There are actually some requirement on the 24MHz crystal. You may check it out in the datasheet and ensure that your crystal satisfied them.
I also assumed that your voltage detector circuit does indeed give you a delay of more than 50ms.
The key thing is that the device must be woken up into the HCI mode, and this is done by shorting BT_UART_CTS_N while powering up.
Can you relax your load capacitors (C11 and C12) to 10pF and observe whether the situation improve?
I have verified that the voltage detector circuit delays enabling the device by more than 50 ms and the issue persists. The crystal I am using is https://ecsxtal.com/store/pdf/ECX-2236.pdf. (24MHz ±15ppm Crystal 8pF 60 Ohms 4-SMD, No Lead )
I also replaced C11 and C12 with 10pF capacitors. The waveform is shown below:
1 of 1 people found this helpful
For chip-on-board design, it's inevitable that you will have to perform some tuning on the crystal. I can suggest the following:
1) swap another crystal
2) change the load capacitors to 12pF/15pF and observe
3) employ unequal load capacitors. You reported above that the amplitude of the waveform was high during start-up, then you may increase the loading capacitor on the Input port to say 12pF and maintain 10pF for the other one.
For the record, our reference design used TXC 8Y24070015 for the 24MHz crystal with 12pF as the external loading capacitors.
Taking a step back, you have apparently built a PCB board and doing some testing, our discussion may be limited to component changes
on the schematic. However I know that issues arose out of less than ideal layout and placement are a different ball game altogether.
Chip on board designs are beyond the scope of this forum. If you still encounter issues, please log in a case at the Cypress URL.