mark

one suggestion: you can restructure the code to make it more portable and readable.

for example, using functions:

void i2c_start(void) {
SSP1CON2bits.SEN = 1;                 //SET SEN  
while (!PIR1bits.SSP1IF);         //WAIT ACK FROM SLAVE
PIR1bits.SSP1IF =0;         //CLEAR FLAG
}

void i2c_write(unsigned char dat) {
SSP1BUF = dat;                         //ADRRESS OF the oled  
while (!PIR1bits.SSP1IF);         //WAIT ACK FROM SLAVE
PIR1bits.SSP1IF =0;         //CLEAR FLAG
}

and your code becomes
i2c_start();  //send the start condition
i2c_write(0x78);
...
i2c_stop();  //send the stop condition

far more readable than your existing code.

also, if you were to change to software i2c, you just need to change the i2c routines and the above code stays the same.

回复【3楼】millwood0  
one suggestion: you can restructure the code to make it more portable and readable.
for example, using functions:
void i2c_start(void) {
ssp1con2bits.sen = 1;                 //set sen  
while (!pir1bits.ssp1if);         //wait ack from slave
pir1bits.ssp1if =0;         //clear flag
}

far more readable than your existing code.

also, if you were to change to software i2c, you just need to change the i2c routines and the above code stays the same.

-----------------------------------------------------------------------
谢谢指正！
我主要是担心函数调用引进时延会带来误码，所以就用执行时间最确定的方式累赘地写代码了。
我写的笨方式的代码在100k没有问题，当时钟到1M也跑得正常，之后放到400k就比较稳妥了。
我确实不知道起始条件和结束条件与通信过程之间的时间间隔允许多大。不过有了1M调通的底，按你说的模式改得好看一点肯定行得通。
再次感谢！

楼主，把你的列子给我们共享一下好吗？