Here's a couple of images of the "Leach Amp"
https://users.ece.gatech.edu/mleach/lowtim/ I built in the mid 80's.
The case was made from scratch. I had access to a machine shop at the time. The cover is entirely perforated.
The rear panel has a power fuse, 2 speaker fuses, speaker terminations. isolated RCA Phono connectors and 2 non-polarized convenience outlets.
There is no power switch. My plans were to add a left clip, right clip, power and over-temp, The protect board had LED drives on it.
The inside:
The 4 x 9,600 uf caps were mounted with custom brackets.
There's a custom toroidal transformer under the PCB toward the front. The transformer is 4 x 35 VAC@3A . The fuseholders for the four 50 VDC supply fuseholders hang off the transformer on another custom bracket. Really close to the toroidal transformer there is another small transformer that supplies 12 VDC for the protection circuit which was designed by me. The PCB was hand done.
The protection circuit also has two isolators mounted near the inputs. There's a relay to disconnect the speakers and a relay that shorts out a line series resistor when every supply is > 2/3 of 50 V approximately. This means that if any individual supply dies (fuse blows), that flameproof resistor dies. The protection circuit could be improved. ZNR's HAD to be added to prevent my supply monitoring opto-isolators from dying.
There is no DC monitoring of the speaker terminals - just an AGX fast acing fuse. Offset is < 5 mV.
The heatsinks are massive. There are 3 diodes mounted to the heatsink. Two are basically a press-fit into holes on the heat sink and one parallel with the body on the heatsink. Having three makes it really easy to mount. They run really cool.
As designed, the boards could mount directly to the heatsink. RCA jacks were added directly to the board. Molex connectors for power and speaker allow me to remove the boards using a cable extender for troubleshooting. The protection flame-proof resistor MUST be bypassed when bringing the amp up with a Variac.
The boards are basically RF constructed with a solid ground plane on the top. Nearly all the resistors are metal film 1%. The bias adjustment is 10 turn. I accidentally made a mirror image of the PCB based on the magazine article and I was able to use the boards by switching the NPN and PNP transistors and making
My initial version had a 18 A ferro-resonant transformer (way too noisy) and the amp sounded better, I'm using a 500 W sine wave voltage regulator now, Measured, I was able to get about 100-125 W into 8 ohms, one channel driven without the regulator.
The parts list said 100 uf 50 V caps and that's what I used and one blew, hence the reason for the protection circuit. One rail died, I don;t think you can beat the sound. The output transistors are rated for a collector current of like 30 Amps. The power supply in the he Audio Magazine article specified was 3 A 70 V CT. This is too small. The speakers mentioned in the article were electrostatic, so the current doesn't matter much. Like a lot of stuff, the case and power supply use up most of the costs.
EDIT: The protection circuit (mains current limiting, AGX fuse, speaker relay, and Opto-FET in series with the input) baically does:
1) DC across speakers - AGX fuse
2) Speaker relay: Turn on/off thumps
3) Opto-FET: turn-on/off thumps, It ramps up the audio over about 10 seconds.
4) Resistor - Turn-on transients. Charging 40,000 uf of capacitance isn't good, so it charges them slowly to 2/3 of 50 V before turning on the speakers, ans starting the audio ramp.