TMS320F28P550SJ: The execution time of the interrupt service routine is much longer when connected to the MCU in CCS

Sorry, I misspoke. The same program runs much faster in CCS, while after directly programming and powering on the MCU, the interrupt execution time is nearly ten times longer. What could be the potential issue here?

Hi Randy,

Apologies for some delay. What optimization level are you using in CCS? What peripheral interrupts are you executing?

Best Regards,

Allison

Hi Allison,

I tried optimization levels 0 to 2, and the results were the same. I am using ADCB1_INT. When running the same code in CCS debug mode, the interrupt execution time is much shorter. However, when using Uniflash to program the MCU and running from a cold boot, the interrupt execution time is significantly longer.

I found that if I place the entire interrupt code in ramfunc, the execution time remains consistent, whether running in CCS debug mode or booting from Flash. What causes this difference?

Best regards,

Randy

Wow, so many information...

We used the same testing method and encountered no issues on the control cards of 280039C/280049C. In CCS, both in debug mode and direct flashing on startup, the interrupt execution time is the same. We are confused as to why using the latest MCU 28p55x/28p65x presents this problem. Previously, we also noticed a significant difference in the time taken to compute the same CRC code between 280039 and 28p55x. The 280039 took only 100ms, while the 28p55x took 600ms. Later, we used ramfuc, which reduced the 600ms to 200ms. This temporarily resolved our issue, but we are still very confused about what is causing this difference...?