Annirak
Head-Fier
- Joined
- Nov 15, 2010
- Posts
- 57
- Likes
- 26
I moved offices a while back. I carefully packed it back in the original box and took it to the new office building. When I plugged it back in and tried it again, something had shifted. It registers as an audio output device, but when I play music... it doesn't play. Instead, the progress bar never moves. It's long past warranty and as near as I can tell, LightHarmonic is no longer operating under LHLabs. Returning it for service does not appear to be an option. So, I did what any EE would do. I opened it up.
My mac automatically selects 96kHz by default when playing a stream. The internal LEDs show 48kHz, x2. when I measure the output of the clock buffer, I get a clean 96MHz signal, so the clocks are working. WCLK (the word clock in I2S) has a convenient testpoint. It's showing a clean 96kHz signal, exactly as expected. All the solder joins look clean--all the solder joins I can see, anyway. The XMOS microcontroller is BGA, so I can't see underneath it.
My reasoning is as follows: If the system is set up properly, the clocks should go:
96MHz OSC -> ES9018 -> XMOS
At the same time, it's a USB Audio Class 2 device, which uses device-sourced clock. That should mean that the DAC requests more samples when it runs out. If the shift clock isn't arriving at the XMOS, then it won't shift out any samples, which means it won't ask for new samples, from the PC, which means the play bar won't advance.
So where is the clock going wrong? Well, the 96MHz makes it as far as the clock buffer. It makes it as far as the ES9018 because the ES9018 is putting out 96KHz shift clock. The solder on the ES9018 looks good, so I doubt it's failing there. But as I said, I can't examine the XMOS since I don't have an X-Ray inspection unit.
Now, I have a hot air rework station. Is it worth trying to reflow the XMOS again? Or is there something else I should try first?
My mac automatically selects 96kHz by default when playing a stream. The internal LEDs show 48kHz, x2. when I measure the output of the clock buffer, I get a clean 96MHz signal, so the clocks are working. WCLK (the word clock in I2S) has a convenient testpoint. It's showing a clean 96kHz signal, exactly as expected. All the solder joins look clean--all the solder joins I can see, anyway. The XMOS microcontroller is BGA, so I can't see underneath it.
My reasoning is as follows: If the system is set up properly, the clocks should go:
96MHz OSC -> ES9018 -> XMOS
At the same time, it's a USB Audio Class 2 device, which uses device-sourced clock. That should mean that the DAC requests more samples when it runs out. If the shift clock isn't arriving at the XMOS, then it won't shift out any samples, which means it won't ask for new samples, from the PC, which means the play bar won't advance.
So where is the clock going wrong? Well, the 96MHz makes it as far as the clock buffer. It makes it as far as the ES9018 because the ES9018 is putting out 96KHz shift clock. The solder on the ES9018 looks good, so I doubt it's failing there. But as I said, I can't examine the XMOS since I don't have an X-Ray inspection unit.
Now, I have a hot air rework station. Is it worth trying to reflow the XMOS again? Or is there something else I should try first?
