TMDSCNCD263: GPT channel configuration

Hi,

If we tried with above value. it going to 200ms. 

Hi Vivek,

Apologies for the wrong caclulation, I have checked again and 0x1388 seems to be the correct value. Can you check with this load value and retry please?

I assume you're using only 200Mhz as the RTI frequency,

Thanks,
G Kowshik 

Hi Kowshik,

Method 1:

If we assuming the RTI  200Mhz mean 

Sys tick happen every 5ns

for 50uS the hex value is 0x2710.

Method 1:

If we assuming the CORE clock 500Mhz mean 

Sys tick happen every 2ns

for 50uS the hex value is 0x61A8

.

I have tried with both value the interrupt not triggered.

Thanks,

Vivek  

Hi Vivek,

The formula for this tick calculation is not right. here's the right one based on the RTI peripheral

(200Mhz/(preload+1) ) *(required time in sec) 

preload here is 1, so for 200Mhz

Then 200mhz/2 = 100Mhz ==>  100Mhz*50uS = 5000 ==> 5000 in hex is 0x1388

Thanks

I have tried with value the interrupt is not triggered. the interrupt triggered once mentioned value these and above 0x4E20

thanks

Hi Kowshik,

I have tried with the same value now interrupt hit the notification function properly as shown below. I tried to find out the exact time with help of pin toggle it is showing 5ms. Could you please correct me anything went wrong from my side and guide us how to capture the notification triggering time in practical. 

I have feed the value for 25uS 50uS. 

Thanks 

Vivek

Hi Vivek,

My idea is - This could be due to the IO toggling as the GPIO IP is located far from the R5 core so the flip channel takes longer than expected. But I'm unsure if this the exact reason. Let me find a more accurate way to measure this.

Thanks

Hi Kowshik,

 Any update on accurate way to measure the timing? thanks

Hi Vivek,

It seems there is a way to much accurately measure this. We can use the CycleP_counter (which counts the R5 cycle counts) i.e., the resolution increases much more. Also there is something called as HALTEN, which means this halts the RTI peripheral when you are in Debug mode i.e., JTAG connected. This way we can run the peripheral once, and when the interrupt is hit we can note the values before and after and subtract them to see the increment count.

You can enable the HALTEN, you can use the code I shared below 

void Write_MMR(unsigned int mmr_address, unsigned int mmr_value){
    unsigned int * p_mmr;
    p_mmr = (unsigned int *) mmr_address;
    *p_mmr = mmr_value;
}

Write_MMR(0x50D01008,0x01234567); //Unlock MSS_CTRL region
Write_MMR(0x50D0100C,0xFEDCBA8);  //Unlock MSS_CTRL region

Write_MMR(0x50D00454,0x1); //Enables the RTI0's Halt enable

Thanks

Hi Kowshik,

Thank you for the above Input.

We have configured the timer as continues mode. 

Test method:

We have enabled the notification and started the timer. the notification function triggered at once. Could you please guide us for continues triggered anything need to modified? 

Thanks

Vivek Arivazhagan

Hi Vivek,

What is the value of the GPTIMER_100_uS ?? The configuration you shared should be enough for the continous mode of operation.

Thanks,
G Kowshik

Hi Kowshik,

Hi Kowshik,

Thank you for your input for measuring accurate time.

As we requested related to the periodic interrupt triggered for continuous mode. 

If we invoked Gpt_EnableNotification(), Inside the notification function then only periodic triggered happening otherwise, its notified at once.

For periodic notification for continuous mode. Could you please correct me. the way of invocation is necessary or not for periodic triggered or we need do anything in software. 

Hi Kowshik,

We have done look like same, but notification triggered at once. It is not happened periodic. Please verified below attached image.

Review and guide us on this issue.

Note: here we configured OS interrupts with CAT 2 (Source ID - 92, priority - 15)

Hi Vivek,

What is the frequency at which you are running your RTI at? It seems the macro says 500Mhz_100uS ? It seems like you're running the gpt at 500MHz, is that correct? 

Also what are the RTI instance you're using for the OS? 

Hi Kowshik,

Sorry for the macro confusion. we are running at 200Mhz. So, we are feed the value of 0x2710 for 100us. we are using RTI1_INT1 instance.

H Vivek,

You are suing RTI1 instance for the OS? Please note to use the first 4 instances for the Usage i.e.,

RTICHANNEL0-3 --> To the OS

RTICHANNEL5-11 --> In your aplication. 

When using vector OS, we have limited the entire RTI0 timer instance to be used by Vector OS and the other instances are left for usage to the application.

Thanks

Hi Kowshik,

Exactly right RTI0 instance used by Vector OS. so, for GPT I am using RTI1 instance.  In RTI1 we have four interrupts (91,92,93,94). I have configured for 4 channels with. two continuous mode and two one-shot modes.

when we verified all the notification function. the notification function triggered at once. but we expected continuous mode notification function triggered based on the time(periodic) correct.

Please provide your input and guide on the periodic mode.

Thanks

Hi Vivek,

Hmm, this is even weirder. Please perform the below tests and let us know what you observe,

1. Once you receive one interrupt and the system stops, try starting the timer again and see if you're getting one more interrupt.

2.Configure all the channels to continous for the RTI1 and test again if you're receiving continous interrupts. 

If you observe the same results after above tests then I recommend you to approach Vector Tech support because you're observing these issues after integrating the veOS.

Thanks

Hi Kowshik,

Let me check and come back.

Thanks

Hi Kowshik,

Please find the Observation below

1. Once you receive one interrupt and the system stops, try starting the timer again and see if you're getting one more interrupt.

     a. If we start the timer again, we received one more interrupt.

  b. If we call enable notification inside the ISR notify function, we are receiving continuous interrupts.

2.Configure all the channels to continuous for the RTI1 and test again if you're receiving continuous interrupts.

If we configured all the channels to continuous mode. we are receiving one-time interrupts only for all the channels.

Do we need to clear any notification flag for continuous interrupts mode?

Thanks

Vivek Arivazhagan

HI Vivek,

Thanks for the clear tests. Can you ensure the in the OS module you have registered these interrupts as IRQ interrupts and not FIQ's?

Hi Kowshik,

Yes, we registered these interrupts as IRQ. Please find the evidence below snapshot.