Originally Posted by scootsit
I found that page, but couldn't figure out where I got the 15v bias. I'm using 12FK6 tubes, do I'm not sure why I biased so high. Thanks!
Yes - for the 12FK6 and 12AE6 tubes, the best bias is 13.5V. This assumes that you have the power supply output adjusted to 27VDC. So, that gives you an equal voltage swing up and down around 13.5V.
The 12FM6 - as stated - seems to run better at 15V. Although that gives you unequal voltage swing with a 27VDC power supply setting, there are really very few times one would use that entire swing. So, it seems to work a bit better overall. Obviously, if you could set the voltage supply to 30VDC, then it would work even better. However, years of testing with very high current 24VAC walwarts (up to 2 Amps), 29VDC seems to be the absolute limit before the LM317 loses regulation (and thus, the PS gets noisy). We settled on the conservative 27VDC setting recommendation to account for worldwide variances in line voltage and walwarts.
You might be able to get 28-28.5VDC power supply voltage if you try, but you run the danger of losing LM317 regulation and making the amp noisy. You can see this yourself by adjusting the PS trimmer. Simply keep turning the voltage up until it stops increasing. To be absolutely certain you haven't exceeded the regulation capability of the LM317, dial it back by about 2VDC. The LM317 needs to burn about 1.5VDC to maintain regulation, so if you dial back 2V to be absolutely safe, it often ends up at 27.5 - 27VDC.
The primary basis of the voltage increase from 24VAC to 27-29VDC is through rectification and cap smoothing. While an over-abundance of current capability will buy you some more voltage, I think it's in the fractions of volts or less. All that pretty much means you need to implement your own transformer if you want 30VDC on the PCB, or - accept a very slight chance of clipping on the positive side of the voltage swing from 15V to 27VDC.
The discussions on the 12FM6 occurred only through forum posts, either in here somewhere or on diyforms.org back with the original revMH Millett Hybrid. I can't remember who originally discovered the optimum tube curves for the 12FM6, but Pete Millett even mentioned higher biases to reduce tube distortion in his original Audio Xpress article:
"Build A Low-Voltage Tube Hybrid Headphone/Line Amp"
Look at pages 28-29 (pages 9 and 12 of the PDF excerpt) for Pete's discussion on raising the bias to reduce the distortion, at the expense of lower output (less room before clipping).
Edited by tomb - 3/31/17 at 8:08am