Hi everyone,
I am bringing up a custom board utilizing the MIMXRT1064CVJ5B processor. The board is powered by an external 5V SMPS power supply fed into a 3.3V LDO regulator. I am using a CMSIS-DAP / DAP-Link debug probe to program and debug via the SWD interface inside MCUXpresso IDE.
I am encountering a highly specific, repeatable power state crash when attempting to flash my code.
The Symptoms:
- Instant Power Drop on Connection: When the board is running independently, it works fine and the power LED stays fully illuminated. However, the exact millisecond I connect my DAP-Link probe harness to the board's SWD header, the 3.3V rail collapses, the power LED turns completely off, and the board powers down. This happens before executing any commands or clicking "Debug" in the IDE.
- Momentary Power Recovery and Crash During Debug: If I leave the probe connected and click the Debug button in MCUXpresso, the board suddenly gets powered back up and the LED turns back on. However, right when the flashing sequence begins, the system freezes at 3% progress with a Wire ACK Fault and a bus hang-up at memory location 0x2000A750. Immediately after this crash, the board powers down again and the LED turns off.
LinkServer Debug Console Log Output:
============= SCRIPT: RT1064_connect.scp =============
RT1064 Connect Script
DpID = 0BD11477 APID = 0x04770041
Disabling MPU
Configure FlexRAM for 768KB OC RAM, 128KB I-TCM, 128KB D-TCM
Finished
======================================================
...
Writing 1322080 bytes to address 0x70000000 in Flash
70008000 done 3% (32768 out of 1048576)
request to clear DAP error failed - status 5
After error Nn(05). Wire ACK Fault in DAP access -
Failed to read address register in DAP - Nn(05). Wire ACK Fault in DAP access
failed to send op Terminate message - rc Em(17). Debug port inaccessible after access at location 0x2000A750
Target error from Commit Flash write: Em(17). Debug port inaccessible after access at location 0x2000A750
What I have tried so far:
- Hardwired BOOT_MODE[1:0] to 01 (Serial Downloader mode) via physical solder pads to isolate application interference.
- Target memory spaces are properly mapped to the internal 4MB QSPI flash baseline region at 0x70000000 within the IDE linker properties.
- The exact same DAP-Link probe seamlessly programs an official NXP MIMXRT1064-EVK kit using identical code and configuration parameters.

Why is the connection causing an immediate power drop, why does the board temporarily wake up during debugging, and why does it consistently deadlock at the 3% flash mark? Any guidance on how to reliably program the chip with this layout behavior would be highly appreciated.
Thanks!