Hi all, I'm working on a single-ended design using a 6945 subminiature tube (datasheet here). The only part that I'm having trouble with is the selection of an appropriate screen resistor in order to avoid exceeding the max screen dissipation. I've read Merlin Blencowe's preamp book, and while he recommends including a screen stopper to limit screen current at saturation, he doesn't go into how to select an appropriate value. I've also read Kuehnel's power amp book, but he sidesteps the issue entirely, since his sample single-ended design is within the dissipation limit at saturation.
So, let's take the sample Class A (single tube) pentode application in the 6945 datasheet. They use a 110v screen, -10.6v bias, 150v plate, and a 7k load. At quiescence, the screen dissipation is 0.8 mA, which gives a screen dissipation of 88 mW, within the max rated dissipation of 330 mA. If the screen is connected to the B+, we'd need a 50K screen resistor to drop the 40 volts from 150 to 110. All this seems fine.
Now though, suppose we overdrive the amp and hit 0v on the grid. With a 7k load, the load line will intercept the 0v line at an approximate plate voltage of 10v. If I'm reading the "Average Screen Characteristics" graph correctly, the screen current at 10v would rise to about 22 mA. The would be a dissipation of around 2.42 W, which is way over the limit. However, it seems like the added screen current through the dropping resistor would drop the screen voltage quite a bit, and therefore limit the current. How much, though, I don't know.
In truth, I imagine that 50k would be quite safe, given the traditional value in Fender amps was around 500 ohms. However, I was wondering if there is an formula to determine an optimum value. Furthermore, Blencowe recommends bypassing the screen to the cathode with a capacitor, and I'm unsure if including that would complicate the selection of a resistor.
So, let's take the sample Class A (single tube) pentode application in the 6945 datasheet. They use a 110v screen, -10.6v bias, 150v plate, and a 7k load. At quiescence, the screen dissipation is 0.8 mA, which gives a screen dissipation of 88 mW, within the max rated dissipation of 330 mA. If the screen is connected to the B+, we'd need a 50K screen resistor to drop the 40 volts from 150 to 110. All this seems fine.
Now though, suppose we overdrive the amp and hit 0v on the grid. With a 7k load, the load line will intercept the 0v line at an approximate plate voltage of 10v. If I'm reading the "Average Screen Characteristics" graph correctly, the screen current at 10v would rise to about 22 mA. The would be a dissipation of around 2.42 W, which is way over the limit. However, it seems like the added screen current through the dropping resistor would drop the screen voltage quite a bit, and therefore limit the current. How much, though, I don't know.
In truth, I imagine that 50k would be quite safe, given the traditional value in Fender amps was around 500 ohms. However, I was wondering if there is an formula to determine an optimum value. Furthermore, Blencowe recommends bypassing the screen to the cathode with a capacitor, and I'm unsure if including that would complicate the selection of a resistor.
I don't think you want 50k for a screen resistor, I wouldn't go over 1k5 and honestly I probably would use anything from 100 to 470 for a single ended 6945 guitar amp, For bigger class AB amps I use between 470 and 1k5 depending on the bottles used. I don't have much experience with this tube FWIW.
Also the plate will never get down to 10v, the DC resistance of the primary is quite small.
Also the plate will never get down to 10v, the DC resistance of the primary is quite small.
Except that this particular design has a screen voltage that's 40 volts below the quiescent plate voltage. Doesn't it need to be that big in order to drop it down? I was under the impression that the 470 - 1k values that you typically see are for cases where the screen voltage is very close to the plate voltage.
The screen resistor is an addition to the power supply design, whether the screen voltage is close to the same voltage as plate (typical with a choke) or less.
I think you are thinking they are one in the same.
For example look at the Fender 5E3, the original design did not use "screen resistors", just a 5k dropping resistor in the power supply which brings it about 50v less than the plates. If you look at any modern 5E3 circuits including the Fender "57 deluxe" they add 470R "screen resistors".
I think you are thinking they are one in the same.
For example look at the Fender 5E3, the original design did not use "screen resistors", just a 5k dropping resistor in the power supply which brings it about 50v less than the plates. If you look at any modern 5E3 circuits including the Fender "57 deluxe" they add 470R "screen resistors".
You're right - and this is kind of the crux of my problem.
So with the 5e3, the screen voltage sits between the plates of the 6V6's and the B+ of the 12AX7. The problem I've run into with some of the submini tubes is that the max plate rating for the output tubes is pretty much the same as the rating for the preamp tubes. If I were to adopt the 5e3 topology with a 110v screen, I would have to put the preamp B+ below that, which doesn't leave a lot of headroom at all. It seems like my only other option is to use a large dropping resistor between the preamp B+ and screen, or add another section to the power supply, which would amount to the same thing.
Let's assume though, that I'm able to adopt something like the '57 deluxe topology with the 470R resistors. Is there a way to determine whether a 470R resistor would be sufficient to keep it under the dissipation limit? 0.33W is not a lot to work with, and I don't know how to calculate the worst-case dissipation.
So with the 5e3, the screen voltage sits between the plates of the 6V6's and the B+ of the 12AX7. The problem I've run into with some of the submini tubes is that the max plate rating for the output tubes is pretty much the same as the rating for the preamp tubes. If I were to adopt the 5e3 topology with a 110v screen, I would have to put the preamp B+ below that, which doesn't leave a lot of headroom at all. It seems like my only other option is to use a large dropping resistor between the preamp B+ and screen, or add another section to the power supply, which would amount to the same thing.
Let's assume though, that I'm able to adopt something like the '57 deluxe topology with the 470R resistors. Is there a way to determine whether a 470R resistor would be sufficient to keep it under the dissipation limit? 0.33W is not a lot to work with, and I don't know how to calculate the worst-case dissipation.
With a plate at 200v and screen at 150 if the grid goes to 0v then it looks like you would have 8mA. .008x150=1watt, this is triple the .33watt max limit. So you would need to drop 50 volts so 50/.008=6250. Next common value up would be 6.8k.
Not sure if this is correct? Hopefully somebody else will chime in that knows what they are talking about.
Not sure if this is correct? Hopefully somebody else will chime in that knows what they are talking about.
Unfotunately it's a nonlinear problem and not amenable to pure algebra. It could be solved graphically, except you'd have to create the graph of Ig2/Vg2 yourself. Not impossible until you realise that the load is reactive and therefore the screen current will change wildly under different conditions!
You're left with a cut-and-try approach. A value of 1k to 2k-ohm is probably fine, since thousands of guitar amps have proved themselves with this value (and all pentodes behave more-or-less proportionately the same regardless of their bottle size).
You're left with a cut-and-try approach. A value of 1k to 2k-ohm is probably fine, since thousands of guitar amps have proved themselves with this value (and all pentodes behave more-or-less proportionately the same regardless of their bottle size).
Thanks Merlin!
I just finished your preamp book a month or so ago - it was terrific. Thank you.
I'm thinking I might just breadboard the circuit, and slowly increase the input while monitoring the screen current and voltage.
I do have one question for you, though. On your single-ended output stage page, you talk about sliding the screen down to put the 0V grid line where the -8 line had previously been. Did you choose -8 simply on the position of the knee, with the assumption that if the loadline comes close to the knee, the dissipation will be under control? Or was there another factor that led you to choose that level of compression?
I just finished your preamp book a month or so ago - it was terrific. Thank you.
I'm thinking I might just breadboard the circuit, and slowly increase the input while monitoring the screen current and voltage.
I do have one question for you, though. On your single-ended output stage page, you talk about sliding the screen down to put the 0V grid line where the -8 line had previously been. Did you choose -8 simply on the position of the knee, with the assumption that if the loadline comes close to the knee, the dissipation will be under control? Or was there another factor that led you to choose that level of compression?
Putting the load line a little under the knee is more of a tonal consideration. That's how most guitar amps run, and people seem to approve. It has nothing in particular to do with dissipation- that's what the screen stopper is for. For hifi use you would probably put the load line through or a little above the knee for lower distortion.
Yep - I understand about the tonal considerations. In the passage I'm referring to, though, you're calculating the size of the screen stopper needed in order to avoid exceeding the screen dissipation limit.
That's down in the Screen Voltage section of your Single-Ended Output Stage page. What I'm wondering is how you determined that sliding the screen voltage such that the 0V grid line moves to -8 would be sufficient protection.
I do have one other question, as well. In your Preamp book, you recommend bypassing the screen to the cathode with a capacitor in order to avoid negative feedback. Would doing so affect the ability of the screen stopper to sag the screen voltage?
That's down in the Screen Voltage section of your Single-Ended Output Stage page. What I'm wondering is how you determined that sliding the screen voltage such that the 0V grid line moves to -8 would be sufficient protection.
I do have one other question, as well. In your Preamp book, you recommend bypassing the screen to the cathode with a capacitor in order to avoid negative feedback. Would doing so affect the ability of the screen stopper to sag the screen voltage?
Oh I see. Well there's a lot of "eyeballing it" where power amp design is concerned. The main thing is that by putting the load line through the knee, the screen current does not shoot up suddenly during clipping. It climbs more slowly, since the grid curves incrementally approach the 0V grid curve, which is 'kinder' on the screen. The -8V is roughly where the load line passes through a knee (OK maybe a bit below, I guess I could have chosen -10V or -12V if you want to get picky about the knee!).
No, it would make very little difference. But with power valves the screen bypass cap is (usually) also an HT smoothing cap, so it is invariably bypassed to ground rather than to cathode.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.