driving 3-500X modulator with no transformers

[Opcom]

This is just a study right now, with the goal of fidelity similar to a BC transmitter.

1. eliminate audio transformers except modulation transformer.
2. provide for biasing each 3-500Z separately
3. provide for same kind of balanced 'voltage divider' feedback as used in BC transmitters
4. all tubes please..

Complications:
3-500Z are not easy to drive. 100 peak volts and 700 peak mA to the grid.
Feedback has to be sent to a low level 'voltage' stage.
DC coupling issues complicate the cathode follower circuit.
Push-Pull signal imbalance should be reduced.

It's overly complicated now but should be able to be simplified. A hard part is making a 200V peak to peak or more signal to drive the 6080's with. Usually a transformer would be used. Two 12BQ6 tubes as a long tailed pair should provide the gain and high voltage as well as being a phase splitter.

[N2DTS]

14 tubes, I like it.

[WZ5Q]

Howdy Patrick,

Very interesting project. I will like to see how it unfolds.

I am going 180 degrees in the other direction, as I will be using absolutely NO feedback on the Mod Tubes.
Instead, I have very high quality Tribute, Lundahl, & Peter Dahl iron for the Input-Phase Splitter/Interstage/Driver transformers and will be using Directly Heated Triodes in push-pull for all the stages of the Speech Amplifier/Driver.

If you need some 6080's for your project, I think I got a big stash of them stored in a box some where.

73,

[Opcom]

Could be a contender for 'most tubes in a speech amp'. I really don't want the final circuit to be so complex. I do not have access to those audiophile brands of transformer but there are others here good enough if used at reduced power.

There are at least two ways to design something:
1.) Trying to visualize the most accurate circuit then decide where to compromise.
2.) Just put it together with the most simple circuit then decide where to improve it.
The second way works but not given good results when built.

The first section before the 12BQ6s is a copy of the GAP/R op-amp, and the resistors must be precision, low noise, and stable.
risks:
1.) High impedance 1V audio input jack on a full size transmitter, but a 600 Ohm line could help there.
2.) Cheating by buying a used GAP/R unit means it's 60 years old and may be out of tolerance, so is any time saved?
3.) The capacitors on the plates of the 3-500Zs were never used in the BC rigs and look like trouble spots. The residual DC without them may upset the op-amp section even with its common mode rejection.

There's a limit to the space in the transmitter. Whatever is done would have to fit on a 19" rack panel 5-6" wide.

The power supply is possibly impractical.
1.) It's easy to just assume the typical mindless power supply but +/-300V, +400, -200, and four filament supplies is going overboard for mounting space.
2.) Heater to cathode voltage rating on the 6080 is +/-300V. The drive is +/-200V max. So, should there be separate heater transformers or not?

Long Tail Pair 12BQ6s as high voltage amp
1.) The 12BQ6 was considered because there's a bunch of them on hand and a big tube is not needed.
2. ) 12BQ6 will take a lot of plate voltage, but its screen only needs 150V unlike 807s and other high voltage beam tubes.
2.) It would be better to run 12BQ6s on 500-600VDC rather than on 400V sharing the 6080 plate supply, but I cheated already.
3.) In a long tail pair the beam tube or Pentode, with higher voltage gain than a Triode, just looks like a better phase splitter.
4.) The 1.3k plus 8K cathode return resistance seems low for a so-called long tail pair, but the 12BQ6s each use 25mA (Pd=6W) in the diagram there.

- -Which takes us to the disadvantage of swinging the plate voltage down to 100V, and to advantages of using a 500-600V supply and swinging only down to 200V.
1.) When run on 400V, the lower plate swing limit need be 100V in order to provide a 300V peak to peak plate swing for the 6080 grids, and before 100V is reached on the 12BQ6 plate, the G2 current increases to 25mA, and that upsets the circuit more than if it had remained fairly constant, meaning it is in the distortion area.
2.) If the 12BQ6 plate is run at 500 or 600V and 20 or 15mA, the disturbances in G2 current can be avoided. Above a plate voltage of 200V, the G2 current looks steady at 15mA.
3.) The answer to this is that the screen current should be kept out of the cathode return path so as not to dilute the feedback signal generated by the long tail. For this reason, a floating 150V regulated screen supply was added there.

I think I have enough 6080's for now there's a box of them, and some 6336's if I want to do that. That section is a copy of the CQ 1955, page 15 article.

I have a very good reason for wanting to use feedback from the modulator plates. I have used no feedback at all, yet. When I sweep the modulator with a resistive load or operating normally with RF into a dummy, I have a poor frequency response and a peak around 1KHz. It can be seen wither on the 3-500Z plate or on the secondary but not on the grid. Although listeners say the rig sounds great, when pressed to be critical the friends with better-fidelity and modern receivers say it has a 'boxy' sound meaning this peaky response. This has been traced to the 1940's RCA 1KW modulation transformer. The original BC design had the same ladder type feedback going back to the cold ends of the split secondary input transformer.

I think the next evolution of this would be to:
1.) try to be rid of the op-amp, but still have the 65V p-p signal for the class A 12BQ6 grids. No 12BQ6 grid current is to be drawn.
2.) find a way to stabilize the 12BQ6 idling current (current source?) - the circuit has a problem in that the tubes are run near cut off and at low currents. They may be hard to control due to variations or aging.

[Opcom]

Allowing one transformer to sneak into the circuit to take care of the input voltage gain and handle feedback seems to do away with half the tubes and power supply voltages.

The attached circuit is surely only a small compromise. Assuming both would work, no one could ever hear the difference over the air and it is much more like the old BC transmitter circuits.

All of the feedback is between the transformers, not through them. There is no mismatched inter-stage transformer driving class B grids.

I need to revise this some more. A model of the 12BQ6 for LTSpice would be useful. That stage is a question mark.

I learned from the RCA 1-K schematic that the 2E26 used in the same way needs much less bias to keep a low plate current with the same plate voltage. Learned a lot of stuff so far on this little exercise.

[WZ5Q]

Good quality Iron sure makes things nice and symmetrical, as long as there is no evil feedback. 
Sounds like your really crunching the numbers over there Patrick.

snip...
I have a very good reason for wanting to use feedback from the modulator plates. I have used no feedback at all, yet. When I sweep the modulator with a resistive load or operating normally with RF into a dummy, I have a poor frequency response and a peak around 1KHz. It can be seen wither on the 3-500Z plate or on the secondary but not on the grid. Although listeners say the rig sounds great, when pressed to be critical the friends with better-fidelity and modern receivers say it has a 'boxy' sound meaning this peaky response. This has been traced to the 1940's RCA 1KW modulation transformer. The original BC design had the same ladder type feedback going back to the cold ends of the split secondary input transformer.
snip...

That's interesting...
Do you have a picture of the sweep?
What was the model of the RCA the mod iron came from?
How is the mod iron secondary configured with the reactor?

I have an oil filled one from a BTA-1R in the K1JCL Homebrew Rig (4x1 modded by 3-500z's) here.
I should have that rig on the air this coming winter, should be interesting to see how it plays.

73,

[N2DTS]

Seems like you are having fun.
2e26 in audio service?
I have a bunch of 6080's but only used them in a voltage regulation circuit.

How much audio power do the 3-500z's need to produce?


Its likely smart to not put the feedback around the transformers with all their phase shift.

I do not think any mod transformer/modified hy zing circuit is flat.
The cvm-5's I had were very poor, the Thordarson was not too bad, the RCA/Harvey Wells I have now is the best I have had, but you can hear the difference with the screen modulator, the EQ needed is much different/less.
But if you put feedback around the iron, you have to limit it in frequency and level.

Small amounts of space and good iron do not mix, and hifi circuits of any power always get very complex...

Can the power supply be remoted?
Its easy to get various voltages from bridge AND center tap setups and VR tubes and dropping resistors.

3-500Z,s seem like a hard tube for audio...

[VE3AJM]

I've been intrigued by the 1955 CQ article on the cathode follower circuit to drive class B grids, the Gates BC-1 807 cathode follower, the 6BG6 circuit, and your circuit. I use a pair of 3-400zs because I had a bunch of them, as the modulator in my GPT-750. I used the Eico 730 as the audio driver. I bypassed the inboard Eico mod transformer and used a 30 watt vintage Hammond 1703 class B driver transformer that is tapped in from both sides in a Williamson connection. I loaded the secondary. I used a 9V or there abouts zener diode in the modulator filament centre tap for bias. The 3-400s idle at 90ma. My scope pattern looks very good and I limit my audio to about 8kc total, but if I could eliminate the driver transformer completely, that would be better. I use a 1:1 1kw mod transformer out of a broadcast tx and heising reactor. The Clegg Zeus VHF AM tx uses a cathode follower audio driver to 811A class B grids. I think its a 6BL7 or 6BX7 dual triode, little brother to the 6AS7 or 6080.

I remember Richard W3NCC in Pittsburgh PA used a pair in his HB modulator and transmitter, and they sounded great late night 75m.

Have you had any issues using a common filament transformer for the cathode follower 6080s with isolation etc. What is the modulation transformer etc. that you have available to use. Thanks.

Al VE3AJM

[W9ZSL]

Just a suggestion.  You should check out the modulator system for the Gates BC-1T transmitter.  The driver is a pair of 807's where their cathodes are wired directly to the grids of the 833 tubes with nothing but a 47 ohm resistor between them.  Feedback is via a ladder on a PC board across the 833 plates and is returned to the audio input transformer.  It is quite similar to the schematic you have.

[VE3AJM]

Just a suggestion.  You could check out the modulator system for the Gates BC-1T transmitter.  The driver is a pair of 807's where their cathodes are wired directly to the grids of the 833 tubes with nothing but a 47 ohm resistor between them.  Feedback is via a ladder on a PC board across the 833 plates and is returned to the audio input transformer.

If you have a look at the schematic that he provided here on this thread, he has used a very similar negative feedback network off the plates of the 3-500z modulators back to the input. Just no audio input transformer used in his circuit. I found an Altec 1570b audio amp using class B 811As in the output, driven by a pair of 6W6s. It looks like cathode follower drive as well.

Al VE3AJM

[WD5JKO]

I love this thread.....

The first schematic in this post appeared to have a little from the late Fred Nachbaur in his creation called "Dogzilla". The use of NE-2 lamps in DC coupled tube amplifiers is quite interesting.

Full Schematic:

http://www.dogstar.dantimax.dk/tubestuf/graphics/dz-full.gif

Full featured web pages:
http://www.dogstar.dantimax.dk/tubestuf/dzindex.htm

Jim
Wd5JKO

[Opcom]

The modulation transformer is from a RCA 1-K transmitter. That transmitter uses two 833 tubes to modulate two 833 tubes.

Mine is two 3-500Z modulating a 4-1000 with this same transformer. It should be a good match if the voltages are about the same.

The audio sweep testing was done with a resistive load and no DC in the secondary, and with normal operation of the final. There was no important difference with or without the reactor, perhaps because I am not running the almost half amp of current needed for a whole KW carrier output, only that needed for 350W.

I did not take pictures of the scope patterns. I do not use a reactor at this time. There is not room in the transmitter for it because of the large plate transformers and I don't like things with high voltage lying on the floor behind it. I don't want to remote the power supplies for this transmitter.

The 3-500Z only have to produce 600W at the most. I have tested and gotten over 700, but the 30W driver was unable to make more due to the CG-512 class B driver transformer being a very poor match for 3-500Z grids. The original tubes in the modulator were 304TL or TH.

The CG-512 wants a load of about 4x the impedance that the 3-500Z grids look like. That is why I am using a 150W amp for the speech amp. It's already amazing how good it sounds to others with that bad kludge going on.

The 3-500Z is a special tube to drive, but only because the best iron for it is rare or maybe does not exist. The peak grid current at 100V peak grid volts is something like 700mA. This do not so closely agree to the two sets of Eimac curves (1968, 1980), being Chinese tubes anyway who knows what.

Edcor makes a 140VCT (RMS) speaker-to-line transformer. That would be a possible choice. It is like any 70V or 100V transformer. I belive this model has to be special ordered. It is center tapped, so 140V RMS is 200V peak to peak, perfect for the 3-500Z at full drive level.

I show a common heater transformer because the heater to cathode rating is +/-300V. I biased it to +100V in the diagram, just had to put the heater volts somewhere.

In the BC-1T, I wonder what the peak grid current is. For the sake of understanding the plate supply, what  is the B+ for the 807 plates (TP2-1 or TP2-2)? Obviously the 807's can take this duty but I feel sorry for their cathodes.
Nachbaur; too bad he's gone.

[W9ZSL]

The 807's in the BC-1T operate at 600/650 volts nominal.  The modulator tubes operate at 80 MA  plate current for the pair with no audio applied for the 1KW version.  The cathode follower 807's with 580 V plates and 195 v screens runs 70 VDC cathodes.  Couldn't find any info on the grid current form the manual. Bias voltage on the 833's is 65/70.