Lift off

[LRRockBox]

So I started on designing my first amp. Not being someone that believes in half measures, I decided to look into power supply design for starters.
I've decided that I want to build a class A amp, delivering 100W into 8 Oms, that's absolutely stable to 1 Ohm, doubling power as impedance goes doewn. So it needs to deliver 800 W way down there.
I've already run into the first snag, heat dissipation. at 800W, I would need to dissipate 1600W. At 100W this is a more manageble 200W, but still, no mean feat ! I'm not in favour of using fans, and using liquid nitrogen might prove a tad cumbersome.
Question : how do the big boys solve this ? I have read about 'anticipator' style designs, that adapt bias according to input signal. Has anyone already done work on this, are there diy solutions available that sort this out.
Input ! More input !!!

Thanks in advance,

Peter R.

[sreten]

No sensible design is still class A into 1R loads.

Your best bet is to design for say nominally 100W class A
into say a 6R load and then let it enter class B for lower
impedances.

This is ~ 3.2A standing current and ~ +/- 45V rails.

You''ll probably need 6 or 8 devices per channel for 1R loads,
and of course a whopping transformer and lots of PS capacitance.

As you can see dissapation will be ~ 300W per channel,
so lots of heatsinking also.

Personally I'd go for optimally biased class aB. With 8 devices
per channel and 0.1R emitter resistors an EF stage would
have around 800mA of standing current, this is enough to
almost say the first 5 watts are class A.

sreten.