The AM Forum

Educate me, please...

I've heard it said that "you can not plate modulate a Pi-Network final"

Why not?

HUH?

Most everything that feeds coax uses a PI net output.

Someone is pulling your leg, or just misinformed.

73DG

Well, I make a call and actually the answer is "no one said it -couldn't- be done - it's just that it's not commonly done with Triodes.  It's hard to neutralize them."  To which I replied "just because it's difficult, doesn't mean it can't be done, right?"

My idea is to plate modulate a single 450TL that's built up as a Class B grid-driven Linear. (home-brew).  I've been steered away from plate modulating that thing for a while.  I never knew why.

So... what would it take to neutralize a single 450TL in a Pi-Network circuit so that it could be plate-modulated?

Yup! 833s are very....... uncommon finals! ::)

Triodes abound in AM transmitters and linear amplifiers.  Broadcast gear was full of them.  Lots of ham rigs with 3-500's 211's, 811A's, 450th, eTc.

Finding neutralizing difficult is only be bad design & dumb layouts.  Even circuits with the touted pentodes that didn't need neutralization were squirrely without a bit of it.

Go forth and neutralize, read the radio books of our forefathers and see how it was done.

Pi-nets were virtually unknown before the war as everyone used open-wire feeders and the resulting high impedance was easier to match with link coupling or other balanced networking.

When coax became readily available everyone went to the easy solution for impedances under 100 Ohms...the PI net.

73DG

Hi Geoff,

Neutralizing triodes for use with pi networks is easy. The only "trick" is to use a balanced grid tank circuit. Shown below is a schematic of an RF deck that I built using parallel triodes with a pi network output. The principal is exactly the same with a single triode. Use a center tapped/center link grid coil and a split stator grid tuning capacitor with the rotor grounded. Bring the neutralizing cap down to the "bottom" side of the coil/cap junction. Bring the grid bias up to the center tap of the grid coil. There is an old post by K9ACT that you could look up in the archives that shows the schematic of an 8000 triode with pi network output and grid neutralization as well. Look up the thread "Mystery grid drive".

It's interesting that you mention wanting to use a 450TL with a pi network output. One of the homebrew rigs I got at the W0ZUS auction is exactly that - 450TL with pi network output and grid neutralization. I haven't had it on the air yet, but it should be back on the air later this year I hope.

I should clarify one thing. The schematic below does not show a center tapped/center link grid coil. This schematic is of my first attempt at building this circuit. The end link coil did actually work, but a center tap and center link is better. Also, the values of capacitance shown are for a single band 160 meter RF deck. Obviously values are different for other bands.

One more thing, I think you want to operate the 450TL in class C for plate modulation with Class B modulators.

Ron W8ACR

For just a single band Rons idea works well however I refuse to use plug in coils as I rarely stay put on any band.

For 40M and down you likely can get by without neutralization with a clean layout and in efficient Class C. Keep the input from even seeing the tube and just feed a grid strap thru the shield. With toroids for the input coils you have that battle mostly licked anyway.

Swamp the input with a real N.I. resistor, feed with a pi net just like they do a GG amp, put 10-20 Ohms real N.I. right at the grid pin and hit it with 100-150W. Hang a parasitic suppressor off the anode cap.

Use copper strap for all RF wiring, no little wires allowed, to minimize inductance.

Play with the pi net step up ratio and the swamping resistor until you are using all the available drive. If necessary use a toroid as a step up transformer in conjunction with the pi-net. With just a few low bands it should be OK.

Add input and output bandswitching and enjoy. The input can be a simple phenolic switch ganged to the output and using 12VDC to control a DPDT DIP relay per band. Those relays and mica trimmers plus some fixed silver micas for 160/80 will handle the power.

Everything I build is pi-net with 1 to 4 tubes, often 160-40M and 20-10M and rarely is neutralization needed and even then a copper strap next to the tube can be adjusted for sufficient in-out isolation to be stable on 10M...all done with no HV and fleapower drive into a pad to a SA, or receiver.

Carl

Ron, I believe this was the 450TL HB that you got at Dewey's estate sale. I salivated over it, but I didn't have the time that day to figure out how to move it. It had some monster iron on the second shelf. That thing was made like a tank!

Thanks for the input, Ron.  Actually, I said
"My idea is to plate modulate a single 450TL that's built up as a Class B grid-driven Linear. (home-brew)."

I can see where the statement misleading (yet it seems others misconstrue what's said and assume what their mind wants) ... but this particular 450T RF deck is already made up.  Pictures this afternoon. 

There's a binding post on the back side of the chassis in which to apply bias voltage.  There's, of course, RF input, AC line (for filament xfmr) and RF Output.  Since it's currently grid-driven, changing the bias will change the class.  I just like the idea of being able to crank up the Vari-AC to make 3kV DC out of the power supply and load the thing up for bear, then back off on the DC voltage until it's relatively 'tame' ;-)

y'know... I'm liking the idea of a single 450TL modulated by a pair... 

pretty much the same as a single 833 modulated by a pair... just more filament power (and a little more drive)
 

I assume you are referring to tetrodes and pentodes.  Except for grounded-grid operation, neutralisation is necessary for triodes.  Otherwise, even if you have it swamped down enough to prevent self-oscillation, it's still regenerative, probably enough to destroy modulation linearity, resulting in one of those wide signals the slopbucketeers piss 'n moan all the time about.

I have even heard of neutralising a pair of triodes in the push-pull class B modulator.  Supposedly it improves audio quality, although I have never tried it.

One problem  with grid neutralisation is that the grid-to-ground capacitance of most triodes is much higher than the plate-to-ground capacitance, and tends to unbalance the circuit, particularly when operating multiple bands.  The best solution for this problem is to add a balancing capacitor to the opposite side of the circuit, from the end of the coil to ground, to balance out the grid-to-ground capacitance. It should be a small variable cap, with maximum capacitance about 30% greater than what the tube chart says the grid-to-ground capacitance is.  Neutralise the stage at the lowest frequency band first.  Then switch to the highest frequency band.  Chances are, the neutralisation will be off a bit.  Re-neutralise by adjusting the balancing cap, not the neut. cap.  After 2-3 repetitions of this procedure, you should arrive at the  combination of settings of the neutralising and balancing capacitors that will hold neutralisation throughout the frequency range of the transmitter.

This also works with plate neutralisation in the case of a single-ended amplifier using a tube with high plate-to-ground capacitance, or when running two or more tubes in parallel.

In any case, do not expect to achieve quite as perfect a result with grid neutralisation as you can achieve with the same triode using plate neutralisation.

John - Yes, that's the one. Parallel 450TL's modulated by 833's. I ended up going back to Edgemont three more times after the auction to get everything that was left. The 450TL rig took all day to disassemble, and yes, it had a lot of very heavy parts. I took a lot of photos and notes. Hopefully it will go back together without too much grief. Looking through Dewey's notes, he apparently ran the 450TL rig at 3200V and 800mA!! I'm going to throttle it back just a bit. I will probably actually run it with just one 450TL. A single 450TL can easily put out up to 850 watts carrier in plate modulated service. That way, I will have a spare final and a clear conscience. I will probably also lower the plate voltage to a more manageable level - say 2000-2500VDC. This will allow me to use a plate transformer that I can actually move by myself. Lastly, at a lower power level, I will probably use different modulator tubes - 805's or maybe 250TL's. Hope to have it back on line this summer.

Geoff - I think I understood your original statement to say that the 450TL is already set up as a class B amplifier, and that you wanted to plate modulate it. That was my point. I'm not professionally trained in electronics, but I thought that class C operation was mandatory for an  RF amplifier being used in plate modulated AM service. A class B RF amplifier cannot be plate modulated in the usual way with a modulation transformer. I don't remember why, but I think any old ARRL handbook provides an explanation. Class B RF amplifiers must be grid modulated, a tricky proposition, although if done properly they can have excellent audio, and it does eliminate the plate transformer. As you said, it's a simple matter to change the grid bias and hence the class of operation.

Ron

I agree.  You cannot satisfactorily plate modulate a class-B amplifier and expect to get linear modulation.  It must be class-C.  Usually a matter of increasing the grid drive and bias. The tube has to be driven to full saturation and then some with the rf drive.  A good check is to back off on the grid drive a little, and see whether or not the plate  current and rf output drop off.  If so, it needs more rf drive.  You should be able to drop back the grid drive by at least 20% or so before the rf output and plate current begin to substantially decrease.  By definition, a class-B amplifier is NOT driven to saturation (except when it's used for SSB during a QuaRMtest ;D).

Furthermore, the class-C amplifier should be grid driven, not cathode-driven in the grounded grid circuit.  The G-G configuration wastes driving power by passing a large portion of it on through the amplifier stage to the antenna.  This also prevents 100% modulation in the negative direction, since this feed-through is always present, no matter how negative the modulation voltage drives the plate.

In linear service, one advantage to the G-G circuit is that the amplifier doesn't have to be neutralised if everything is well shielded and isolated.  Neutralisation of a triode is always necessary for grid-driven class-C service.

So there!

73DG

Probably could mean it probably wont work or it probably will work. Spoken like a true politician ::)

Im betting it probably will work on 160 and 80 with attention to details or at least worth a try.

Good thing you aren't using a coaster brake.

IIRC the T-368 with tetrode final has no neutralization in the final stage.

'  By definition, a class-B amplifier is NOT driven to saturation (except when it's used for SSB during a QuaRMtest ;D). "

You forgot class "D" operations on the 11 meter band. These lin yar pillform amplifiers are quite capable of being swung nicely. The good ones will even swing Bird watts and will not pull down volts to the single digits.
10 4


kLC

Have you examined your trapezoid pattern with a scope?  Do you have a straight-sided triangle?

I said "probably" because there are other factors to consider.  Your tube(s) could be a little on the weak side, and the regeneration from lack of neutralisation might bring them closer to appearing linear.  Some tubes I have tried in the final showed substantial curvature, enough that they wouldn't make even 100% positive even though I had plenty of audio.  But that would be very poor engineering to use those crappy tubes, then introduce regeneration to straighten out the trapezoid curve by compensating for the drop-off in emission.

Guess they (whoever that is) never heard of a BC-610.

The BC-610 uses the same circuit I used with my first 304TL rig back in 1963.  Plate neutralised with a balanced link coupled plate tank circuit using a split stator tuning capacitor, and unbalanced parallel tuned circuit feeding the grid.  Both the 250TH and 304TL have very low plate-to-ground capacitance, so the minimum capacitance of the split stator easily swamps out any unbalance caused be the capacitance of the tube. OTOH, a high-capacitance triode like the 833A might not hold over 20-160m so well.

My HF-300 rig uses a 211 to drive the final.  That also is a high capacitance triode.  I used the balancing capacitor off the opposite end of the balanced plate tank circuit of the driver stage as described in my previous post, and neutralisation holds perfectly from 160m through 20m.

The best way to do it with a PI-network is to use the balanced version.  The BC-339 military rig uses that circuit with a pair of 833As. Push-pull cross-neutralised final with split stator plate tuning capacitor, balanced split tank coil with symmetrical taps for changing bands, and split stator loading capacitor. The rig is fully bandswitching and designed to work directly into a balanced feedline with no external tuner.  Unfortunately, it was intended for CW/RTTY use only, so no modulator is included.  A few hams have successfully built up plate modulators for theirs. The balanced PI-network is also described in the early west coast RADIO handbooks.

Of course not, it was brought forward as a discussion item and not to be dismissed out of hand because of the NIBY syndrome. Curiosity may prompt someone to actually build and test.

What is the NIBY syndrome?


That can be done without too much of a problem, using a tetrode or pentode with a careful layout. A triode is a another story altogether.

Not In My Back Yard >:(

73DG

I thought that was NIMBY.

Him too!

I believe this is why Stu AB2EZ modulates the exciter as well as his 2 X GS-35 linear - it resolves the issue of feed through RF from the exciter since it is modulated as well

Al

That would work.  But even in grounded grid configuration, the amplifier still should be run in fully saturated class C mode if it is to be plate modulated, not linear.

Agreed.

Well, the input has been wonderful and I thank each and every one of your for your input.  I think I'll just run the 450T deck in it's constructed configuration for the SSB rig.  In the process of building a universal power supply, complete with meters and a Vari-Ac, and use it to power either the SSB amp, or the 250TH x 450TL rig.
The final is near complete - just a few minor repairs.  The power supply and modulator are being built from scratch.  The big transformer I discussed in another post, a pair of 872A's, 30H @ 500mA choke to meet L_(crit) and somewhere around 25uF @ 5kVDC (total) should filter any hum out.  Critical reactance calculates out to somewhere between 20~40Hy - make have to string a pair of chokes together to get there.

That said, anyone have any 5kV oil-filled caps they wish to part with?  ;D

Don

I think this would be a good topic of discussion / documentation.  There are a lot of linears out there that might lend themselves to this application.  Your point of still running the final in saturation while applying some modulation to the linear is an interesting one.

Most commercial ham linears dont have the Tune cap spacing to handle Class C AM and barely make it on SSB.

On top of that some tubes in use likely wouldnt even hi pot high enough.

Ok... here's some pics of the 450T amp...

two more...

That's a good-looking layout there.

yeah, I think it'll work.  I just need to get some time to do it. 

I wish there was a way to bank time... ~sigh~

You might want to run those nice caps through the dish washer.
You can also take them apart and clean them one part at a time. Just be careful with the ceramic and make sure the soft washers get reinstalled.