Other Parts Discussed in Thread: PCM3168, PCM3168A,
G'day,
I have a noise issue with a PCM3168A design even when the input data (Din) is silence. The design is typical of that supplied within the datasheet. The DAC outputs seems to randomly jump around as shown. What's confusing, the noise appears to be digital switching. (Not what one expects to see from a DAC output). Although there also appears to be some ramping. The observed noise is present on all channels and is unlikely to be caused from cross coupling as the noise signal is differential across the DAC+ and DAC- outputs. There are also no other digital signals running near the DAC traces. When the device is held in reset, the noise disappears. The noise is also present before the PCM3168 is initialized via the I2C.. albeit the edge rates are much more defined (i.e. square).. The noise is also superimposed on the top of a valid DAC output when driven from a valid Din.. I've attached a scope reading from DACA+ (I can confirm that DACA- is the complementary.). The CODEC settings:
- Software control configured for Left Justified TDM (using all six ADC input channels and only 6 of the 8 DAC output channels. All channels are unrelated independent streams).
- Fs = 8kHz (note: the datasheet does not explicitly use this sample rate within the detailed specifications, however, it is listed as the minimum usable frequency within the Electrical Characteristics)
- Data size: 8 Bits (adhering to the Left Justified TDM, 8 MSBs with the remaining 16bits usable (or 24 given there are 32 BCLKs) padded with 0).
- SKI = 2.048MHz the clock is generated by a standalone 2.048MHz oscillator.. this drives both the PCM3168 and the clock domain of an STM32F7xx SAI (Sync Audio interface).
The output looks the same from an unused DAC channel (not connected to anything) as one of the used channels. (Connected to a BP 150Hz - 4kHz 2nd order AC coupled differential filter).
Apart from the noise, the design seems to work as expected... can anyone please shed some light?
Possible thoughts:
1. Issues with running Fs = 8kHz.
2. Some sort of syncing issue.. although it seems to work quite well when working and the said noise can be seen riding on a valid output..
Dan.
