| View previous topic :: View next topic |
| Author |
Message |
Torello
Joined: 29 Sep 2006
Posts: 120
|
| Spi master SS (slave select) bug? |
 Posted: Mon Jul 30, 2007 9:20 am |
 |
|
Hi,
Has anyone encountered this one??
I've been searching for days now why spi_write(x) some times does not give out value X. If it does not, the value is mostly '0' or sometimes some high bits disappear. I could see that on a scope It was driving me crazy.
Investigation of bad solders ect. did not solve it. Then I suspected SS or slave select, although this pin should not function under Master mode.
In my test setup this pin is free and has a piece of wire attached (scope debug pin). It was not used and still configured as an input. So it can pickup 50/60Hz hum.
Then I made this pin an output_low -> problem gone! So my conclusion is that the hum somehow interferes in the SPI cicuitry
Has anyone encountered this issue? Maybe my chip is bad.....
Regards,
Edwin |
|
 |
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Mon Jul 30, 2007 12:07 pm |
|
|
1. Post your PIC.
2. Post your compiler version.
3. Post the device that the PIC is talking to.
4. Post the list of connections between master and slave. |
|
|
Torello
Joined: 29 Sep 2006
Posts: 120
|
|
| Posted: Mon Jul 30, 2007 12:31 pm |
|
|
I don't think it is a software bug.
The pic is a PIC18LF4620. The problem exists when communicating to a flash mem. SST25VF016 (16Mbit). |
|
|
ckielstra
Joined: 18 Mar 2004
Posts: 3680
Location: The Netherlands
|
|
| Posted: Mon Jul 30, 2007 12:45 pm |
|
|
| Quote: | | I don't think it is a software bug. |
Writing this line has taken you more time than it would have to mention your compiler version.
Please answer all PCM's questions. You are the one who wants to be helped, so put some effort in it. Right now we don't have enough to go on. |
|
|
Torello
Joined: 29 Sep 2006
Posts: 120
|
|
| Posted: Mon Jul 30, 2007 1:27 pm |
|
|
My apologies, your right.
The code is a bit big to post. So I'm already busy simplifying my code to the lines that do the trick. And also testing it on a 18LF2685. I'll post it asap. I'll test the same code tomorrow on the other 18LF4620.
Compiler version PCWH 4.047
ICD-U40 v2.79
Regards,
Edwin. |
|
|
Torello
Joined: 29 Sep 2006
Posts: 120
|
|
| Posted: Mon Jul 30, 2007 2:21 pm |
|
|
I can reproduce the same on a pic18LF2685 running on 5v, internal oscillator on 4 Mhz. You do need a digital scope to see it. Attach Sclk and Do. Trigger on Sclk. Watch the output data being distorted when you touch PIN_A5 with your finger. The latter doens't happen when Pin-A5 has been made an output.
I'm thinking of applying an asynchronous TTL signal to this PIN_A5. Touching it with your finger might be a little overkill although a piece of wire did the trick in my application.
| Code: |
#include <18F2685.h>
#FUSES NOWDT,INTRC_IO, NOPUT, NOLVP, Nobrownout
#device ICD=TRUE
#use standard_io(A)
#use standard_io(B)
#use standard_io(C)
#use delay(clock=4000000)
void main()
{
long cnt;
short t,j;
int8 i,k,l;
char c;
setup_oscillator(OSC_4MHZ | OSC_INTRC );
setup_spi(SPI_MASTER | SPI_XMIT_L_TO_H | SPI_CLK_DIV_4);
input(PIN_A5); // WARNING! CAUSING SDO PROBLEMS IF HUM IS PICKED UP !!
//output_low(PIN_A5); // do this and problems are gone.
while(true) {
delay_ms(500);
spi_write(0x06);
spi_write(0x60);
}
}
|
|
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Mon Jul 30, 2007 3:12 pm |
|
|
I looked at the code in the .LST file for your version, and I didn't see
anything wrong. The start-up code is setting the analog pins to be all
digital i/o.
Here is another person with your problem, on the Microchip forum.
http://forum.microchip.com/tm.aspx?m=254485&mpage=1
It may be a hardware bug.
If you have to leave it as an input pin, then put a pull-down resistor on it.
I didn't see an errata on your PIC for this problem, but there might be
one on another PIC. You could look through all the erratas that mention
certain keywords, in an attempt to find one that may show the problem.
Here's a Google search string that shows a lot of the erratas:
| Quote: | | site:microchip.com "slave select" errata SPI Master SDO |
|
|
|
Torello
Joined: 29 Sep 2006
Posts: 120
|
|
| Posted: Mon Jul 30, 2007 3:28 pm |
|
|
Thanks for the tips.
I did hook up an async TTL signal and indeed it screws up the output of the SPI. Maybe I'll report it to microchip. For now I don't need this pin, so ill output it low.
Pffft.... I'm happy to have found this bug; took me days! 
Regards,
Edwin |
|
|
ckielstra
Joined: 18 Mar 2004
Posts: 3680
Location: The Netherlands
|
|
| Posted: Mon Jul 30, 2007 4:37 pm |
|
|
Just as a general design note: never have open pins configured as input for this will cause a high power consumption.
Normally CMOS circuits don't use current, except when switching. Now it happens that open inputs tend to float at half the power supply voltage and start oscillating at this voltage. The high oscillating frequency could cause a current draw of tens of mA.
Several discussions can be found on the internet about what is the best thing to do with unused i/o-pins. It seems like there is no 'best solution' for all situations, but leaving pins open and configured as input is always considered bad. |
|
|
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
