6146 Plate modulated with VR 150 regulated screen supply

[n4wc]

One handbook suggests only a series choke/reactor in series with the screen supply. 
Just break the plate supply for the modulator and the screen will "move" with the plate.
This is a 6146 (40w input).   Any suggestions will be appreciated.
Bill Cook  N4WC

[The Slab Bacon]

To properly plate modulate a class c tetrode final you have to put a little swing on the screen as well as the plate. The usual way to do this in a smaller transmitter is to tap down the screen voltage for the final from the modulated plate voltage with a dropping resistor.

There have been other scheems for this as well. Many older military (and civillian) rigs had a tertiary winding to put a little monkey swing on the screen. These usually
had a seporate screen supply.

On larger finals one school of thought used to be to send the screen supply voltage through a screen reactor (choke) so the screen could somewhat "self modulate" from the swinging of the plate voltage. (if you are using a seporate screen supply)

On a smaller transmitter, tapping down the screen supply from the modualted plate voltage is the simplest and easiest way to do it. 

"Hard regulating" the screen supply is usually done to the modders for linearity.
I hope this helps you understand it.

                                                        The Slab Bacon

[W2XR]

My suggestion is to go with a screen reactor. Once you know the impedance presented by the screen based upon the screen voltage and screen current, you can easily calculate the screen Z. The screen current should  be determined at the operating point of the modulated amplifier when it is loaded to the desired plate current, as the loading has a very pronounced effect upon the screen current.  Increased loading =reduced screen current.

A good rule of thumb for  the required inductance is about 10 henries of inductance for every 1000 ohms of screen impedance; this will enable the final amplifier to fully modulate at low (<40 hz.) audio frequencies. Obviously, the current rating for the reactor must be based upon the screen current drawn.

Another point to consider is keeping the screen reactor above ground potential; this minimizes the effect of the reactor to shunt high frequencies to ground, based upon the distributed capacitance of the winding/core to ground within the reactor. A couple of ceramic stand-offs, etc.,  will do the trick.

The -3 dB point for low frequency roll-off based upon the screen reactor inductance can be easily calculated using the simple formula for inductive reactance.

Good luck with your project!

73,

Bruce

[W2XR]

Further to my previous post on this subject, I neglected to add a caveat; if you do elect to go the screen reactor approach, you will need a separate power supply for the screen! For a small 40 watt rig, the added complexity of this additional supply may not be justifiable. In my homebrew 2x 4-400As modulated by class B 833As, I have a separate variac'd screen supply that I can ramp all the way up to 800 volts if I desire for class AB1 linear service, but for class C plate modulated operation, the screen voltage is typically held around 500 VDC for 4-400As. The variac also makes tune-up a snap; just reduce the screen voltage when tuning the final for resonance to hold the plate current to a safe value, and gradually increase it as you load up the amplifier.

73,

Bruce

[WU2D]

There is a fourth method where you use a resistor (not a divider) to feed the screen from the unmodulated side of the transformer (B+) instead of the modulated side. Supposedly the screen tends to modulate as a result of the screen impedance increasing with the plate voltage and decreasing with the plates voltage.

Mike WU2D

[Bacon, WA3WDR]

Hi Bill,

A number of us have experimented with the screen grid modulation in a 6146 plate-modulated RF final.  Yes, some screen swing (about 60%) is necessary for good modulation linearity, and it is often not done right.  With tubes like the 6146 up to the 813, series resistor tricks work.

A resistance in series with the screen grid allows the screen voltage to vary along with screen current.  Screen current in turn varies with plate voltage - the higher the plate voltage, the lower the screen current.

So a resistor in series with the screen will allow the screen voltage to be higher when the plate voltage is high, and lower when the plate voltage is low.  Hence the screen will "self-modulate."

This is how a screen grid self-modulates with a series inductor, too - the inductor allows the screen voltage to vary momentarily, such as over an audio waveform.

The screen not only varies in current along with the plate voltage, but this variation has an impedance and a desired instantaneous voltage as well.  This is why the screen inductor does not cause massive undermodulation at low audio frequencies and massive overmodulation at high audio frequencies.  As long as the inductance of the screen choke is high enough, and loading from its laminations and stray capacitance (and screen bypassing) are not too much at high audio frequencies, the screen will modulate itself fairly well.

A self-modulated screen grid works pretty well, but it could use a little more swing than it gives itself.  So the right modulation transformer screen winding, or a resistive divider giving some modulated and some unmodulated voltage to the screen, is better.  The best modulation linearity is achieved with WA1KNX Dino's low-distortion screen modulation circuit.

Class C Optimization for Ultra Low Distortion
From the "Kleenex" - Dean, WA1KNX
AM Press/Exchange, Issue 71, May, 1989
http://www.amfone.net/AMPX/71.htm

(Note - the online schematic in this article has been corrected.  The original print version had an error.)

The same basic article updated at the AM Window:

Screen Modulation Optimization for the DX-100
by Dean, WA1KNX
http://www.amwindow.org/tech/htm/scrnmodd.htm

See also:

THE SELF-MODULATED SCREEN GRID
Bacon, WA3WDR
AM Press/Exchange, Issue 98, January 1992
http://amfone.net/AMPX/98.htm


If you use dropping resistors to produce the screen modulation, you need to have a clamp tube on the screen, in case you lose RF drive to the control grid.  In all designs, you should have some protective bias on the control grid.  Additional self-bias on the control grid also helps modulation linearity.

A fixed screen voltage from a screen choke and a screen supply, with protective control grid bias, acts very much like a good clamp tube system - plate current does not go wild if you lose grid drive.  With the fixed screen voltage that you get with a screen-choke system, plate current variations are not quite as horrible when you tune up, too.

By the way, the usual method of connecting a single dropping resistor from modulated B+ to the screen will overmodulate the screen, causing a significant kink in the modulation characteristic at about 85% negative modulation.  It was this nasty nonlinearity that got our attention in the first place.

I have heard bad things about series screen resistors with more powerful tubes that have significant screen grid secondary emission.  With those tubes, it is best to use a stiff screen supply and a screen choke or a separate screen winding on the mod transformer to control the screen voltage, and for the very best results with a screen choke, add a little 'oopmh' to the screen modulation level somehow - maybe a resistor and capacitor in series up to the modulated B+, etc.

  Bacon, WA3WDR

[WU2D]

Lookie what I found stuffed in the APACHE stuff that I got from Peter N2IDU.


Mike WU2D

[WA1GFZ]

Wow Guys, I observed this distortion in the V2-CDC I'm playing with.
Here is a question for you guys.
I have a dropping resistor off the high voltage supply with 2 10 watt series zeners to ground. This is the regulated supply for the 807 modulator screens. I wonder if I could tap off the bottom zener to get final secreen voltage that would be regulated. I would only need to lower the dropping resistor value providing additional 20 ma for the final screens.
Would a regulated voltage be better or would it take a third winding on the mod transformer to monkey swing it? gfz

[Bacon, WA3WDR]

Hi Frank,

I've only used the dropping resistor approaches for applying the modulation to the screen.  The regulated voltage and inductor or mod transformer winding is probably easiest on the tube, because it is a little easier on the tube during tuneup.  If you already have a clamp tube, then the KNX two-resistor and equalizer approach is great for linearity, and it is as good as it gets, from DC to the highest audio frequencies.  A mod transformer screen winding would be nice, but the ratio is important.  The series inductor is not perfect down to DC, but with a high-inductance choke and a little extra oomph from a resistor and series cap to modulated B+ it should be fine.

My approach of just moving the screen resistor to unmodulated B+ is near perfect, and good down to DC.  Your CDC - I think that's a PDM or some other super high quality thing, so you should optimize the screen - you should use at least the unmodulated B+ screen resistor circuit, or ideally the KNX design, or maybe the extra-oomph screen choke.

I remember KNX Dino talking about that kink at 85% negative way back around 1970.  I knew about it, but I accepted it.  One night many years later, I noticed what happened to screen current as the plate voltage was varied, and I realized that a screen would self-modulate with just a resistor to unmodulated B+, and I tried it and it worked.  I had never really thought about how a screen choke worked.  Out of curiosity, I tried a big electrolytic bypass cap on the screens, and wow did the modulation get distorted.  The screens really need some modulation.

[WA1GFZ]

TNX Bacon,
I was looking at Dean's circuit and I like it. I think I will give it a try.
the CDC is a rig I just got and plan to do minor mods. Then I plan to rip into the pdm rig and add an 813 to the final because I keep taking out finals with the high level of PDM modulation. The tubes get the heck beat out of them. An 813 will give me twice the dissipation. I also want to run a bit higher voltage to get the power back to 100 watts out. gfz

[Bacon, WA3WDR]

Makes me think of putting big tubes in my Viking II.  Why not?  Man, I've hacked that thing up one side and down the other.

[WA1GFZ]

My hacked rig was hit by lightning and had to replace the harness so it is on its second life. I may put a bigger driver in it like a 5763 or 6cl6. The CDC is too clean to hack.

[WU2D]

Frank,

Do you know the only radio that uses a 6CL6, a 12BY7 and a 5763?

Hint: They are not all drivers.

Mike WU2D

[WA1GFZ]

Heath SB 401 6cl6
Rice boxes TS820 etc 12by7
Valient 5763

Mike,
I was hoping you would know the innards of the art13 mod transformer but I can easily determine if the screen winding used the same size wire as the secondary. Just put them in series and run a low DC current through them and measure the voltage drop across each winding. Then jack the current to 250 ma and measure it again. If the ratio of voltage changes then they have different size wire.  gfz

[The Slab Bacon]

Frank,

Do you know the only radio that uses a 6CL6, a 12BY7 and a 5763?

Hint: They are not all drivers.

Mike WU2D

I think its a griefkit Apache. 6CL6 buffa, 5763 driva, and 12BY7 audio driva, if I remember corectly.
                                                 The Slab Bacon.

[wa1knx]

you know Bacon, I did a search on my call and found hundreds of
hits on me because of that article, including 2 main stream sites
in iran! jeez not sure i liked that   the screen circuit was optimized
for 6146's, in the dx-100. I remember well, I heard stuff in my audio
I couldn't put a finger on. I set up a trap pattern, from the plate
low side choke, and a probe and there was the ugly stuff to see! the
article outlines what I found to fix.

If your using different tubes, ie 813, 4-400, to look at a trap pattern and optimize
for each tube because the tube geometry will be different. Anyway,
the fix cleaned up my dx100 audio, esp bumping up the grid drive
to clear up a bump near 95-99% negative modulation. the mystery
distortion gone. its worth the effort to look and not casually feed the
screen.

[WU2D]

Yup the good old Apache using them all, argueably in the best functional positions.

I wonder if it makes a difference?

Johnson would have stuffed in 3 6AQ5's and called it good.

Mike WU2D   

[WA1GFZ]

I scored an interesting new modulation transformer used in the ARC 58.
the ARC 58 used 28 volt 6146s.  3 in the final and two in the modulator.
5500 ohm, 235 ma primary, 2000 ohm, 300 ma secondary and 200 ohm 20 ma.screen winding.
It is a nice sealed unit made by Stancor for Collins. sealed unit good to 50,000 feet.
It looks like it will fit inthe same  space the stock V2 mod transformer takes.
This seems to have ratings more suited for the voltage the V2 runs compared to the art 13 iron.
Not bad for $5, my big marlboro purchase.

[W2XR]

Hi Bacon & Deano,

I was the one who suggested that this fellow use a screen reactor in his HB 6146 rig, based upon my positive experience with this. I am currently using one in my 2x 4-400A rig with good results.

However, the posts made by you guys about taking the screen voltage off of the unmodulated side of the mod xfmr secondary is interesting and intriguing. For me to do this, I would have to take the 2600 VDC plate voltage and drop it down to about 500 VDC for the 4-400A screens, and the screens draw around 70-80 ma in my rig with the final loaded for normal operation. The dropping resistor in this topology is dissipating nearly 170 watts of power, and this is a big ass resistor. I can afford to burn off this power, (although honestly I would prefer not to as it really is wasteful), as the PSU in the rig is quite oversized, but my question is this; is the modulation linearity better with the screen reactor approach, or the approach you have described?

Mack/N4VGB, in a previous post stated that, "Everybody says "you gotta modulate the screens of tetrodes". Makes me wonder how Collins sold so many 20V, 20V-1 and 20V-3 1KW BC transmitters? Screen supply on these is tapped from the center of the load resistor string with an added 2K resistor before the screen, no modulation. In fact, there is an instantaneous drop in voltage at this point during modulation peaks. Does wonders for linearity."

Could someone clarify this topology for me, as it is not clear. What load resistor string???

Interesting question. I have always used the screen reactor in my rigs, but I'm always open to anything that makes a better sounding signal.

Thanks & 73,

Bruce

[wa1knx]

hi Bruce,
       I'd say set up a trapazoid test and look at what you have.
in a sense its a nit pick, but I could hear the distortion that got
cleared up back then. eg if you found for example the screen was
overly self modulated with the choke, then you say could tamp it down
with a resistor across the choke.

deano

[Bacon, WA3WDR]

Frank - that ARC-58 transformer is interesting.

Deano - I just realized that I have been mis-spelling your name!   Yeah, I used to see in my stats report that my website got a few hits from Seychelles.  Who the heck is in Seychelles?  Now I get some hits from Pakistan and the Phillipines.

Bruce - a screen inductor is not bad advice.  There are plenty of rigs out there that use them, and they sound fine.  And the power savings can be substantial in a high power rig, as you noted.  I believe that a properly sized screen inductor will work about as well as a screen dropping resistor from unmodulated B+, although there is a low frequency limit, and there are minor distortion issues because of nonlinearities in the iron core of the screen choke.

A screen choke has a low-frequency cutoff point, and below that frequency there is substantial distortion as a result of phase shift and insufficient screen modulation with the plate modulation.  If that frequency is low enough, then there is no real problem.  If not, then you need a bigger screen choke, or an additional choke in series with the one you have. 

In that Collins 20V series, the resistor string is probably a series string of bleeder resistors on the main power supply.  I suspect that the main filter capacitors are also connected in series, and are connected to the resistors at both ends - if so, the resistor tap point that is used for the screen grid voltage would be filtered, and it would not move significantly at audio frequecies.  Then the 2K series resistor would be where the varying screen current turns into modulating voltage.

Take a look at the trapezoid pattern on a scope, using an actual sample of modulated B+ for the horizontal, and the RF output for the vertical.  If it is linear, then you are good to go.  If it has significant curvature on its sides, then you might want to give it more screen audio by adding a large resistor in series with a good-sized capacitor from modulated B+ to the screen.  You want the capacitor to be equal in reactance to the series resistor at a frequency well below your desired bottom end, maybe 1/4 of the lowest audio frequency.

If you see a sort of oval shape on the sloping sides of the trapezoid pattern, it could mean that the screen choke is too low in value.  It could also mean that you have too low of a capacitor value coupling the sample of modulated audio to the scope.

The dropping resistor designs are good down to DC, but as you note there is power wasted in the resistors.  The screen choke design has a low frequency limit, but it supplies the necessary DC power to the screen much more efficiently.  It probably would benefit from a little extra audio level to the screen by adding a resistor and capacitor in series from the screen to modulated B+.

An active screen modulator could provide response down to DC more efficiently than the passive resistor designs.  Maybe someone will design an active amplifier for applying modulation to the screen grid of a modulated stage, possibly a digital one with a switching circuit similar to the kind that is used to apply modulation to the plate.

A modulation transformer with a screen winding would power the screen efficiently, and it would be good down to the lowest frequency that the transformer can handle.  The optimum screen turns ratio would be different for different tubes, but a slight level 'mismatch' would probably not be disasterous.

   Bacon, WA3WDR

[WA1GFZ]

I given the final screen a bit of thought. As the screen voltage is increased the tube gain goes up. So as Dean found you don't need to double the voltage to make a clean output. So the idea of a little more gain makes sense. Looking at this ARC58 transformer It only has 200 ohms on the screen and 2000 ohms on the plate. Sounds like the there will not be a lot of audio on the screen.
Income tax day has me out of radio motivation. fc

[Bacon, WA3WDR]

Frank - the voltage ratio is the square root of the impedance ratio, so 2000 ohms to 200 ohms would be a turns ratio and a voltage ratio of about 3.16:1.  If it's 6146s, then you usually have about 600V plate and 150V screen, and +/- 600V plate would translate to 1/3.16 of that, or +/- 185V p-p onto the screens that run at about 150V DC, and that will seriously overmodulate the screens.

If the ARC-58 is basically three 6146 modulated by two 6146, then that screen winding impedance is wrong - unless they are running very low plate voltage on the finals, which is possible in equipment intended for aircraft use.  With 300V on the plate, the 6146 would want about 105V screen, and +/- 300V/3.16 = +/- 95V screen which would not overmodulate them clear into cutoff, but that's still too much concurrent screen mod on the 6146.

I guess you should put some AC on the plate taps of that mod transformer and confirm what appears on the screen and output taps.

[WA1GFZ]

Hi Bacon,
Yes, I ran the numbers and got the same voltages. I would think the plate voltage was lower since it was an aircraft radio. I remember my LaPointe days when used to trip over ARC58s. It looks like the transformer will drop right in. I could also do the two resistor method using the third winding. gfz

[Bacon, WA3WDR]

This screen stuff is tricky.