Screen and plate modulation

[AB2EZ]

Hi!

As some of you know, I have been experimenting with external plate & screen modulation of a couple of classic, low power, late 1950's rigs: an Ameco AC-1 (5 watts at carrier, single 6V6 power oscillator) and a Heathkit DX-20 (25-30 watts at carrier, 6CL6 oscillator and 6DQ6A output stage).

The results have far exceeded my expectations in terms of linearity and flatness of modulation frequency response. In fact, there is no FM-ing of the AC-1 under modulation.

In the process, I have thought more deeply about the details of plate modulation of a tetrode or a pentode.

I have come to the conclusion (new to me, probably obvious to many readers of this bulletin board)  that when "plate" modulating a tetrode or a pentode... you are primarily screen-modulating it, with the modulation of the plate voltage contributing primarily to improving the linearity of the plate current v. screen voltage characteristic, and significantly increasing the overall efficiency of the modulated stage.

I have also observed / concluded that while modulation of the plate voltage does help to improve the linearity of the plate current v. screen voltage characteristic... improving the fidelity of the screen modulating voltage waveform (linearity and frequency response of the actual voltage on the screen(s) of the modulated rf output tube(s) v. the modulating audio signal) is more important than the fidelity of the modulated plate voltage waveform.

Therefore, it appears that using a signal to modulate the screen voltage that is derived from the modulated plate voltage (for example a dropping resistor between the plate and the screen) may not be the best approach... particularly when there are limitations in the modulator and/or the modulation transformer that limit the fidelity of the modulated plate voltage waveform ... particularly a low frequencies, where the modulation transformer is experiencing saturation effects.

It appears that even if the modulator and/or modulation transformer are producing relatively large amounts of distortion, and even if the modulation transformer is exhibiting saturation effects at low modulating frequencies... one can obtain high fidelity modulation by using a separate modulating voltage (not derived from the output of the plate voltage modulator) to modulate the screen. For example, this separate screen modulating voltage could be derived from a separate push-pull modulator, in parallel with the main push-pull modulator (i.e., driven by the same driver stage), with a modest-sized modulation transformer for that separate push-pull modulator to drive the screen grid(s) of the modulated tube(s). Since this separate screen voltage modulator would supply a much smaller current than the plate voltage modulator, the screen modulation transformer could be much smaller than the plate modulation transformer, while still producing much less saturation at low modulation frequencies. In effect, at low modulation frequencies, the screen would be modulated with good fidelity even though the plate might not be modulated with good fidelity. As a result, one would still obtain good fidelity of the modulated rf output, with some loss of efficiency.

Stu
AB2EZ

[AF9J]

Hmmmmm,

It sounds like my Globe Scout does this.  It is described as Modfied Heising Modulation in the case of my Scout 680.  In effect like regular Heising modulation, it works on the plate (please correct me if I'm wrong about this), but they also mention in the circuit description doing some modulation of the screens.

73,
Ellen - AF9J

[AB2EZ]

Ellen

Thanks for the comment.

Actually, the Globe Scout does not do what I am suggesting.

Looking at the schematic of the Globe Scout, I see that the screen of the 6146 is fed through a direct connection to the (unmodulated) HV supply, via a 20k ohm resistor. No modulation is applied directly to the screen voltage via this resistor or any other path, other than through parasitic coupling effects and interactions between the plate voltage and the screen voltage that occur within the tube itself as electrons pass through the screen on their way to the plate.

I also noticed that there are mods posted on the Web for the Globe Scout which change this by directly coupling the screen of the 6146 to the modulated plate supply of the 6146 via a resistor or a resistor in parallel with a capacitor.

I'm looking at the schematic of the Globe Scout Deluxe... which, given the mods referred to above,  is similar to a Globe Scout.

The Globe Scout Deluxe design (and the Globe Scout design as well) employs true Heising modulation, as you pointed out. It uses a 5 Henry choke (which many people would refer to as a Heising choke in this application) to feed DC to the plate of the 6146 rf output tube and also to the plate of the 7027 A modulator tube. This choke represents a high enough impedance to modulating frequencies above around 300 Hz (~j9425 ohms at 300 Hz) so that the load on the modulator tube is approximately equal to the plate output impedance of the 6146 (i.e., its plate voltage / plate current).

This design avoids the need for a modulation transformer... although it also gives up the opportunity to use a modulation transformer to match the modulator tube's output impedance to the 6146's plate output impedance.

The modulated screen voltage for the 6146 is derived from the modulated plate voltage via the 50k ohm resistor R12. This is a standard approach (used in many plate modulated transmitters, although the value of the resistor will depend upon the tube being modulated, and upon some of the design tradeoffs being made by whoever designed the rig). If I had a Globe Scout Deluxe, I would remove the 50k ohm resistor R12, and replace it with a voltage divider (i.e., I would replace R12 with a 25k ohm 20 watt resistor, and I would place a 10 watt resistor from the screen pin of the 6146 to ground, having a value of around 30k ohms ... but that is a digression from the subject at hand.)

The focus of my post (this post) is on whether or not one could achieve improved modulation fidelity by using a separate modulator (not the 7027 A shown as V-6 in the Globe Scout Deluxe schematic) to drive the screen of the 6146. This separate modulator would be driven by the output of the speech amp, in parallel with the 7027 A plate modulator, and would be optimized for the specific purpose of modulating the screen of the 6146. For example, it would have a relatively low output impedance (e.g., maybe around 5000 ohms or less) compared to the 50k ohm resistor feeding the screen of the 6146 in the Scout. As a result, it would be able to do a better job of driving the screen's varying input impedance... to force the voltage on the screen to accurately track the modulating audio waveform.

Please note: For a tetrode or a pentode (e.g., a 6146), varying (i.e., modulating) the plate voltage has very little effect on the plate current, unless the plate voltage drops below the screen voltage. Thus, to "plate modulate" a tetrode or a pentode, you must modulate the screen voltage (one way or another)... and therefore a plate modulated 6146 is really a screen modulated 6146, with the modulated plate voltage helping to improve the linearity of the screen modulation process, and to reduce the average plate dissipation (heat) that the tube must handle.


Best regards, and thanks again for the comment.

Stu
 

[Bacon, WA3WDR]

Interesting point.  I never thought about applying higher quality modulation to the screen than the plate, but it makes sense.

The plate current of a tetrode or pentode is substantially independent of plate voltage.  So if we only modulate the plate voltage, we will only get about 2X positive peak power at 100% positive, not 4X as we should.  To get much variation in plate current with modulation, it is necessary to modulate the screen.

We want the plate current to behave like a resistance with plate voltage.  Hence we apply the correct amount of screen modulation, and it needs to have good frequency response.

Plate power is plate volts times plate current.  If the plate modulation is a bit distorted, there will be less output distortion with high quality audio on the screen because although the modulated plate voltage is not quite what it should have been, the screen audio is good, so the plate current variations will be pretty much as they should have been; therefore the modulated output will be more like it should have been.  Had we used the distorted plate audio for screen modulation, the plate current variations would have been distorted as well, and output distortion would have been worse.

Very interesting idea.

[AB2EZ]

Bacon

Thanks for the kind words.

As a further thought... perhaps I am mistaken, but if you modulate (at audio frequencies) the plate voltage of a tetrode or pentode  Class C rf amplifier amplifier... with a fixed screen voltage, with a fixed amount of  rf applied between the grid and the cathode, and with a fixed r.f. load (i.e. the tank circuit is not changing)... then you will not modulate the output at all!

More accurately...you will modulate the output very little until the modulated plate supply voltage is reduced to too low a value to maintain the existing rf voltage swing. This is because the current on each RF cycle will be the same even though the plate supply voltage is being modulated, and the rf output is proportional to the square of the current flowing through the (fixed) rf load on the tube.

Thus, without modulating the screen voltage (one way or another), modulating the plate voltage will have no effect on the output of the transmitter (again, provided the modulated plate supply voltage does not drop below the level needed to support the rf signal on the plate of the tube at carrier level).

This is why I believe that what we call "plate modulation" is really an improved form of screen modulation. By modulating the plate supply voltage, we increase the efficiency of the r.f. output tube(s) by reducing the (unneeded) plate voltage to the just the level required to maintain the r.f. signal level at the output of the tube. This would be analogous to modulating the plate voltage of a linear amplifier to keep its efficiency at 66% at all output levels.

Best regards
Stu

[AF9J]

Thanks for correcting my misassumption & clarifying the explanation of the modulation scheme of my Scout 680 Stu.    You explained it far better than the manual did.

73,
Ellen - AF9J

[W1GFH]

As some of you know, I have been experimenting with external plate & screen modulation of a couple of classic, low power, late 1950's rigs: an Ameco AC-1 (5 watts at carrier, single 6V6 power oscillator)...

Hey Stu, I think you may have mentioned this on the AC-1 Yahoo group, but there was not much of a response. I'd love to see photos and a .wav file of your AC-1 modulation experiments, if possible!

[Bacon, WA3WDR]

As a further thought... perhaps I am mistaken, but if you modulate (at audio frequencies) the plate voltage of a tetrode or pentode  Class C rf amplifier amplifier... with a fixed screen voltage, with a fixed amount of  rf applied between the grid and the cathode, and with a fixed r.f. load (i.e. the tank circuit is not changing)... then you will not modulate the output at all!

More accurately...you will modulate the output very little until the modulated plate supply voltage is reduced to too low a value to maintain the existing rf voltage swing. This is because the current on each RF cycle will be the same even though the plate supply voltage is being modulated, and the rf output is proportional to the square of the current flowing through the (fixed) rf load on the tube.

True - if the delta-I does not change with plate voltage, then the RF output should not change either... I think.  What I saw when I held my 6146 screens steady with a large capacitor, like 20 uF (if the cap was good), was -some- positive, and mostly negative modulation, when it should have been heavily upward from the voice polarization.  Now, the plate current can not hold steady at very low voltages, so you'll get negative modulation at some point.  But how I got upward at all, I don't know.  Of course, the plate resistance was not infinite, but it seemed like maybe 40% positive, where it was usually about 180% or more.  Maybe it was the control grid self-modulating according to plate voltage, but the screen shields a lot of that.  I'll have to do some experiments.

There is a load impedance issue, but I am starting to agree that the current is more important than the voltage for the output modulation linearity.

Very, very interesting subject.

[AB2EZ]

Bacon

I suspect that the Class C current pulses (that occur once per RF cycle when the grid goes positive enough) have a bit more area... and therefore contain a larger amount of current at the fundamental RF frequency... when the plate voltage is modulated upward... but that is just a guess at this point.

Joe: I've recorded a .wav file of the modulated Ameco AC-1's RF output, as received and demodulated by my off-air monitor... which I can send you: as is (25 MB) or as an MP3 file. Please respond directly to me at  sdp2@verizon.net to let me know which you prefer. I will also be posting some information (next week) on my web site regarding the details of how I am plate/screen modulating the Ameco AC-1 and the Heathkit DX-20.

Ellen: You're welcome!

Best regards
Stu
 

[Bacon, WA3WDR]

Yes, that makes sense.  You need enough plate voltage to support the output signal at any particular instantaneous level.  Too much instantaneous plate voltage, and you lose efficiency.  To little, and you run out of supply volts and distort.

It's not perfect, because pentodes and tetrodes only approximate current sources.  But it's close.

[kf6pqt]

I will also be posting some information (next week) on my web site regarding the details of how I am plate/screen modulating the Ameco AC-1 and the Heathkit DX-20.

I'm eager to read about it, post the link here when its ready!

-Jason kf6pqt

[W1GFH]

Stu's diagram.



I heard his mp3 too. The audio sounds squeaky clean and broadcast quality.

The 60W audio amp seems slightly overkill. But imagine checking into a 75M AM fone roundtable: "Rig here is an Ameco AC-1 driving four 811's in grounded grid..."

[AB2EZ]

Joe

Thanks for posting the JPEG, and thanks for the comments on the AC-1's audio.

To avoid confusion:

A. The modulation method I am using with the AC-1 is the standard approach. In this case, I am using a 15k ohm resistor from the modulated B+ to the screen on the 6V6 (or 6L6). The screen RF bypass capacitor is .001 uF. Thus, the approach that I am using with the AC-1 does not employ the concept of a separate screen modulator.

B. If anyone wants a cleaner version of the diagram, I will be happy to provide it. Please contact me directly: sdp2@verizon.net. I can send it as .ppt or .pptx, or JPEG, or GIF... so please specify which you would like. If you want the MP3 file of the audio produced by the AC-1, as received by my off air monitor (1.1 MB), let me know.

C. To make it easier to read the diagram: the modulation transformer is a Hammond 1629SEA single-ended audio output transformer, which I purchased new (at a discount price) from Radio Daze. I have used it with both my AC-1 (approximately 7.5 watts of audio power needed) and also with my DX-20 (approximately 30 watts of audio power needed). The audio amplifier is driving the 8 ohm winding of the transformer. The 6200 ohm winding of the transformer (rated for up to 100 ma of unbalanced DC) is placed in series with the B+ line of the AC-1. The audio amplifier that I am using is one side of a  Samson "Servo 120a", which has an output impedance of less than 4 ohms... so the AC-1 "sees" a modulating source that has an impedance of less than (6200/8) x 4 ohms = 3100 ohms. The transformer is delivering approximately 300 volts peak to the AC-1 (i.e., equal to the B+). The turns ratio is the square root of (6200/8) ~ 28. So... the amplifier is delivering 300/28 ~ 11 volts to the 8 ohm input winding of the transformer. The impedance seen by the amplifier looking into the input of the transformer is around 6000 x (8/6200) ~ 8 ohms... where the plate impedance of the modulated tube is ~300 volts/.05 amps = 6000 ohms. Thus the audio power being delivered by the amplifier (at 100% modulation) to the input of the transformer when it is delivering an audio sine wave signal is 0.5 x 11 volts x 11 volts / 8 ohms = 7.5 watts. [Note: to protect both the audio amplifier and the transformer... under circumstances where I disconnect the AC-1 from the output of the transformer, but I leave the amplifier on, with audio input still applied... I have an additional 10 ohm resistor connected across the output of the amplifier. Thus the amplifier is also delivering additional power to this extra load]

Stu