The AM Forum

Hola!

I need a 50 watt class hi-fi exciter to drive my pair of 4-1000As linear amplifier.  It needs about 15 watts of drive to do 400 w of carrier.  I figure a 50 watt, 200w pep AM exciter would make a very clean 15 w carrier at 200w pep. (> 200% modulation peak capability)

The series modulated design has my attention since it requires no mod iron, thus no phase shift.  I like the design of Bob WB9ECK's rig below. I also included my old 6AQ5 series modulated rig below that.  I still have this rig and works FB, but only about 5 w output.

I was thinking that a single 813 final series modulated by another 813 would do nicely with 2KV on the whole thing. In my 6AQ5 rig, I noticed the series modulator tube ran close to class A and dissipated much more power than the class C final, as expected.  I don't understand how the series tube can be smaller than the final, thus using a big tube for the modulator. (an 813)

Did I say that the 813 is one of my favourite tubes?   ;)


Any comments or suggestions?

T



Bob / WB9ECK's design:
http://www.amwindow.org/tech/htm/wb9eckseriesmod.htm

K1JJ 6AQ5 design:
http://www.amwindow.org/tech/htm/series.htm

From some of the discussion I remember on here, you need a tube with a low mu and a low plate resistance that can handle the current that the final is drawing for series modulation like that to work properly, plus it needs to have enough dissipation. Kinda overkill for the power level you want to run, but an 833 sounds like a good choice. A better idea might be a GM70 or a pair of them, Russian triode that looks a lot like an 813 without a plate cap, and has specs similar to an 845 (higher dissipation for the GM70, 125 watts). I'd go for the graphite plate version for what you want (there's also a copper plate version that is hard to find and the audiophools desire it over the graphite GM70).

I guess the tube, or the filiment transformer has to withstand the high voltage in these designs.
Some tubes are not rated very high in the voltage the filiment to cathode can withstand, and some filiment transformers are not rated for 4-1000 kinds of voltages...

Another method is to screen modulate a tube directly with transistors, so it can 100% modulate the RF deck.


W2IMX sent me the design he is using on a pair of 4-400's, uses 3 transistors and takes line level audio in.
Everything is adjustable, so you can set it to 99% neg mod, and any positive mod you want, with no iron in the mix.

Note: these types of designs call for some method to limit the high frequency audio if you do not want a 40,000 Hz wide signal...

Thanks for the info, guys.

I plan to wind my own filament transformer on an old Variac core. It will be easy to isolate it for 5KV.  I've done this before.

That 4-400A final with transistor screen modulation sounds interesting. Could you post the schematic?  I would think some 11N90 900V MOSFETS would work FB and I have a bunch here.

The problem with screen modulation is the plate loading is critical and needs careful adjustment to be clean.  But so does a series modulated rig.

This exciter could also double as a low power rig running 50-100 watts when needed.  This is a step above a barefoot ricebox but not as much hassle as a big rig.

T

Sent it to you on email.
The transistors are horizontal output devices used in old TV sets.

On the air, it sounds really clean and the modulation looks good.

John runs it on a pair of 4-400 broadcast pull outs that have almost no emission left, at 200 watts carrier.

The interesting thing is that it will work on almost any rig, you just find the operating points, zero carrier, 25%, and 100% and adjust the pots on the board for the tube in use.

John reports it takes negitive voltage on the screens to cut most tubes off.

I heard it on the air and it works, looks good on the scope, and since John has limited or no audio processing, it goes out to 20Kc...

Depending on how it's done, the series modulator should behave more like plate modulation, just applied to the negative side of the tube. The trick is to keep the grid to cathode voltage from changing with modulation, if it does then you are basically grid modulating the final. Floating the grid-cathode circuit along with the cathode circuit will do this (similar to a tube PWM final).
(maybe you already know this, I don't know. But those schematics almost look like that's kinda what's going on)

I received the circuit email.  Very cool, Brett, Tnx.

You said most tubes need some negative voltage to cut them off.  Does this mean that instead of grounding the circuit as shown on the schematic, it needs a negative supply to float this common lead above ground?

I'm thinking maybe a single 813 would do the trick.  I wonder what pep output a single conservatively screen modulated 813 will do with 2KV? 


The advantage of the screen modulated config over the series is it requires just one tube and the cathode / grids do not have to float at HV.



T

I was wondering about that, since John said most tubes need negitive voltage to cut them all the way off.
The design says to use 1.5 times the peak positive voltage needed on the screens, but I do not think it can go below zero volts...
But why set the resting power to 25%, 50% would be normal, no?


He also said the rig was 70% efficient, and I think it would do lots of power out with new tubes.

He was in the broadcast business in the NYC area, so some here likely know him.
His building of the unit looked top notch.

I would think the resting power at 25% would give the required 4X power for 100% modulation.  

He may be suggesting 150% screen voltage swing to allow for 150% modulation headroom, I dunno.

Maybe 70% efficient at 100% modulation, like a linear, but much lower at dead carrier conditions.  Still, the stage is in class C, which is good.   Some air with a chimney-blown 813 would allow for at least 175w of plate dissipation.

But whatever it works out to be, testing it on the bench with all sorts of parameter changes will find the best power and IMD sweet spots.

Thanks again for the schematic and I will definitely consider this method for a small 813 50 watt exciter.

Is it OK for me to post the schematic here for the other readers?

T

I do not see why not.
John seemed pleased anyone was interested in the sircuit and screen mod.
He used to work on big broadcast screen modulated rigs, and thinks they are great...

Tom,

You may want to see what Frank - KB3AHE did on his 4/1 final with his screen modulated scheme.  I seem to remember that he applied audio to the grid during modulation to improve its performance.  In his case he used I believe an ART-13 mod iron which had the second winding.  You could do exactly the same thing with a solid state solution so no transformers would be in the picture.  What you have is essential a series pass element and the output will follow which is on the gate or base depending if you use bipolar devices or FET's.  I would go with the FET's.   A high impedance bias network on the gate and that would then allow audio to be applied easily to modulate the bias voltage.  

Joe, GMS      

Joe,

Yes, I remember Frank adding some control grid monkey swing too.  I'll bet that looking at a THD test on the SDR would show when JUST the right amount of additional control grid swing is added.  That, proper drive, grid bias and proper plate loading should make the modulated output look sweet.


The Tron has a lot of BC experience with scream modulation too. I'll have get his take on it all.


Here's the solid state screen modulation circuit by W2IMX:

 

While I was in qso with him, I turned the screen voltage off on the 3x4D32 rig and still had some plate current, even though my bias is way high.
Not a lot of current, but some, with 1250 volts on the plate.
Every tube type will be different I suppose, I have not tried that with the 2x 4x150 rf deck yet.
The 4x150 / 4cx250b type tubes have some very different screen setup, in AB1 service, you can run 1000 to 2000 volts on the plates, keep the bias and screen voltage fixed, and have the same resting plate current.
I assume the screens have a lot of control over plate current.

I also have not tried it on the 813 rig, I may try it tonight.

Yep, I agree.

Coincidently, earlier I tried turning the screen voltage down to zero with my plate modulated 4-1000A rig.   The power dropped about to 1/3 and the plate current also dropped down.  But it will still needs more negative screen voltage and/ or grid voltage swing to cut the tube off.

So looks like we will need to add some negative supply voltage in the  control circuit.

Did you mean Tron?  What did he say about the subject?

T

No, Frank, W2IMX.
We had a nice qso about it for over an hour on 80 meters I think.

With the 3x4d32rig it was only 10 or 20 mills, and I am not sure I had any power out or not...

He was running 4000 volts on the 4-400's, he said he had to to get any power out of them, they had nothing left (emission).
I do not have any 4-400's or I would have sent him some.

He was talking like the circuit would go negative, but I do not see how it can. 

I was looking at Frank AHE's 4X1 screen circuit attached below.  His does not use a negative swing on the screen either.  Maybe he is making up for it with the grid swing from the transformer.

He doesn't use a grid leak resistor - full fixed -150V bias.   Interesting.

Back in the early 70's both Chuck and I tried that transformer circuit with a 4X1. We called it "composite grid and screen modulation."  


BTW, on the W2IMX circuit:  

A good friend sent me an email and suggested for improvements...  

"That circuit needs resistors Q1 base to ground and Q2 base to ground. Without them the circuit will be temperature sensitive so gain will change with temperature changes. Also the pot wiper leads could use some series resistors  so you can put too much current into the bases. FETs would be better."

Those changes would be easy enuff. I'll see what he suggests for values.

**  I do wonder what a negative screen current will do to the IMD. Maybe instead of negative screen, the control grid needs more negative swing to produce a cleaner cutoff without crossover distortion..



T

I did a test on the 2x 813 rig.
2000 volts on the plates, with no screen voltage, I get 120 ma resting current and 60 watts out.

Here is a reply I got from John:



Hello Brett,

Yes, most power tetrodes will produce RF output without screen voltage
as you saw for yourself in your test. The screen must be driven negative
towards the direction of cut-off :
 
"WITH RESPECT TO THE RESTING QUIESCENT VALUE OF 25 % POWER"

This resting voltage is always "POSITIVE"  So, when you begin to swing in the
negative direction "FROM THIS POINT", you're still in the "POSITIVE DOMAIN"
as far as the screen is concerned. It becomes a "BIASED" sinewave who's
zero crossing line is not zero but a positive voltage.

This part of the negative going voltage is derived from the audio voltage swing
at the output of Q2, emitter follower. Therefore, NO negative voltage supply required.

Regards,
John / W2IMX
 

 

Hi Brett,

Thanks for the info...

Well, the mystery still wanders thru my head - if the tube is still putting out power at zero screen volts, how can it be cut off when this circuit will not go below zero volts?   ;D

I started getting parts together for an 813 version.  But then I started thinking about a simple class D 50 watt rig that could do 300 w pep.  I could use 11N90s in a band switched final and build a simple class A modulator with 11N90s too.

I know Jay / W1VD has a working homebrew class  D version - and aware of the class D "Senior" commercial rig.  That may be the easiest and cleanest approach for a small exciter for the 4X1 linear amp.

I would use my new 20 watt DDS VFO as a driver.  After having success with the 11N90 VFO IPA design, I am inspired to continue on in that solid state mode.

Still thinking.   Anyone have a simple, working, homebrew solid state class C or D and simple modulator circuit like I described?  Band switched 160 - 40M, 30 to 50 w carrier?

Here's Jay's version for a monoband RF deck - quite nice!

http://www.w1vd.com/375wattclassD.html

T

Tom,

I can see the temptation of going over to Jay's class D design with modulation generated with PDM and as we all know it sounds excellent.  

Sometimes the hurdles or question that remain unanswered can be a wonderful source of education.   So you may want to tackle those and continue while learning.  Any of my experience with screen modulation required a waveform with transverses below zero volts to achieve 100% modulation since zero volts on the screen produces some amount of output power, hence a sine wave approaching zero volts will not achieve 100% negative modulation.  Frank's transformer schemes does that due to the nature of the AC being superimposed on a DC voltage just like plate modulation does while using a modulation transformer.  To do that from a direct couple schemes require a that the AC swing goes below zero if the tube requires that to cease conduction.  

Remember that the screen is very non linear.  A low impedance driver that can deliver the current under modulation is required if one is to achieve clean audio at the screen during modulation.  Cathode Followers, Emitter Followers or Source Followers are capable of doing that very well.    

Joe - GMS    

Not sure how it works, but Franks circuit seems to modulate 100% in both directions, per what i saw on the scope, the the pictures of the waveform he sent.
That with a pair of 4-400's, which are likely no better at screen mod then any other tube.

It might be a good project for me this winter...

http://www.ing.unitn.it/~fontana/IN3IEX%20AM%20MODULATOR.GIF

By combinig various circuits I developed a very effective screen grid modulator.
It works very well with my T4XB. I suppose that the voltage for the 6AU6 could be increased up to 500 V for tubes requiring 500V max for the screen grid. Update the zener diode voltage according to your supply voltage, maybe update the 6AU6 with a 6AQ5.
It can drive the screen grids of big tubes if the MOS and its heath sink can support them.
With a "touch" of controlled carrier the energy balance of the TX will improve....including that of the 10kW linear. 

Giorgio

" Frank's transformer schemes does that due to the nature of the AC being superimposed on a DC voltage just like plate modulation does while using a modulation transformer." 

It works because of what I just said above.  I have cut and pasted if from my previous posting. 

Joe, GMS

I was talking about the other circuit from John with no transformer.

If you are going to use a transformer, it might as well be a mod transformer, no?

Yes, Frank/AHE is able to get a superimposed "negative" voltage by adding in the swing of the transformer.  I don't see how John is doing it with his circuit, however.


I'm still considering ideas here.

BTW, Frank/ WA1GFZ had an interesting suggestion. Why not screen modulate the 4X1 linear itself?

It's an interesting situation: My two 4-1000As GG linear amp uses a regulated 450v screen voltage, has diodes in the cathode for grid bias and the control grids themselves are bolted to ground.  Both the grids and screens are at RF ground.

Maybe I can add modulation to the screen regulator circuit. I imagine the proper amount of diodes in the cathode would push it into class C. (I already have a class C tap with the diodes)

The only "problem" is it would not hit -100% negative modulation since it is in GG and the cathode pass thru power is unmodulated. But that would make a great built-in negative peak limiter, right?   ;D

I have 4KV on the 4X1s.    I'm not sure if there is any advantage to screen modulating it vs: driving it with an RF exciter other than the tube is in class C for better efficiency. But I can now run it biased into "linear" class C when on AM because of the carrier putting the tube into conduction. (Technique discussed in another thread)

Any thoughts?

Giorgio: That is a cool screen circuit for the Drake T-4 guys. Maybe Steve can archive it with the other T-4 circuit on the AM Window site.

T

According to some experimentation that Stu, AB2EZ posted on here some time ago about plate modulating GG amps, they have no problem reaching 100% negative modulation, it's 100% positive where they run into problems, and from what I remember, the reason is that when the modulation hits 100% positive, the input impedance of the amp is half of what the impedance is when the amp is only running a dead carrier, so the driver can't deliver enough drive to the amp which chokes off the positive peaks. When the amp is hitting 100% negative modulation, the input impedance is an infinite impedance, so the driver is really not able to deliver any power at all anyway. I don't know if that would be different with screen modulation, since that is a form of efficiency modulation, but it might be an interesting experiment.

Did you get my email with the pictures?
If John was using the circuit he sent, he was modulating 100% in both directions, as his scope picture shows.

Hmmm... well that kills that GG idea then, cuz I want to be able to modulate positively above 100%.  Tnx for the info.

Brett: I see that John is able to hit -100% negative looking at his scope pattern.  So the question is, we found that both of our rigs are not cut off when the screen is at zero, so how does this circuit go below zero volts to make this happen?   It looks like both Q2 and Q3 are in series with the screen voltage to the tube screen.  When both are turned on there is full screen voltage on the screen. When they are off there they produce a high resistance in series with the screen.  Wouldn't your rig and mine still be putting out residual power as we measured?

T

http://amfone.net/Amforum/index.php?topic=26603.0

And yet another good screen modulated design by WA1QIX for the DX-60.

T

If you modulate the driver in phase with the amp, then the driver is gonna be delivering peak drive when the amp is hitting peak modulation, but like I said, that was for a plate modulated GG amp, I don't know if a screen modulated GG amp would be different or not.

I edited my previous message..
I also tried a combination of screen grid modulation of the finals and screen grid driver modulation with various intensities. The best result was obtained with no modulation to the driver tube.
Anyway nothing is better and more efficient than plate and screen grid modulation with a modulation transformer unless you adventure though PWM.
 

I received an explanation as to how the screen W2IMX circuit works. The audio AC comes through Q1 and Q2 and is capacitively coupled thru Q3 - and is superimposed on the screen DC bias. This is something like how Frank's transformer adds superimposed audio.

This allows there to be a negative voltage swing in reference to ground to turn off the tube completely.

I am still considering this circuit modified with MOSFETS to drive an 813 screen.  A screen modulated 813 should put out about 40 watts of carrier which would be perfect as an exciter.

T

An 813 should do about 75 watts of carrier, no?
One does 300 watts, 2000 volts, 200 ma, 400 watts in, 300 out, 100 watts plate dis, on a 125 watt plate tube.

Since plate modulation increases the plate voltage to 2x the resting plate voltage, what would happen if you ran them at much higher voltages, would the power go up?

What will be the power output if you ran an 813 at 3000 volts and 100 watts plate dissipation?

In class C plate modulated service, you run the tube at 2000 volts, 200 ma, 100 watts plate dissipation, then double the plate voltage, and take it down to almost nothing.
Does not screen modulation do the same thing, but with current?
Full power at 100% modulation, almost no power at 99% negative modulation.

So, what if you ran the 813 at 4000 volts on the plate, same as 100% modulation (or less) with plate modulation?
4000 volts, 100 watts of plate dissipation at 25% power output, what would the power be?

Brett
N2DTS

you have to remember that with screen modulation, like any other form of efficiency modulation, the efficiency with a resting carrier is gonna be low (around 30% to 33%, something like that) and the peaks are created by incresing the efficiency of the amplifier. So with a dead carrier, that 813 has to have enough disipation to dump 66% or so of its input power as heat. CCS voltage limits for any form of efficiency modulation (based on the datasheet ratings for class B AM service and grid modulation service) seem to be 2000 volts and 100 watts of plate dissipation. So the max input that you would want to run the 813 at for screen modulation service would be 150 watts (remember, this is just an exciter for an amp, running the tube conservatively will work fine) which works out to 75 milliamps at 2000 volts, and would give about 50 watts out.

Yep, that's pretty close to what I have seen in the real whirl.

Actual screen modulated rigs appear happy at an output carrier of 30% of their plate dissipation. For example, an 813 at 125 w diss would make about 40 watts carrier out. I think an 813 is more like a 150 w tube, esp with air, so it can be a little  higher.

A 4-1000A is also more than 1000w diss tube with big air, so a screen-modulated 300 watt carrier is easy - Probably more like 450 watts is OK max, just like linear operation on AM.

I think it is mainly a power dissipation issue limitation regardless of the voltage run on the plate.  I would find the high voltage sweet spot vs: current for best IMD and go from there.

I am leaning towards the single 813 with this screen circuit at this point... Maybe even two 813s in parallel to do a nice 80 watts when barefoot, dunno.   Two tubes backed off would be cleaner driving the linear too.   ;D  The linear is AB1 GG, and a clean -40dB 3rd IMD, so a worthwhile effort to use a clean driver perhaps.  Bear in mind that most ricebox drivers are only about -30dB, so they wud not do justice.

T

Another problem with screen grid modulation is that it is necessary to tune with full screen voltage, which is not the screen voltage without modulation. Therefore an AM, TUNE switch is necessary.
Years ago (before experimenting) I was very open to various alternatives to plate and screen grid transformer modulation. Now I can say: if you can have a modulation transformer, use it. This is the best you can do. Maybe use feedback to improve fidelity.
Series (plate and screen grid) modulation looks fine but you need 4000V for the two, better three 813 tubes: 2 x 813 //  class A series modulating a 813 class C. Overall efficiency is quite reduced respect to a similar transformer modulation.

In conclusion, looking for the best possible audio and ignoring modulator efficiency my choice would be the series modulation: 2 x 813 //  class A series modulating a 813 class C.
Class A audio, class C RF, no audio transformer. 100 W carrier, 400 pep. Why not?

Yes, that is some good advice - obviously gleaned from experience.

I'm still actively weighing all the possibilities  and the pair of class A 813s series modulating a class C 813 RF final is appealing too.

Although I do have the mod iron available, I want to stay away from a higher powered 813 plate modulated rig cuz I already have a plate modulated 4-1000A rig.  I really need a low powered clean exciter to drive my other 4-1000A linear amp.  This exciter could double as a "late afternoon" barefoot rig too.

Another idea is to use a single 4-400A as the series tube modulating the 813.  I have the parts for this too.  At 1.5 KV across the final it would be backed off and very clean at exciter power levels.  (I would be using a 4KV supply, borrowed from the 4-1000A linear.)  I also have a 2KV supply, if useful.

Question:  Does a class C series cathode modulated rig need critical plate loading adjustments just as a screen modulated rig does?  A plate modulated rig's loading is not critical for clean audio peaks. Since series modulation is similar to plate modulation, I was hoping it was not (or less) critical.  

I mean, when we QSY or change antennas with a screen or grid modulated rig, we almost have to put a tone thru each time to retune the rig to optimize for cleanest audio peaks.

T

I am almost certain that this design: http://www.amwindow.org/tech/htm/wb9eckseriesmod.htm behaves like a classical class C plate and screen grid modulated transmitter. Tune with no tone and everything will be fine. I was very tempted to build it, but I was able to find a modulation transformer and that project was abandoned only because of overall efficiency.
When I use the screen grid modulated (modified) T4XB with the linear amplifier, if the modulation is not good it is difficult to find the origin of the problem: TX4 mistuned or linear amplifier mistuned? My linear amplifier is not exactly 50 ohm input, how can I tune the T4XB independently?  Solution: put T4XB in tune mode (SSB + tone) and tune  4 controls for the maximum. 4 controls is not "quick" compatible.... Go with series modulation WB9ECK like, less problems with the amplifier. You might also tune while speaking !
Bob says: "Now you can have a nice small 25 watt transmitter/exciter requiring only four tubes and NO modulation transformer. Hi-fi and no iron required. What's not to like?!? The transmitter output is adjustable from zero to twenty five watts output with 100% modulation. If you run about 15 watts the positive peaks will be somewhere around 115% while the negative peaks remain below 100%.  Works great for driving a grounded grid linear." This is it.

Giorgio

What could a pair of 6080's or 6B4's cathode modulate?
Or, what would make a low distortion class A amp for a class C final?

I hear some very good screen modulated rigs on the air, but maybe they do not change things often, just run it where its parked like a broadcast rig.
I would tend to do the same, just park it on 7290 and I would be good to go, but class A series modulation might be even cleaner?

100 watts makes a good signal on 40 meters with a good antenna, so I wonder what would be needed to series modulate something at 100 watts carrier level?

I see no reason to build another RF deck, just a modulator would be easy.

I have three RF decks that work on 40, 2x813, 2x4-150a, and 3x4d32.
600, 300, 300 watts carrier.

What would two (or four) 6080's be able to modulate?

I have a small chassis and need a winter project..


Brett
N2DTS

A cathode modulated rig should be easy to properly tune/load... just as a plate modulated rig.

When tuning/loading a transmitter (regardless of the modulation type), the objective is always to adjust the RF load impedance from plate to ground so that, at peak output power ... on each RF cycle... the plate-to cathode voltage swings up to 2x the (modulated) B+, and down to zero. If the RF plate load impedance is too large, then the tube will saturate when the plate-to-cathode voltage drops below zero (resulting in distortion in an AM or linear amplifier application). If the RF load impedance is too small, then the tube will operate inefficiently.

With screen modulation (where the B+ is not modulated), you have to adjust the tuning/loading with the transmitter operating at peak power so that the RF load impedance causes the plate RF voltage swing to be 2 x B+ in amplitude. Otherwise the tube will saturate on modulation peaks. The consequence of this is that the tube will be operating very inefficiently at output levels below the maximum (where the amplitude of the RF output voltage much is less than the B+).

With plate modulation (which also always includes modulation of the screen-to-cathode voltage if you are using a tetrode) or cathode modulation (which also results in modulation of the screen-to-cathode voltage at the same time as you are modulating the plate-to-cathode voltage) the plate-to-cathode voltage (i.e. the modulated B+) increases and decreases with the amplitude of the RF voltage from plate to ground. Therefore, if everything is working properly, you can tune/load the amplifier at carrier level (i.e. the amplitude of the RF swing is the same as the B+), and it will also be tuned/loaded correctly for any value of the modulated B+. I.e. if the B+ (voltage between the plate and the cathode) is modulated to 2x its value at carrier, then the RF output swing will also be 2x as big... and you will still have the correct loading/tuning (i.e. RF output voltage amplitude = to the modulated B+)

Of note: the principle of adjusting the B+ to match the output level (in order to maintain high efficiency) is basically the same as is employed in an audio Class H modulator. A Class H audio amplifier has higher efficiency than a Class A audio amplifier because you keep the B+ at a lower level unless the amplitude of that audio voltage tries to exceed the B+. If it does, you increase the B+.

Stu    

We appreciate your FB info, Stu!

It looks like the series modulation is the better way to go from a precision tuning perspective.  I like to optimize a rig on the bench and be done with it.  The screen modulation might be a QSY hassle.

I was reading your older posts about using plate (or cathode) modulation and adding in a separate very clean screen modulation.  Normally I use a screen choke. Some use a resistor from the plate. But both of these methods are dependent on the plate / cathode modulation source. You indicated it could be cleaner if the screen used a separate source.  

Question:  Do you feel a super clean class A series cathode modulator would be clean enough (cleaner than a mod transformer system) to eliminate the need for a separate screen audio source?   Or do you think a screen choke with class A series modulation would be as pristine?

Another idea:  To eliminate the need for screen modulation, what if we used a triode instead of a tetrode for the RF final?  And also for the series modulator?    I have two 3-500Zs available and they are very linear at 1500V .... or what if an 813 was used with the grid/screen connected as a triode?   Or will the elimination of the screen just mean I now have to modulate the driver stage or add some grid audio to linearize it?

My objective here is to build the cleanest (40-80 watt carrier) exciter I can, regardless of power level or efficiency.  (-40dB 3rd order IMD)


T

The discussion in this thread has me thinking in two directions....   Low power series modulation followed by a linear amplifier, or high power series modulation, perhaps PP 304TL or PP 250TH modulated by two, three, or maybe even four 4-400As. 

For a low power solution, there are not too many challenges.  But with the higher power, filament transformer voltage breakdown ratings are critical.  Should the 4-400s be at the B+ side?  This option creates additional driver isolation challenges, compared with 4-400 cathode modulation.  But neither of my PP finals are designed to have the filament circuit elevated via series modulation at the cathode (filament).  Modulating either of these finals without a mod transformer sure sounds interesting.

On the other end of the scale, an interesting QRP series modulated transmitter is the HP 606A RF Generator.  This one uses series modulation, and my experience is that it is one of the cleanest AM transmitter/modulator implementations I have seen.  It has some other features that might just apply to the low power exciters discussed in this thread.  I share below an excerpt from some data from HP/Agilent marketing data.  The 606 schematic and detailed information is readily available on the web......  (I added bold italics to the interesting features.)

Here is the link and an excerpt of the text:  http://www.hpmemory.org/timeline/art_fong/hp606_01.htm

HP 606A/B, the Existential Signal Generator
 

The Signal Generator is a calibrated transmitter. Intended primarily to test system receivers. In the design, test and maintenance of any communication system, radar, satellite, Police radio, or wireless system receiver, there are phases when you need precise simulation signals that behave like the transmitter. Signal generators give you tunable signals, AM or FM or pulsed or digital modulation, and a very precise settable output level WAY down to -127 dBm, which reveals just how sensitive the receiver is.

In all those decades of early signal generators in the 1940s, 50s and 60s, some of which used Klystrons for the source, some like the HP 608 and 612, used tunable circuits with tunable power amplifiers for the output stage, they weren't ideal but they were pretty good. As you tuned frequency, set on channel, the frequency drifted a little, the output power at each new frequency was different, meaning you had to set the output power each time. And the modulation simulations were not so perfect.

And then, in 1959, along came Art Fong's HP 606A. It wasn't time for semiconductors yet, but it was time for an almost perfect design. Art chose to use capacitance tuned oscillators and power amplifiers. The vacuum tube oscillators gave LARGE voltage tank voltages, meaning a very low single sideband phase noise on the carrier. The drift was low but the HP 606B/8708A furnished a phase locking means to stop drift. The power amplifier had feedback on the power monitor, so you could just tune anywhere, switch frequency bands, and the power output remained rock solid.

The modulation function on the power amplifier had an additional high performance characteristic. It was a highly-linear detector at the output, which was part of the AM feedback loop, and forced the modulation envelope to match EXACTLY the input modulation audio. All the way up to 100% modulation. Wow. Now THAT was a signal generator. The customers LOVED that generator for its convenience and accuracy.

Future HP generators like the HP 8640 came along as semiconductors took over, gave us internal phase locking, microwave generators with YIG-tuned microwave transistors, phase-locked, digital synthesis, nanosecond pulsing, and digital (I/Q) modulations to match the amazing new communications and radar systems technology. But Art Fong's generators took our RF, VHF and microwave instrumentation through several decades of advances in our technology world, from the 1940s to the 70s.

Tom

Both traditional methods of modulating the screen voltage in a plate (or cathode) modulated transmitter seem to work pretty well (the output sounds good to people who receive it) if adjusted properly. The value of the screen voltage dropping resistor has to be selected to match the characteristics of the tube (e.g. in a Johnson Ranger it is 30k ohms; while I have used 15k ohms to plate modulate an Ameco AC-1).  Likewise, the screen choke method(the "self modulation" method) seemed to work very well in my KW-1. [The screen choke causes the screen voltage to vary with the modulated plate voltage, because of electrons that accumulate on the screen when the plate-to-screen voltage drops; and which leak off the screen, through the choke, when the plate-to-screen voltage rises].

As you know, your vocal tract produces a set of harmonics that respond to the harmonics in the periodic pressure wave being produced by your vocal cords... and a little extra distortion introduced in the audio chain and modulation process may make your voice sound a little different... but need not introduce any objectionable artifacts.

On the other hand... even if you modulate the screen-to-cathode voltage directly, with a separate, low impedance voltage source (in addition to modulating the plate-to-cathode voltage)... the modulated RF envelope will still not be a perfectly linear replica of the input audio. If you look at the tube characteristics of the tube you plan to modulate... you will see that the output current is not exactly proportional to the screen voltage over a wide range of output currents (but is approximately so). Likewise for plate modulating a triode.

When I experimented with using a separate screen modulator (in addition to modulating the plate voltage) in a DX-20, it was educational... and I did find a sweet spot in the ratio of screen modulation percentage to plate modulation percentage, that produced very clean sine wave RF envelope in response to a sine wave audio input. But, after that, I went back to using a screen voltage dropping resistor (I don't remember what value worked best... but I think it was different from the stock value of the DX-20 screen dropping resistor).

When I plate modulated my high power, grounded grid, triode transmitter, I found that it was necessary to modulate the driver transmitter (my Johnson Ranger) simultaneously with modulating the plate voltage on the high power triodes. For optimum linearity (but not perfect linearity) I found that I needed about 20-30% modulation of the driver transmitter's output, while plate modulating the high power, grounded grid, triode transmitter to around 125% on positive peaks.

I think that it you really want a driver with very clean (very linear) modulation, you need to think about something like a small Class E transmitter with PWM.

Note, that your "linear" amplifier is going to introduce some non-linearity with respect to the RF envelope of its output signal v. the RF envelope of its input signal... so trying to make the driver increasingly "clean" will produce diminishing returns.

Stu




 quote author=K1JJ link=topic=35375.msg272550#msg272550 date=1387138835]
We appreciate your FB info, Stu!

It looks like the series modulation is the better way to go from a precision tuning perspective.  I like to optimize a rig on the bench and be done with it.  The screen modulation might be a QSY hassle.

I was reading your older posts about using plate (or cathode) modulation and adding in a separate very clean screen modulation.  Normally I use a screen choke. Some use a resistor from the plate. But both of these methods are dependent on the plate / cathode modulation source. You indicated it could be cleaner if the screen used a separate source.  

Question:  Do you feel a super clean class A series cathode modulator would be clean enough (cleaner than a mod transformer system) to eliminate the need for a separate screen audio source?   Or do you think a screen choke with class A series modulation would be as pristine?

Another idea:  To eliminate the need for screen modulation, what if we used a triode instead of a tetrode for the RF final?  And also for the series modulator?    I have two 3-500Zs available and they are very linear at 1500V .... or what if an 813 was used with the grid/screen connected as a triode?   Or will the elimination of the screen just mean I now have to modulate the driver stage or add some grid audio to linearize it?

My objective here is to build the cleanest (40-80 watt carrier) exciter I can, regardless of power level or efficiency.  (-40dB 3rd order IMD)


T
[/quote]

There is another method of modulating the screen in plate modulation of tetrodes, that is to use a dropping resistor from a fixed supply.
I do that in some of my rigs, the screen supply is adjustable, and i set it for about 100 or 200 volts extra, and drop the 100 or 200 volts in a series resistor.

I like this method because screen voltage is limited, it can never get to the plate voltage level and the screen is self protecting.
It self modulates as well.

I think you get some odd stuff going on when you use a dropping resistor off the modulated B+ voltage, and many rigs do best if the screen voltage is got from a combo of modulated and unmodulated B+.


When I did tests on my rigs, they seemed very clean if I adjusted things right.

Well, I do change my mind a lot when considering a new idea, but have come to a decision.  The rig will be a class C  4-400A series modulated by a class A  4-400A.  Pictured below are the parts thrown together in the staging area.

The reasons for the choice:  

1) It will do an easy 125 watts of carrier, thus a good lower power barefoot stand-alone rig.

2) When backed off to 25 watts it should be a super clean driver for the 4-1000A linear amp.

3) The tank tuning and loading for optimum performance will be as easy as a conventional plate modulated rig.

4) I already have a 4KV supply that will put about 1500 volts across the RF final at carrier (2500V across the modulator) and have enough positive peaks when the 4KV is across the final under modulation.  This is an increase of about 166% in voltage.

5) With the quiet Variac-controlled muffin fans blowing air from underneath thru the chimneys, it should be a silent running rig.

6) A bonus:  Frank WA1GFZ has designed a nice MOSFET driver circuit that will drive the 4-400A series modulator grid directly. It can't get better than that. The circuit still needs some work and will be available after I test it in the working rig.

I am missing one 4-400A chimney (SK-406) and two turns counters for the vacuum variables. Anyone have these parts for sale?

I found an old Peter Dahl 7.5 V filament transformer in the junk box that is good for 10KV insulation for the floating 4-400A RF final filament HV isolation requirement.  I will Variac it down to 5V.   The series modulator tube is at ground and uses a normal 5V transformer.

Except for some scrap yard aluminium I need to pick up, I've got about everything to build the rig.

It will have the normal white panel with black lettering.   No name yet!    


** Another thought - If I get tired of the class A modulator tube heat, I could later drive it as a PDM modulator and add a low pass PDM filter.  Not really a big deal to do at that point.

 T

Continental does that in one of their AM BC rigs.  Pair of 3-500's in the RF deck with a single 3-500 as a PDM  switch tube.  Bob, 2ZM has one on the air and it sounds fantastic!  Steve, QIX had a tube PDM rig on in the seventies if I remember correctly. 

Glad you make a decision and now let the fun begin!

Joe, GMS

I have been following this thread with great interest.  It seems you have the plate and screen modulation issues resolved regarding the class C final.  Not much discussion has addressed the class A series modulator.  I am considering doing something similar with my PP triode RF decks, so I am thinking more about the modulator section.  (I am considering placing the modulator tube between B+ and the RF deck, but that is irrelevant to the present discussion.)

Running some rough numbers, if you are looking at 25 watts out, at around 60 % overall RF section efficiency, that is around 42 watts input.  125 watt output is just under 180 watts input at 70%, and just under 210 watts input at 60% efficiency.  If you run 1500 volts on the RF stage, at 200 watts input, the series string will run 133 milliamperes with no modulation, and in class A the modulator average plate current will not vary from this value.

The modulator tube will drop 2500 volts at 133 milliamperes,  dissipating approximately 333 watts during the entire transmission.   This is well under the 400 watt rating, but will need significant air.  At 42 watts RF input, 28 milliamperes would result in the class A modulator tube dissipation of about 70 watts, assuming you keep the voltage split at 1500 for the final, and 2500 for the modulator.  Some control of the bias on the modulator will be required to alter the power level, because changing the loading on the final will result in different plate current, thus requiring adjustment at the modulator bias if it is desired to keep the final plate voltage around 1500 volts.  (It has been a long day, please excuse me if the numbers above are not accurate.)

This brings three questions to my mind.  First, where are we in the characteristic curve of the class A 4-400A modulator, and are we going to see linear modulation at either the higher or lower power level, or both?  Second, might it make sense to evaluate running the 4-400A series modulator in triode, rather than tetrode mode, to attain more linear modulation?   Third, might it make sense to run two 4-400A series modulators in parallel to split the overall dissipation between two tubes, thereby allowing the operating point to be adjusted for maximum audio linearity?  Or possibly use one or two 3-500Z tubes for the modulator? 

I realize all this is a moot point if you go to a PDM modulator for efficiency and audio quality, (where a single 4-400A will do nicely) but I think there is a lot to be discovered when planning the class A series modulator in this rig.  In the wintertime, the extra dissipation of the modulator will help keep the shack warm and cozy, but I think some evaluation of the characteristic curves and audio linearity should be evaluated before settling on the chassis layout.  Perhaps some feedback from rectified RF output might also be beneficial to provide the cleanest audio signal with the series Class A modulator.

As you have stated previously, the fun and challenge of making the rig operate as clean as possible starts after the initial build is completed.  Thanks for sharing the information on this planned build, it has been inspiring and educational!

Hi Rick,

Thanks for taking the time to give us your thoughts.

Yes, that is a good idea to consider running TWO 3-500Zs as the series modulator for the single 4-400A final.  The 3-500Z  is a VERY linear tube when running at reduced voltage (1500-2500V) and will also eliminate the regulated screen supply requirement. In addition, two tubes will make cooling much easier for air flow.

I have a pair of 3-500Zs here and might use them.  Later, if I go with PDM, I can always unplug one.

I would think in class A the 3-500Z would be as clean as any PDM modulator, maybe even better. But we are cutting hairs at that level.

I once considered your idea to use the series modulator (or PDM modulator) in the plate circuit too. Would you use some kind of fiber optic floating method or something else?

** BTW, Frank's new MOSFET driver is including a negative feedback loop that will sample some audio from the plate of the series modulator tube and feed it back to an early ~1v stage.   That shud be sweet.   I'll bet that the 3-500Zs will need only a small positive DC bias on the grids to put it into class A.



Joe:  Yes, Steve had a 450TLs X 4-1000A  PDM rig for many years.  I've had two homebrew tube PDM rigs over the years.  They were both two 4-1000A finals. One used a 3CX-2500 in the PDM modulator.

I ended up tearing the last one down cuz of the safety concerns for the HV floating screen and grid supplies as well as the PDM filter coils being so large and exposed.    Instead, this time I will use a floating grid leak and screen resistor instead of floating supplies. Also, later I may do some experimenting with compact toroidal PDM filter coils.  We'll have to see.

Stu: Based on your last post I will go with the B+ to screen resistor as mentioned above.


T

    Did this original requirement change as this thread evolved? Seems like a Tim-Tron modified Ranger using 6550's in the modulator, and one of them new ER/Hammond modulation transformers would do the deed, and do it well. How did this evolve into dual 3-500Z modulators, and a 4KV plate supply?  ???

Jim
Wd5JKO

Hi Jim,

I wondered the same thing. These rig plans always seem to grow as we dream on. I slept on it and looked at the parts this morning and realized I'm just building up another "big" rig.    I already have a big rig and need a 20 watt hi-fi exciter for the 4-1000A linear.

Today I looked again at WB9ECK's 6LF6 pair of 40 watt sweep tubes (80 watts dissipation) series modulating an 807 in class C.  I could drive the 807 directly with my new DDS VFO using a toroidal transformer, no input tuning.  His simple one tube (12AU7)  audio driver would drive the 6LF6s.  It would need just an 1100 volt HV supply and a 300V low voltage supply. (40V screen voltage using a zener)     This should be good for about 20-25 watts carrier out and clean.

I could build it very compact with small bread slicer variables and associated parts. I am running out of room in the shack... ;)

http://www.amwindow.org/tech/htm/wb9eckseriesmod.htm

The planning and dreaming is half the fun.


T

Tom, I was typing a reply when you and Jim corresponded.  I agree, the heavy metal is not required for the exciter, but you mentioned that it would be nice to be able to crank it up to 125 watts or so.  I think it would be rather complicated to make the series modulated transmitter work properly for that low to high range of power, while keeping the audio section linear.  I also believe you would need some negative, not positive, bias on the 3-500Z to achieve the voltage split you described between final and modulator.

When looking at placing the series modulator above the final, I see more challenges than it may be worth.  I wanted to try this because the two finals I have restored do not lend themselves to isolation with a large series modulator on the negative supply side. 

With the series modulator on top, it functions a s a cathode follower, and would require massive swing on the control grid to attain decent modulation percentage.  Also, coupling audio to the input would be a challenge to isolate, considering the high voltage at the input.  This is easy to do at RF, but not for full range audio response.   So I may just shelve the idea and use conventional modulation transformers in those rigs.  It is, however, interesting to consider building a bigger rig as you were describing, floating the RF section at HV.

The smaller scale solution looks like it has merit and will fill the bill for your requirements.  GL in the implementation and testing.

Rick,

Yes, maybe the cathode approach is the way to do it.

Well, I committed to the project. It's going to be quad 6LB6s series modulating a pair of 807s.  Quad 6LB6s are good for 160 watts dissipation total. I will have a fan on them too.

I like the idea that I can do this with a 1200 volt supply. Reasonably low voltage compared to big rigs.

Today I found a nice chassis and laid out the parts. It will use air variables, so is reasonably compact.

Gonna  take some pictures once the tubes and sockets arrive and get installed.

Here's the circuit again. Planing to add some audio negative feedback from the modulator plates back to the 12AU7 preamp.   The 807s will be driven by the solid state DDS VFO and use a toroid input to eliminate tuning hassles.

http://www.amwindow.org/tech/htm/wb9eckseriesmod.htm


I've never built a 6LB6 / 807 series modulated rig before, so it should be fun getting it working.  Again, this may turn into a pdm rig later on - it wud be easy to add on.

T

The goodies are shipped:

Tom, that seems like a conservative and practical approach. 

In addition to the audio feedback to the driver or preamp stage, you might want to consider playing with some DC feedback as well.  This could function similar to an electronic voltage regulator, with two benefits.   First, it would make the DC voltage applied to the final RF stage relatively constant as you do your tuning and loading.  Second, it could help keep the class A series modulator at the center of the linear portion of the curve, thus minimizing distortion.  Basically, you would use one pot to adjust the operating point of the parallel modulators, and another pot to adjust the desired amount of feedback.   My thinking is that one of the most significant challenges in class A series modulation is keeping the modulator as linear as possible, and the key is the correct bias.  This will vary if the voltage across the modulator tubes is allowed to vary indiscriminately with changes in final plate current.

Another feedback option is audio derived from rectified RF.  Again looking at the circuit for the HP 606A RF generator, it may provide some ideas to make the series modulator extremely clean and stable.  I am not suggesting cloning the circuit, but have a look at how they designed a tube differential amplifier to mix the feedback with the incoming audio signal. 

If I were to build a rig such as this, I might breadboard a section of the modulator, and load it with a fixed resistor equal to the modulating impedance of the final amplifier, and play around with operating points, voltage drops, feedback, etc.  This will enable you to determine just how much "linear" swing you can get, positive and negative from the quiescent operating point, with your proposed modulator.  You could even do it initially with one tube, a test oscillator and input coupling transformer, and a variac on the HV supply.  Then extrapolate the results to estimate how four tubes in parallel might perform.  Minimal test setup and possibly lots of data to come closer to the final design before the full build.

  Here is another series modulator discussion with DC feedback:

http://amfone.net/Amforum/index.php?topic=27856.0

  This differs in that the modulator is on top, and solid sate.

Jim
Wd5JKO

For a continuation of this thread discussing the final configuration  -  quad 6LF6s PDMing quad 6146s:


http://amfone.net/Amforum/index.php?topic=35443.msg272874#msg272874