Trouble Initializing AD4696 over SPI – Errors & Setup Questions

Context:
I'm trying to initialize the AD4696 ADC using SPI on an STM32 (running at 400 MHz). Below is a high-level summary of the initialization function I’ve written, followed by a few questions about behavior I'm observing.

Summary of Functional Steps in AD4696_Init()

This function initializes and verifies the AD4696 ADC through a retry mechanism. Here’s the sequence of operations:

• Retry Loop
  Attempts initialization up to 3 times if any critical checks fail.

• SPI Bus Setup

  • SPI peripheral is initialized.

  • SPI bus is reset to start clean.

• Configure SPI Speed (for Initialization)

  • SPI speed is set to a lower, more reliable rate during setup.

• Hardware Reset

  • The ADC is hardware-reset and given time to stabilize.

• Clear Device Status

  • Internal status registers are cleared to remove stale error flags.

• SPI Communication Check

  • The ADC is probed to confirm it's responding to SPI reads.

• Vendor ID Check

  • Vendor ID low and high bytes are read to confirm the device identity.

• Enter Configuration Mode

  • Verifies whether the ADC is in configuration mode.

  • If not, issues a command or register write to enter it.

  • Re-checks mode status after the attempt.

• Scratch Pad Test

  • Writes a test value to a scratch register.

  • Reads it back to confirm communication integrity.

• Health Checks

  • Reads the SPI status register.

  • Checks for interface and general fault conditions.

• Enter Conversion Mode

  • Puts the ADC into normal operation (conversion) mode.

• Switch to High-Speed SPI

  • Attempts to raise SPI clock speed for regular data acquisition.

  • If unsuccessful, continues at the slower rate.

• Success or Retry

  • If all checks pass, initialization completes.

  • If not, the routine retries up to 3 times before giving up.

Questions

1. Is it safe to continue if the SPI_STATUS register shows errors like CRC_ERROR, SCK_ERR, or INV_WRITE, even though the scratch pad test passes?
   I still see these errors after multiple initialization attempts, but basic communication seems fine otherwise.

2. I only plan to read from the ADC (sensor data). Are these errors worth debugging thoroughly, or is it acceptable to log them and proceed?
   Writes aren't critical in my use case.

3. What are the recommended SPI frequencies for:

   • Configuration mode?

   • Conversion mode?
     I’m currently running both at 25 MHz. The scratch pad test succeeds at this rate.

4. Why is the Vendor ID high byte returning inconsistent or incorrect values, while the low byte always returns the expected 0x56?
   Could this be a timing issue or something else?

5. Am I overengineering this?
   I'm relatively new to this, and I want to be sure I'm not overcomplicating what could be a straightforward process.

Any insights or advice would be greatly appreciated. Thanks!