We are developing a new audio product with the TAS5756 in 2.0 mode with HybridFlow 1. I followed the reference design as closely as possible including output inductors.
SPK_GAIN_FREQ is set to 2.6V (20dB, 8x fs).
I’ve run into a strange issue.
Most of the time, it works. But sometimes when the TAS5756M comes out of power down (e.g. after there was no sound for a while) it switches to the lowest switch frequency , like the SPK_GAIN_FREQ is set to zero.
As the inductors are only 4.7uH, the lowest switch frequency causes a large idle current to go through the inductors and everything heats up quickly. Stopping and then resuming the music (to make it power down) usually makes the TAS5756 go to the correct setting.
The I2S signal is fixed 48 kHz signal from a ComTrue CT-5302 chip, using it’s MCLK signal. (The CT-5302 is an ASRC chip.).
The class-d switch frequency is checked by looking at the amplifier’s output signal on an analogue scope screen.
Some observations:
- When audio stream becomes all digital zero, the amp will switch off (automute?)
- When the audio stream starts to have non-zero audio, the amplifier becomes active. Most of the time (98%) it starts at the correct frequency & amplification. Sometimes it starts like the SPK_GAIN_FREQ pin 9 has zero voltage on it (20dB, 4x Fs). Stopping the audio (press pause on source) and then resume will most of the time make the amp go back to the correct settings.
- When playing really loud (some clipping) with real music & real speakers, sometimes the amp will jump to 4x Fs during playback.
- When using the resistors to select 6x Fs, 26dB (5V on the SPK_ GAIN _FREQ. pin 9), the problem is still there. When it switches by itself to 4x FS, the amplification falls back to 20dB. I can hear this clearly, during very loud playing, suddenly the sound becomes less loud. On the scope I can see the waveform jumping to ca. 400khz.
What I tried:
- Tried to use a PVDD of 12V, 15V, 20V
- Tried another sample board to make sure the TAS chip or any other component on the board was not defective
- Used a lab bench power supply to supply voltage to SPK_ GAIN _FREQ, pin 9 to rule out problems with the resistor network & EMI influence.
- Rechecked the PCB layout, but it looks very close to the evaluation board. I have only 0.1uF + 10uF MLCC on the PVDD, but there is a big low-ESR EC cap nearby.
- Repeatedly assert SPK_MUTE, which also puts the amp to power down. Also here in about 2% of the cases it comes back with the wrong switch freq.
- Looked at the 3.3V voltage with a scope to see if any variation.
Anyone else ever saw this? What more can I do to troubleshoot?
Rob Smit
