I2C Slave with a PIC

[aratti]

I need this because i build it from the status of four bits of Port A

Any idea?

Ralf, you can try this:

Select Case PortA

Case 1
i2caddress CON $01
Case 2
i2caddress CON $02
case 3
i2caddress CON $03
.
.
.
.
Case x
i2caddress CON $x
End Select

It should work.

Edited:

No! it doesn't work sorry.

Al.

[Dave]

ralfmayr, Here is what I have been using for years. It uses the base address for the first of 3 PCF8574's. I then calculate the address depending on the one I need to gather or send data to.

UNIT0 CON %01000000 'U5's CONTROL BYTE (PCF8574) I/O EXPANDER

'**************** I2C I/O ROUTINE (PCF8574A) I/O EXPANDER ************
READ_WRITE VAR BIT 'READ DATA or WRITE DATA FLAG
EXPAND_UNIT VAR BYTE 'CONTROL BYTE FOR I2C
DEVICE VAR BYTE 'SELECTED I/O EXPANDER DEVICE
OUTPUTS VAR BYTE[3] 'RELAY OUTPUT BYTES
SWITCHES VAR BYTE[3] 'SWITCH INPUTS BYTES

'************************************************* *******************
R_WI2CS:'READ DIGITAL INPUTS FROM or WRITE DIGITAL OUTPUTS TO PCF8574'S
'************************************************* *******************
EXPAND_UNIT = UNIT0 'COPY BASE ADDRESS OF PCF8574'S
EXPAND_UNIT = (EXPAND_UNIT | (DEVICE << 1)) 'SELECT WHICH DEVICE
IF READ_WRITE = 0 THEN 'READ INPUT STATUS FROM PCF8574'S
I2CREAD SDA,SCL,EXPAND_UNIT,[SWITCHES(DEVICE)]
ELSE 'SET OUTPUT STATES TO PCF8574'S
I2CWRITE SDA,SCL,EXPAND_UNIT,[OUTPUTS(DEVICE)]
ENDIF
RETURN


Then to use it first set the device index.
Next set the read_write flag.
Then finally call the routine.....

Dave Purola,
N8NTA

[aratti]

Ralf, you can try the case select using the proper addresses (not my example). You must correct your code as per the line in black bold and comment out the line in red bold. The code compile so it should work. Longing to hear from you on the final test.

Al.

' Initialize I2C slave mode

Select case PortA

Case 1
SSPADD = 2
case 2 
SSPADD = 3
case 3
SSPADD = 4

end select

SCLDIR = 1  ' SCL must be made an input before enabling interrupts
SDADIR = 1   
'SSPADD = I2Caddress ' Set our address
SSPCON2.7 = 0 ' General call address disabled
SSPCON = $36 ' Set to I2C slave with 7-bit address
SSPSTAT = 0
SSPIE = 1 
SSPIF = 0 
RxBufferIndex = 0
TxBufferIndex = 0

'Initialization Done!

 
GoTo main

i2cslave: ' I2C slave subroutine
SSPIF = 0 ' Clear interrupt flag
IF R_W = 1 Then i2crd ' Read data from us
IF BF = 0 Then i2cexit ' Nothing in buffer so exit
IF D_A = 1 Then i2cwr ' Data for us (not address)
IF SSPBUF != SSPADD Then i2cexit ' Clear the address from the buffer
readcnt = 0 ' Mark as first read
GoTo i2cexit

[ralfmayr]

Hi Al,
my mistake, it works now.
I made a mistake in calculating the address from four bits of a port :-(

This works:

main_loop:
...
i2caddress = (addr0 * 16) + (addr1 * 8) + (addr2 * 4) + (addr3 * 2)
SSP1ADD = I2Caddress
...
goto main_loop

...in the moment i try to implement the slave routine into a 18LF14K22...
It does not work in the moment, it is the same like on a18F6722, there it
works great.

Regards,
Ralf

[aratti]

i2caddress = (addr0 * 16) + (addr1 * 8) + (addr2 * 4) + (addr3 * 2)
SSP1ADD = I2Caddress

Very good Ralf !

Al.

[ralfmayr]

Hi Al,
success!
The slave routine runs in a 18LF14K22.
I read back via i2c 20 values (a/d values and digital i/o) and i can set
several parameters with two bytes, 1st is array index, 2nd is value.
The handler runs via interrupt, in the moment i am cleaning a little bit,
but i am very happy with it.
Regards,
Ralf

[ralfmayr]

Some Remarks:

1. It is necessary to set the "DEFINE I2C_HOLD 1" in the master to get a
better behaviour of the timing. I checkrd it with the logic analyzer and it is
exactly like in the original "Phillips-Spec"

2. In the section
...
while STAT_BF = 1
i = SSPBUF ' Dummy Read inserted => clears BF Flag
wend
...
it is necessary to inlude the dummy read to get it work on
a 18F6722 and a 18LF14K22 (refer to MCHIP App Note 734, page 17)

3. Address range ist tested from 00h to FFh, all work well.

4. New address setting works (in the main loop) by only setting SSPADD
register with new value without reseting the complete handler.

Regards,
Ralf