We have a custom board based on the imxRT1021 EVK, and the 24MHz oscillator won't start.
The team has already iterated through resistor and capacitance changes as suggested by AN3208 with no success.
We are able to replicate the issue on both Rev A.3 and Rev B.1 imxRT1021 EVKs by pulling jumper J37, which decouples "MCU_DCDC_IN" from "MCU_DCDC_IN_3v3". We see a valid looking POR sequence and voltage come up on the XTAL points, but no oscillation. This makes us bring our attention to the power supply and power up sequence.
Is there anything in the power-supply timing or iMxRT1021's PMU that would inhibit startup of the 24MHz oscillator? For example, if there is any issue with the DCDC_IN or VDD_HIGH_IN timing or quality that would subsequently cause the downstream LDO to not start the 24MHz circuit?
I'm not able to find documentation directly associating DCDC_IN behavior or power on timing with start of the 24MHz oscillator circuit for the imxRT1021.
Solved! Go to Solution.
FYI - our custom board uses a different 32KHz oscillator circuit than the EVK. It was being overdriven at 2.2V. It is not directly stated in the documentation, but apparently an invalid drive level on the 32kHZ circuit inhibits the 24MHz oscillator kickoff. Reconfiguring to use the built-in 32kHz source allows our system to come up.
So basically, if the 24MHz clock doesn't come up, after checking the usual suspects of resistance, capacitance, POR timing, and supply rails - check the OTHER clock.
FYI - our custom board uses a different 32KHz oscillator circuit than the EVK. It was being overdriven at 2.2V. It is not directly stated in the documentation, but apparently an invalid drive level on the 32kHZ circuit inhibits the 24MHz oscillator kickoff. Reconfiguring to use the built-in 32kHz source allows our system to come up.
So basically, if the 24MHz clock doesn't come up, after checking the usual suspects of resistance, capacitance, POR timing, and supply rails - check the OTHER clock.