Screen or Plate Modulation? Which way should I go?

[w4bfs]

many excellent ideas .... I also am a fan of the G2daf circuit to derive screen voltage ... It will likely require a volt'g quadrupler for the 4x1 ( this is based on ssb use) and will use 75 or so watts drive just for this ... I have tested a quadrupler using schottky diodes and found reasonable performance 160 - 10 meters just using a bifilar 1:4 un-un .... i like the notions of using less iron ... I am also considering the bias shift Heising (see parallel thread) modulator ... I especially like the fact that so many are willing to share their ideas ...many thanks .. 73 ...  John

[w8khk]

John, I have met very few folks who knew what the G2DAF amplifier is, but they did know about the G2DAF receivers!

When I was at Davis Monthan in Tucson, I was running SB-101 barefoot, and of course I was collecting parts for a linear.  One of my ham friends gave me two 4-400A tubes to get me started, and he handed me a photocopy of a G2DAF schematic, hand drawn.  I still have that original paper.  He said: "Before you go building screen supplies, have a look at this."   Seemed too good to be true. 

No neutralization because the input swamping resistor was mounted right at the tube socket.  Negative grid bias cuts off the amplifier, and with no drive, no screen voltage.  Very simple PTT switching.  I remember modifying an old xmit cap for plate tuning, adding spacers to double the spacing.  I used 2 6AU4 tubes for the doubler, and it worked fine the first time, never modified the rig since building it in 1968!

I have been thinking of using 4X250s (four of them) to make a small G2DAF linear in an SB-400 cabinet, complete with power supply.  I suppose this amp might not even need a doubler.  I have not thought of using a 4x1 to build the linear, I just want something on AM soon.  I spent months restoring a 32V2, but it still is not right, but that is another story.

I would be interested in seeing more data on your DAF design.  That would make an excellent thread.  Wonder how well the DAF would do on AM???

[w4bfs]

I have wanted to do a multiple 7034/4x150/4cx250 anp for years ... I worked for the us navy in the 60's at a site that was starting to rebuild/test TWT's.  there was a reg hv supply using a 4cx250 as a series pass reg and to simplify cooling and insulating it it was oil cooled ... it should be possible to do this and might eliminate chief problem using these which is plentiful blower noise ...specific heat MUCH higher for liquids ...... G2DAF should be a natural with this tube as well ... I even made some cooling jackets for a related tube the 8072 ...maybe one of these days... 73 ... John

[w4bfs]

Rick ... got to thinking about your fil V only 4x1 test ... something doesn't seem right as I don't recall them getting excessively hot ... If you are pressurized, what is your air delta P and is it close to the min that the tube spec calls for ? ... if it is, then what are the fil Amps and Volts? ... one of fil rods may be misbehaving (assuming Eimac), the Amperex fil looks like a fish net  ... finally (pun intended) fil color?

[w8khk]

I have a stash of 7035s, the 26.5v fil version of the 4x150.  According to spec sheet, it has a diss rating of 250w and can handle 2kv on the plate.  Never thought of oil cooling, that could prove very interesting, and quiet!  Would not even need any special sockets.  Something to think about.

The temp of my 4x1 with fil only might be subjective, I did not measure it.  Also did not measure the delta p, but I will before I put any plate voltage on it.  Just had a test jig with socket and fan, all air thru the socket, none around it.  I think the filament is ok.  I measured 21.5A at 7.5vac.  Have a 10v 30a transformer on a variac.  I also performed a simple "edison effect" test, just to see if it had emission.  Connected a milliameter between electrodes.  Fil to plate, 0, fil to control grid, 4ma, fil to control and screen grid, 7 ma, fil to control, screen and plate, 9.5ma.  So it is "emitting".  I am going to punch some holes in the chassis around the socket amd test delta p before and after, and if it does not meet Eimac minimum spec, a bigger blower goes in.  BTW, it is a GE tube.  Used, from a SK in SC, sold by ham club via epay.  Has very slight discoloration on glass in line with holes at top of plate, so I know it was used, but I doubt it was abused.  If it is toast, I will probably go with 4-400s for rf and mod, but I think it will be ok.

[Tim WA1HnyLR]

Hi Rick ,
Very interesting thread here. Whether to go with plate or screen modulation to me depends of what you have for plate tank components . If you have the right stuff I would plate modulate the 4-1000 . If you have piss weak tank components I would go to screen modulation.It would be a shame to limit the maximum power you could run by the modulation scheme. A 4-1000 will make alot more power being plate modulated than screen modulated. The question is do you want a kilowatt transmitter or a 250 watt transmitter? The so called legal limit can be defined in a number of ways. FACT: 1000 watts am modulated 100% with a sinusoidal tone is 1500 watts raw bulk rf power delivered to the antenna. Some people perceive this to be running over the legal limit. I disagree. The fact is it is FeeCCee law that one must run the minimum power to maintain communication. Having a transmitter that can easily fold back to 250 watts is convinient .The Stancor mod transformer does have rather poor high end audio response . One must make full use of primary and secondary to make sure you are not subjecting the transformer to uduly high audio voltages under operation.Protective Spark gaps on the primary are a MUST. Use of the 50 Hy mod reactor with the Stancor mod transformer may be a waste of time being that the A-3899 was designed to handle unbalanced DC. Being that you have 4-400s or 4-250s I would use them . With 3Kv you should be able to get a comfortable 1Kw out of the equasion. Nothing beats power when the chips are down and the slopbucketeers are quacking up your backside.Screen modulation has it's benefits too. Using cathode follower screen modulation with 3Kv on the plate you should get a clean 250 watts with good modulation headroom. Some years back I designed a cathode follower screen modulator for a 4-400 using a 6DQ5. I almost finished it when the fire of '92 badly damaged the transmitter. I never picked it up after that. A good screen modulated transmitter will have unparalleled fidelity compared to the plate modulated scheme with what you have for components. I would plan to build a plate modulated transmitter with a screen modulator as a back up modulator as well as a modulator to use when you don't need a kilowatt of power. Good luck
Tim WA1HnyLR

[k4kyv]

Use of the 50 Hy mod reactor with the Stancor mod transformer may be a waste of time being that the A-3899 was designed to handle unbalanced DC.

Years ago I ran a UTC VM-5 mod transformer with a UTC LS-103 50 Hy reactor.  The final ran 2000 volts @ 500 mills, the maximum rated current for the LS-103.  I noticed a slightly better waveform on the scope with the reactor, and reports said the reactor made it sound better, although there was not a tremendous difference.  The big advantage of using the reactor was that it practically eliminated the talkback from the VM-5.  Running unbalanced DC caused it to sound like a small  speaker was inside the transmitter cabinet, with the resultant acoustical feedback squeal.  With the reactor, the talkback was almost imperceptible.

[w8khk]

Hello Tim,

Thanks for the detailed response and advice.  I believe the RF components I have will be sufficient to run a KW plate modulated.  Very heavy duty tuning and loading caps, and the coil is 1/4 inch copper tubing, 3-1/2 inches in diameter.  The band switch is from a BC-375 tuning unit.  B&W 800 plate choke.  All similar to my G2DAF linear parts, which never had a problem at full strap.

I am looking at capacitor input, full wave center tapped supply for the final, with 100uf 4kv oil cap.  For the modulator, choke input full wave center tapped supply, using the same plate transformer. which is fed via a 240 vac 20 amp GR variac.....  I calculate that this will provide 2500 volts or so for the modulator when the final is at 3KV full strap.  The modulator plate impedance is around 10,300 ohms, and the final should draw 333 ma at 3KV for 1KW input.  This is around 10,000 ohms dc impedance as seen by the modulator, so I should be able to use the 3899 tranny at 1-1 ratio, using entire primary and secondary, with spark gaps. 

Since I have the reactor, I will probably leave it in, just to take the DC IR drop off the mod tranny, but I could always short it to see if it makes any difference in the audio quality.  I would expect screen mod with no iron in the path would have much better high-end, and I plan to play with that; but with the plate mod I expect to attenuate the high end in the audio chain.

For plate modulation I expect a suitable reactor in series with the screen would be sufficient; I am thinking a high-end audio output transformer here, perhaps a UTC LS-58.

Screen and bias supplies will be electronically regulated and swamped, adjustable so that the biass may be set to work with any reasonable plate voltage, meeting the FeeCee power reduction requirements.  The regulators will include audio modulation, so via front panel controls I may adjust any desired level of grid and screen modulation.  A relay will short the mod tranny secondary and reactor while removing filament voltage from the 4-400 modulators, when using grid modulation. 

This rig will use solid state rectifiers, no xenon or MV tubes.  I plan to use a string of 10 amp, 1000 piv diodes.  I have been told that it is not necessary to use equalizing resistors and capacitors with the newer quality diodes.  Not sure if this is true, I will find out either by experiment or advice...   I will of course design conservatively - lots of diodes, MOVs on the input, and step start circuitry. 

Sorry to hear the fire killed your screen mod rig, would be interesting if you were able to finish it.   One of the things I want to do is determine actual bias levels, and modulation swing needed on control and screen grid to get linear modulation, and to sweep the frequency of the rig in both grid and plate mod modes.  I have no idea what swings will produce a linear modulation, but at least I have an idea what biass values to start with, thanks to Slab Bacon's input.

73, Rick

[w8khk]

 The big advantage of using the reactor was that it practically eliminated the talkback from the VM-5. 

Hello Don,

I wonder if the Stancor 3899 has a reputation for talk back?  If it does, I assume the reactor would help.  I am thinking about shock mounting the mod tranny and reactor to minimize this effect. 

This brings up one other question I have had for some time.  I wonder if I should insulate the mod tranny and reactor from ground, and let them float, to reduce the risk of high-voltage breakdown between windings and core.  Of course I will wire the mod tranny secondary to float at PS output DC level, not at ground potential. 

There will be no safety issues floating the iron, as the equipment will be fully enclosed and interlocked, and sufficient clearance between iron components.   How would you do it?

Thanks for your input....

73,
Rick

[Steve - WB3HUZ]

FACT: 1000 watts am modulated 100% with a sinusoidal tone is 1500 watts raw bulk rf power delivered to the antenna.

Only if you have it hooked to an antenna. 

No matter what it's hooked to, with the setup you will be producing 1500 watts of average power. But since the FCC is only concerned with Peak Envelope Power (PEP), vis-a-vis the amateur radio power limit, this fact is irrelevant.

[W2XR]

For plate modulation I expect a suitable reactor in series with the screen would be sufficient; I am thinking a high-end audio output transformer here, perhaps a UTC LS-58.

Screen and bias supplies will be electronically regulated and swamped, adjustable so that the biass may be set to work with any reasonable plate voltage, meeting the FeeCee power reduction requirements.  73, Rick

Hi Rick,

Be sure that the LS-58 output transformer (or whatever impedance device you select for a screen audio choke) can handle the unbalanced DC and still maintain the value of inductance you require to provide the desired low-frequency response. The LS-58, like most audio output transformers, is designed for push-pull operation, and cannot handle the 50 to 100 ma. of unbalanced current that the screen of a typical high-power tetrode tube or tubes will pull in class C plate modulated service, without a radical drop in it's inductance.

In my homebrew rig, which consists of a pair of 4-400As plate & screen modulated by a pair of class B 833As, I use a 30 hy choke as my screen reactor. It is mounted above ground on four ceramic stand-offs, to reduce the distributed capacitance losses to ground, thereby improving the audio high-frequency response of the screen circuit. Based upon the screen impedance (ratio of screen voltage to screen current) when the final is loaded to normal plate input, the predicted/calculated -3 dB low-end response with this value of inductance is about 30 hz. This is more than adequate for really good high-fidelity quality.

I see no point in regulating the screen supply when plate modulating tetrodes or pentodes. It is not done in commercial broadcast transmitters, as there is no technical advantage in doing so, and will only serve to drive the cost, complexity, and reduce the reliability of the design. Besides, once you stick the screen reactor at the output of a regulated screen supply, the series impedance and DCR of the reactor will negate any power supply regulation.

Additionally, you do not want to have a stiff or regulated bias supply for the grid bias for the final amplifier, when utilizing plate modulation. A portion of the bias voltage (about one-third of the required operating bias for class-C plate modulated service) must come from the fixed bias supply (for protection of the final amplifier tube or tubes in the absence of grid drive), and the remainder must be derived from a grid-leak resistor that is in series with the bias supply and grid circuit. This arrangement provides for much more linear modulation of the final amplifier, and further details can be found in any the ARRL Handbooks from the golden age of AM.

Good luck with the new rig. There is nothing in the world like running a high power transmitter that you designed and built yourself. Well maybe I'm somewhat overstating this............let's say almost nothing in the world.

73,

Bruce

[W2XR]

 The big advantage of using the reactor was that it practically eliminated the talkback from the VM-5. 

 

This brings up one other question I have had for some time.  I wonder if I should insulate the mod tranny and reactor from ground, and let them float, to reduce the risk of high-voltage breakdown between windings and core.  Of course I will wire the mod tranny secondary to float at PS output DC level, not at ground potential. 

There will be no safety issues floating the iron, as the equipment will be fully enclosed and interlocked, and sufficient clearance between iron components.   How would you do it?

Hi Rick,

I float the modulation transformer and modulation reactor in my rig above ground, primarily to eliminate any potential failure issues of winding-to-core voltage breakdown. Each device is mounted on ceramic stand-offs so they stand way above frame ground. The modulation transformer is also protected on both the primary and secondary sides with spark gaps.

Very cheap and easily implemented insurance, considering the difficulty nowadays of sourcing good quality modulation iron.

73,

Bruce

[w8khk]

Bruce, thanks for the suggestions.  I will definitely float the iron on porcelain. 

I had not thought about the unbalanced DC in the hi-fi audio transformer used as a screen choke, guess I better look further for some appropriate iron here too. I also had not thought about distributed capacitance losses, good point indeed.  I planned to provide regulated bias to run as either screen or plate mod, but with all the discussion it seems to be getting a bit complicated, so I should probably stick to either one or the other. 

Based upon Tim's point on heavy duty RF components, I went back to the book and looked at tuning capacitor spacing.  I have 1/4 inch spacing on the cap, according to Orr this is good for up to 3500 vdc for cw or ssb, with no dc on the cap. 100 percent modulated would double the peaks, and it appears I would need to go with either 1/2 inch spacing, or use a vacuum cap instead.  I wonder if this is being too conservative..... 

Thanks also for the advice on the combination of fixed and grid leak bias.  On the rigs my grandfather ad father  had, only grid leak bias was used, but they were built in the 30s and 40s.  Certainly safer to have fixed bias to protect the rig in the event of loss of drive. 

Still many options to sort out.  But all the suggestions here encourage me to do more research and try to get it right the first time.  With all the screen issues that have come to light when plate modulating tetrodes, I may still go the route of screen/control grid modulation on this rig, and get on and join the fun.  Still have several other boat anchor rigs to restore, and it just may be easier to plate modulate the triodes- either the pair of 450THs or 304TLs.  If I use the mod tranny in a rig that could be screen modulated, then I will be short a tranny for the triode rig....

BTW, still the open question on whether I should bypass each diode in the string with a cap and resistor, or whether that is no longer needed with modern diodes.  If no advise, I will try it without any equalization components, if I zorch a few, they are cheap.

73, Rick

[w4bfs]

Hi Rick ... If the diodes are from the same production lot, generally no eq/surge parallel stuff needed ... modern mixed production diodes do well with just eq res ... snubbing caps/mov's should generally go across windings for spike protection, if you can find parts with sufficient ratings ... easiest thing is just to use more diodes in series, if you can tolerate the V drop ...

I'm still considering my refill of a Gates 250gy.  I have a Dahl mod xfmr from Gates BC-1 (833's mod by 833's) rated for 750 W audio.  With a ratio of about 1.86:1 and a tertiary winding, the plate mod option is very attractive.  I am currently considering 2 rf decks (one mf and the other hf) to give 10-160 mtr coverage.  Pretty unorthodox, I guess.  The strange thing is that the stock panel hole location supports this notion.  73   John

[W2DU]

Hello Don,

Seeing your mention of the VM-5, and Rick's mention of TZ-40s, challenged my long term memory back to 1933 when I was first licensed as W8KHK. From then until 1936 I operated 40m cw exclusively, first with a Hartley '10 osc, and later adding a 203-A. In 1936 I began grid modulating the 203-A as my first foray into AM on 160m. But in 1937 I built a new rack-mounted rig with a 59 tri-tet osc, 6L6 amp, followed by a 35-T driving a 250-TH, modulated with four push-pull parallel TZ-40s and a VM-4.

I guess the main reason for writing is that you said your VM-5 gave talk back, while mine gave no talk back at all, even when running full tilt.

I regret it like mad now, that I deep-sixed the entire TZ-40 mod unit with the VM-4 when I retired from RCA in 1980 in NJ and moved to Florida. At that time I didn't think I'd ever use AM again. How wrong I was. However, I saved the four TZ-40s, which Rick now has and plans to use, even though they're more than 70 years old!!!

I'm going to attach a coupla pics of the 250-TH rig I built in 1937.

Don, I want to thank you for the excellent advice you've give Rick. In the earlier days I was Rick's Elmer, as well as his brother Bill's, W2WM, but now it's the other way around, I'm now learning from them. Rick wouldn't tell you, but I will. When he retired from Agilent a coupla years ago he was one on Agilent's top-level computer engineers, coming up with many innovative techniques in LAN and global networking used on their own world-wide network.

Walt, W2DU (Rick and Bill's Dad)

[W2DU]

The other pic of the 1937 250-TH rig was supposed to be attached to the above post, so here it is.

Walt,W2DU

[w8khk]

Hi Rick ... If the diodes are from the same production lot, generally no eq/surge parallel stuff needed ... modern mixed production diodes do well with just eq res ... snubbing caps/mov's should generally go across windings for spike protection, if you can find parts with sufficient ratings ... easiest thing is just to use more diodes in series, if you can tolerate the V drop ...

Thanks, John.  I have over 100 of the 10 amp 1kv piv diodes, all from the same lot, and plan to use at least ten, maybe twelve, in each leg of the center tapped full wave supply.  Voltage drop is insignificant considering the plate voltage, which must be reduced from maximum available via a variac, to stay within ratings of the other components, and the FeeCee limitations.

FB on the Gates BC rig.  When I first started working on my BC rigs, I had a notion that I should keep everything as stock as possible, but the more I get into it, the practical modification route wins.  Moving to 75 meters requires lots of changes.  Interesting idea to have two separate RF sections for different band groups.   Lots of room in the old BC cabinets for customization!   I had a slightly different approach in mind - since my second RCA is missing most of the iron, I thought I would set the two side-by-side, and use the HV PS and Mod from one, to drive the RF sections in each one, just by removing filament power from the final that is not in use.  Switching from one band to another would be very simple.

73,
Rick

[KX5JT]

What oh what has become of this project?

[w4bfs]

big projects are sorta like pregnancy .... a big commotion at first ... a 9 month wait ... then BOOM !

[w8khk]

What oh what has become of this project?

Thanks for the interest, John.  The project is not dead, just sidetracked for a while.  It will definitely be screen modulated, primarily because the plate tuning cap plate spacing is not sufficient for plate modulation.  If it is not considered a sacrilige, I will probably feed unmodulated RF from a Kenwood TS-830.

Several familiy illnesses have kept me out of town for weeks at a time.  Another delay is the "ready to operate" Valiant I picked up in Huntsville this summer, turned out it was in need of major repair, so I stripped it and it is almost completely rebuilt with many mods and upgrades.  Will finish this project before going back to the 4 by 1 rig.  Hope to have it finished soon enough to enjoy the heat output during the winter!

[WA1GFZ]

Hey Walt,
The girls must have loved that flip in your hair...

[w4bfs]

hey Rick ... I have almost finished development work on my screen modulated Viking II .... came up with a way to run straight screen mod or 3 or 6 db of controlled carrier in a manner similar to bias shift Heising .. winds up being a hybrid design ... hoping to test soon .. 73 .. John

[w8khk]

Sounds cool, John.  I have not yet touched the Viking II that followed me home from Huntsville.  I would be interested in what you are doing with yours, maybe I will try it too, after the Valiant and 4x1 are finished.