PWM 呼吸灯教程

Minimal reference for driving the LCKFB Tianmengxing MSPM0G3507 onboard PB22 LED with PWM.

This example is based on the verified 26testproject5 smoke test. It is not a complete CCS import project. It is an agent-readable reference showing the .syscfg and C patterns that were validated on real hardware.

Clock Note#

This example uses the 80 MHz clock-tree pattern summarized in ../../references/hardware_validation_notes.md:

• CPUCLK: 80 MHz
• HFXT: 40 MHz input on PA5 / PA6
• SYSPLL: enabled
• ULPCLK divider: 2
• MFCLK gate: enabled

This differs from the older empty_project and led_blink examples, which are 32 MHz baseline examples.

PWM Setup#

• PWM module: /ti/driverlib/PWM
• Timer instance: TIMG8
• Output channel: CCP1
• Output pin: PB22
• Generated macro: GPIO_PWM_0_C1_IDX
• PWM period: 1000 timer counts
• PWM clock frequency observed: 2.5 MHz

The verified generated header contained:

#define PWM_0_INST             TIMG8
#define PWM_0_INST_CLK_FREQ    2500000
#define GPIO_PWM_0_C1_PIN      DL_GPIO_PIN_22
#define GPIO_PWM_0_C1_IOMUX    (IOMUX_PINCM50)
#define GPIO_PWM_0_C1_IDX      DL_TIMER_CC_1_INDEX

Files#

• example.syscfg: 80 MHz clock tree and PB22 / TIMG8_CCP1 PWM setup
• src/main.c: DriverLib PWM breathing pattern
• manifest.json: machine-readable summary for example selection

Application Pattern#

The successful breathing pattern:

• Sets the first compare value before starting the timer.
• Updates GPIO_PWM_0_C1_IDX, not channel 0.
• Avoids compare values exactly equal to 0 and period.
• Uses about 100 steps per half-cycle.
• Uses delay_cycles(800000) at 80 MHz, about 10 ms per step.

Observed behavior:

• About 1 second from bright to dim.
• About 1 second from dim to bright.

Failed Attempts#

Initial slow version:

#define STEP_SIZE  100
#define DELAY_CYC  80000000

This produced a very slow fade because each brightness step waited about one second.

Second attempt:

for (int duty = 0; duty <= 1000; duty += 5)

This could make the LED appear off or glitchy on the verified board. Avoid the exact 0 and period boundaries for this smoke test.

Build And Flash Notes#

gmake clean all exposed a generated CCS makefile issue in one test project: Debug/makefile included ../device_linker.cmd even though SysConfig generated Debug/device_linker.cmd.

That is a CCS project/build-file state issue, not a PWM SysConfig failure. Do not patch generated makefiles as the default fix. Regenerate the CCS build files, rebuild in CCS, or use the clean generated linker input path when doing a one-off CLI validation.

For automated flashing after 80 MHz clock-tree changes, use DSLite System Reset:

dslite.bat -c path\to\MSPM0G3507.ccxml -e -r 2 -u path\to\project.out

If DSLite can enumerate the J-Link core but cannot connect to the target, close stale debug sessions or stop stale DSLite / JLink / JLinkGUIServer processes, then retry.