SPI - No Data Signals

v38zy · Joined: 10 Mar 2005 Posts: 6

Hi,

I am having serious problems trying to get the SPI to work. I have a PIC 18F4431 microcontroller, and the CCS C Compiler V3.212. All that I am trying to do is write some data out onto the SPI, but no data appears at the output (SD0) pin.

Currently I am using the default pins:

C3 = SCL
C4 = SDI
C7 = SDO

Using a scope, I have been able to determine the following:

1) The SCK signal IS present on pin c3, and is working correctly (i.e. outputs 8 clocks every time that I call spi_write).

2) The data out signal is always high (i.e. 5V), and does not change no matter what I write using spi_write().

3) If I try to use the alternate SPI pins:

D3 = SCL
D2 = SDI
D1 = SDO

then I don't even get the clock signal (on D3) anymore.

Here is the code that I am using:

C File:

void main()
{

setup_adc_ports(NO_ANALOGS|VSS_VDD);
setup_adc(ADC_OFF|ADC_TAD_MUL_0);
setup_spi(SPI_MASTER|SPI_L_TO_H|SPI_CLK_DIV_16);
setup_wdt(WDT_OFF);
setup_timer_0(RTCC_INTERNAL);
setup_timer_1(T1_DISABLED);
setup_timer_2(T2_DISABLED,0,1);

while(1)
{
spi_write(0xF0);
}
}

Header File:

#include <18F4431.h>
#device ICD=TRUE
#device adc=8
#use delay(clock=40000000)
#fuses NOWDT,WDT128,HS, NOPROTECT, IESO, BROWNOUT, BORV27, NOPUT, NOCPD, STVREN, DEBUG, NOLVP, NOWRT, NOWRTD, FCMEN, NOWINEN, T1LOWPOWER, HPOL_HIGH, NOWRTC, NOWRTB, NOEBTR, NOEBTRB, NOCPB, LPOL_HIGH, PWMPIN, NOMCLR, FLTAC1, SSP_RC

This seems like it could be a problem with the compiler.... but I have no idea why or what I can do to fix it. Does anyone know what I am doing wrong?

As well, if I connect the clock signal to the clock pin (C5) on another PIC which is configured as slave, the clock signal voltage drops from 5V peak to about 1V peak...

Thank you,

Andy

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

The first thing I see is that your #use delay() statement is set for 40 MHz,
but that implies that you're using either the 4x PLL mode (with a 10 MHz
crystal) or the external clock mode. But then you've got "HS" in your
#fuses statement. You would need either "H4" or "EC" instead.
My suggestion is, don't try anything fancy while in the early stages --
just run it at 10 MHz, in HS mode. Do this: #use delay(10000000)

Also, when you tried the alternate pins on Port D, did you remove the
"SSP_RC" fuse setting ? That setting puts the SSP pins on Port C.

v38zy · Joined: 10 Mar 2005 Posts: 6

Thanks for the quick reply Smile

I am using a development board which has a 40MHz crsytal on it, I can't change that. The clock is connected to the CLKI pin on the PIC.

I tried changing the fuse from HS to H4 and EC as you mentioned, but it didn't resolve anything. I even tried using the internal oscillator by setting fuse to INTRC and #use delay(8000000)... but I got the same result.

Yes, I did change SSP_RC to SSP_RD when I was trying to use the alternate pins on port D.

Any other suggestions?

Thanks,

Andy

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

Is the documentation for your development board available on the net ?
Can you post a link to it ?

Guest · Posted: Thu Mar 10, 2005 6:39 pm

No There is no documentation it is not publically available. It was built for the students to use at the University which I am attending.

There isn't much to it, it is just the PIC device, a power connector, the ICD connector, and a bunch of 10 pin headers which are connected to each of the ports on the PIC.

Andy

v38zy · Joined: 10 Mar 2005 Posts: 6

The previous post about the board was written by me... not sure why it shows up as Guest.

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

I don't have your version of the compiler, but I compiled the following
test program with vs. 3.188 and saw a problem with the TRIS setup.
Can you compile this program with your compiler, and post the .LST file ?
Then I can see if there's a problem.

When you post it, don't show your CCS user number. It's the number
right after the version at the top of the .LST file. Example:

CCS PCH C Compiler, Version 3.188, xxxxx

I don't know if giving that away makes it easier to for someone else
to logon and download the compiler, but why risk it ?

v38zy · Joined: 10 Mar 2005 Posts: 6

Ok, I compiled the program using the code that you provided. Here is the .lst file that was produced (minus the user number).

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

There's a bug in the setup_spi() function. It's using the SCK and SDO
pins as if this was a 18F452. The problem is that in the 18F4431,
Microchip moved the SPI signals to different pins.

The code is setting up pins C3 and C5 to be outputs, and you really want
it on pins C5 and C7. So, the minimal fix would be to just set the TRIS
on C7 to be an output. See the revised test program below.

v38zy · Joined: 10 Mar 2005 Posts: 6

I can now see the data being sent out on C7!!! Thank you soo much, I was beginning to get pretty frustrated with this problem.

Is this the solution to the problem if I use the alternate pins on port D as well? Here is the .lst file if I add the fuse SSP_RD... Do you know what I should change?

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

v38zy · Joined: 10 Mar 2005 Posts: 6

Great! It works perfectly now Smile

I really really appreciate your help PCM programmer, I never would have figured that problem out myself. People like yourself are what make these forums so useful Smile

v38zy · Joined: 10 Mar 2005 Posts: 6

One final question:

When I set up the slave, I assume that I will have to make similar modifications... Here is the .lst file for the slave, and I would like to use port D... Is there somewhere that I can get a listing of all the registers within the PIC so that I can make the necessary changes?

dyeatman · Joined: 06 Sep 2003 Posts: 1924 Location: Norman, OK

The listing of all the registers (and LOTS of other good info) is found in the PIC datasheet.

You get the datasheet from the Microchip website: www.microchip.com

Mark · Joined: 07 Sep 2003 Posts: 2838 Location: Atlanta, GA