22:50:03 +0000
> From: km1h@juno.com (km1h @ juno.com)
> Subject: Re: [AMPS] Grid capacitors
> Date: Thu, 10 Jul 97 21:55:58 +0000
Carl,
I know you get upset with technical explanations, but let
me try to explain this. I'll try to keep it simple.
That circuit was a mistake. Bill Orr made great efforts to get
manufacturers to use the circuit, some agreed readily while others
agreed reluctantly. Some manufacturers simply copied other
manufacturers and installed the circuit without knowing how it
worked.
The circuit is easy to analyze, even if you aren't comfortable with
engineering. First, consider the grid system diagram I drew. With
only 8 or ten pf at one end of ALL the inductance, how much change
in frequency do you expect by adding 700 pF or so at the other end
vs a dead short at VHF?
You can try that on paper, or in a VHF amplifier if you like. The 8
to ten pF end dominates the VHF resonant frequency, the ~700 pF
change at the other end of the inductance has almost NO effect at
VHF. As a matter of fact it actually INCREASES the series impedance
to ground at the frequency where 3-500Z's generally become unstable.
This can be PLAINLY seen with a network analyzer, or if you own a
MFJ-259 and a HF/VHF detector you can do your own crude transmission
loss measurements. If you care to know how I can explain it, and you
can observe the cathode to plate feedthrough loss for yourself.
Since the grid shields the anode from the cathode, that feedthrough
loss is a direct indicator of the tubes internal grid impedance. As
that loss increases, the grid is better grounded (or operating at a
lower impedance). I'm sure you can follow that thought.
In all cases I've ever measured, adding these caps INCREASES the
feedthrough on all frequencies except one. The one frequency it helps
isn't even where the tube is unstable... and the improvement is very
minor, less than a dB or two as a general rule. It takes a value of
about 20 pF to significantly move the loss at 180 MHz, where the
3-500Z normally oscillates in a good layout with very short and
direct grid grounds.
But the ~700 pF of total stray and lumped C does lift the grid
from ground at lower frequencies. It lifts it most on 160, where the
reactance is about 125 ohms. In that case the grid voltage division
is the ratio of cathode to grid IMPEDANCE compared to the impedance
of 125 ohms.
Under no drive conditions, the reactance to ground is in series with
the grid to cathode capacitance, about 7 pF or so. You have a 12.5 k
ohm reactance in series with 125 ohms, so the grid is fairly
well grounded and phase shift is OK.
The problem occurs when drive is applied. The cathode to grid
impedance drops greatly, and varies over the drive cycle. During the
grid current peak the impedance might get as low as 20 ohms or so.
Now the divider is perhaps 20 ohms in series 125 ohms, and the grid
floats up away from ground. The grid is nearly tied to the cathode on
160 during drive peaks, and the long time constant of a 700 pF cap
holds the grid voltage up unsymmetrically as cathode to grid
impedance increases as the signal crosses the peak.
The action isn't greatly dissimilar to the G2DAF circuit's uneven
RF loading.
The problem centers around the fact grid impedance changes
greatly with grid current, and so the feedback that is SUPPOSED to
be negative actually changes phase as well as level over the entire
RF cycle. You can actually see the IMD increase when the grid is
floated in this circuit with a three tone test. Under certain tuning
conditions IMD gets better, under others it gets worse. The problem
is it gets WORSE more often than better, and unlike the Collins PA
did, Heath, Drake and Kenwood (who Chinese copied the SB-220) NEVER
provided a tuning meter that measures proper resonance.
On ten meters, the grid is grounded through an entirely different
impedance value, the capacitors look like less than ten ohms. On ten
meters, the phase and amplitude of the feedback also changes with
drive level, but the change is less and different than on 160. The
PA works differently on all bands because the feedback changes.
It doesn't take a rocket scientist to figure this out, just a little
careful thought.
Ask yourself this, why would any engineer add a circuit that
destroys gain flatness while increasing regeneration? I can tell you
the answer, they took someone's word for it without "checking the
circuit out". Heath planned on removing the circuit after we did IMD
tests, but opted for a different PA design instead to conform with
the 1500 watt limit. That PA, the Warrior II, was designed
prototyped but never marketed because Heath found kit sales
were nearly dead. The normal buy was 1000 pcs, and it would have
taken years to sell them.
The proper way to add negative feedback is to add an unbypassed
resistance in series with the drive path. That resistance lowers IMD
levels, as well as improves gain flatness and stability.
Cast whatever personal attacks you like, it's poor engineering to add
a circuit that decreases stability when a simple resistor would work
better and enhance stability. No matter who used them, the
mica grid caps are a bad idea in this application. It not only works
out that way on paper, it MEASURES that way in life on real test
equipment.
A second bad effect is if the tube arcs, the grid can float up. The
grid can reach full anode voltage, and the voltage then dumps into
the cathode circuit where it can hit the exciter through the input.
There have been cases where manufacturers using this circuit have had
to buy customers new radios when the PA tubes arc.
Not a pretty thought when profit margin on the PA sale is only about
15% of retail price.
> Now you come along to defend an alternative design which IMO is far
> inferior.
You have your opinions. I'm not defending anything. Anyone here can
do what they like. They just need to be careful, some of the advice
is pretty poor. I'll leave it up to everyone else to figure out who is
making up stuff by guessing, and who knows the facts.
>The Ameritron AL-82 with directly grounded grids is a known
> firecraker....that has been discussed at length here in months past; your
> design I assume.
Yes, it is my design.
> The Amp Supply LK-500 with directly grounded grids is another fine
> example of instability....did you have a hand in that too?
No, that was Mr. D.J. Had.
> How about the Viewstar and also the B&W PT-2500A....more
> fireworks with direct grid grounding.
That was Paul Herevnik, of Vectronics fame. And DJ Had "designed" the
Ten Tec 3-500Z's.
All that aside, you operate under the belief the extra problems are
due to instability. They are not.
The extra headaches come from operating the PA's at higher voltage.
Problems increase by the cube of the voltage increase. Not only does
the higher anode voltage tax the tubes more, it stresses the
bandswitch and other components.
> Is the problem because of the grids or elsewhere in the design?
Elsewhere. It is the voltage. Tube manufacturers will confirm this,
if you know George Badger ask him. Tubes with higher voltages are
almost exponentially more headaches, especially glass envelope tubes.
All the stuff you and Rich blame on "parasitics" is mostly arcs due
to outgassing and seal leakage. The peak anode voltage under perfect
operating conditions is almost two times the dc voltage, while when
mistuned it can be several times the dc voltage. This also happens
with exciter transients, or momentary glitches in the antenna.
Two to four times 2500 volts is 5-10 kV, two to four times 3600 volts
is 7200-14,400 volts. Many brand new 3-500Z's fail at 13 kV peak
voltage. Mistune or glitch a 3-500 with 3500 volts or so on the anode
and you can expect a loud pop and a blown meter protection diode.
But then if you want 1500 watts from a pair of 3-500Z's, that's
the voltage you must have. The voltage has to be at least 3400
volts full load to deliver 1500 watts with 100 watts of drive. There
ain't no free lunch. You pay for it up front or down the road. High
power is expensive, either initially or down the road in parts. The
trick is getting the balance as good as possible.
Lower the voltage of any of the 1500 watt amps you mentioned to Drake
L4B or SB-220 levels and tank heating and tube arcs will be greatly
reduced. All of a sudden the "firecrackers" as you call them make
the other old "battleships" look like junque. An AL-82 would be
virtually indestructible at Heathkit voltages, and the PT2500 would
be also..
Comparing apples to oranges is silly. The Drake and SB-220 are 600
watt CW output designs, three of the other PAs you mentioned were 1500
watt CW.
> I dont know what you mean about using a GDO to measure tube
> internals....never heard of that one. The only GDO mentioned here
> recently, that I am aware of, was in reference to a Plate Choke.
Rich was speaking about using a GDO on the grid lead.
> Other than the stupid and boring battle between you and Rich, where
> neither one gives an inch, I find little of value passing back and forth
> after the first week or so. After that its all reruns.
I suggest you use the delete key. It wouldn't hurt to edit out some of
you personal comments either while you are at it. I'm sure many
people find you as stupid and boring as you do others.
> Directly grounding the grids does not help a SB-220 at all.
> Maybe you are just upset that Heath did not take your "advice" and decided to just
> update the original circuit a bit in the SB-221 and HL-2200.
No, they always took my advice just fine the first time I gave it. I
reported directly to the amateur radio PLM (product line manager),
who reported to the president. Some things I learned about AFTER
making decisions. None of us are born knowing everything, we all
learn from experience. I didn't discover the IMD and stability flaws
until the Warrior II design was nearly completed.
> My measurements with 6M only SB-220's is that the parasitic is around
> 110-120MHz not 180MHz. It took some careful attention to position the
> suppressor values so that harmonics of 50MHz were not smoking the
> resistor and yet stopped the parasitic cold.
I have no idea what a Radio Kit modified SB-220 acts like, but with
the grids DIRECTLY grounded the oscillation is about 180 MHz.
> Just so you dont think I am just picking on you Tom, I have another msg
> ready for Rich.
I know you treat everyone the same, don't worry about me thinking you
unfair with anyone. I cherish your friendly warm fuzzy always trying
to help demeanor. ;-)
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
Search: http://www.contesting.com/km9p/search.htm