09/04/2026
With the ground loop issues resolved, I moved into deeper testing to see what was limiting performance.
One thing that surprised me was just how much the power supply was affecting the amplifier.
When running the amplifier from a lab supply, I measured an SNR of around 109 dB. However, when switching back to the original PSU, performance dropped noticeably. The main issue was ripple and supply noise (primarily around 100 Hz) coupling into the signal path.
It also became clear that I had not spent enough time ensuring truly clean rails for the preamplifier and input stages.
For this reason, it was back to the drawing board. In this newest iteration, it had the three big improvements:
1. Added RC decoupling networks:
These help isolate the small-signal stages from the high-current output stage, reducing ripple and preventing supply noise from modulating the signal.
2. Introduced dedicated regulated rails (+15V, +24V, -15V):
This provides much cleaner and more stable supply for the preamp and control circuitry.
3. Improved thermal design of the rectifier stage:
Added large copper pours on both sides of the PCB with thermal vias to better dissipate heat from the bridge rectifier diodes and improve long-term reliability.
This iteration should significantly improve noise performance. This means there should be no noise at all coming from the speakers when the amplifier is turned on but no audio is playing. Stay tuned!
