CCS/CC2650MODA: Fast GPIO toggle

Try to refer to to do micro second delay.

Hello Pritesh,

Is pull configured on the IO? That might explain the difference in rise and fall times.

BTW, It also might be possible for you to implement this interface in the sensor controller if it makes sense.

Thanks Chen for sharing the article. I will try it out.

Hi Eirik,

The current configuration for the IO is open drain with no pull. I need IO to be open drain to connect device to 1-wire network. Out of curiosity I did tryout push-pull configuration but with same result. It seems some thing in the IO controller is blocking calls.

Ultimately I do want to implement this in sensor controller but doesn't sensor controller use/share the same IO controller module on the SOC?

Hi Eirik,

Any update on the issue? Were TI engineers able to replicate the issue that I am having? I really need to know if this is a limitation of CC2650MODA and in that case we have to replace CC2650MODA with an alternative.

Hello Pritesh,

Can you show me how you configured the pin?

What is the value of ONE_PIN?

Hi Eirik,

Below is the parts of the code that assigns PIN ID and get a handle for the ONE_PIN. In the process, I actually figured out where additional execution block is happening in my application. One of the requirement for the protocol is to able to generate a negative falling interrupt when master assert the line and produce delays base on timer as compared to waits. The edge triggered call back with just timer stop/start calls is blocking task for 32us. Even with  call back routine empty, I measured task block of 20us. Is there a way I can reduce call back routine delay? Without GPIO enabled for negative edge trigger I am getting delays of 0.125us for both high to low and low to high transitions.

/* Board Specific header file */
#define Board_BP_ONE_WIRE              Board_BP_Pin_J2_12

/* One Wire */
#define Board_ONE_WIRE                 Board_BP_ONE_WIRE

/* Application Task File */
#define ONE_PIN			       Board_ONE_WIRE


// Define pins used by one wire slave as open drain input/output
const PIN_Config aOwPin[] = {
    ONE_PIN     | PIN_INPUT_EN | PIN_GPIO_OUTPUT_EN | PIN_GPIO_HIGH | PIN_NOPULL | PIN_OPENDRAIN | PIN_DRVSTR_MAX,
    PIN_TERMINATE
};
PIN_State  hOwPinState;
PIN_Handle  hOwPin;

void edgeTriggerCallback(PIN_Handle handle, PIN_Id pinId) {
    //one_falledge_flag = 1;
    //timer1_stop();
    //timer1_start();
}

// Get handle to One Wire Pin
hOwPin = PIN_open(&hOwPinState, aOwPin);
if (!hOwPin) {
  // Handle allocation error
}
// Edge trigger
PIN_registerIntCb(hOwPin,  edgeTriggerCallback);
PIN_setInterrupt(hOwPin,  PIN_ID(ONE_PIN)| PIN_IRQ_NEGEDGE);

Hi Eirik,

Should edge trigger interrupt callback take 20us to enter after an edge occurs on the IO? Or is there something I am missing?

Hi Eiric,

The interrupt latency of about 20us is really affecting to generate the protocol I need. Is there a solution that TI engineers can provide to reduce interrupt latency?