ESB acknowledgment time-out value - clarification

[kammutierspule]

I would like to understand the timeout values for wait for acknowledge on ESB protocol

Example, from the source code:

#define RX_WAIT_FOR_ACK_TIMEOUT_US_1MBPS        (73)        /**< 1 Mb RX wait for acknowledgment time-out value. Smallest reliable value - 68. */

It considers the timer timeout for receiving an Ack to be 73us

If I consider

Preamble (8bits) + Address (5*8bits) + LengthField (6bits) (Note: S1 should be skip at this point) + CRC 16bit

That gives 70 bits that should be 70us at 1MBPS

I understand from the comment that it could work even with 68us but that would be lesser than the value I was expecting

Could you explain the ACK timeouts defined for ESP protocol?

Isn't the timeout value too small and risk ?

[ovrebekk]

Hi Mario

The timeout timer is stopped as soon as the ADDRESS event of the RX packet occurs, it doesn't wait for the END event. 

In other words the timeout doesn't have to be large enough to cover the entire length of the packet. 

Best regards
Torbjørn

[kammutierspule]

I understand the idea, I believe that was the described behavior on the nRF24L01+ datasheet.
but looking on the nrf_esb.c (I'm looking to SDK 17.0.2) I can't find where it deals the things that way.
Could you point me in the esb sourcecode what makes it work as you describe?

[ovrebekk]

Hi Mario

If you mean the act of stopping the timer on the ADDRESS event, this is handled by a PPI channel, as configured on line 573 in nrf_esb.c:

    NRF_PPI->CH[NRF_ESB_PPI_TIMER_STOP].EEP  = (uint32_t)&NRF_RADIO->EVENTS_ADDRESS;
    NRF_PPI->CH[NRF_ESB_PPI_TIMER_STOP].TEP  = (uint32_t)&NRF_ESB_SYS_TIMER->TASKS_SHUTDOWN;

This ensures that the ADDRESS event will immediately stop the timer, without having to wait for the software to respond to the event. 

Best regards
Torbjørn

[kammutierspule]

Thanks a lot, I understand now!