Reeltarded wrote: ↑Wed Feb 01, 2023 10:40 pm
…
I think I might swap to 180Rs.
The resulting crossover distortion will probably look like the shoulders of Atlas
Self biased outputs pretty much have to be biased close to class A or else that’s what happens.
I went through a similar thing with my last amp which was Class AB Cathode Biased. I had it at 70% and it just sounded meh, now biased at 95% it is worlds better. I was actually going to write an article on this as I did a fair bit of research but a house move got in the way, so I'll try and sum it up now.
When you start to drive a cathode biased class AB amp the bias voltage increases pushing the bias colder. If you use the fixed bias 70% rule then under load your amp will actually be biased colder. At clean levels the bias shift isn't too bad but if you monitor the bias in overdrive state you'll find it shifts A LOT. So we bias hotter to make up for it. When biasing my amp I checked it at different drive levels using a scope and meter while also doing A/B tests (no pun intended). I just tried to minimize the bias shift as much as possible while making sure the valve was happy and not melting itself while sounding good at the same time.
Now, when researching this I did look at a lot of datasheets, some cathode biased class AB amps are specced at 80%ish idle bias, some even biased cold at 55% and some at max dissipation. I don't know about the max dissipation spec but my theory on the colder bias settings is that they are catering for nice clean amplifiers (Think hifi clean, not guitar amp clean). They only expect you to drive the power amp hard enough to produce the highest clean output (with a tiny bit of distortion, 3-5%). So the bias won't shift too much. But in guitar amp land our version of clean actually has a touch more distortion and we love to crank our amplifiers into overdrive. So that's why we have to push our bias that bit hotter.
I do have another part about using fixed bias datasheet spec to determine your cathode bias but I don't know if it's entirely correct.
Littlewyan wrote: ↑Thu Feb 02, 2023 9:57 pm
… some cathode biased class AB amps are specced at 80%ish idle bias, some even biased cold at 55% and some at max dissipation. …
Bear in mind that bias can be tweaked to operate closer to class A. But another approach is to increase the load impedance. Might that align with the 55% example?
It’s about conduction angle, and the change in current from idle to max output, rather than idle dissipation per se.
Littlewyan wrote: ↑Thu Feb 02, 2023 9:57 pm
… some cathode biased class AB amps are specced at 80%ish idle bias, some even biased cold at 55% and some at max dissipation. …
Bear in mind that bias can be tweaked to operate closer to class A. But another approach is to increase the load impedance. Might that align with the 55% example?
It’s about conduction angle, and the change in current from idle to max output, rather than idle dissipation per se.
The amp in question has a 6k6 primary for a quad of EL84s.
martin manning wrote: ↑Thu Feb 02, 2023 1:41 am
Let's see exactly where it is now. What have you got for voltages, plate, screen and cathodes?
Primary impedance 6k6
Plate voltage 333V
Screen voltage 317V
Cathode voltage 10.20V
Littlewyan wrote: ↑Thu Feb 02, 2023 9:57 pm
… some cathode biased class AB amps are specced at 80%ish idle bias, some even biased cold at 55% and some at max dissipation. …
Bear in mind that bias can be tweaked to operate closer to class A. But another approach is to increase the load impedance. Might that align with the 55% example?
It’s about conduction angle, and the change in current from idle to max output, rather than idle dissipation per se.
The amp in question has a 6k6 primary for a quad of EL84s.
martin manning wrote: ↑Thu Feb 02, 2023 1:41 am
Let's see exactly where it is now. What have you got for voltages, plate, screen and cathodes?
Primary impedance 6k6
Plate voltage 333V
Screen voltage 317V
Cathode voltage 10.20V
Bias is 120R/220uF per pair of EL84s
This is 114% plate dissipation (wrt 12W). I'd go with 100%.
pdf64 wrote: ↑Thu Feb 02, 2023 11:55 pm
Bear in mind that bias can be tweaked to operate closer to class A. But another approach is to increase the load impedance. Might that align with the 55% example?
It’s about conduction angle, and the change in current from idle to max output, rather than idle dissipation per se.
The amp in question has a 6k6 primary for a quad of EL84s.
martin manning wrote: ↑Thu Feb 02, 2023 1:41 am
Let's see exactly where it is now. What have you got for voltages, plate, screen and cathodes?
Primary impedance 6k6
Plate voltage 333V
Screen voltage 317V
Cathode voltage 10.20V
Bias is 120R/220uF per pair of EL84s
This is 114% plate dissipation (wrt 12W). I'd go with 100%.
6P14P-EV aren’t a bad option either. They have 14 watt plate dissipation.
Though resistors are cheaper than valves. You might have to increase the cathode resistors more than you think though.
I should have qualified the 114% as including the screen current, since I used Vk and Rk to get Ik. Actual plate dissipation may be closer to 100% as it stands. Setting 100% Pa using cathode current seems to be a reasonable approach in most cases.