Schematic capture and bias voltage simulation in OrCAD. The load simulates an open circuit. I chose R10 to be 370 Ohms so that the bias current through the driving transistors (in the next picture) is close to 120 mA, as suggested by the designer of the amp.
Schematic capture and bias currents simulation in OrCAD. The offset at the output is 1.368 * 1,000,000,000 = 1.368 mV, not bad. Also note that 120.4 mA bias current.
Max output voltage is about 26 V, corresponding to a maximum 1.2 V max input, which is perfect for maximum line-level input.
RMS and peak power.
The simulated frequency response shows 8 Hz - 60 kHz @ -3 dB (16.3 voltage gain), or 18 Hz - 33 kHz @ -1 dB, which matches to the design specification well.
Schematic of the power supply. I used an on-delay relay to avoid the transient thump that lower-quality amps make sometimes on boot. I used 35 V rails instead of the 40 V rails because that's the transformer I had handy. The fan is run off of another set of 12 V secondaries that the transformer has. I stole the low power rectifier and filter cap as an assembled PCB out of a 12 V wall adapter.
I designed the PCB in Cadence Allegro PCB designer. I had footprints for the most typical components, but for the high power resistors and most of the transistors I had to make my own footprints from either modifying footprints or creating my own pad stacks. The PCBs were milled in-house at Rose-Hulman. My first attempt at the boards failed: I populated an entire board and went to solder on the output MOSFETs and noticed they were upside-down. I had a second set milled and they worked great.
The tested frequency response. I didn't test at a frequency low enough to show the lower -3 dB points, but the -1 dB points were similar to simulation, and the midband gain on average is about 23 g(V), which is also pretty close to simulation.
Completed channel, ready for mounting in the chassis. The heat sinks are cut from stock I had lying around. I didn't do any calculation, just figured that with enough ventilation that they'd be big enough, so there's a 2" hole cut in the bottom of the chassis to allow the fan to pull air up through the heat sink blades.
Complete enough to run tests on. No front panel or controls. I just hooked up a function generator to the inputs and scoped the outputs into a 9 Ohm 500 W load that I have for testing amps.
Completed interior, front panel installed. The small PCB over by the fan is the rectifier that I stole out of a wall adapter. The green device that's mounted to the rear is the on-delay relay for the output. I wanted to leave plenty of room on the front panel for analog VU meters in the future.
Interior of the finished product to date.
audio amplifier

I built this 2 x 50 W/channel professional audio amplifier from scratch for a class at Rose-Hulman. I used the schematic from and modified it a little. I reused a lot of parts in construction from other projects and the PCBs were milled in-house. Overall the project was pretty low-cost: about $200 finished. I’d like to add some analog VU meters to the front soon. The original specs are 90 W into 4 Ohms, 30 Hz - 20 kHz : -1 dB frequency response, THD @ 1 kHz = 0.01% @ 20 W.

Brian Cherbak
Engineer at MindTribe Product Engineering San Francisco, CA