It wasn't too bad. I just had to use multiple vendors to get all of the parts. Bill was something like $300. The amp was borrowed and used in a disco and money was "no object" because of the "borrowed" aspect.
The toughest one was the Leach Amp. I made a mirror image of the board and did not realize it, I was able to use the board by moving the NPN's to the PNP spots and cutting and moving only two traces for the bias regulator. That was amazing. It went through a few iterations. I used a CVT with like a 20 A secondary that hummed to much and I used 100 uF 50 V caps on the board that EXPLODED. The supplies were 50. Won't do that agian.
So, I moved it to a 2 RU home built case and ordered a custom 35 VAC x4 at 3A toroidal transformer. In hindsight, it should have been higher current. The construction article said 3A total and I doubled it. 40,000 uF of capacitance total on 4 supplies. Thus I needed a slow turn-on circuit and protection. Nothing really fancy, but I switched the NPN and PNP outputs by mistake and only had to replace one resistor. Nothing but a 3A AGX fuse for speaker protection. But, if any of the amp rail fuses pop, a metal oxide resistor will pop.
Now,, it runs with a 500 W Sorenson AC sine voltage regulator as front end. I bought that for $100 in NY. Worth about $2000 when new. It gets improved bass with the regulator.
I added input and output thump protection. The input protection was necessary, because the delay was based on all of the filter caps getting to 2/3 final voltage. That's when the metal oxide resistor got shorted out. It's very unusual to have rail fuses, but a "power OP amp" on one rail isn't nice.
The input has a FET-Optocoupler in series and the audio ramps up logarithmically which is really cool.
The amp has an unheard of un-rolled off frequency response of 0 to 800 kHz. Not a typo. it's rolled off to be 0.5 Hz to 40 KHz. Built on a ground plane with 99% metal film resistors and a 10 turn pot for the bias adjust. The heatsink runs really really cool.
I never got around to a "clipping indicator" or a high temp sensor or improving the turn-on circuit. I made a clipping indicator or a work project that was really cool using a bi-color LED and therefore it detected both positive and negative clips and extended the duration. Turn-on improvement would have been something to prevent the resistor from popping. i.e. If all of the supplies weren't up to 2/3 max, then shut down the amp. The bigger problem is room in the enclosure.