I have a TPA3122 built up on my breadboard according to the datasheet simplified application circuit (link: http://www.mouser.com/ds/2/405/slos527a-88277.pdf
) with a couple of changes. I only have the left channel hooked up as a single-ended load, and am using 1uF electrolytic caps as the output filter caps instead of 0.68 uF, due to the limitations of what's in my box. I am powering this with a 12V wall supply.
This setup was working fine, sharing space on the breadboard with a 358 op amp, 7805 regulator, and an ATmega328P microcontroller. For a long while it worked just fine. However, recently a strange problem has occurred since I rearranged the op amp and uC on my board, not touching the TPA3122 (at least I think).
Whenever I plug the board in, the speaker clicks about twice per second. I've verified multiple times that power is getting to all the right places, and when I measure the voltage on LOUT with my multimeter, I get around 0 V instead of the expected 6 V. MUTE and SD are 0V and 5V, respectively (attached to the microcontroller). The problem persists even if I tie these signals directly to GND and the supply. GAIN1 and GAIN0 are connected to GND. All the grounds are connected together, and I have 0.1uF and 470uF bypass caps on both sides of the TPA.
The only thing that will stop whatever is happening is to unplug the 1uF filter cap I have installed. Once one of the leads comes out, sound immediately starts getting through to the speaker.
Just in case the 1uF might be too big, causing the short circuit protection to kick in, I put two 1uFs in series, to give 0.5 uF of capacitance. Same problem happened.
Does anyone know what could cause this problem, and why unplugging the filter cap would fix it? I want to eventually put this onto a PCB and in a box, so I need to understand what will make this issue go away.