Proved inverted inputs and pullup mode

[Beerlesklopfer]

Hi Bertrand,

this is the output of the "InvertedInputs.ino" sketch. So it is proved to work correctly.

Best gregards Joerg

INTERRUPT PORT A FIRED

INTERRUPT PORT B FIRED

Checking setting port mode

------------- SUCCEEDED --------------

Checking setting pin mode

INTERRUPT PORT A FIRED

INTERRUPT PORT B FIRED

------------- SUCCEEDED --------------

[blemasle]

It seems to me that there is a mistake in your code. Feel free to correct me If I'm wrong :)

[blemasle]

Why did you invert the iodir assignation logic between OUTPUT and INPUT ? An OUTPUT pin must be cleared according to the datasheet, so iodir &= ~_BV(pin); was the correct way of handling this.

[Beerlesklopfer]

I did it first as you suggested, but the the result was just inverted as expected. This way solved that problem. My wiring was as described in the new example.

[blemasle]

I'm lost with that change. The code you're suggesting was initially the one I wrote, before it was corrected by #4.
I'll try to have a deeper look at this myself this weekend and sort it out once and for all. If that's true, then the datasheet is lying ^^

[blemasle]

So, I finally had the chance to get an mcp23017 out of the closet and test the library again.
I stick with my version of iodir, and this is very easy to prove :

I have a bunch of input switches on port B (active high), and some leds on port A.

#include <Wire.h>
#include <MCP23017.h>

#define MCP23017_ADDR 0x20
MCP23017 mcp = MCP23017(MCP23017_ADDR);

void setup() {
    Wire.begin();
    Serial.begin(115200);
    
    mcp.init();
    mcp.portMode(MCP23017_PORT::A, 0);         //Port A as output
    mcp.portMode(MCP23017_PORT::B, 0b11111111);//Port B as input

    mcp.writeRegister(MCP23017_REGISTER::GPIOA, 0x00); //Reset port A 
    mcp.writeRegister(MCP23017_REGISTER::GPIOB, 0x00); //Reset port B

    mcp.writeRegister(MCP23017_REGISTER::GPPUA, 0x00); //No pull ups on port A
    mcp.writeRegister(MCP23017_REGISTER::GPPUB, 0x00); //No pull ups on port B

    mcp.pinMode(7, INPUT);         //Pin 7 as input

    uint8_t conf = mcp.readRegister(MCP23017_REGISTER::IODIRA);
    Serial.print("IODIRA : ");
    Serial.print(conf, BIN);
    Serial.println();    
}


void loop() {
    uint8_t currentB;

    currentB = mcp.readPort(MCP23017_PORT::B);
    mcp.writePort(MCP23017_PORT::A, currentB);
}

IODIRA : 10000000

In the example above, the led on pin 7 is not configured as an input, so it does not lit, whereas every other led do lit. This is with the current version of the library. Integrating your changes on IODIR gives

IODIRA : 0

And all the leds are working (which proves that pin 7 is not configured as an input.

[blemasle]

Solved by v2.0.0. Feel free to reopen the discussion if you still disagree on the resolution. But in this is case, please read the comments on your example first. Your sketch has mistakes that clearly indicates you misused the library :)

Thanks for the suggestion, glad to see it finally made it to the library ! 👍

[blemasle]

Same remark as above :)

[Beerlesklopfer]

me too 👍

[blemasle]

The second parameter of port mode use an uint8_t, each bit represents the direction of the corresponding pin. As the MCP23017 use 1 for input, and 0 for output, here you're actually setting the port A to be an output port entirely.

Suggested change

	mcp.portMode( MCP23017_PORT::A, INPUT, 0xFF, 0xFF);
	mcp.portMode( MCP23017_PORT::A, 0xFF, 0xFF, 0xFF);

Would have produced the result you were looking for.

[blemasle]

Same as above, OUTPUT on the arduino is 1, to here you're only setting the first pin of the B port as an input.

Suggested change

	mcp.portMode(MCP23017_PORT::B, OUTPUT);
	mcp.portMode(MCP23017_PORT::B, 0x00);