Know when the softdevice / Interrupt took control

[yuval]

Hi,

I'm developing a unit which i need from time to time to send a packet throw I/O lines (F.E.virtual uart)

However the packet bit timing is very critical.

As far as i understand, when using a softdevice, the softdevice can interrupt me whenever he needed (& also, other interrupts)

that's ok,  but i need to know that the softdevice interrupted me (or other interrupts occered), ruined my packet timing and i'll take care of it and resend the packet again.

What i thought of is making a flag and check at the end of my packet transmition it's status (if it still 0 - then no interrupt were made)....

So, Where shoud i put the "flag=1" ? is there a centrer place that i can raise this flag and know that the soft device / Other interrupt took control/were made ?

B.R

Yuval

[yuval]

Thanks,

But this not working for me since i'm the slave and not the master, so i don't know when the other transmitter sends it's data....(And i act according to the data)

Another question,

If i set a timer event for every 200 micro seconds, (f.e. Inside the event i have 1 line of code - advancing  a counter)

How much time will be spent in going into the event, advancing the counter and return back to the regular program?

B.r, Yuval.

[Einar Thorsrud]

Hi Yuval,

Why do you absolutely need to handle the UART transaction immediately? Does the master device not support flow control?

I don't have numbers for how long time your timer event would take, but it should not take many instructions per time.

[yuval]

Hi,

No, There is no flow control. I can't know when it transmit to me.

About the 2nd question: what do you meen by "not take many instruction"?

Is it up to 10us? 50us? 100us?  I don't need very aqurate number from you but i need to know more or less how much time it will take...

[Einar Thorsrud]

Hi,

What I meant is that reading the timer value and setting a variable does not take many CPU instructions. Then you have the interrupt latency which adds a bit. We are definitely talking about far less than 10 μs in this case.