Hi,
I'm trying to understand what happens in a circuit such as this one (AC30 power section) when one or more valves fail.
Some of my thoughts:
1. Because of the shared cathode resistor, bias will become significantly hotter. This could encourage cascading failures.
2. The power supply would have somewhat less DC load, also increasing power dissipation by the rest of the circuit. If the quiescent cathode bias current is 50mA and the power supply has 100ohm internal resistance then this would be a ~5V increase, which seems relatively minor.
3. If one bottom valve went open-circuit, then for small signals, instead of the rest of the valves seeing load impedance 4Z, they would instead see 3Z. For larger signals, during their respective class B portions of the cycle, the top valves would see 2Z and the bottom valves would see Z. It is this last part that seems most problematic to me.
4. If both valves on one end went open-circuit, I don't think the other end would have so many issues, although power dissipation might still increase due to seeing 2Z instead of 4Z during the class A portion of the cycle (but NFB due to the plate voltage has less effect during this time anyway). That being said, the output would be clipped on one side.
5. If one valve on each end went open-circuit, connecting 16 ohms to the 8 ohm tap seems like it would restore normal conditions (aside from the above biasing and sag concerns).
I also have some questions:
1. Is my analysis correct?
2. Are cascading failures a real issue, or would you have to go a long time without replacing a dead valve for it to become one?
3. Using 4*R31 and C16/4 separately for each cathode would cause output valves with variations in their characteristics to each settle at their own bias voltage. Does this make an audible difference?
4. Is using the above methods (separate Rk, remove one valve from each end, and connect 16 ohms to 8 ohm tap) a viable way to implement half-power operation?
I'm trying to understand what happens in a circuit such as this one (AC30 power section) when one or more valves fail.
Some of my thoughts:
1. Because of the shared cathode resistor, bias will become significantly hotter. This could encourage cascading failures.
2. The power supply would have somewhat less DC load, also increasing power dissipation by the rest of the circuit. If the quiescent cathode bias current is 50mA and the power supply has 100ohm internal resistance then this would be a ~5V increase, which seems relatively minor.
3. If one bottom valve went open-circuit, then for small signals, instead of the rest of the valves seeing load impedance 4Z, they would instead see 3Z. For larger signals, during their respective class B portions of the cycle, the top valves would see 2Z and the bottom valves would see Z. It is this last part that seems most problematic to me.
4. If both valves on one end went open-circuit, I don't think the other end would have so many issues, although power dissipation might still increase due to seeing 2Z instead of 4Z during the class A portion of the cycle (but NFB due to the plate voltage has less effect during this time anyway). That being said, the output would be clipped on one side.
5. If one valve on each end went open-circuit, connecting 16 ohms to the 8 ohm tap seems like it would restore normal conditions (aside from the above biasing and sag concerns).
I also have some questions:
1. Is my analysis correct?
2. Are cascading failures a real issue, or would you have to go a long time without replacing a dead valve for it to become one?
3. Using 4*R31 and C16/4 separately for each cathode would cause output valves with variations in their characteristics to each settle at their own bias voltage. Does this make an audible difference?
4. Is using the above methods (separate Rk, remove one valve from each end, and connect 16 ohms to 8 ohm tap) a viable way to implement half-power operation?
Answers to your thoughts:
1. Yes, bias will get hotter.
2. Total current draw will probably decrease, so less dissipation in total.
3. If it is Class A, then each remaining valve will see 3Z. If Class B then 2Z (as usual) and 1Z. This could cause less effect than you might expect.
4. I don't understand you.
5. If you have half the valves driving half the load then conditions will be unchanged.
Biasing four output valves from one cathode resistor is quite poor engineering - one might call it cheapskate engineering.
1. Yes, bias will get hotter.
2. Total current draw will probably decrease, so less dissipation in total.
3. If it is Class A, then each remaining valve will see 3Z. If Class B then 2Z (as usual) and 1Z. This could cause less effect than you might expect.
4. I don't understand you.
5. If you have half the valves driving half the load then conditions will be unchanged.
Biasing four output valves from one cathode resistor is quite poor engineering - one might call it cheapskate engineering.
Yes, I wasn't clear. I meant more power per valve, potentially contributing toward overloading them.
I meant the apparent load would only differ during the class A portion. But during the class A portion, the current is not that much, and so differences in the load impedance seen by the plate would not necessarily change the plate voltage by that much, and so the effect on dissipation per valve might be negligible (I guess this also requires assuming the load is resistive).
Last edited:
Oh, yes, I see why. And only the impedance as seen by AC changes when a valve dies in the split-resistors example. So it cannot become worse.
Thanks!
Last Vintage AC30 I restored (a 1963 Copper Front Panel job) I put 2 x 470 Ohm 2W metal film resistors in parallel (for 235 ohms) with a 47uF/63V bypass cap on each of the 4 output tubes.
The owner says he could not tell any sonic difference from when it had the single 50 Ohms with 220uF bypass.
That runs the EL84/6BQ5 at a more conservative dissipated power point and deletes the need for tightly matched output tubes.
Building an AC30 Clone now and have done the power section that way on it too.
Hint:
The Output Tube Cathode connections are a low impedance point and so not subject to noise/interference pickup. You can safely extend cathode connection wires to a convenient place to mount the cathode bias resistors and bypass caps. They don't have to be right next to the tubes.
Cheers,
Ian
The owner says he could not tell any sonic difference from when it had the single 50 Ohms with 220uF bypass.
That runs the EL84/6BQ5 at a more conservative dissipated power point and deletes the need for tightly matched output tubes.
Building an AC30 Clone now and have done the power section that way on it too.
Hint:
The Output Tube Cathode connections are a low impedance point and so not subject to noise/interference pickup. You can safely extend cathode connection wires to a convenient place to mount the cathode bias resistors and bypass caps. They don't have to be right next to the tubes.
Cheers,
Ian
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.