Question about DT's Interrupts

[tazntex]

I am trying out Darryl Taylors Interrupts but I have a few questions, I am using a 16F628A, and Interrupt on change of Portb pins 7-4.

The first question is do I still need to set up the INTCON=%00001000, or does DT's Interrupt's take care of that.

The second question is in the main program I am trying to get the processor to sleep until there is a change on portb.

@ INT_ENABLE RBC_INT ; enable external (INT) interrupts
siesta:
@ Sleep
@ nop
GoTo siesta

After spending the last week searching the forum for some clarifications, one mentioned that portb needed to be read to see if a change occurred. How do I go about just reading the port to detect a change? What I mean is, yes I can read portb and store it as a variable but if I am constantly comparing then I won't be sleeping right? Or is there another way?

From what I gathered through the PBP manual and the forum, I need to clear the interrupt flag, intcon.1=0 but from what I can tell from DT's Interrupt's it resets the flag. Is this correct?

According to the 16F628A datasheet I only see the use of one NOP so I believe this is correct if not please correct me.

Another question is regarding the enabling the interrupt before the label. This is puzzling to me how when the interrupt is serviced then returning to the main program how would the interrupt be re-enabled. So I was thinking this:

ISR:
'go do something
trisb = $f0
portb = $1
@ INT_ENABLE RBC_INT ; enable external (INT) interrupts
@ INT_RETURN

Is this right?

And my last question is if my program is to start with let say loop: and my main will be the one called Siesta, always keeping it sleeping until a change on portb 7-4, then should I change @ INT_RETURN to @ INT_RETURN siesta?

Thanks for the help

[HenrikOlsson]

Hi,
1) You don't need to enable the interrupt thru the INTCON register - DTInts does that for you.

2) You must read PortB in the interrupt handler to end the mismatch condition. If you DON'T read the port in the handler the interrupt will fire again as soon you exit the handler you're stuck in an endless loop. Constantly reading and comparing it in the main routine would defeat the whole purpose of interrupts ;-)

3) If you, when you "declare" the interrupts set the clear flag parameter to YES then you don't need to clear the interrupt request flag "manually" in the ISR.

Once in the interrupt handler you don't need to do anything in particular (when using DT-Ints). The @ INT_ENABLE RBC_INT is used to enable the interrupts at the beginning (ie. setting the appropriate interrupt enable bit in the INTCON register etc) and again IF you've manually disabled the interrupts.

As for the last question I don't think it matters much, it will either return to the beginning or it will end up at the NOP and then goto to the beginning which puts it to sleep.

Hope I got that right, and that it helps.
/Henrik.

[Acetronics2]

Hi,Tazntex

Sleep can work in two manners ...

1) whithout interrupts

ahaaaaa ... here, you just have to re-arm the interrupt flag (RBIF )- but not valid IOC interrupt ( RBIE )

in this case, the program stops @ sleep line and, when awaken, continue from there. here, the NOP added is necessary ...

2) With interrupts enabled ( RBIE valid )

the program stops @ sleep, but jumps to location 4 ( interrupt header ) when awaken ...

so, both cases can be used ... and you can choose the one you have better !!!

just depends on what you want to do.


For the example you show ... interrupts are not necessary at all ...

just place the sleep command at the beginning of the program ..., and if awaken, program will continue with treating your input ...

then a " GoTo siesta " will re-enter sleep ...

can be placed in a whole " While -wend " infinite loop too ...

Alain

[Bruce]

Here's an example of wake up from sleep on change without global interrupts enabled - or an interrupt handler.

Without global interrupts enabled, and an interrupt handler, the response time is a LOT faster since there's no context save/restore needed.

@ DEVICE INTRC_OSC,MCLR_OFF,LVP_OFF,WDT_OFF
 
DEFINE OSC 4
DEFINE NO_CLRWDT 1   ' don't need it cleared when it's disabled in config
 
PortVal VAR BYTE
LED VAR PORTB.0      ' alias LED on RB0
IntFlag VAR INTCON.0 ' alias RBIF interrupt flag bit
 
PORTB = 0          ' LEDs off
TRISB = $F0        ' upper 4-bits inputs, lower 4-bits outputs
OPTION_REG.7 = 0   ' internal pull-ups on
INTCON = %00001000 ' global ints disabled, int-on-change enabled for wake up
 
' PORTB should always = %1111xxxx when no switch is pressed
' the lower 4-bits can be anything since these are outputs.
 
Main: 
  GOSUB CheckState   ' check for switch press before sleeping
                     ' and clear interrupt flag bit
 
  @ SLEEP            ' sleep until switch is pressed
 
  TOGGLE LED         ' do stuff here immediately on wake up
  PAUSE 250
  TOGGLE LED
  PAUSE 250
  GOTO Main
 
CheckState:
  WHILE (PORTB >> 4) != 15 ' read port & wait for switch release
     PAUSE 20        ' short debounce period
  WEND
 
  IntFlag = 0        ' clear int-on-change flag bit
 
  RETURN
 
  END

And if you do use interrupts, having something similar to CheckState in the beginning of your interrupt handler helps avoid another interrupt when a switch is released.

I still recommend you learn to use DT_INTS though. It's pretty cool stuff...

[jellis00]

Here's an example of wake up from sleep on change without global interrupts enabled - or an interrupt handler......
And if you do use interrupts, having something similar to CheckState in the beginning of your interrupt handler helps avoid another interrupt when a switch is released.

I still recommend you learn to use DT_INTS though. It's pretty cool stuff...

Bruce, an application I am working on needs to awaken from sleep mode when a USB cable is connected. I was using the Darrel's INT_Handler/USB_Handler from his DT_INTS-18 for servicing the USB interrupt to try to do this. However, DT told me in one post that you can't use his USB_Handler to awaken from sleep mode and I should sense the connection of the USB cable with an RB change interrupt and then in the ISR for the RB change do an @RESET to somehow start the USB service. I think I know how to use the RBC_INT from the DT_INS-18 to sense the USB connection but I don't understand how to setup my code to then start the USB service on the sense of the USB connection. If I place the @RESET statement in the RBC_INT handler, what do I have to have in my Main Loop to get the USB service going (application must then dump EEPROM data that has previously been logged over the USB interface to PC which I already have working) and then have it return to sleep mode after the USB is disconnected??

[Darrel Taylor]

However, DT told me in one post that you can't use his USB_Handler to awaken from sleep mode and I should sense the connection of the USB cable with an RB change interrupt and then in the ISR for the RB change do an @RESET to somehow start the USB service.

Yes I did.
But let's pretend I didn't.

You can put the PIC to sleep, and it will re-enumerate when re-connected.
The main oscillator is stopped in sleep mode, so the USB module's clock also stops.
You can also place the USB module in SUSPEND mode to reduce the current from the USB module. That's done in response to an IDLEIF, currently used to determine un-plugged state in DT_HID.

When the USB cable is plugged in, you'll get ACTVIF interrupts (a signal to RESUME from SUSPEND mode).

So you don't even need to connect USB +5V to an INT pin.
The USB module can detect the changes.

It's just that you have to do it. Nothings automatic.