Problem getting counter value with HLC interrupt

You have counters, edge detection and FSMs. You should be able to do all of your logic inside the CLB tile to get the time between two falling edges. Also you have ECAP which can do the same thing.

As far as your specific design, you want to push the Counter value to the FIFO, then generate an interrupt and read the FIFO in your C28x CPU interrupt handler routine.

Not exactly sure where the issue is in your design but which CLB example have you reviewed and used?

Nima,

I can't use ECAP because It uses my processor and I need it for something else in parallel. I currently use counters, edge detection and FSMs but as I said, I don't have the value that I expected in my interruption function.

And yes, I tried to get my value by the way you described and also by reading directly the counter register, but for both ways I don't get the value I expected.

I used exemples 11 and 13, based on interrupt tag and push_pull.

Abdoul.

Go through example 7 and the application report associate with it please. 

Nima

I went to this example and read the report associate but it didn't really help me as this exemple show how to use the PULL instruction.

Here are screenshots of my interupt function and my HLC program:

Abdoul

Sorry, I clicked on resolved by mistake...

I can read it but now, the only values that I get are 0 and 1 randomly, both with the FIFO and by reading the counter register. I still can't read the value 55 that i'm supposed to have.

When is the HLC triggered to transfer the COUNTER value into the FIFO?

I checked the push_data_en signal of my HLC and I actually found that the push instruction was delayed by a few cycles. When it occurs, the counter value is indeed 0. But, as I move the counter value in R1 register earlier it should not be a problem to me, should it?

I understand what you are saying to me and I can stop my counter as you said but there is still a problem. My input signal is an external signal, and it will keep scrolling whether the counter is enabled or not. So if I disable my counter, I could have a wrong decoding of the next value, as it will start counting too late.

What I suspect is happening is your values are either being overwritten or the counter value continues to increase/overflow/reset while your HLC is running and executing. So you should really change the logic to when a capture has occured, stop the counter, execute the HLC instructions, reenable the CLB COUNTER.

How fast is your signal?

Okay I see what you mean and I understand your suspicions and the approach that you advise me.

As I said my signal looks like a PWM but the period between 2 falling edges determines my value so I don't have a precise signal period. I don't know if you know Sent protocol frames but I will not enter the details here because it will be too long to explain but I count up every 3 µs and the 12 firsts count up return the value 0, then the value is normally incremented every 3 µs. The problem is that, if i miss the first count for exemple, I will have a value reduce by one... Even worst, if the value that I'm supposed to get is 0 (12 counts up), and i miss a count, my signal will be declared wrong...

Maybe I can try to decode my frames by reducing all the verification by 1, it will let me stop the counter to get the value and reset/restart it during the first count.

I don't know if I am clear enough for you but let me know if there is a point I poorly explain.

The minimum amount of time I need is 36 µs.

So, I remove the PUSH and the MOVE instructions from my HLC and I change my interrupt function this way:

But it is still the same values, only ones and zeros...

Maybe I did something wrong but, to me, on line 187, I stop the counter, then on line 189 I read my value and finally on line 190, I restart the counter.

I finally succeded to read the correct value after stopping the timer in the CLB Tile design.

Thank you very much for your help

Fantastic :) !