Returning to the am high level modulation

[KD1SH]

  Okay, looking at the Eimac notes, it's a little clearer now. When Eimac talks about obtaining screen voltage from the unmodulated B+, the assumption is that this voltage is sufficiently higher than what the screen would require, such that a very large value dropping resistor would be needed.
  The Eimac specs linked to - for the 4-65 in this case - show a max plate voltage of 2500 volts and a max screen voltage of 400 volts. That's 16% of the plate voltage. In the "typical operations" parameters, they're calling for only 250 volts on the screen, so now you're down to only 10% of your plate voltage; thus you've got to drop 90% of your plate voltage across that dropping resistor. At only 16ma of screen current, that's a pretty high value dropping resistor - about 140K I think.
  As K9MB pointed out a few posts up, such a high value dropping resistor will allow the screen to self-modulate. Personally, I think I'd feel better using a choke along with that resistor, just to insure that the filter caps in the B+ supply aren't going to shunt away your higher audio frequencies, but with such a high series resistance you'd probably be fine.
  Now here's the rub: using your idea of supplying the screen directly from a fixed, regulated source of much lower voltage, with either a vastly lower dropping resistor or no dropping resistor at all, there's no appreciable series impedance to prevent audio frequencies from being shunted directly to ground. You will have in effect clamped your screen firmly to a fixed DC level. Fine business for a linear amp, but for plate modulation, not so much. Your scheme would work with a series choke, though, assuming the reactance at audio frequencies is high.
 

[K9MB]

  Okay, looking at the Eimac notes, it's a little clearer now. When Eimac talks about obtaining screen voltage from the unmodulated B+, the assumption is that this voltage is sufficiently higher than what the screen would require, such that a very large value dropping resistor would be needed.
  The Eimac specs linked to - for the 4-65 in this case - show a max plate voltage of 2500 volts and a max screen voltage of 400 volts. That's 16% of the plate voltage. In the "typical operations" parameters, they're calling for only 250 volts on the screen, so now you're down to only 10% of your plate voltage; thus you've got to drop 90% of your plate voltage across that dropping resistor. At only 16ma of screen current, that's a pretty high value dropping resistor - about 140K I think.
  As K9MB pointed out a few posts up, such a high value dropping resistor will allow the screen to self-modulate. Personally, I think I'd feel better using a choke along with that resistor, just to insure that the filter caps in the B+ supply aren't going to shunt away your higher audio frequencies, but with such a high series resistance you'd probably be fine.
  Now here's the rub: using your idea of supplying the screen directly from a fixed, regulated source of much lower voltage, with either a vastly lower dropping resistor or no dropping resistor at all, there's no appreciable series impedance to prevent audio frequencies from being shunted directly to ground. You will have in effect clamped your screen firmly to a fixed DC level. Fine business for a linear amp, but for plate modulation, not so much. Your scheme would work with a series choke, though, assuming the reactance at audio frequencies is high.
 

Very well summarized. You made all the points.
I also favor the regulated screen voltage feeding a 10-12H choke. That way, you have a stable low impedance DC supply with a very high audio impedance,
The only other thing is to choose the screen bypass so that it keeps the screen close to ground for RF while not cutting the high frequency response too much. Published values are between .002 and .005uF I think.

[Opcom]

Hi Mike, long time no hear!

833A's are a great tube, too bad they look like a skinned rabbit to me.

849's and 851's (aka hen's teeth) are lot more elegant.

But I digress...

73DG

A nice alternative is the 3-500Z. I'd put its looks somewhere in between the above mentioned and they are inexpensive. I used to think they were hard to drive in modulator service but I was doin it rong using a GC-512. Edcor provided an off-label but perfect solution.

Don't discount some of the nicer screen grid tubes. The 860 and 861 are very coat-and-tie in appearance.

PJ

[Opcom]

Y'all have covered 99.9% of everything by the time I got around to answering. Good job! Sleeping late has its advantages.

Some things:
480mA for the screens means several tubes. Five 4-1000 or Qb5/1750 in parallel will be interesting due to plate capacitance vs tuning circuit, possible instabilities, unpredictable parasitic elements from wire lengths etc. What to do? Perhaps mount them in a circle with plate choke in center and PI-L above?
Would it not be better in some ways to use a single tube? This is a monstrous transmitter considering a single plate-modulated 4-1000 or Qb5/1750 easily make a 1KW carrier. The 4CX5000A comes to mind and can be rebuilt. It doesn't glow unfortunately.

===================

Just want to add to some opinions of each screen grid voltage modulation method:

1.)
The resistor from unmodulated B+ is merely the cheapest way to modulate the screen. It is not always the best way depending according to the tube characteristics.

The inexpensive solution of the resistor works better when the plate voltage is much higher than the screen voltage. Here we have, in the 4-1000 or Qb5/1750, some 10:1 ratio of plate to screen voltage @ carrier.

Small tubes like the 807 have a ratio of more like 2:1 and so the ratio between the screen grid voltage dropping resistance and the actual load presented by the screen grid is quite lower, not allowing as much screen voltage modulation to take place.

2.)
Choke use with separate G2 supply:
The use of a separate screen voltage supply in the case of low P:G2 voltage ratio tubes discourages the high screen grid voltage dropping resistance because by the time enough audio voltage was developed across the resistor due to screen current, there would be too little DC screen voltage left to operate the tube at the proper plate current.

In that case a choke is great, having a low DC resistance but a quite high impedance at audio frequencies. Little of the G2 DC voltage is lost across the choke but much AC voltage is developed across the choke from the effect of the fluctuating screen current. So, the DC screen voltage level is maintained properly while enough audio voltage is added by the choke to modulate the screen.

There is some math that can be done for the choke value depending on DC screen voltage & current supply and the lowest audio frequency one wishes to use. The choke scheme can be applied to any screen-grid tube if a separate screen voltage supply is used. I didn't look up the formula but it shoud be in books and books.

Here is Eimac general knowledge:
https://www.cpii.com/docs/related/22/C&F3Web.pdf

Here is an example of two different chokes with a 4-1000A:
http://www.arizona-am.net/PHOENIX/KB7OCY/gm/globe_master_man03.html


3.)
The separate screen winding on the modulation transformer serves the same function as the choke solution, except that it avoids the frequency-dependent phase shifts inherent in the choke modulation scheme, so that the screen modulation is always closely in phase with the plate modulation. It also avoids frequency-dependent screen voltage modulation effects that arise because of the choke reactance, so that the screen modulation depth is constant over the audio range.

4.)
The cathode-follower type of screen grid modulator is not common but the adjustments provided for linearity by adjustment of the G2 modulating audio voltage independent of the plate modulation audio levels, and carrier setting via G2 voltage are interesting. Maybe it is a good replacement for the much-bemoaned non-existing screen winding on mod iron.

================
A side note only:
I found it interesting that among two functionally identical FAA type AM transmitters covering 220-400MHz with two 4x150s @890VDC plate, making a 100W carrier, that the modulator one version used 811As with Freed mod iron, and had a screen mod winding, and the other version had 4-65s with Collins mod iron and used the cathode follower type of regulator. Maybe this means exactly nothing. Do not detract from this topic, but the manual is here:
https://www.bunkerofdoom.com/lit/mil_uhf_01/index.html

[KD1SH]

RF power tubes are like wives: if you insist on having more than one, keep them in separate houses and don't let them know about each other .

Would it not be better in some ways to use a single tube? This is a monstrous transmitter considering a single plate-modulated 4-1000 or Qb5/1750 easily make a 1KW carrier. The 4CX5000A comes to mind and can be rebuilt. It doesn't glow unfortunately.

Yes, I was thinking about that, too: tubes like the 6146, 807, 4D32, etc... present a different challenge, with their lower screen/plate voltage ratio.

Small tubes like the 807 have a ratio of more like 2:1 and so the ratio between the screen grid voltage dropping resistance and the actual load presented by the screen grid is quite lower, not allowing as much screen voltage modulation to take place.

[K9MB]

RF power tubes are like wives: if you insist on having more than one, keep them in separate houses and don't let them know about each other .

Would it not be better in some ways to use a single tube? This is a monstrous transmitter considering a single plate-modulated 4-1000 or Qb5/1750 easily make a 1KW carrier. The 4CX5000A comes to mind and can be rebuilt. It doesn't glow unfortunately.

Yes, I was thinking about that, too: tubes like the 6146, 807, 4D32, etc... present a different challenge, with their lower screen/plate voltage ratio.

Small tubes like the 807 have a ratio of more like 2:1 and so the ratio between the screen grid voltage dropping resistance and the actual load presented by the screen grid is quite lower, not allowing as much screen voltage modulation to take place.

I think that these high gain small tetrodes are naturals for a straight dropping resistor from the plate voltage that is modulated. The lower natural value will not lower the amount of modulation and a coupling capacitor in parallel with the dropping resistor will form s capacitive divider with the screen bypass capacitor and that ratio can be used to adjust the amount of audio coupled to the screen in phase with the modulating audio on the plates.
Bigger tubes like 4-xxx and 813s, etc will have plate voltages near 2kv and regulated screen with choke becomes a better option, IMO.
No one size fits all, I think..

[AMLOVER]

Would it not be better in some ways to use a single tube? This is a monstrous transmitter considering a single plate-modulated 4-1000 or Qb5/1750 easily make a 1KW carrier. The 4CX5000A comes to mind and can be rebuilt. It doesn't glow unfortunately.

Αs the time passes and we become old, we get pleased more with looking and less by using what is sitting on the shelves.
The glass tubes have a very 'hot' look in comparison with the ceramic ones.
The challenge to tame them when they are in big groups is also much more attractive.
Looking to put some salt in my retirement life, decided to recall the memories of youth.
In that time we used to build 6 to 8x807s modulated by 2 to 4x811s, later 6 to 8x813s modulated by 4 to 6x811s and before the ceramics some of us tried the 4-400s and 4-1000s.
For few years i used ceramics in high level am with 4cx3000, 2xGU100B and lately in AB1 linear (4cx15000).
My mind though is always traveling back and dreaming the glass tubes with the warm almost orange-red filament lighting.
The last two years I have also touched my tongue into the honey of the super efficient PWM, in very high carrier level, too.
The efficiency of the PWM and the top signal reports of the powerful linear couldn't quench the fire of the glass tubes filaments in my youth memories. They are always there and still flaming.
Decided to return and found out those QBs in very low prices 40-80euro per piece as far tetrodes aren't fortunately attractive to HiFiers.
Having the 10kvdc supply from the linear that is built by 6 transformers in series, decided to cut it in the middle and create two hv supplies with 3 transformers in series each in order to use one for RF part and the other for AF part.
The modulator I have is 2:1 (2x1200t to 1200t), the Rload of 4xQBs in B class push pull with 4Kv is 8K and they'll offer about 4500w peak which has to be 50% of the class C final consumption.
Using 6xQBs in the class C rf part at 4kv/2A will have an Rfload=2K and so I believe that all parties including the modulator will be happy. My reactor is only 15H/3A but I'll go with this even small I think in inductance.
I have all the main parts except of 2 sockets SK510 that I'm looking for in the net.
The idea to use part of the plate high voltage for feeding the screen came from stacking 3 transformers in order to obtain 4kv plate voltage. I will go with that but using the screen reactor, too.
I had never problems with parasitics or self oscillating by parallelizing any number of tubes.
I always use 47-100 ohm resistors to each grid and each screen plus the parasitic filter (resistor & coil) on each plate.
It is not a monstrous transmitter in comparison with one tube as far the lethal voltages are remaining in the same level. The currents are only multiplied with the need of stronger wiring and heavier duty parts.
Could you please inform me where is the place in the forum to ask for the sockets I need? I couldn't find it and my question was moved from the technical info department.
Thank you.

Stefano

[w3jn]

Post an ad in the "wanted" section here http://amfone.net/Amforum/index.php?board=34.0

Are you in Greece?

[K9MB]

I just put this link on his other thread for some in Greece
Good price too. 85Euros each

https://antreas555.webs.com/socketsrfparts.htm

[AMLOVER]

w3jn

Yes, i am in Greece. I remember you in the past you were listening to us sometimes between the broadcast and the lower amateur band. Are you still in Athens?

http://amfone.net/Amforum/index.php?board=34.0

I have done it, thank you.

https://antreas555.webs.com/socketsrfparts.htm

K9MB

I have answered you about this webpage in the other topic.
Thank you.

Stefano

[w3jn]

Stefano, No, I left in 2012.  I loved my time there, one of my favorite places in the world!

You're correct, I loved listening to the pirate/unlicensed cacophony above the broadcast band.  Broadcasting music, or diatribes, and/or multi-party QSOs.  Most had excellent signals and audio.  You couldn't get away with that here in the US.

[AMLOVER]

w3jn,

Things have changed a lot from 2012. The romance is lost, PWMs with ugly looking but excellent modulation are now the leaders of the band.
Only few of us have remained old times lovers of that sweet cacophony and return back to the filaments in the first chance.
I wish you all the best.
Thank you.

Stefano