This is very strange - amp problem, help needed

[SoundmasterG]

I've been working on a Bogen CHB100 that I've been modding. The circuit right now is as follows:

Merlin's parallel triode circuit with switchable triodes and gains on the input > Vox T/S > volume control > 6BR8A triode > 6BR8A pentode > cathodyne inverter using 12AV7 > cut control > output stage using four 7868's.

What is happening is that if the amp is cold and is turned on, I can monitor the signal as it passes though the amp and everything appears to work as it should, and the voltages all check out as something that should function mostly well on all the stages. Then as the amp warms up, the cathode on the 6BR8A will lose it's bias and go to zero, and then of course the signal stops going through the amp. I'm using Merlin's suggested values in his book for an EF86 on this pentode....100k plate resistor, 560k screen resistor bypassed with a .1uf to cathode, 680 ohm cathode resistor with no cathode cap. While it is working I can measure the screen voltage as being around 40v and the plate as around 103v, cathode about 1.3v....should be good enough for a starting point right? I've tried different 6BR8A's, all NOS and not used...all of the tubes test fine on my B &K 747 tube tester and they all perform well initially until things get warm, and then they all have the same problem. I've replaced the 680 ohm cathode resistor and tried a 1k too, even with just clip leads instead of soldering in place, and I've tried a resistor sub box also....all do the same thing. The socket performed flawlessly in the amp when I had another preamp design in there. I do get continuity from each side of the pin, and when the tube is in, the pin is making contact.

What is going on here? I'm stumped at this point as to what is happening.......

Greg

[HotBluePlates]

Then as the amp warms up, the cathode on the 6BR8A will lose it's bias and go to zero, and then of course the signal stops going through the amp.

Do you mean bias voltage goes to zero, or tube current goes to zero?

Also, do you have a schematic for your present circuit?

[SoundmasterG]

I mean that the cathode voltage of the pentode goes to zero, and the plate and screen voltage go to the rail. I don't understand how a cathode can go to zero anyway with a functioning and permanently mounted resistor in between the cathode and ground? The signal coming into the pentode is a nice and clean sine wave with no distortion or artifacts present.

I'm not actually sure if the heating up is the problem, as it seems to be ok while the master volume control on the output of the pentode is at zero. Once the master is turned up, then the signal and AC voltage on the plate of the pentode begin to slide down to where eventually they go away, and then if I check the DC voltage at the cathode, I get a zero reading. I should think a 1M load is enough on the grid of the following stage right?

All I have is a hand-drawn schematic that has been through about 20 changes so far, from preamp design to phase inverter design. :) I could draw something up on the computer as far as what it is now, but it is pretty standard as far as the circuits. Imagine a parallel 12AX7 on the input that can have the signal to one or the other triode section turned off at a time, or they can both be run in parallel. They have dissimilar cathode resistor and cap values for different gain characteristics. This is on the input of the amp. After that, a standard Vox tone stack, then a volume control, then the triode section of the 6BR8A, then the pentode section of the 6BR8A, then a master volume, then the phase inverter, which is the cathodyne using the 12AV7. The triode section of the 6BR8A has a 56k plate load, a 1k5 cathode resistor, no cathode cap, and a .022uf coupling into the pentode section. All grids have local grid stoppers...I believe I have a 100k on both the triode and pentode grids.

Greg

[eleventeen]

If the cathode goes to zero or ground and the plate goes to the rail, the simple conclusion is that the tube simply isn't conducting anything, eg; it's an open circuit. E = IR. If I [current] equals zero, then the R [cathode resistor] can be arbitrarily large and E [voltage drop across the cathode resistor] will still equal zero.

If every part of the input sine wave to that stage is below cutoff for that tube, then the tube will always be cut off. Maybe you need some kind of DC restoration, or, stated differently, maybe the "zero" crossover level for your input sine wave is way negative relative to where it needs to be.

As usual, I'm working off my best understanding of the symtoms you've described.

[HotBluePlates]

... the simple conclusion is that the tube simply isn't conducting anything ...

Agreed. If current drops to zero, the voltage across the cathode resistor goes to 0, and there is no drop across the plate load, so it goes to the rail voltage.

I disagree on the input signal part, because there is a coupling cap. Since there is no d.c. reference, the input signal would pull the grid up as much as down. Anyway, as long as the grid input signal is not more than the bias voltage, I don't see an issue there.

You have a 0.1uF cap from the screen to ground. What's the chance that's it's leaky? If it were, screen voltage could eventually drop to zero, which would likely kill plate current. I thought of some other hair-brained ways the tube current could be killed, but as I think through them they don't seem like they would keep the tube off after it reaches zero current.

I can't really comment on whether parts values contribute to the problem at all, since I don't know your supply voltage. I do notice the data sheet implies that screen current is about 1/3 of plate current. I'd think you'd want the screen resistor around 3-4 times the plate resistor, to keep the 2 voltages similar. But being a little low probably isn't hurting. What I do not know is whether there is a critical screen resistance (like 40v) where you kill off tube current.

If the screen bypass cap is not leaky, reduce the screen resistor and bring screen voltage back up to ~100v. See if the tube suddenly works.

[SoundmasterG]

Well I seem to have found out what the problem was. The grid leak (1M) going to ground in front of the pentode was open. So I guess when the master volume was opened, allowing the pentode to conduct, it would quickly reach cutoff and then everything was leaking out the grid, even so far as to run the cathode to zero? After I replaced that part, it is stable now for the screen, plate, and cathode voltages, even after some testing and passing signal through. I lowered the screen grid resistor to 470k and that moved the screen to 46v and the plate is at 134v, cathode at 1.2v. This is with a 300v supply. I can vary the supply between around 300v and 250v for the preamp, though I didn't try it at 250v yet since I got the pentode working better. I'll have to fiddle with part values and supply to get the voltages more in line, though I'd read that the screen should be .3 to .4 times what the plate is on a preamp pentode, so if that is true, then the voltage on the screen is in the middle of that range currently. Now that the pentode is working more correctly, it has changed some other aspects of the rest of the preamp and what was stable before now has some issues. I believe some noise is getting injected at the front of the amp on the parallel stage as I'm seeing some artifacts on the scope and hearing some fizz. So back to more tweaking to clean things up. Thanks guys for the help!

Greg

[HotBluePlates]

The grid leak (1M) going to ground in front of the pentode was open. So I guess when the master volume was opened, allowing the pentode to conduct, it would quickly reach cutoff and then everything was leaking out the grid, even so far as to run the cathode to zero?

Oops. That's the other way for cutoff to occur. This is a variation of what eleventeen was thinking.

Okay, grid reference resistor is open, so there's no d.c. path from grid to ground. Electrons travelling from the cathode to plate mostly pass through the open gaps in the control grid, but a few strike the grid itself. The grid reference resistor provides a leakage path for those electrons back to the cathode (though ground). If that resistor opens, the electrons sit at the grid and at the grid side of the coupling cap. They accumulate, eventually giving the grid an increasingly negative charge. Once that negative charge gets big enough, the tube is biased to and beyond cut off.

I'll have to fiddle with part values and supply to get the voltages more in line, though I'd read that the screen should be .3 to .4 times what the plate is on a preamp pentode, so if that is true, then the voltage on the screen is in the middle of that range currently.

You can neglect what I said about changing components for the pentode, since you found the actual problem.

The screen voltage doesn't need to be a certain ratio of plate voltage. It should be more than 0v, and more than plate voltage is awkward. So use something between those extremes. It is common in preamp pentodes to have the screen voltage somewhat below the plate voltage; you don't see that much in output pentodes because you generally don't want plate current limited in an output stage.

Bottom line, G1 and G2 set plate current in a pentode, it's just that G1 is more effective at doing it. You have a supply voltage, you have a load resistance. Assuming you want an roughly equal positive and negative output swing, pick a point somewhere in the middle of the loadline. Note the plate voltage and current, and the nearest grid bias line. Adjust G1 and G2 voltages to land at your desired point.

That's not a complete picture of how to design a pentode voltage amp, but is the basic idea. The curves probably spec a value of screen voltage. If you use less screen voltage, the plate current will be lower. Normally, a pentode has a fairly consistent ratio for how cathode current splits between the plate and screen. It might be 5:1 in one tube and 3:2 in another. You look at typicalconditions listed on the data sheet to get a feel for what the ratio is. You note the plate current for you chosen operating point, then note how much below the supply voltage your screen needs to be set. Based on your noted current ratio, you figure the needed screen resistor value by finding the amount of screen current (maybe screen current = plate current / 3.5), and using ohm's law to get your needed resistor value. Select your screen bypass cap to have Xc of 1/10 to 1/100 of the screen resistor value at the lowest frequency of interest.

But you're free to tinker now that the pentode is working properly.

[SoundmasterG]

Unfortunately there isn't much info for the 6BR8A on the data sheet...no curves that I can find anyway. The 6U8A datasheet is similar though, and I found some info that should help there, even though I don't really understand how to do a load line yet. One person on another forum mentioned that due to the high plate current of the triode (13.5ma) and pentode (9.5ma) in this tube, I should lower the plate, screen, and cathode resistors down to get more current, and get the screen voltage up closer to the plate on the pentode, and then I'll get better tone out of it. I did change the screen resistor down to a 470k, keeping the plate at 100k and cathode at 680 ohms, and it sounds pretty fizzy, but after I fixed the issue with the grid leak resistor, it sounded like that at a 560k screen grid resistor too. So if I change down to smaller values, then the wattage has to go up of course. I wonder how far I should try going down? Looking at the curves for the triode I figured that I should aim for having about a -4v on the cathode, a 300v supply, and plate volts around 225v or so. I don't know what values of resistor choices will get me there though, or what wattage they will need to be. I'm not sure where to start on the pentode yet, but going lower down seems to be the way to go. I would like to get the triode and then the pentode overdriving just a little when the volume on the amp is around 8-10. How to interact the output of the pentode with the driver for the phase inverter...well I'm not sure what I should aim for there yet. I have the 2k7 on the cathode and a 56k on the plate on that stage, with a 350v supply like you and PRR recommended....but I guess one stage at a time. 

Greg

[HotBluePlates]

Unfortunately there isn't much info for the 6BR8A on the data sheet...no curves that I can find anyway.

Most of the time, I don't like short-form data. And because of that, the only data sheet site which I like is Frank's. If you do a search and you see a sheet with a large file size, that means a more complete sheet. So forget the suffix for a moment and look at the 6BR8 on Frank's. Now ya got curves.

One person on another forum mentioned that due to the high plate current of the triode (13.5ma) and pentode (9.5ma) in this tube ...

I used to not know what PRR meant when he said the data sheet has show-off conditions. They often show conditions at very high tube current, because it makes the Gm higher, rp lower and mu (if listed) higher. All around, the tube looks like a better performer. Makes sense, why not show everything your tube is capable of?

Just because the tube can operate at high current doesn't necessarily mean you have to operate it at high current.

I don't really understand how to do a load line yet.

Looking at plate characteristics, you draw a line that is tipped up on the left side of the graph and down low on the right side. The line defines the operating area set by the load resistor. If the tube passed zero current, there is no voltage drop across the load resistor, and plate voltage is equal to the supply voltage. That's 1 end of your line: supply voltage, 0mA. The other end of the line is the entire supply voltage dropped across the load resistor. There is no voltage left to be across the tube, so this end on the line is at 0v, and supply volts/load resistance.

For a preamp stage, your operating point must lie on this line; you don't have to do anything special for it to be there, it will lie on the line. You pick if you want high current and low plate voltage or vice versa, and whatever point you pick will be on or near a grid bias line. If your screen voltage equals that shown on the sheet, and you apply the bias shown on the sheet, you'll idle at the plate (-to-cathode) voltage and plate current listed, within the limits of tolerance of the tube characteristics.

I wonder how far I should try going down?

What follows this tube? what is the bias voltage of that stage? have you measured the output of the pentode with a typical input? With an input that causes the fizziness you're talking about?

If you go too high current, the plate voltage drops low. You might be able to swing a big output in 1 direction but not the other. That's distortiona nd weak output. Same is true, but different, if you run a too low current/high plate voltage and the same given supply voltage. You have to strike a happy balance. That's somewhere in the middle, and really ought to be determined by how big a signal you have to feed the next stage.

[SoundmasterG]

Thanks for the load line primer. I'll have to sit down with the plate curves and see if I can figure out how to use it.

Currently after this tube right now is the 1M master volume pot, and then the driver stage for the cathodyne inverter using a 12AV7. I have considered going with a LTP, but for now I am sticking with the cathodyne to see where it gets me. With the volume full up before the pentode, and the master full up after, I am getting 84v AC out of the pentode. I don't recall what I get when it is outputting a clean sine wave right now, but I do have to turn the volume down a ways to get a clean sine wave out of the pentode. Everything before the pentode is clean even when full up right now. The stage after the pentode is a 12AV7 gain stage with a 2k7 cathode resistor. The voltage on it's cathode is around 4v. I get around 41v AC and 32v AC out of the two legs of the phase inverter, and bias is -19.5 so I am getting full power as far as I can tell, even though the pentode is clipping a bit at that point.

Greg