SPI help

[das]

I am having SPI issues trying to use a MCP3551 (A/D converter) with a PIC16F877. I use the setup_spi(MASTER | H_TO_L | DIV_4) and use the format to read with byte_1 = spi_read(byte1); (continuing to clock for the other 2 bytes). I am assuming the data that I want is byte1, but both byte_1 and byte1 return 0. Can anyone help... even when I tried bit-banging the SPI I still get zeros. Please help - I need to get some sort of a solution working today... Thanks.

[ckielstra]

I'm not really sure how you have set up your program. As always, post a very short but complete example program showing your problem, this makes the discussion much easier and gives you quicker and better responses.

[das]

Here is a short snippet from the code...

if(input(PIN_C4 == 0))
{
delay_us(10);
byte_3 = spi_read(0);
delay_us(5);
byte_2 = spi_read(0);
delay_us(5);
byte_1 = spi_read(0);

value = make32(byte_4, byte_3, byte_2, byte_1);

This was the latest that I wrote (that failed). You think I should use:
if(input(PIN_C4 == 0))
{
delay_us(10);
byte_3 = spi_read();
delay_us(5);
byte_2 = spi_read();
delay_us(5);
byte_1 = spi_read();

value = make32(byte_4, byte_3, byte_2, byte_1);

[das]

Changing to the following:

if(input(PIN_C4 == 0))
{
delay_us(10);
byte_3 = spi_read();
delay_us(5);
byte_2 = spi_read();
delay_us(5);
byte_1 = spi_read();

value = make32(byte_4, byte_3, byte_2, byte_1);

Still results in a return of 0 for all the bytes...

[Ttelmah]

You don't show what you are doing to the CS line?.
You _must_ have a pull-up resistor on the SD line.
The sequence needed, is to drop CS, raise it, then drop it again, and raise it again. You then read the SDI pin, and when it drops, data is available. You don't need any delays between the reads.
I suspect you don't have a pull up resistor on the SD line, so it is being seen as 'low', when it is actually in the high impedance state (conversion not yet complete).

Best Wishes

[das]

When I initialize (at the start of the program) I set CS to low (and have a pull down resistor on it. I don't currently have a pull up or pull down resistor on DO. I messed around with putting one there, but there were no change in results. I put the delays in because initially when I was running this bits 23 and 22 were both 1 (indicating that there was an SPI communication error - from the datasheets).

From what the data sheets said, it did not appear to need a pull up or pull down resistor on the Data Out pin.

[Ttelmah]

You are running the chip in continuous conversion mode. Unfortunately, this is a pig to drive from the PIC. The reason is that the 'status' doesn't become available, until an extra clock is generated on the clock line. After transferring 24bits, the line remains as the last data bit, until the extra clock is sent. You are going to have to turn off the SPI peripheral, and generate one extra clock on the SPI clock line, then you can use the data line to detect the status. Without this, the status is not available.
Using 24 clocks, you need to operate the CS, to get the status.

Best Wishes

[das]

I am trying to operate it in single conversion mode right now. Do you know if anyone has ever written a driver for this?

[ckielstra]

[ckielstra]

[das]

I will try that. Right now, here is the current code:

#int_RDA
#define SPI_MODE_0_0 (SPI_L_TO_H | SPI_XMIT_L_TO_H)
#define SPI_MODE_0_1 (SPI_L_TO_H)
#define SPI_MODE_1_0 (SPI_H_TO_L)
#define SPI_MODE_1_1 (SPI_H_TO_L | SPI_XMIT_L_TO_H)

int byte1, byte2, byte3, byte4;
int byte_1, byte_2, byte_3, byte_4;
int32 value;
int i;
int completed, loop;



RDA_isr()
{
puts("RDA was tripped");
}



void main()
{
int loop = 1;
OUTPUT_HIGH(PIN_D0); //Start with CS high

setup_spi(SPI_MASTER | SPI_MODE_0_0 | SPI_CLK_DIV_4 );

delay_ms(5000); //Delay 5 seconds
puts("Magnum Torque Meter Initialized");
while(TRUE)
{
enable_interrupts(INT_RDA);
enable_interrupts(GLOBAL);
loop = 1;
OUTPUT_LOW(PIN_D0); //Start the conversion
delay_ms(10); //Delay 10 ms for the conversion to complete
do //Do...while loop
{
if(input(PIN_C4) == 0) //Check to see if the conversion is completed
{
byte_4 = spi_read(byte4);
byte_3 = spi_read(byte3);
byte_2 = spi_read(byte2);
byte_1 = spi_read(byte1);

value = make32(byte_4, byte_3, byte_2, byte_1);
for(i = 0; i < 32; i++)
{
putc(bit_test(value, 31 - i));
}
OUTPUT_HIGH(PIN_D0);
completed = 1;
}

if(input(PIN_C4) == 1)
{
delay_us(1);
} //Closing if PIN_
}while(loop = 1);
delay_ms(10000);
}
}

[ckielstra]

Some general remarks:
- When posting code on this forum please use the 'code' buttons. It will help to preserve the layout of your program and makes it easier for us to read the code.
- When debugging a problem like this, get rid of all other program code that might interfere. Especially enabled interrupts might cause unexpected side effects.
- Never use a printf or put instruction in an interrupt handler unless you know and accept the side effects. Receiving a character on the UART will take the same amount of time as sending a single character. You are receiving one character and transmitting 10-fold, this will block receiving and cause buffer overflow. The UART will stop responding at receive buffer overflow (the third character). Resetting the UART requires specific actions or adding the ERRORS directive to the #use RS232 directive.


Then about your program:
- Move the enable_interrupt() functions out of the while loop. You never disable these functions, so why add the overhead?
- I made a mistake in an earlier post. For the spi_read() to generate a clock signal you have to pass the spi_read() function a value, however this doesn't have to be a variable. Just any constant value will do for generating the clock signal, for example spi_read(0).
- while(loop = 1); Change this to '=='
- In SPI-mode 0,0 you are supposed to read 3 bytes, not 4.
- You said you want to operate in single conversion mode. If so, then your handling of the ChipSelect line seems incorrect (read Ttelmah's note on this).

[das]

I refined some of the code shown above - and made changes to keep it in constant conversion mode. I am still seeing all outputs as zeros. Any other ideas? Current code is:

#int_RDA
#define SPI_MODE_0_0 (SPI_L_TO_H | SPI_XMIT_L_TO_H)
#define SPI_MODE_0_1 (SPI_L_TO_H)
#define SPI_MODE_1_0 (SPI_H_TO_L)
#define SPI_MODE_1_1 (SPI_H_TO_L | SPI_XMIT_L_TO_H)

int byte1, byte2, byte3, byte4;
int byte_1, byte_2, byte_3, byte_4;
int32 value;
int i;
int completed, loop;



RDA_isr()
{
puts("RDA was tripped");
}



void main()
{
int loop = 1;
OUTPUT_LOW(PIN_D0); //Start with CS high

setup_spi(SPI_MASTER | SPI_MODE_0_0 | SPI_CLK_DIV_4 );

delay_ms(5000); //Delay 5 seconds
puts("Initialized"); //Output string for debug
while(TRUE)
{
enable_interrupts(INT_RDA);
enable_interrupts(GLOBAL);
loop = 1;
delay_ms(10); //Delay 10 ms for the conversion to complete
do //Do...while loop
{
if(input(PIN_C4) == 0) //Check to see if the conversion is completed
{
byte_4 = spi_read(byte4);
byte_3 = spi_read(byte3);
byte_2 = spi_read(byte2);
byte_1 = spi_read(byte1);

value = make32(byte_4, byte_3, byte_2, byte_1);
for(i = 0; i < 32; i++)
{
putc(bit_test(value, 31 - i) + 48);
}
putc(10);
putc(13);
loop = 0;
}

if(input(PIN_C4) == 1)
{
delay_us(1);
}
delay_ms(10000);
}while(loop = 1);
}
}

I know that you said in a previous response something about the
if(input(PIN_C4) == 0)

Can you elaborate?

[Ttelmah]

Your code, now generates 32 clocks. This won't work.
You need to either generate 25 clocks, and not operate the CS, or generate 24 and operate the CS.
Generating 32 clocks, will result in the bytes being shifted, and a garbage result.
The way to generate the extra clock, is to simply toggle it in software.

[das]

Sorry - I didn't realize about the code posting. I disabled what you recommended. Looking at the Mode 0 0 it generates 4 bytes - not three based on the Figure 5-7. That would seem to work better for the continuos conversion anyway as it will send out the 25th clock cycle.