Multiple Interrupts PIC16F917

januszkl · Joined: 10 Sep 2007 Posts: 2

1. Does 16F917 have the ability to service multiple interrupts?
The problem I encounter is that my TIMER1 ISR cannot be set as the highest priority interrupt. I tried the '#INT_TIMER1 HIGH' directive but get an error, 'Invalid interrupt directive requires #DEVICE HIGH_INTS=TRUE'.

I need the time to increment even if there are other 'things' being processed at the time.

When the button on port RB0/INT is pressed and held down the timer stops counting. I need the time to increment even while the button is pressed (EXT_ISR() is being processed), or even if the hell brakes lose.

2. Why capacitors are required on the Vdd power supply pins?

Ttelmah · Guest

HIGH_INTS, only applies to _18_ family pics.
The same applies to the 'FAST', and 'HIGH' keywords.

The 16, can 'service multiple interrupts', but only ever one at a time. (In fact this is true of any 'single processor', but on chips like the 18x chips, an interrupt can interrupt aother event, on the 16, it can't).
The way interrupts work on the 16 chips, is that when an interrupt 'event' occurs (three things needed - the interrupt flag goes 'true', the interrupt enable for this source is set, and the global interrupt enable is set), the code will execute a 'call' on first cycle of the next machine instruction, to address 0004. This call is special, in that it automatically _clears_the global interrupt enable bit, preventing any other interrupt from triggering. Then the code has to save all registers. This is done automatically for you by the compiler. Then the code has to check _which_ interrupt bit is set, to trigger the interrupt. Now the order in which this is done, determines which interrupt is serviced _first_, if multiple sources have triggered. This is set by the #priority keyword (if this is not present, it is determined by the order the individual handler routines are defined).
Once the source is identified, _it's_handler routine is called. Then the flag is cleared - beware though that some events like 'serial data available', and 'port B changed', _require_ the hardware event to be cleared in the handler, or the flag bit will immediately reset. The registers are restored, and a special 'return' instruction is used, which resets the global interrupt enable, ater it executes. If another interrupt flag is still set, or when the next interrupt flag triggers, the whole squence will trigger again.

Now, comments.
The big problem is that you are sitting waiting in the RB interrupt. This is basically a 'no no'. _All interupt handlers, should always be kept as short as possible_. The other problem is that you are writing to the display in the handler, which will imply that all interrupts will be disabled in all display operations in the main. Don't do this...

The way to solve this is:
int1 key_pressed=false;

Guest · Posted: Mon Oct 22, 2007 11:19 am

That is one of the best supports I have got regarding MCU's. Thank you for your explanations.

I hope you can assist me further and explain your comment about setting timers.

Ttelmah · Guest

You should be concerned about the PIC being a source of noise. It can affect _itself_, and also things like the LCD. LCD's are also often quite 'noisy' (the line scans can create some quite nasty clock patterns).
Capacitors should draw basically nothing (leakage currents out in the femto amp range typically), having a poorly smoothed supply, can result in unexpected signal triggers, and make consumption _worse_...
The problem is that the point when you arrive in the interrupt, is not 100% deterministic. If a key even, just happens to be already being handled when the timer triggers, there will be a small extra delay. There will have been anything between perhaps 35, and 100 instruction times, from the moment of the timer 'event', and reaching the line where you set the timer. This _will_lead to errors.
You don't really need to set the timer to a count. There have been efficient routines posted here in the past (using binary wrapping), which allow an accurate RTC, to be done with a timer that runs at a really 'odd' ratio to the required time. These wll give accurate results.

Best Wishes

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

rnielsen · Joined: 23 Sep 2003 Posts: 852 Location: Utah

I, almost always, place a cap next to each IC in my circuits. You never know what stray noise there might be out there. You have your PIC running at around 20MHZ with this crystal thingy buzzing. It's bound to be creating something that a neighboring part is going to be annoyed by. Kinda like a little gnat flying next to your ear that you can't seem to swat.

It's not just the PIC that you should be worried about but it's all the other components that you should be concerned about too. Caps are cheep, especially surface mount ones, and really don't take up that much real estate.

If you're going to be doing any ADC stuff, you'll want to use caps to decouple things or you could have a lot of noise there. Ground planes work well to help with noise.

You have analog grounds and digital grounds. You'll learn to keep these two separate or the digital grounds could make the analog ones noisy.

Just a few things to consider. Clear as mud now?

Ronald

Guest · Posted: Tue Oct 23, 2007 1:34 pm

Thank you ALL for your good advise. I will do some PICing this coming weekend and let you know about the progress on this project.

"To catch a PIC you have to think and act like a PIC."