[00940]

It's been a while I didn't get a soldering iron hot but I received this morning the parts needed for my next project and it feels nice. Good feelings have to be shared 

 
 
Nothing fancy here, it's kind of an opportunity build, based more on what I've got on hand than careful planning. A yamaha integrated just died and it offers me a 36-0-36 xformer (+a weak extra 17v winding), a big convenient heatsink, connectors and a nice case. Here's the plan:
 
 
Power supply:
 

 
The pseudo regs provide decent ripple reduction (accurate voltage values aren't very important here) and a slow start.
 
 
Amp:
 

 
R7 stands for the 12au7 heaters, r2-r9 are a 500k trimmer to set the output voltage at exactly 12.6v and the small transistors might be bc560c or bc337-40 depending on what I find in my boxes. The input stage runs at 3ma per tube or so and the plates sit at about 75v (is there somewhere a study of how different bias for the sohaII input stage sound ?). I might need to add a relay to avoid thumps at startup, I'll see.
 
 
I'll work on it tomorrow and during the week-end. Hopefully, I'll tell you how it ended on Sunday night. Till then, if anyone spots a mistake in there, it might save me some headaches
 
 
 
 
Edit:
 
Power supply done and working properly. I added a pair 0.33r resistors in between the diodes and the first cap of the main power supply, to smooth things a bit (it limits spikes in diodes). Nothing fancy in terms of parts but solid BC and rubycon caps all around. The ugly duck:
 

 

 

[Goobley]

Looks pretty decent
 
I don't know if you saw my thread recently talking ('Opulent Student Ideas' or similar), where I've been looking at something similar. Your design looks very solid to me. However the question that I would ask is: I imagine that R5 represents your headphone load, if this is the case and you have ~300R headphones then everything should be cool. If on the other hand you plan to power low impedance high current headphones (30-50R difficult cans) then I would suggest increasing the size of your output capacitor (LTspice's analyser will tell you where your -3dB point is anyway). I'm sure it's not an issue for the higher impedance cans anyway. I expect you'll need a relay as you have decent sized coupling caps, or just unplug headphones before shutting down and a good minute after the amp is turned on.
 
With regards to different SOHA II biases, I don't know of any studies, however a 12AU7 is a 12AU7 as this is fairly standard valve implementation (nothing odd like the EHHA). This tube is well known for not being very linear at low current and low (everything is relative voltages). A figure I've heard quoted a few times is 130-150V@8mA to get to linear. Obviously thats quite high for a headphone hybrid. I'd be inclined to read a few datasheets for this tube, build it at your panned bias, then have a play!
 
Good luck with the build
 
PS, what do you gain from having so much capacitance on the bottom of that mosfet? I think you're using it as a capacitance multiplier but as far as I'm aware they're supposed to be low-pass filters attached to the base of a semiconductor, I'd be inclined to reduce the capacitance and then add a resistor in between the zeners and the caps (11, 13, 4). Whilst a fairly high value like 20-50k would slow start-up even more, I reckon that you'd find that it's have an even bigger effect on ripple in that supply

 

[00940]

Yes, I've seen your project. You're more ambitious than I'm trying to be here though 
 
The amp will probably be used for my brother's hd600, so the output cap should be big enough. But 220uf should be enough even with 32r cans. With the feedback around the output cap, the -3db point is at about 12hz and we're only 1db down at 20hz.
 
As for the cap multiplier. You're right than adding a resistor in between the zener and the caps would lower the ripple, no reason not to add it and reduce the caps from 3 to 1. On the other hand, keeping 220uf there, going further than 5K doesn't really help the ripple due to the simplicity of the setup. So I'll just have a rc filter of 4.7K/220uf after the zeners. Thanks for the catch.