Hi All
as title.
I am nRF52 DK PCA10040.
Chip select is nRF52832.
I measure out from wfe to Timer Interrupt time is 1.3us.
Is it correct? and How can I reduce the delay time from 1.3us to 0.3us or less?
Austin
Hi All
as title.
I am nRF52 DK PCA10040.
Chip select is nRF52832.
I measure out from wfe to Timer Interrupt time is 1.3us.
Is it correct? and How can I reduce the delay time from 1.3us to 0.3us or less?
Austin
Hi Austin,
the specification says that startup time for internal ocsillator is 3us, so 1.3us is quite a good result. To achieve faster response, you can try to force HFCLK to run continuously by keeping some peripheral running (for example, timer at 1MHz), but power consumption will increase significantly.
Hi Dmitry
1. System already in HFCLK
2. Timer at 16MHz
Also, the question is why enter Timer interrupt so slow?
In my imagine, it should be less then 0.3us.
I need more time to execute more instructions in once interrupt.
Austin
Ahh, I missed out that your interrupt already comes from timer... HFCLK is running.
Did you write your own low-level interrupt handler, or use app_timer handler? What is an interrupt delay if you remove WFE instruction?
For nRF52832, CPU wake-up time (TIDLE2CPU)=3us. Other tickets (for example, this and this) have reported 10-12 us delay when running from flash and 3-4 us when running from RAM. Especially, look at the second link, an oscillogram for "running from RAM" case - the delay between hardware PPI-triggered event and interrupt handler is ~1.1us, so I don't think you can get much less.
Ahh, I missed out that your interrupt already comes from timer... HFCLK is running.
Did you write your own low-level interrupt handler, or use app_timer handler? What is an interrupt delay if you remove WFE instruction?
For nRF52832, CPU wake-up time (TIDLE2CPU)=3us. Other tickets (for example, this and this) have reported 10-12 us delay when running from flash and 3-4 us when running from RAM. Especially, look at the second link, an oscillogram for "running from RAM" case - the delay between hardware PPI-triggered event and interrupt handler is ~1.1us, so I don't think you can get much less.