Thats great, thanks!

Ok, I'm getting ready to make this, but don't really know where to get a 50VCT transformer. Would a 50VA transformer be enough for 25 audio watts? Thanks.

Posted a better copy:
http://www.diystompboxes.com/analogalchemy/sch/diodepwr.html

This is how "diode compression" is added to the amp. It does a similar thing to this pedal:
http://www.diystompboxes.com/analogalchemy/sch/vulcan.html

The first waveform shows the amp with the diodes disabled (shorted with a jumper) being heavily overdriven at the input. The second picture shows the effect of the diodes. The amount of peak compression depends on the value of the 680K resistor, which in this situation can be from 470K - 1M.

The diodes are biased so that they aren't quite on, but not quite off either. In this situation, the diodes resist signal peaks as they rise, producing the more curved output.

This is the latest circuit, which "probably" will bias itself alright. Went back to the basic circuit, which is probably good enough for government work.

It looks like a strange amplifier, but in reality it's not that much different than anything else. The 1uF/4.7K is a positive-feedback network which increases gain.

Different supply voltages are easily accomodated by changing both 22K resistors to be around the value of the supply voltage. So a 50-0-50 volt transformer would use 47K, etc.

I had a Dual Rectifier years ago, and the high-gain clips sound close from what I remember. But it's been a LONG time. 

A little update on the amp project:

Have been doing a lot of simulation work, and it looks like I can make the "diode compression" trick work with the Depalma amplifier. This will be a very easy build (~40 parts), but I'm pretty sure it needs to be capacitor-coupled because of possible DC offset problems.

If the extra peak compression is worth bothering with (the Depalma amp already soft-clips) then it's very easy to make this switchable, or even variable.

I've made a cheap 12" combo test rig & chassis for the project already, just need to build the thing when I get some parts in.

Sorry again about the files, will keep them available in the future.

Sorry guys, my computer was ruined in a flood, so I can't replace the images. The good news is that they probably weren't worth looking at anyway

Thanks for the reply, but I don't feel it's a waste of time to work on something new. I've already tested this and know that it's a great sounding circuit.

Back to the short protection, the zener diodes work in simulation, so I'll just do that and see if the amp blows up when I short the leads.

Ok, I am working on a design similar to this:
http://depalma.pair.com/Analog/analog3.jpg

I've built a couple of cap-coupled jobs, now working on a direct-coupled version in simulation. I want to add short-circuit protection, and I'm down to three options:

1. Forget it, and use the caps/single supply.
2. Use typical VI limiting and hope it works.
3. Place zener diodes from output to driver bases, hoping to limit the current to a known maximum.

Also, I can't guarantee that the bias won't shift very slightly with temperature. It's not much at all, but possible to get 1/8V or something if it gets really hot.

My first two builds (one 6W/channel stereo and one for headphones) sound nice and punchy, so I feel it's a design worth exploring a bit more.

Built and tested this:

http://www.diystompboxes.com/analogalchemy/tmp/newamp.gif
http://www.diystompboxes.com/analogalchemy/tmp/clip.gif

I have to rebuild it for several reasons, but the initial test was a thumbs up. This amp sounds great (check out the soft-clipping on the simulation).

The 2nd stage is prone to thermal runaway so requires good heatsinking. The output is centered with the 10k trimmer, and the quiescent current with the 1k trimmer, like most other amps. The part running off the bottom is the short-circuit protection circuit. Negative feedback is through the 470k resistors.

Edit:
I don't have a scope so I can't do much more testing than that. 4 ohms didn't really work in my proto (hum/oscillation), but I mounted the transistors/board pretty close to the fuse/power transformer. The heatsinking was probably inadequate anyhow. I also think better transistor choices can be made than the relics I was using.

http://www.diystompboxes.com/analogalchemy/tmp/amp7.gif

I planned on doubling up the outputs but it's shown now. I raised the 100 ohms back to 470 ohms because of the higher rails. The output centers up better (at least on the simulator) this way.

I decided to build it (nicely this time, heh). Should get about 100W, give or take depending on what values are used for the 470 ohm resistors.

Thanks for the correction and other information. Will check into adding a limiter to it.

few changes, but not done:
http://www.diystompboxes.com/analogalchemy/tmp/amp6.jpg

The fuse is there as a reminder that it needs output protection of some sort. Will probably put this on the back burner for awhile.

Went back to the drawing board (not complete):
http://www.diystompboxes.com/analogalchemy/tmp/amp5.gif

Basically a giant opamp. Probably needs some capacitors here and there. Automatically adjusts for rail voltage, just need to adjust the Vbe multiplier. Thoughts and ideas welcome.

I settled on a 10k volume, but I agree it's not the best situation. The preamp will have a buffer before the tone stack, hopefully that will help.

Owell, it's a cheapskate amp but was a good learning experience.