Rotary Encoder and the LP

[RobG 1,892]

So here is my final (for now) encoder implementation, most reliable 2 pin so far.

1. Increased capacitors to 1uF electrolytic (did try to debounce in software with WDT, but results were not so good.)

2. Encoder A is P1.6, B is P1.7

3. No other pins will use interrupts on P1

4. Only one switch can interrupt at any time, this eliminates most of the problems with rocking and bouncing

5. I really need to get something better than PEC11, about 90-120 deg of the turn produces horrible results.

 

 

#include "msp430g2553.h"
//required funct.asm

#define APIN BIT6
#define BPIN BIT7
#define DPINT P1IE = 0;
#define CLRPFG P1IFG = 0;
#define LOWERNIBBLE BIT0 + BIT1 + BIT2 + BIT3
#define UPPERNIBBLE BIT4 + BIT5 + BIT6 + BIT7

#define ENABLEPIN BIT4
#define REGISTERSELECTPIN BIT5

#define sendData(data) send(data, 1)
#define sendInstruction(data) send(data, 0)
#define initDisplay() sendInstruction(0x3C); sendInstruction(0x0C); clearDisplay(); sendInstruction(0x06)
#define clearDisplay() sendInstruction(0x01); _delay_cycles(2000)

void binaryToASCIIScale(unsigned int n, unsigned char * digits, unsigned int scale);	// asm prototype

unsigned char d[5] = {0,0,0,0,0};

void sendDataArray(unsigned char data[], char length);
void send(unsigned char data, unsigned char registerSelect);

char charIndex = 0;
char valueIndex = 0;

int counter = 500;

void main(void) {
   WDTCTL = WDTPW + WDTHOLD;

   BCSCTL1 = CALBC1_1MHZ;
   DCOCTL = CALDCO_1MHZ;

   P1OUT = ENABLEPIN;
   P1DIR = ENABLEPIN + REGISTERSELECTPIN + LOWERNIBBLE;

   P2SEL &= ~(BIT6|BIT7);
   P2DIR = UPPERNIBBLE;

   P1IES = BPIN;
   CLRPFG;

   _delay_cycles(10000); // delay to allow display to settle down, might not be needed
   initDisplay();
   _bis_SR_register(GIE);
   P1IE = BPIN;

   while(1) {
       sendInstruction(0x80);
       binaryToASCIIScale(counter>>1, d, 0);
       sendDataArray(d, 5);
       _bis_SR_register(LPM0_bits);
   }
}

void sendDataArray(unsigned char data[], char length) {
   charIndex = 0;
   while(charIndex < length) {
       sendData(data[charIndex]);
       charIndex++;
   }
}

void send(unsigned char data, unsigned char registerSelect) {
   P1OUT &= 0xF0;
   P1OUT |= data & 0x0F;
   P2OUT &= 0x0F;
   P2OUT |= data & 0xF0;
   registerSelect ? (P1OUT |= REGISTERSELECTPIN) : (P1OUT &= ~REGISTERSELECTPIN);
   P1OUT &= ~ENABLEPIN;
   P1OUT |= ENABLEPIN;
}

// Port 1 interrupt service routine
#pragma vector=PORT1_VECTOR
__interrupt void Port_1(void) {
   DPINT;                  // disable interrupts on port 1
   if(P1IFG & APIN) { // interrupt came from switch A
       CLRPFG;           // clear interrupt flags
       if((P1IES & APIN) ^ (P1IN & BPIN)) { // XOR IES of A with level of B to determine dir
           counter--;
       } else {
           counter++;
       }
       (P1IN & BPIN) ? (P1IES |= BPIN) : (P1IES &= ~BPIN); // set IES of B based on it's current level, safer than toggling
       P1IE = BPIN; // enable interrupts for switch B
   } else {                  // interrupt came from switch A
       CLRPFG;           // clear interrupt flags
       if((P1IES & BPIN) ^ (P1IN & APIN)) { // XOR IES of B with level of A to determine dir
           counter++;
       } else {
           counter--;
       }
       (P1IN & APIN) ? (P1IES |= APIN) : (P1IES &= ~APIN); // set IES of A based on it's current level
       P1IE = APIN; // enable interrupts for switch A
   }
   _bic_SR_register_on_exit(LPM0_bits);
}

[kenemon 29]

did you guys see this one?

 

http://cgi.ebay.com/ws/eBayISAPI.dll?Vi ... K:MEWNX:IT

 

hell of a deal, but no push switch

[kenemon 29]

Hey Rob, please help me find funct.asm

[RobG 1,892]

Here it is funct.rar

[Mac 67]

... you've come up with simplest and most elegant solution I've seen.

@Mac: Please don't heap praise on me. I stand on the shoulders of others on this topic. In this case, I located the sample code on this blog entry. I only adapted it to CSS from mspgcc.

Then thank you (Zeke) for passing along a good idea. I implemented it (on the circuit below) by adding a few lines of code to my parallel switch state logic and it's working surprisingly well. Since the switch state logic is designed to filter out all but a "new press" switch state, it works extremely well for filtering out all but a single A or B encoder transition between detents. Check out this excerpt from my test program (a small eight line mux' routine fills the 'swnew' sample variable);

 

    /*                                                                  *
    *  Mike McLaren's parallel switch state logic (plus encoder)       *
    *                                                                  *
    *  swnew  ___---___---___---___   inverted active lo sample        *
    *  swold  ____---___---___---__   switch state latch               *
    *  swnew  ___-__-__-__-__-__-__   changes, press or release        *
    *  swnew  ___-_____-_____-_____   filter out 'release' bits        *
    *  flags  ____------______-----   toggle flag bits for main        *
    *                                                                  */
       swnew ^= swold;             // changes, press or release
       swold ^= swnew;             // update switch state latch
       swnew &= swold & 0xEF;      // filter out 'release' bits

       if(swnew & 0x20)            // if encoder B "new press"
       { if(swold & 0x10)          // if opposite direction
           swnew ^= 0x30;          // toggle up/dn switch flags
       }                           //
   /*                                                                  *
    *  test for encoder "new press" and bump "count" accordingly       *
    *                                                                  */
       if(swnew & 0x20)            // increment 'count'
         count = _bcd_add_short(count,0x01);
       if(swnew & 0x10)            // decrement 'count'
         count = _bcd_add_short(count,0x99);

I'll post a demo' program in a new thread. That is, if anyone is interested.

 

Cheerful regards, Mike

[zeke 693]

I'll post a demo' program in a new thread. That is, if anyone is interested.

 

Don't think like that. Of course we're interested!

 

:thumbup:

[gwdeveloper 275]

Good work to everyone on all of this. I have a question about simplifying the debouncing of a rotary encoder... Could a hardware debouncer such as the Maxim MAX6817 be used to get a pure, clean digital signal from a rotary encoder?

http://www.maxim-ic.com/datasheet/index.mvp/id/1896

 

I got a few of the 6817s as samples for a device I'm working on after seeing this post http://dangerousprototypes.com/2011/09/25/debounced-breadboard-keypad/ on dangerousprototypes' blog. Tonight, I was thinking it might work on a rotary as well.

[bluehash 1,581]

I would go for a hardware debouncer if it weren't too expensive. You should whip out a booster pack.

[zeke 693]

I believe RobG already implemented a hardware debounce circuit.

 

I wonder which post that's hiding in?

 

Oh! It's this one. Back on page 2 there are a couple hardware debounce circuits.

[gwdeveloper 275]

@zeke thanks, man. It should probably work with the MAX6817 then with no issues and less parts count.

 

@bluehash That's a great idea. I have a few of the MAX6818 octal debouncers as well. Maybe a shield that has a the 6818, one or two rotary encoders and a few buttons all hardware debounced?

---- Honestly, I'm still struggling through making my own schematics and boards in Eagle. Is there a file or post somewhere that contains a blank shield template to start with? I've already created the part and footprint for the 6817 if anyone wants it.

[gordon 229]

Unless someone beats me to it (do! please do! pretty please with sugar on top!), I'll try to glue together a LaunchPad device in the coming days.

[gordon 229]

Here's a quick one. Haven't even made a printout to compare to the device, so quick one . Should be more or less OK, though. Includes the main LaunchPad obstacles on tDocu, might come handy if you plan populating the bottom as well.

 

I'll eventually add a reduced 5cm x 5cm package as well, as it is a popular cheap option.

 

(BH: .lbr is still not allowed)

 

Edit: the current version can be found at viewtopic.php?f=10&t=1684#p11477

[SugarAddict 227]

Fixed some misalignment and increased pad sizes to be comparable to the LP pads.

TI_launchpad.zip

[gwdeveloper 275]

Thanks guys! I'm going to finish off the MAX681x library with a single and octal and design a booster 'shield'. (or Sword as I read it in another thread )

[gordon 229]

design a booster 'shield'. (or Sword as I read it in another thread )

I thought we called them Payloads :shh:

 

There are tiny bits of enhancements still to come. I did them already, now waiting for SugarAddict to come around, check them then post the updated lib (I'll probably be fast asleep the whole weekend); you may want to wait for that .