Screen Inductor minimum value for clean self modulation?

[K1JJ]

What are the guidelines for calculating the minimum inductance for a screen circuit using plate modulation? [For a decent low end]

I'm wondering if my 8H @ 300ma choke is barely enuff inductance for 80 cycles and a 140ma @ 500V,  4-1000A screen circuit.

I seem to remember a rough rule of thumb like X Henries for every X thousand ohms of screen impedance?? And screen impedance  =  screen voltage/ screen current ?


Thanks.

T

[Bacon, WA3WDR]

I don't know.  I think I heard that 5 henries wasn't enough in a Globe King and 10 henries worked better, but that's about all I know.

There is an impedance associated with the audio that the screen will generate across the inductor.  This is why lows modulate at all, and it is also why the top end doesn't overmodualte.  That impedance will determine the lowest inductance that should be used.  It is probably not equal to the screen voltage divided by the screen current.  It would have to be measured, or derived from a complicated delta-V/delta-I calculation using the grid-plate-screen charts for the tube.

If you modulate with 1KHz or so, and then measure the AC voltage across the screen choke as you put various resistances across it, you can figure out what the source impedance of the screen is.  (My wild guess is about 1K.)  Then figure what inductance is needed to keep the choke reactance comfortably above the screen impedance at the bottom end.  Or try different screen choke values and see how the final modulates at the low end, and check the frequency response at the screen compared to the plate.  Actually that's a better way to do it.

[K1JJ]

Thanks for the info, Bacon.

Well, what prompted me to axe is that if you calculate the impedance of an 8 henry choke at 80 hz, I believe it came out to be something like 1800 ohms or so?  It's easy to do on the computer web calculators.

Anyway, if a screen is 1800 ohms and so is the choke, at that freq you wud see a split in voltage division. Not a linear modulating situation I wud think. I'd imagine it wud be like 10K when above 500 cycles, etc. Seems kinda non linear compared to a solid resistor in there. But the power waste is really bad wid the resistor.

So, guess I'll just do the measurements when the rig is running and see what 8H does in the screen. My gut feeling is that 20H is needed for decent low end response.

73,
T

[Jim, W5JO]

Tom the screen is seeing votage variations of the DC type.  Impedance here does not matter as much as in the mod transformer.  From an old Handbook using a single 813 with the separate screen supply and voltage at 400 volts, it calls for 7 hy at 50 ma.

The GK 500 series uses 10 hy (I think) at the appropriate current level and screen voltage at 400 volts from a separate supply.  So anything between 8 and 15 hy should do anything you need.

[Bacon, WA3WDR]

I think Jim is right... it's not critical matching situation, it's more like the value of a Heising choke.  You just don't want the reactance of the choke to start loading the screen down.

The reactance of an inductor is 2 PI F L.  So 8 Hy at 80 Hz is 2 * 3.1416 * 80 * 8 = 4021 ohms.  If the screen impedance is about 1K, this isn't tooooo bad of a load.  Of course, the more the merrier.

The screen self-modulates because it grabs more loose electrons when the plate isn't snatching them away like crazy.  So the screen current tends to be the inverse of the plate voltage.  If you put a resistance or a reactance in series with the screen, then the screen voltage goes up when the screen current is lower, which is when the plate voltage is high, and that's what you need!  So the screen tends to self-modulate.

The screen current is also affected by the screen voltage.  So if you put a big reactance in series with the screen, the screen voltage will not go crazy, because as it rises, the screen draws more current, and this acts like a resistance.  If this resistance is much lower than the reactance, it will dominate the screen voltage swing.  But the reactance of an inductor is lower at lower frequencies, so when you get down below some frequency, the screen modulation starts to get inadequate, and this shows up as poor positive modulation.  Also the phase starts to shift, so the waveform peaks start tilting.

I don't think the screen voltage-current characteristic is linear.  I suspect the current rises fast when screen voltage is a little higher.  This would make the dynamic screen impedance lower than the screen E/I product would suggest.

[K1JJ]

Bacon, Jim -

Yes, it's just like a plate eising choke, I would agree.  I figure that the impedance of the final will have a bearing upon what the Heising choke minimum value needs to be -  the same for the screen.

ie, A very high impedance final would need more inductance in the Heising choke than a low impedance final, right?  so, I figure the same with the screen choke.

If the ARRL Handbook suggests 7 H for a screen choke for 813's, I'm betting it's for their classic 300hz low end audio.   I talked with Tron/HLR about it on 75M last night. He said from his experience,  20H is about right for the 4-1000A and 8H was too small.

So, looks like I'll have to find some more room to series one in.  I will try a single 8H first and make some measurements, but won't be disappointed if I need more.

73,
T

[Jim, W5JO]

T

The only place the impedance is critical is the final plate, not the screen.  The screen is varying and  only aiding the plate in it's job as Bacon described.  Remember this is high level plate modulation and not Heising. 

The choke in the screen is supposed to be high inductance (what ever that means).  What happens is the increased voltage on the plate will increase the current on the plate and decrease the screen current.  When this happens the  magnetic field of the chole will collapse inducing a more positive voltage on the screen.  Therefore the choke should be chosen for about 7-10 hy on a 2KV plate circuit.  That is a fairly high value of inductance.

 You only care that the voltage on the screen be sufficient for the tube in question.  So design the screen supply to give the voltage & current necessary for operation with the resistance of the choke or resistor in series with the screen

Interestingly the new plate modulated AM transmitters modulate the power supply not the final.  Audio input is to the power supply.  I sure would like to see the theory and a diagram on those new ones.  More power, more power supplies.  Interesting concept.

[K1JJ]

The only place the impedance is critical is the final plate, not the screen. 
The choke in the screen is supposed to be high inductance (what ever that means). 

Jim,

I think there must be a MINIMUM inductance for a screen choke in order to have the screen modulating for a particular set of low freqs. No, the inductance value is not critical like a resonant circuit, but if you put in,say, only two Henrys, the LOW END "modulation contribution" of the screen would be lacking.

So it goes to say that there IS a minimum value you need for a particular low end of frequencies.  The choke must develop a high impedance at the modulating frequency in order to work. Otherwise at say, 20 cycles with a  choke too small for the job, you have literally a piece of wire and almost NO screen modulation contribution at all. 

Personally, I feel 7H is too small for the 30-60 cycle range. Remember that even though our voice may only hit 80 cycles, we need the rig to be clean down below that by some amount.    I mean, if the choke is only 500 ohms at 30 cycles and 10K at 1,000 cycles, that sounds like too much of a variation to me. The screen would be modulating more heavily at the higher than lower freqs. In contrast, a fixed resistor coming off the modulated plate to the screen would always be the same impedance at ANY audio frequency and produce a more linear contribution throughout the full audio spectrum to the overall plate modulation scheme.  Too bad it sucks big power.

ie, As far as I know, you want as stable an impedance at ALL audio frequencies as possible. This can only be accomplished with a choke that has a minimum acceptable impedance at the lowest audio frequency to be used. 

T

[KL7OF]

Very interesting thread....I have a question......How much of the total modulation is contributed by the screen? 

[Jim, W5JO]

The purpose of the choke is to provide the collapsing field no matter what the frequency of the modulating tones so it does not care what the audio section passes or the modulation transformer is doing. 

The only trick here is to provide enough inductance to place the positive on the screen.  This choke will add about 10% to the total modulation because of the fact as the plate voltage drops the tube will cut off before it can reach 100%.  Adding the choke will allow all things to work properly and the tube to conduct enough to reach 100%.  I do not remember a formula for calculating the value of the inductance but most examples I have seen show the choke, at minimum, 7 hy.  That will provide enough magnetic flux to perform the magic. 

Now you might say, what if I use a 20 hy choke.  The tube can't absorbe any futher modulation so it does the 100% or 120% the power supply, modulation transformer and audio section gives and that is it.  You can play with it but I would put in a 8-10 hy and go with that.

[WA3VJB]

I always thought "more = better" when it came to screen inductance.
Sounds like there's a point of diminishing returns, with no risk of having too much.

I've got 30 hy in my homebrew rig that Kevin, WB4AIO helped me complete over Christmas break from college in 1976-77. Pair of 810s modulating a pair of Amperex 7527A. 2250V Ep, and Escr 325V at 45ma from a separate screen supply (seen above the RF tank).  I notice some lag in indicated screen current from voice peaks, as expected, and have noted increased low end response with lower distortion.



You might also want to run by Gary/INR what he was using in the screen of the 4-1000 rig.
That's always been a standard reference.

[K1JJ]

Wow, Big G always takes such great photos, Paul. Love that 4X1 - 833A's. Why did he ever get rid of it, I dunno.   In fact I heard Tom/KBW on yesterday using it and sounded great.

Jim, I understand what you're saying about the tube not caring about the inductance, as long as it has the collasping field to handle the lowest particular frequency. That's a good thing, cuz it means that the varying impedance will not affect modulation as the frequency goes higher - great.  I was worried about that. 

Well, figgering the min inductance will be simple... I'll just try 8H at first, which is easier to fit in physically, and jumper in another 8H and see what effect it has.  My guess is it will extend the low end down for a flatter response.  I will do it first in the 813 rig and then in the 4X1 rig to see what different current/screen loads make too. Will report back here. BTW, I do remember about a 10% increase in ability to hit 100% when adding a choke in the past. My first 6146 rig in 1969 did not have a screen resistor tied to mod B+ and I couldn't figure out why it would not modulate heavily until someone told me...

Though, what stands out to me is the Tron/HLR said it takes 20H for a broadcash type low end - for a 4X1 screen..  I've axed him questions for over 35 years and it's hard to believe, but I've never found him to be off the mark. Not once.     

I'll run some tests in a month or so. Got all of the rigs now laid out with finished brackets and starting to mount the smaller parts. Then the wiring... Then the testing and arcing.

T

[Jim, W5JO]

T, I will be interested to see your results.  Remember the higher the current in the plate, the more inductace in the screen lead, that is why HLR said that.  If I were running a single 4-1000 at 2500 Ep and 500 Esg, I would use about 12 -15 hy.  20 would not hurt and could help you if you want 50 cy low and 6 Kcy high out. 

My only cavet is that below about 300 cy you use power that will not help you on noisey low signal reports.  In those circumstances you would want your response to be 300 to about 4-4.5 K to maximize you communications.  Using 20 hy will not comprimise your signal either way with either choice if you have an equalizer.  You simply adjust that for the conditions.

Broadcast years ago used  small freq. response to maximize communications over long distances to accound for noise and fades at the receiver.  As the rules changed to allow local stations on those Class A & B frequencies, local recievers were less subject to phase reversal and fading and to get the 125% modulation they increased frequency response on both ends and increased the size of the choke to maintain the modulation at the very high level.

[w3jn]

noisey low signal reports.

That's the funniest thing I've read all day 

I don't think Tom has had one of those since his gotham vertical days 

[K1JJ]

Here's some info I got from Eric, N2BRC:

From page 290 of the 41st Edition – 1964 “The Radio Amateur’s Handbook”
 
“…The choke L1 is the audio impedance in the screen circuit: its inductance should be large enough to have a reactance (at lowest desired frequency) that is not less than the impedance of the screen. The screen impedance can be taken to be approximately equal to the d.c. screen voltage divided by the d.c. screen current in amperes.” 
 
If this is correct, then it's kinda what we suspected. 

Data:

A 4X1's screen impedance is  500V/140ma = 3571 ohms.

813 tube screen =  400V/40ma = 10K.

It will not be easy to obtain that much inductance to generate the required inductive reactance down at the real low frequencies so that L1 = Screen impedance.

A 20H choke at 30hz is about 3700 ohms. At 60 hz it is 7200 ohms and so forth. So, looks like 20H is needed for the 4X1 and MUCH more for the 813 rig.

Does this make sense?

Here's the inductance reactance caluclator I used on the web. Use 20,000 millihenries for 20H.

http://ourworld.compuserve.com/homepages/Bill_Bowden/XLC.htm

T

[Jim, W5JO]

Well T my 21st edition of the same handbook does not provide that formula but does talk about modulating the screen with a choke or resistor and cap duo.  However any product can be reduced to math.  I am sure that if you follow that formula it will perform properly.

My problem with the formula as stated is the impedance calculation for the 813.  Should the formula apply universally, then one could not modulate the screen of an 813 under those conditions.  I am sure that it can be done but how to provide an overacrhing formula is by me.  Bacon did a good job with that and his results, as I recall, followed yours.

I believe it could be calculated according to inductance formulas for flux density, but am just shooting in the dark.  I hate math and find it boring to sit and try to calculate all those unknowns for hours.  I can do it, but hate to do so.

My 1956 ARRL handbook has a 500 watt transmitter with a single 813 in the final and the 7 hy choke I mentioned earlier, so I would assume it will work.  So after you run the tests, please post so we can see what the results are.  I will bet the 4-1000 will modulate fine with about 15-18 hy and the 813 would with 7-10 hy.  Let us know.

[Bacon, WA3WDR]

A tetrode or pentode tube plate-modulates very poorly, unless the screen grid is also modulated along with the plate.  It modulates somewhat, but the positive peaks just aren't there, and the audio is badly distorted.  That is because varying only the plate voltage of a tetrode or a pentode does not affect the plate current much.  We know this effect as high plate resistance.  The plate behaves much more like a variable current load than a resistor.  But if we increase the screen voltage, this increases the plate current.  So, if we modulate the screen somewhat, along with the plate, we can cause the tube plate to emulate a resistive modulation characteristic, and modulate pretty linearly.  So in fact, a linear modulation characteristic is really synthesized by applying some modulating audio to the screen grid, which modulates the plate current so that it varies along with the plate voltage like a resistor would do, and this also produces a much more linear plate modulation characteristic.  It is a happy coincidence that the screen will generate this modulating voltage for itself if we put a high impedance between the screen and the screen supply.

This is interesting - we modulate the plate voltage of the tetrode or pentode, but mostly it is the screen modulation that varies the plate current.  This is why we do not need 100% modulation on the screen for 100% modulation on the plate - because we don't care how much the screen is modulated, we only care about the effect it has on the instantaneous plate current.

One day, I suddenly wondered how a screen choke modulated a screen grid in the first place, and why it did not severely undermodulate the screen at lower frequencies, and severely overmodulate it at higher audio frequencies.  If it had been only a matter of varying current, that's exactly what it would have done.  But it doesn't work that way, so the screen must be exerting significant control on its instantaneous voltage, in response to the voltage on the plate.  The screen must be behaving as a resistance in series with something that resembles a shunt-regulator.  And the knee of this crude shunt-regulator characteristic varies along with the plate voltage of the tube.

Calculation of the dynamic impedance of the screen could be done with detailed tube curves, but probably the best way to do it is with actual measurements.

Probably the first thing to do would be to measure the operating voltages and currents, and the amount of self-modulation versus the amount of plate modulation at some moderately high audio frequency like 1 or 2 KHz.  Then, applying modulation, I would try loading the inductor with various resistor values, and observe the effect on the modulating audio on the screen.  The reactance of the inductor would be very high at this frequency, and the changes in AC voltage caused by the resistive loading would reveal the dynamic impedance of the screen grid.  A resistor that cuts the AC voltage in half would be equal to the dynamic screen impedance.  But I would measure the decrease in AC voltage with a larger resistor and make the calculation, because the effect may not be linear.

I can do this with my Viking II.  It is modified, and it uses a series resistor from unmodulated B+ for screen power, and I can examine the screen impedance by connecting different value resistors in series between the 6146 screens and a large capacitor to ground.  I think I'll do that, some weekend.

I think that the screen grid acts something like a variable-voltage zener diode (with the plate voltage causing the variation), and because of this I think that the dynamic impedance (dE/dI) will be somewhat lower than the effective resistance (E/I).  That works in our favor, because it means that the screen inductor value does not have to be as high to work well.  But it has to be what it has to be.

With a screen choke, I would also measure audio phase of the screen modulation, and modulation distortion, at various frequencies with different amounts of inductance in series with the screen.  I would also measure the actual inductance of the screen choke with DC on the inductor.  I guess a good way to do that would be to put known capacitor values like 0.1uF across the screen inductor, and observe the frequency of the resonant peak with power applied.  Then I can calculate the actual inductance.

[Jim, W5JO]

I think you are pretty much on the mark Bacon.  From what I understand the ability of the tetrode or pentode to perform modulation is restricted by several factors. 
As the plate of the tube nears full conduction or cutoff the varying modulation has already begun to change and because of the time constant of the components, the tube does not reach full cut off/conduction without the screen components. 

These effects are mitigeted by the addition of varying the screen voltage/current which are tied to the plate variations.  Even though the screen has a different power supply, the voltage/current is varied by the movement of the signal allowing additional current by the collapse of the magnetic flux field or discharge of the capacitor/resistor combination.  This brief time provides the tube the small additional time to reach its peak value on the plate.

Now as to the value of the choke, I am guessing it was determined by what was/is available.  If you calculate the amount of flux density needed to provide full cut off or max current on the plate it might be 6.78 hy for an 813.  Choke available might be 5 or 7 hy and the 5 hy won't provide enough current from the collapsing field to the screen for the tube to reach the maximum current value.  On the negative cycle the screen rises to Vps which aids in cutting off the plate over a very brief time.  These are observations you made.

Now as to the size of the choke or the value of the resistor/cap combinations, on a given audio section where the lows are restricted to 400 cy.  This will require X value of magnetic flux to aid the tube in reaching max conductance.  But on the negative part of the cycle the voltage on the screen does not play any part as it is recharging, only on the positive side of the cycle are these parts in use resulting in higher positive modulation.   Since the transit time of a low frequency is longer than than that of a 2 Kc signal it requires more inductance to provide the current over that part of the cycle. 

To me that is not a function of impedance, rather a function of current capacity of the choke/resistor cap combo.

[W3SLK]

I'm enjoying this thread about self-modulation on the screens via a choke. What I would like to know is where do you find one that is designed for the screen instead of the plate? A while back, I was entertaining the notion of modulating a pair of 4CX250Bs, because they were cheap and very plentiful. An article in the early '60's discusses this very notion and that you can't modulate the screen with traditional construction techniques, (i.e. tapping into the modulated B+) The article suggested the only way around it was to use a mod transformer with a tertiary winding. Other than the ART-13, they are like hen's teeth to find. Somehow the idea of a big old smoothing choke doesn't fit the bill either. Any suggestions?

[Bacon, WA3WDR]

I think that the parts will get harder to find as the years go by.  An ideal screen choke would be reasonably sized - an audio-grade thin-lamination core, with a narrow gap and a lot of Henries of thin wire on it, because screen current is not very high.  I am not aware of many other uses for such a choke other than as a screen choke for an AM transmitter.  Of course, a full-sized Heising choke could be used, but that seems like overkill.

An alternative would be a smaller audio-grade choke in series with a larger power-grade choke.  The power-grade choke would be OK up to a few KHz, and then the smaller choke would take over and choke the highs.  In that arrangement, I'm not sure if it would be best to put a small capacitor across the power choke or not.

Another solution would be a simulated inductor feeding the screen - a regulated voltage source, but with very slow regulation, behaving like a choke.  DC would be fixed at some voltage, but AC down to fairly low frequencies could appear on the screens without much opposition from the slow regulation.  Or the regulator could be designed to apply the expected modulation itself.  Maybe there could be a screen-modulator tube to ground, and a pull-up resistor, with feedback down to DC, and some audio from the plate circuit.  Of course, this would have to withstand a good tube arc. This wouldn't be as efficient as a real choke, but it wouldn't be too bad, because screen power is not that high.

[Ian VK3KRI]

Heres a chunk chopped out of Eimacs 'Care and feeding of power grid tubes'  http://www.cpii.com/eimac/cfcontnt.htm

" In anode and screen modulation it is necessary to introduce not only amplitude modulation of the anode voltage,but also to develop 70 to 100 per cent amplitude modulation of the screen voltage for 100 per cent carrier modulation.Modulation of the screen voltage can be developed in one of the folowing three ways:
 
 (a)By supplying the screen voltage through a dropping resistor connected to the unmodulated d-c anode supply.
 
 (b)When a low voltage fixed screen supply is used,a modulation choke is placed in series with the supply.In the case of voice modulation,this is about a 5 to 10 henry choke.
 
 (c)A third winding on the modulation transformer designed to develop the required screen modulation voltage.

 It is interesting to note that in all three cases the screen of the tetrode or pentode tube supplies the necessary audio power. During the portion of the modulation cycle,when the anode voltage is increased,the screen current decreases.If the screen is supplied through an impedance,such as the screen dropping resistor of a modulation choke,the voltage drop in this series impedance becomes less and the screen voltage rises in the desired manner. On the part of the modulation cycle when the anode voltage is decreased,the screen current increases causing a greater voltage drop in the screen series impedance and thus lowering the voltage on the screen of the tube.The screen by-pass capacitor value in the Class stage is a compromise between good r-f bypassing and the shunting effect of this capacitance on the screen modulation circuit."

[KL7OF]

In Ian's post......
Eimac sez "by supplying the screen voltage thru a dropping resistor connected to the UNMODULATED dc anode supply"   ....Does this mean use a dropping resistor to obtain the correct voltage and then series to a choke to modulate the screen? .....If so, how does this compare to using a dropping resistor in the MODULATED anode B+ with no choke......? 

[Bacon, WA3WDR]

The screen grid typically needs about 50 to 60% modulation for linear AM with 100% plate modulation.  A number of AM transmitters connect the screen dropping resistor to modulated B+.  When this is done, the screen is overmodulated, and we see a glitch at about 85% negative modulation and a pinchoff below that.  This was a common mistake, and you can see it in the ARRL AM trapezoid pattern in many of the Handbooks.

If the dropping resistor comes from unmodulated B+, the glitch goes away, and modulation is more linear.  You could put a choke in series with the resistor, but it isn't necessary, because the resistor is a high enough impedance to let the screen modulate as it wishes to.

Screen self-modulation works very well.  However, it is just -slightly- low, and a little assistance improves linearity a bit more.  WA1KNX Dean found that a combination of two dropping resistors was best, one going to unmodulated B+, and one going to modulated B+.  This gives a slight boost to the modulation on the screen, and produces the most linear modulation of all.
http://www.amwindow.org/tech/htm/scrnmod.htm

Some tubes (usually the big ones) have too much secondary emission to use a screen dropping resistor like that, so a screen choke or a third mod transformer winding especially for the screen is required.  When this is done, there is little DC drop through the winding, so the screen supply needs to be lower in voltage.  One advantage of this arrangement is that if the screen supply is regulated, and you apply some fixed negative bias to the control grid, then you won't need a screen clamp tube to protect the amplifier tube if RF drive fails.  You can use a combination of fixed and self-bias on the control grid - you only need enough fixed bias to cut the tube off.

[KL7OF]

Dean's 2 screen resistor scheme makes a lot of sense..... providing more voltage on the neg swing and less on positive... I plan to do some experimenting on my 813s x 813s...I now use a screen dropping resistor from the modulated B+ and it provides more total modulation than when I had it hooked up with a 15 hy power choke in series with a separate screen supply.........In one of Tom's earlier posts in this thread, he stated:   Data:  a 4x1's screen is 500v/140 ma =3571 ohm    813 screen is 400v/40ma = 10K ohm....Is that correct?  I think the 813 should be 1000 ohm....

[K1JJ]

Interesting info.

Yes, I remember pondering Dean's dual method. That DOES make a lot of sense. Why? Because who's to say that by simply adding a choke to either B+ or moded B+ alone, that it's going to be the PERFECT mix of additonal voltage swing to the screen. For max linearity you wud think there is a sweet spot and tweaking could find the optimum level.

But, if you figure the power eaten up by a resistor to the HV supply for a 4X1 it's a gagger. 3000V- 500 = 2500V across the resistor.   140ma X 2500V = 350 W!  And if ya run a little higher voltage like a 4X1 really wants, lets say 4500V, that's 560W wasted whenever the rig is keyed.   

I'll bet this adjustable level thang cud be done with chokes too.  But the problem seems that there is not ENUFF swing voltage available when using self choke modulation, so a resistive shunt or some other method wud not work.  Bacon, can you think of some choke method that is tweakable other than that solid state idea?  For example, maybe adding an additional TINY bit of power from the modulated B+ via a 100K resistor string is all that needed to tweak things, etc.

Dean's method of using a trapezoid seems to have been a very sensitive method for his linearity tweaking purposes.  His DeanX-100 always sounded super clean on the air. Very transparant audio with a pair of 813's! in a DX-100... 

Steve:  Yes, the 813 screen is correct at 10K.  Take  400V / 0.04 = 10,000.   You may have used  0.4 as the current which is 400 ma....

73,

T