Here are "Pete Wood's observations and commentary" on the amps, which I'm quite happy for you to stick in a page somewhere:
If you intend to leave this, or any other item of home-made electronics "on" all the time (and it will sound better if you do) then I recommend constructing them in sealed, earthed metal cases. No ventilation holes on top for smoke or flames to get out of. Mine have never burned in this way, and there's no reason that yours should either, but a university friend's life was seriously endangered when his old NAD amp decided to immolate itself while unattended in the living room. This implies the use of external heatsinks, and NOT using the fancy internal-heatsink-and-fan of the Naim NAP135. We're in the business of having fun here, people, not cremating our families.
The output protection offered by the regulator boards is amazingly good (it will shut down when almost any fault occurs in the amplifier board), but it is not infallible. One of my amps (then aged about 5 years) was subjected to severe mechanical stress by my soldering iron, and subsequently, the +56 V connection into the regulator board came undone. The resulting one-sided PSU could NOT be handled by the regulator, the result was non-trivial damage to the amps (cheap to fix) and two new bass drivers for one of my speakers (not so cheap). It is doubtful any electronic scheme could have saved me in that case; a semi-mechanical over-current detector (ie, an output relay :-)) might have helped.
I found that my amps sounded considerably better when some "audiophile" 4.7 uF Ansar polypropylene caps were placed across the +40V/0/-40V feeds to the amplifier boards. I haven't investigated further, but this kind-of implies that the regulated voltage rails as seen by the amp are noisy. This can hardly be a surprise, as it's a high-current feed and is going through a lot of joints between the output transistor of the regulator board and the output transistor of the amplifier board. The impedance must be significant. It's probably no co-incidence that, in their latest iteration of this circuit (the NAP 500) Naim have chosen to lay out the regulator and amplifier on one PCB, so these can be very close together. (They also provide a separate regulated feed for the early stages of the amplifier itself).
You don't need massive heatsinks for 'normal' listening. When I first built my amps, they got very warm when asked to do any work. As they have subsequently been modified (ie, the current deisgn as detailed in these pages), and the distortion cleaned up (both in the amps and in the preceeding system), the heating has become very much less, they're clearly working much less hard. They will now drive a pair of ProAc Response 3.5s (admittedly not the most onerous of loads) to "very realistic" [ :-) ] levels without becoming more than slightly warm, and mine are best described as inadequately heatsunk.
If you want to turn the amps on and off a lot, I recommend a soft-start circuit in the power supply (it usually goes in the mains feed to the transformer) to limit the inrush current. The inrush of a 530VA toroid feed 2 x 22000 uF is immense, and it doesn't do the capacitors much good. I fitted pretty little mains switches to the backs of my amps, and they're both now burned out and internally bypassed. I simply plug & unplug 'em at the wall, with no soft-start, but it's not ideal.
It's worth arranging the mechanical contruction of the amps so that you can get at both sides of the amplifier and regulator PCB "in-situ". This is not always easy to arrange. Mine is not done this way, which means that to replace a component I have to remove the amp from the unit. To ease this, I use connectors on all the 40V, 56V, 0V and output cables between the boards, which is probably not a good idea from the sound quality point of view.
Refer to Doug Self's Power Amplifier Design Handbook on a regular basis!