CCS/TMS320F280049C: HLC tile problem in the TMS320F280049C

Hello,
I will move this thread to the device forum. The device experts there can help you best. Note that this is a holiday time for many people so responses may be delayed.

Thanks,
ki

ki,

I have no idea what your question is. Can you describe specifically what the problem is with the HLC?

Nima

Hi Nima,

My customer are using CLB tool to generate PWM waveform, I will send the CLB configuration to you by email.

In CLB configuration, use TILE1_COUNTER1 to generate two PWM waveform as below showed. set Match1 value for half of the PWM1 period and Match2 for the others. And Match1 event will trigger HLC, then will run HLC program and move R0 to Match1,  move R1 to Match2.

So in theory change the value of R0, R1 value will change the output PWM frequency, and do the simulation show that it work and PWM period is determined by HLC R0/R1, but from real output waveform show that the PWM period is actually determined by TILE1_COUNTER1 Match1/ Match2 initialization value, which should have only work for one time, and did not updated by HLC R0, R1 value.

For example , if TILE1_COUNTER1 match1_val is 2000, HLC R0_init is 1000, then PWM waveform period in simulation is determined by HLC R0_init(1000), which is correct, but real output PWM waveform period is determined by match1_val(2000). Which show that bypass the HLC.
Could you give some suggestion on how to debug? Thanks.

Strong Zhang

I see in your code that you are generating an interrupt with a TAG of 11. Is that interrupt occurring in your system?

This thread seems to be related to:

https://e2e.ti.com/support/microcontrollers/c2000/f/171/p/870081/3219450#3219450

Please see my last post there. 

Hi Lori,

This issue is difference, in customer code there is no code to update R0/R1,  in CLB, there are have the initialize value R0_init/R1_init, these value should have update match1_val/match2_val, but it did not work.

Ok I will look at this closer and get back to you by end of tomorrow.  Thank you for the feedback.

Strong,

Can you respond to Nima's question - does the CLB interrupt to the C28x happen? 

-Lori

Hi Lori,

There is no CLB interrupt in customer's code, the code that I send is just example code. but the configuration of the CLB are the same.

there is also have another strange scenario.

1. if customer load the example code first, then load their own code,  then everything is ok, and HLC initialization value R0_init will update the PWM period. which means PWM period is determined by R0_init configured in the CLB tool.

2. but if customer load their own code again, then PWM output is wrong, the pwm period is determined by Counter initialization value match1_val.

Read the code and data to look at what are the difference, find that there is no updated value for register address 0x3120 and 0x3130, which should be CLB_INTR_TAG_REG, CLB_DBG_R0. 

If the customer, power cycles the device and loads their application, the R0_init value is not loaded in??

Yes, if power cycle the device and load their application code, the R0_init value is not loaded in,

only load the example code and follow up to  load the application code, then output is ok,

but if load the application code second time, then the output is also wrong.

Zhang, 

Does the application code call initTILE()?

In the generated clb_config.h, do you see the updated init value?

Nima

Hi Nima,

Does the application code call initTILE()?  ---> Yes.

In the generated clb_config.h, do you see the updated init value?---> Yes.