Class C input tuning - Broadband toroid vs: L/C circuit ?

[K1JJ]

Hi All,

For a class C amplifier, what are the advantages to using a conventional L/C link input tuning into an amplifier's grid instead of a broadband toroid?  Is it to make it easier to drive and reduce harmonics only?

Let's say a neutralized, unswamped 4-1000A takes only 10-15 watts to drive with a grid L/C circuit. What do you suppose it will require with an appropriate step up toroidal unun?

I would think that the so called linear L/C  "flywheel effect" is not needed with a class C amplifier.

It would be so much easier to eliminate the input tuning for both construction and band hopping later on. Might even be more stable without the TPTG tendencies of the plate tank and grid tank interacting too. I plan to neutralize the amplifier, so will be easy to drive without swamping.

Bottom line is a little more drive required and less harmonic attenuation seems like a small price to pay for a passive input. Most drivers are clean enuff already harmonic-wise.

Am I'm missing something?  Can it be that easy? Do these Spandex pants make my butt look fat?

T

[WA1GFZ]

Tom,
What happens to the input SWR without the flywheel effect? Then one would think you could just put a resistor load across the transformer to swamp the effect.
I can only think the driver might not like a load that changes from conduction to nonconduction. Heck you could introduce neutralization with a third winding on the core. fc

[Steve - WB3HUZ]

If you can get in touch with Jamie, KF2VM, he's supposedly built transmitter with an untuned unun input on the grid.

[Jim, W5JO]

Dunnoi, Broadband = low Q = harmonics in, harmonics out.  You going to have broadband output tuning? 

[K1MVP]

Hi Tom,
 I am not sure if I am "missing" something here,--but my experience with
PW transmitters,(my 6L6 and 6146) homebrew xmtrs, is I have not had
much luck with getting enough "clean"  drive unless I installed a tuned
LC circuit to the grid of the final stage.
I might have been able to get "drive" without a tuned circuit, but the
final usually would act "finicky", even with neutralizing it.
I do not like "broadband" input circuits feeding a PA stage,--but that`s just
my experience.
                                      73, Rene, K1MVP  
    

[K1JJ]

The output will be a conventional pi net.

I'll probably go with L/C input tuning on the grid after hearing the comments and thinking it over more. I mean, I know that guys do broadband inputs, but it's not real popular.  I'm just trying to think of ways to make bandswitching a little less painful.

In the past I've used a roller inductor and variable cap in the grid using a 50 ohm link in. I think the roller was what made it a pain, along with rolling the plate vac caps around.  Maybe this time a grid bandswitch tapping a coil and a variable cap with presets will make it easier.

I did do a broadband toroidal input with a linear once, and it worked, but for some reason I later changed it back to L/C.

Open to ideas.  Hmmm.. maybe a ganged drive off the bandswitch to the input bandswitch will help. I'm so lazy when it comes to retuning for band hopping, the rig always stays pretuned on one band. That's one reason why I'm building up a few new rigs right now.

T

[W2VW]

Steal the output tank out of an old MW marine rig. You can build a real nice bridge neutralized input ckt from the minidux with all the nice tap clips. Don't forget to space the parts so they don't arc to the chassis. The rectified grid Voltage can get pretty serious. Makes a real good SBE.
IIRC Jamie liked to use a transformer to get the drive Voltage up. No neutralization though. You can get away without neutralization like the T-368 but being Tom Vu will not arr-owl-it.

[w3jn]

I vote for tuned input unless your input sig is REALLY clean (harmonics-wise).  The more crap you amplify, only to be filtered out at the tank, is just gonna reduce efficiency and certainly won't help anything. 

You should be able to do 75 and 40 (or perhaps 160 and 75) with a single inductor and variable cap.  No it won't be optimum but it' should be quite a bit better than nuttin.  Or you could find one of those old National input networks that tuned the C and L simultaneously and did 80-10 w/o bandswitching.

73 John

[WA1GFZ]

Measures has a cool grid drive circuit with just a variable inductor if I remember

[K1JJ]

Measures has a cool grid drive circuit with just a variable inductor if I remember

Since 1972, Chuck/K1KW and I have debated who came up with the "WA1DGK/WA1EKV"  variable input matching circuit first.

For driving any grid driven amplifier, linear or class C" :   Start at 50 ohms with a variable ~1200pf to ground, connect a series floating roller inductor. Then a 2nd ~300 pf variable to ground then to the grid. This will match anything 160-10M 1:1 .  It's really a reverse pi-net if you think about it.  I'm sure it was used before 1972, but it's still fun to argue who was first....

That circuit is used in most of my HB linear rigs. Though, for class C neutralized rigs, I prefer the 50 ohm link into a parallel circuit that gets neutralization at the cold end . It's used in the Tesla 360 as pictured in the url below and I intend to use it in the new 4X1, 833A and 813 rigs:


http://www.amwindow.org/tech/htm/813/813.htm

T

[k7yoo]

On my Multi-Eimac rig (4-65 driving 4X1)  I use the 200 ohm swamping resistor from an old T-bolt on the grid of the 4-65. A small SGC autotuner (L network) is used to match the drive source to the 4-65. This allows instant bandswitching. I could do the same directly to the 4X1 with a higher resistor value and more drive.
NO tune , no bandswitch, no waiting.
The SGC cannot be used directly to the grid without a resistor to present a somewhat stable driving impedance. Other wise the tuner is chasing a moving target and it goes nuts.

[k4kyv]

I shouldn't think harmonic suppression at the grid input of a class-C amplifier would be important.  Even if fed with a pure  sinewave, the tube conducts only at a fraction of each rf cycle, and remains cut off throughout most of the cycle, giving the tube nothing more than a "kick in the ass" at each cycle, and letting the flywheel effect at the plate tank produce the sinewave carrier at the output.  The very nature of class-C operation makes the input signal the equivalent of a short pulse every rf cycle, and  therefore the grid current  is inherently rich in harmonics.  With an untuned transformer input, you would just need to develop enough rf voltage at the grid to drive the tube properly in class-C.  With a tuned grid tank at the input, the flywheel effect of the tuned circuit delivers enough umph to the grid to drive the tube into saturation over part of the rf cycle, even with an anemic rf driver, as  long as it delivers enough rf to do the job, and the Q of the grid tank is sufficient.  With the untuned transformer, the internal impedance of the transformer and driver circuit becomes a factor, since the grid impedance at the peak of the rf cycle is much lower than over the rest of the cycle when the grid of the tube is not conducting, and the grid must still be driven positive enough on peaks to drive the plate current of the tube fully into saturation.

[Bacon, WA3WDR]

It's easier with a grid tank circuit acting as a flywheel.  It gets more complicted with broadband drive, but it can be done.

To duplicate a grid tank, you would just produce a sine wave section or the full sine wave if you prefer, but all you really need is the section of the sine wave that drives the amplifier tube into conduction.

It doesn't really have to be a sine wave, either - but that's another story.

Without a grid tank, you want a polarity-inverting device between the driver plate and the amplifier grid.  The grid is easy to drive negative, but harder to drive positive, because it draws current when it goes positive with respect to the cathode.  But the driver tube pulls down better than it pushes up, so a center-tapped broadband driver transformer would be a good solution.  Then, the driver tube would be pulling down while the amplifier grid is driven positive.  Of course the grid would have the conventional RF choke to a grid resistor, and possibly fixed bias, and capacitive coupling to the driver transformer winding.  Because of the intermittent currents, the transformer should have minimum stray capacitance, and a snubber circuit to control ringing.

Usually we don't use a center-tapped RF driver transformer like that.  Usually we have a single-ended tank, and we couple directly from the plate to the next grid without polarity inversion.  In that design, we need the tank resonance to drive the grid positive, unless we run the driver in class-A - which is not efficient.  But we can use a single-ended broadband driver in class B or C, if we invert the polarity between the driver plate and the amplifier grid.

The driver transformer could have two separate windings, one for the plate and the other for the grid, as long as it inverts polarity between the plate and the grid.

A broadband driver will not work as a frequency doubler.  You would need the flywheel effect to give two positive swings to the following grid with only one downward tug from the driver plate.

[nu2b]

Hi Folks-Good discussion.
One additional possibility is achieving a BBmatch reactively rather than
with a toroid xfmr. The regular PiNet isn't a broadband configuration
but a Pi-L can be.

Take the following 4-element lowpass (-to-bandpass) network
which matches 50 to 2500 ohms and maintains less than
1.062 to 1 swr across 3.5 to 4 Mhz.

Ro=50 ohms
Ls1=5.5uh
Cp2=324pf
Ls3=40.5uh
Cp4=44pf
RL=2500 ohms

The Cp4 cap absorbs the tube grid to gnd capacitance.
Neutralization IS required.
RL at 2500 ohms would be used for initial alignment.
Then for operation in the circuit might be changed to a swamping
resistor (example) of about 7.5K in parallel with
the effective shunt grid resistance over the cycle of about 3750ohms.
This would increase required drive power by 50%.

If we do the input, why not the output?
Just turn the network around to be the output network.
Now there is no tuning at all in the box.
Just adjust the antenna tuner for 1:1 and go to town.

We might even arrange the neutralization to be part of the
Cp4A input cap and Cp4B output cap by using a fat 500pf (guess) to
gnd in series with each and variable cap joining the two fat cap nodes
for a bridge balance gizmo. It's fun to think about.


The above network would be close to an output match for a 5K final,e.g
300V at 60ma (6AQ5 Tom JJ's cakepan rig)
500V at 100ma (807)
750V at 150ma (6146)
1500V at 300ma (813)
etc..

Regards,
BobbyT

[K1JJ]

Thanks for the valuable info, guys!

Bacon and Bobbie, your input was quite informative.

I've decided to go with an L/C input on all rigs.  Wil be using a bandswitched inductor and variable cap.    Link input/neutralization for the class C rigs and simple parallel L/C for the linears.

73,

Tom, K1JJ

[wa1knx]

    Interesting topic, I am kind of at the same point in thinking of how to drive class C broadband.  Wonder if
anyone has used a broadband scheme of late to drive a tube class C.   I have some mail off to Jamie kf2vm.
getting a final to C or better, like a switch mode square wave would be more efficient. so odd harmonics
should be encouraged anyway.

dean

[MikeKE0ZUinkcmo]

If the drive source is good enough to put on the air, then whats wrong with just using a 50 Ohm non-reactive resistor(s) as the input element.  That could easily be followed by a toroid auto transformer if needed.

The input resistor stack was used years ago by National in their 80 thru 10 meterNCL 2000 amplifier.  Plus, you have a built-in dummy load for tuning up the exciter.

Manual on BAMA.

[KD6VXI]

With a 50 ohm resistor, a 4cx250B needs approximately 100 watts of drive. 

With a 3bturn transformer and a 450 ohm resistor, it needs about 10 watts of drive.

That would be the reason, imho.

--Shane
KD6VXI

[wa1knx]

re -1
     yes, something along those lines.   I'm collecting parts for HB xmtr, thinking of 813's or 4-400s
not sure yet. but they need some voltage for class C. like 4-5 watts drive a piece but at 300v peak grid
or so. I'm doodling stuff like a low q pi /  4:1 or 9:1 transmission line xfmr.  but as I remember
ferrites don't like high impedances.  I am sure its been done (?)

[wa1knx]

  Interesting. sleuthing the broadband grid drive led me to Central Electronics, who produced a few
broadband output transmitters years ago. Apparently mysterious  'broadband couplers' were used
that were potted affairs.  following that lead led me back here!

http://amfone.net/Amforum/index.php?topic=20459.msg147599#msg147599

I hope cut/paste works anyhow questions were asked years back HERE about how central electronics
did their broadband amplifier output circuits.  I'm not sure how this line of thinking would work
for class C grid drive but you guys covered it!

[PA0NVD]

I have made an amp with two 4-400 in parallel in grounded cathode. I use a wideband input with a 4 ; 1 configuracion, so 1 ; 16 in impedace It is loaded with  1000 Ohms and gives no input SWR problem at all bands The amp has sufficient drive with approx 20 watts and I drive it well into gridcurrent at 30 Watts. More sensitivity I don't want because almost every transmitter can deliver 100 Watt
I connect the ALC from the rectified output of the linear to regulate the transmitter power. So even at 100 Watts no problem, but I normally limit the TEC-TEC driver transmitter to 30 - 40 Watts.
Indeed worth the simplicity and no negative effects found at all. I use the linear more than 15 years.