Need help interpreting tube plate curve

[Solidus]

Okay, so to give a little background, I have a couple 4-125A tetrodes laying around as I've had a bit of a collector's hobby with collecting various tubes.

The 4-125A is the smallest of a series of air-cooled RF transmitter tetrodes, the others in the series being the 4-400A and 4-1000A. The 4-125A is rated at 125W dissipation, 400 and 1000W respectively.

So, I've decided to develop them into a stereo hi-fi amp. Probably not the smartest thing to do with RF tetrodes, but as they're gathering dust on a shelf, I figured I may as well tinker with them. AF builds with them have been done, although information on them is damn-near nonexistent so I'm working a build from scratch.

I already have a supply capable of supplying 2kV to the plates, but the question of an output transformer comes into play. I could wind my own (I have all the equations and what-not to properly calculate), but the immense amount of winding is a bit of a discouraging factor, not to mention I'd have to rip apart existing trafos for the laminations.

Now, here's my issue:

I used the datasheet to determine that a good starting point (and good in terms of power supply) would be to supply each of the two tubes with a 2kV plate voltage at 200mA, biasing at -100V per the data sheet. If I have any of the math correct, that would give a plate impedance of about 10K nominal.

Problem is, the Hammond output transformer I found with a 10K primary (which is in itself a costly and hefty beast) is only rated to 90mA safe. I'd be fine if I were having to put 120mA across it, but 200mA is asking for a smoke machine.

Point being, I can't use it.

So, what should my next step be? I need help interpreting the datasheet to determine appropriate points.

Here is the link to the data sheet:

4-125A Datasheet

 

[duke37]

You could use two identical Hammond transformers connected in parallel but would they handle the voltage?

The 4-125A will, I believe, need a stiff screen supply which will source and sink current otherwise they go bananas.

 

[john monks]

You can do as you suggest except you must apply a greater negative bias voltage to the control grids. Now you should be OK except if you run your amplifier at near full power all the time you still could cook your output transformer. Another thing you might consider is peak voltage on the primary leads of the transformer. If you are running full power and your load should become disconnected you can damage your transformer. And you might already be running your transformer beyond it's specifications. I wouldn't let that stop me. I used to do that all the time. But what I would to is set up a spark gap across the primary of the transformer in order to limit the voltage across the transformer.

You might want to set your bias voltage so you get about 30-50 ma of plate current per tube and set up a negative feedback circuit in order to reduce your crossover distortion.

I hope you realize the danger of 2kV at 200mA. If you come in contact with that, it was nice knowing ya.

 

[Solidus]

You can do as you suggest except you must apply a greater negative bias voltage to the control grids. Now you should be OK except if you run your amplifier at near full power all the time you still could cook your output transformer. Another thing you might consider is peak voltage on the primary leads of the transformer. If you are running full power and your load should become disconnected you can damage your transformer. And you might already be running your transformer beyond it's specifications. I wouldn't let that stop me. I used to do that all the time. But what I would to is set up a spark gap across the primary of the transformer in order to limit the voltage across the transformer.

You might want to set your bias voltage so you get about 30-50 ma of plate current per tube and set up a negative feedback circuit in order to reduce your crossover distortion.

I hope you realize the danger of 2kV at 200mA. If you come in contact with that, it was nice knowing ya.

So, how do I actually interpret the plate curve? It's grid voltage vs. plate voltage, with another dashed line for grid voltage.

So what do you think would be more appropriate starting point with regards to a power supply spec list?

The idea of the power supply is a monstrosity for this project, being about twice as big as the amp itself.. The amount of components included for safety is immense. As the 4-125A operates with a plate cap, I intend to also place on the power supply interlocks where if the top cover is removed a relay disconnects the B+ trafo primary.

Now, I realize this doesn't help me much and if this thing goes south during testing or construction, all the interlocks in the world won't help me. 2kV and 200mA isn't forgiving. It is quite literally no joke when I say that I intend to do multimeter testing on the supply using an extension cord and binoculars.

You could use two identical Hammond transformers connected in parallel but would they handle the voltage?

The 4-125A will, I believe, need a stiff screen supply which will source and sink current otherwise they go bananas.

Two in parallel might make my weight limit for the project go bananas, as I'm building a stereo amp.

Here's what I was thinking of doing about the screen - either bypassing the trafo and dropping it directly from B+ (it takes 350V 167mA) or using power resistors to jimmy-triode-strap it to the input of the output transformer.

I'm calling this the "Amp from Hell"

 

[john monks]

The way I read the datasheet if you are using class AB1 then if you use a plate voltage of 2000 volts and a screen grid voltage of 600 volts and a control grid voltage of -94 volts then you will get a plate current of 50 ma. I think this is what you want. To start with I would suggest using a potentiometer to adjust the grid voltage to get 50 ma per tube, 100 ma for both.
I am a little apprehensive about using the Hammond transformer because I don't know what the specifications are. But I am assuming that it will handle about 4000 volts from one plate connection to ground. You might just have to try and see.

Do you have a part number for your Hammond transformer?

 

[Solidus]

I was planning on getting components for one channel and adjusting to fit before buying the others for the other channel, that way if it doesn't work as it should I'm only half out.

Would any potentiometer work (i.e. 1/4W trimpot)? I know typically in most tubes the grid itself draws negligible current, but I'm not sure in high-power tubes of this sort where the grid has dissipation values.

The Hammond output transformer I was originally looking at was the 1638SE (I believe that's it). A small kink gets thrown in when you realize it's rated to 30W. However, from the research I've done on design of output transformers, it seems most of the wattage is based on the size of the stack. The transformer is clear on a dielectric test to 2kV and is rated to pass 90mA, so 50mA would be well inside of that. It is a bit of guesswork though.

Also does it change the numbers at all that I was planning on operating in class A single-ended? (2 stereo channels?)

 

[john monks]

Class A operation would change everything because the tubes would be half on when no signal was coming in causing the tubes and transformer to become blistering hot. I don't recommend it.
The pots I would recommend would be 100kohm because they will withstand 100 volts easily.
The screen grids draw current therefore they will dissipate power. They are sitting at +600 volts. The control grids will dissipate almost nothing because they are negatively biased in order to suppress the flow of electrons going through the tube.
The transformer should be able to withstand 4000 volts because this is twice the 2000 volt excellerating voltage. This is needed when one tube is on and the other off.

 

[john monks]

Class A operation would change everything because the tubes would be half on when no signal was coming in causing the tubes and transformer to become blistering hot. I don't recommend it.
The pots I would recommend would be 100kohm because they will withstand 100 volts easily.
The screen grids draw current therefore they will dissipate power. They are sitting at +600 volts. The control grids will dissipate almost nothing because they are negatively biased in order to suppress the flow of electrons going through the tube.
The transformer should be able to withstand 4000 volts because this is twice the 2000 volt excellerating voltage. This is needed when one tube is on and the other off.

 

[Solidus]

Class A operation would change everything because the tubes would be half on when no signal was coming in causing the tubes and transformer to become blistering hot. I don't recommend it.
The pots I would recommend would be 100kohm because they will withstand 100 volts easily.
The screen grids draw current therefore they will dissipate power. They are sitting at +600 volts. The control grids will dissipate almost nothing because they are negatively biased in order to suppress the flow of electrons going through the tube.
The transformer should be able to withstand 4000 volts because this is twice the 2000 volt excellerating voltage. This is needed when one tube is on and the other off.

Well, that would really screw things over fast if I was forced to operate in AB1. Finding a transformer of suitable impedance wouldn't be hard, but the fact that Eimac RF tetrodes of the 4A series tend to over-rate on audio (the 4-1000A, while rated for 1000W RF, can be uprated to 3,500W when used as an audio modulator - and safely), and then knowing how that higher efficiency further uprates the output for Class AB1 & AB2, I would be ending up with essentially a guitar amplifier that could be just south of 350W output. I wouldn't be able to find a transformer that could run that high without going into saturation-land very quickly.

The goal of this project is to get two 125-150W channels that I can use to drive 8 ohm floor speakers. The wattage would work out correctly.

If I were to operate into Class AB, that would mean I'd have to scrap using two seperate output trafos (one for each channel) for one large one with twice the dielectric breakdown rating, as well as having to find one with those criteria that could flow twice, if not slightly more, the current.

To operate the 4-125A a cooling system must be used, so the tube running slightly over operating temperature isn't a significant worry. More than half of my power supply is the fan supplies and interlocks that cut the filament supply if the fans fail.

 

[john monks]

I see what your saying. I just looked up the 1638SE and noticed that the primary does not have a center tap. So that leaves out a push-pull arrangement. So now I'm presuming that you want to have two output tubes, one for the left channel and one for the right channel and both running in class A operation. So in that case I believe you still want a 2000 volt B+ and 600 volts for the screen grid. So now your are wondering about bossing the control grid. So the easiest way to do this Is to use a cathode resistor, maybe about 1.8kohm and about 10 watts.

This is not a very efficient amplifier but it will have very low harmonic distortion.
Is this what you had in mind?

http://www.digikey.com/product-detail/en/1638SEA/1638SEA-ND/2358082

 

[Solidus]

I see what your saying. I just looked up the 1638SE and noticed that the primary does not have a center tap. So that leaves out a push-pull arrangement. So now I'm presuming that you want to have two output tubes, one for the left channel and one for the right channel and both running in class A operation. So in that case I believe you still want a 2000 volt B+ and 600 volts for the screen grid. So now your are wondering about bossing the control grid. So the easiest way to do this Is to use a cathode resistor, maybe about 1.8kohm and about 10 watts.

This is not a very efficient amplifier but it will have very low harmonic distortion.
Is this what you had in mind?

http://www.digikey.com/product-detail/en/1638SEA/1638SEA-ND/2358082

What voltage does that bias to?

And yes, this is exactly what I had in mind. I plan on driving each stage with either one or two 12AU7s, each running pretty hard (not as much as to cause distortion, but to give me about 2-4W driving power). It doesn't matter to me if I have to sacrifice power as long as the amplifier sounds good.

I'm about to use Scheme-It to get a rough sketch of the amplifier circuit, will post when done in the next little bit.

 

[Solidus]

Okay, here it is. Not many of the parts are labeled, as I'm suffering from sleep deprivation right now. I just noticed it says you're from Richmond - I stopped there for gas the other day - I'm down in Merced (central valley).

The diagram and the amplifier build will be dominated by 1N4007 diodes and filter networks. The diagram I just drafted doesn't have any of the fans or interlocks but you should get the point. I labeled relevant test points.

If you can spot the giant easter egg my sleep-deprived self included it may entail a trip to bring this beast up to show you, provided I ever get it completed.

 

[Solidus]

Just had an idea - what are your suggestions as to the possibility of cutting out the 12AU7 preamps, and driving the 4-125As with two LM324 quad op-amps, with the outputs set in parallel to increase the output wattage?

Just thought of that. It would allow me to cut out the tube heaters and parallel them, rectify, filter and slap a few heatsunk 7805 regulators on there to supply the 4-125 heaters. It would simplify circuitry a lot, at least in the amplifier portion.

 

[john monks]

The 4-125 filaments are rated for 5 volts at 6.5 amperes. This is a bit much for a 7805. Normally a separate AC filament winding is used to supply these filaments.
The control grids are to be biased at negative 90 some volts, not positive.
Normally the IC's don't mix well in an environment like this, around hot tubes, and 2000 some volts, with all the electrostatic discharges that are likely to take place.
My suggestion is to use a tube type driver stage for the 4-125's because you may need a driving force of 180 volts peak to peak to fully drive the 4-125.
To bias the 4-125 you can used a resistor between the cathode to ground so that the cathode can become positive relative to the control grid thus simplifying the circuit. And then you should place about a 1000uF capacitor across the resistor so you don't loose much gain.

 

[Solidus]

Well, I would be using a stack, probably 5. The 7805 is rated to 2.2A max heatsunk.

Oh well, I figure it'll be easier to trafo supply, rectify and filter to supply. The 78-- series are good for small projects but if you use them to supply large power they get pretty dirty. All that extra voltage turns into heat so I'd have to find a way to mount them directly to the chassis itself most likely.

That's why I haven't found myself venturing into directly-heated valves in the audio spectrum before, they're rarely standard voltages and unless you like the sound of buzz, you need a pretty complex filtering network to prevent any ripple from passing.

I don't mind using a larger resistor to bias the control grid. Keeping the cathode at perfectly ground zero is fine for me, that is, unless it adversely affects the performance of the circuit.

Do you have any recommendations for driving tubes?

Originally I was planning on using a 12AU7 with both sections coupled in parallel to supply the control grid with about 2W, but I rethought that on the basis that the 12A series valves are popular for guitar amplifiers because they're known to be "drive-y" tubes, that is, they're very easy to push into distortion. This is why many pre-amps for guitar amps use a 3- or 4-section cascade to ram quite a bit of harmonics into that circuit. This isn't a push-pull circuit so I wouldn't have the safeguard of second harmonic cancellation like I would on other low distortion designs.

Now, seeing as the 12A series is mostly rated (the 12AU7, AT7 and AX7) for max 1W dissipation per section, to demand 2W from there would be pushing it too close to plate saturation for comfort. I want an audio amplifier that I can listen to and enjoy music on, not make growling notes on my guitar. It doesn't have to be hi-fi but I don't want distortion if I don't have to have it.

Now I was looking at the 6CL6 pentode, and it seems like it would work well as a driver - 2.8W max dissipation, and I could triode-strap it for better sound quality. It is a bit harder to drive so I'd need to bias it pretty carefully, though.

Let me know what your thoughts are.

 

[john monks]

The driving requirements for the 4-125 is 0 watts according to your datasheet for audio. I don't believe this is realistic but I believe it would only be a few milliwatts because practically no current is going to be on the grid lead. What little current there would be is some stray electrons bombarding the grid from the cathode but this is very little. So you are essentially driving an open circuit. So your only concern is that your driver amplifier can supply about 180 volts of swing. I never tried it but I think a 12AU7 can do that.

You can drive the heater of the 4-125 with AC. You can take about a 100 ohm potentiometer and place the ends across the heater and connect the wiper to the chassies. Then you can remove any buzz by simply adjusting the potentiometer for a null point. That way you can forget about the 7805's

In the past I tried connecting tetrodes and pentodes as triodes and my results were poor. I think the output impedance was more that you get with a regular triode but I am going by memory so maybe it would work. You want the output impedance of the driver stage to be as low as possible so as to avoid high frequency roll-off.

Bottom line I think your are getting closer but I am still confused why you want to use sugh a great big tube like the 4-125. I used 6146's when building power amplifiers when I was a kid and I got very good results. The voltage is lower and they turn on very hard.

 

[Solidus]

So you're saying that I would not have to push the 12AU7 anywhere near cutoff to get good drive into the 4-125.

That makes sense, knowing how the values derate. When I looked in the datasheet last I saw the maximum driving power being 5W, and there was a chart on driving power vs. plate voltage vs. output power. However, one has to remember that is well into the tens of megahertz.

Well, if I can do it with a 12AU7 I will do. It is a very familiar tube to me and I've worked with it before. The triode equations are well-tested and very familiar; there is no real unexpected outcomes if it is calculated in that manner. I may bump up to the 12AX7 as it is such a well-used tube there is literally no end to the data on it. Being a high-mu tube it would provide plenty of drive for the economy.

I decided to use the 4-125A mainly to do something with the couple of them I have around from collecting. I've built a 6L6 amp using schematics that were fairly hard to go wrong with, so I wanted a bit of a challenge.

I'm still in my first year at UC Davis, so I guess it may be the idea of me wanting to go where most people don't, but the unconventional idea of it is what sealed the deal for me. Plus, the tubes are unusual, and rarely seen to anyone who isn't a radio engineer or ham. The sight of seeing those massive tubes light up and function is more of a conversation topic if nothing else.

 

[john monks]

I'm jealous. I understand fully with what you are doing because I'm doing similar things. Now specifically a 4-125 is usually used as a transmitting tube, sometimes for AM and sometimes FM broadcast. So in that case you want to use the tube in the most efficient maner possible, as a switch or more commonly called class C operation. So in that case you want to drive the control grid very hard. You would drive it with another tetrode through a tuned circuit to get the grid to turn the tube on and off as completely as possible. So you would drive the control grid positive relative to the cathode and that means that the grid will be drawing current from the cathode or you would drive the grid to -100 or maybe -200 volts to get the tube off. As an audio amplifier the drive requirement is not nearly great because you rarely turn the tube completely on or off. So the control grid is almost always negative in regard to the cathode. So I think you are on the right path. You just need to come up with some kind of negative supply for goading the tube.
One more thing in case your are interested. You don't necessarily need to use 2000 volts to supply the B+. just 200 volts or maybe a little more should be enough to make you amplifier work. You will just have greatly reduced power.

 

[Solidus]

I'm jealous. I understand fully with what you are doing because I'm doing similar things. Now specifically a 4-125 is usually used as a transmitting tube, sometimes for AM and sometimes FM broadcast. So in that case you want to use the tube in the most efficient maner possible, as a switch or more commonly called class C operation. So in that case you want to drive the control grid very hard. You would drive it with another tetrode through a tuned circuit to get the grid to turn the tube on and off as completely as possible. So you would drive the control grid positive relative to the cathode and that means that the grid will be drawing current from the cathode or you would drive the grid to -100 or maybe -200 volts to get the tube off. As an audio amplifier the drive requirement is not nearly great because you rarely turn the tube completely on or off. So the control grid is almost always negative in regard to the cathode. So I think you are on the right path. You just need to come up with some kind of negative supply for goading the tube.
One more thing in case your are interested. You don't necessarily need to use 2000 volts to supply the B+. just 200 volts or maybe a little more should be enough to make you amplifier work. You will just have greatly reduced power.

Now this means that I would bias the tube differently so the idle point sits at a much higher voltage with respect to the cathode, correct? That does mean, however, that I would have to configure the bias supply to feed current to the grid, as that does factor in grid dissipation as well. Am I on the right track? All the circuits I've ran have been Class A or AB1.

How would valve dynamics change if I was to run them at 800V? What would my operating points be in that case?

Also, if you want a bit of my ponderings on the idea of triode-strapping tetrodes and pentodes, I've done a little bit of experimenting and a whole lot more reading but nothing terribly much.

Triode-connecting a pentode seems to work very well sonically if it is of the power pentode/kinkless tetrode type. 6L6, 6550, KT66/88/120 reportedly work very well in this configuration. However, it makes sense that the drawback being you a) have to run it at a much lower voltage due to the tied screen grid limitation, and b) you have to flow a lot more current. I believe you also have to, in many cases, calculate a resistor for the screen grid tie to prevent its dissipation from exceeding safe limits.

That being said, while I've read a lot on kinkless tetrodes connected as triodes, I'm not too sure how it would work for small/medium signal valves, returning to my 6CL6 musings earlier. The 6CL6 in most of the ways I read about it is used for its linearity often as a small signal oscillator that feeds more powerful valves. Most of the times I hear about it being in AF circuitry it's being used as a very small signal (<5W) amplifier push-pull stage.

Nor do I know how true tetrodes would fare connected in this manner. One of the benefits of having the suppressor grid is the lack of the tetrode kink as well as more linearity in the rest of the tube characteristics. I'm not sure how it would fare without.

That would make sense about the impedance being different, possibly higher. It's a trade-off, the other major trade-off being significantly reduced power.

There is always the Ultra-Linear configuration, although I've never tried that as it opens up a whole can of worms with a 40% screen-tapped output trafo, as well as requiring that the PP pair be biased more carefully and the phase inverter had more attention paid to it. Most guitar amps the phase inverter is simply a triode with a Rk = Rp. I believe in the UL configuration it is more crucial than that.

Besides, returning to the question of "why the 4-125A?", the project is a lot more unique when I realize I may be one of the only guys building an amp like this, and knowing the circuitry is original is half the pride in having the thing.

 

[john monks]

Normally in a tetrode or a pintode the bias requirements are directly proportional to the screen grid voltage. For example if the screen grid voltage of a 4-125 is 600 volts the the control grid DC bias should be around -94 volts. If the screen grid was 300 volts the the DC bias on the control grid should be about -47 volts. The reason for this has to do with space charges within the vacuum of the valve. The control grid characteristics has little to do with plate voltage. So my suggestion would be ,if you want to run lower voltages, is to reduce them all by the same ratio with the exception of the heater voltage. It must remain around 5 volts or damage to the valve will occur.

One thing about running tetrodes or pintodes as triodes. A tube like a 12AU7 has a mu or AC gain of about 70 if I remember right and a 12AX7 has a mu of about 100. These valves are purposely designed to have lots of gain. I don't know what a 6V6 would give you if hooked up as a triode, maybe 10 or 20. I don't know but certainly would be much worse than most triodes designed to be voltage amplifiers.