sampling timing help

[ashrau]

Hi,

I need to sample an analog signal on channel 0 of PIC18F4220. My problem is that i need to take 104 samples, and each of them needs to be spaced exactly 192 microseconds apart. This is for a frequency filter, so the time b/w samples has to be exact.

I am unsure as to how long the a/d acquisition and conversion takes, so that i can accordingly set the delay between samples so that the total delay is exactly 192 us.

Any code along with explanation of what is going on will be greatly appreciated. I would also appreciate a simple explanation of how I can find out the time it takes for the a/d acquisition and conversion process. I am relatively new to microcontrollers. Thanks in advance.

ashrau
_________________
ashrau

[Sydney]

I would setup an rtcc interrupt every 192us, which would be fairly exact. Eg with an 8mhz xtal, and a prescalar of 2 the rtcc will increment every 1us. Or you could clear the rtcc before the conversion, and wait until the rtcc reaches 192 before looping to the next conversion. Either way it doesnt matter how long the conversion takes.

[SherpaDoug]

If you still want to know the time the A/D conversion takes, the very best way is to read the datasheet. It takes a little time to study it but you learn just what is going on.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[Ttelmah]

Realistically, Sydney is right, except possibly for chosing the RTCC.
Provided you are using a PIC, that has a timer2 module, far easier to use this, since you can program it to automatically interrupt/reset at a specified number of counts.
Trying to 'cycle count', to get exactly 192uSec, will involve you in working out not only the ADC timing, but how long every memory transfer takes etc.. Easiest way, is probably to use the MPLAB smulator, and just run the stopwatch function in this,then you can tell exactly how long everything takes if needed.

Best Wishes

[RLScott]

[Guest]

Yes. However I was more worried about the whole approach to cycle counting to get the 192uSc total. It is too easy to miss things like (for instance), it taking longer to access RAM, after one point in the data array, as an extra bank switch appears. Though ccle counting is very essential when using some low level applications, it is easier to get the whole thing accurate, without cumulative errors, by using the interrupt flag (remember you don't actually have to use an _interrupt_ - a search hee will find lots of examples on doing this).

Best Wishes

[Sydney]