There are two things I have observed when sending data out via SPI. One case is when I am blasting data out on the buss and when I am slowly sending data out manually by using the debugger.

All my settings are correct for the chip select and I have CSAAT=0 so the line should pull high after the last transfer. I also set LASTXFER high in the tx register to ensure that chip select will be deasserted after data has been transferred. I have verfied that all the data looks good on the scope.

1.)

The chip select is supposed to idle high, go low , and then back up when data has been sent. If I send out data quickly I never see my chip select being pulled high after each transfer. I just stays low for all the data. Even though I have all the settings set correctly.

2.)

To verfify that I have not messed something up with the above I fire up the debugger and I step through each command one at a time. I watch CS carefully during each transfer on my oscilloscope and I see the CS pin now being pulled low and then back up high! This is excellent! The problem is that when I keep stepping through 1/2 of the time I don't see the CS pin being pulled down!

For example this is similar to what is happening:
spi_pt->SPI_TDR = 0x010e00026; // I'll see a good CS toggle
spi_pt->SPI_TDR = 0x010e00045; // No CS toggle at all
spi_pt->SPI_TDR = 0x010e00081; // Good CS toggle
...


I've looked at the AT91SAM7S64 errata sheet and I don't see anything under SPI that relates to my issue.

Thanks for any info!

-Henk

if this is exactly what you have in your source code:

you have an extra zero between e and 2

jihwan,

Thanks for post.

I have it correct in my debugging code:

spi_pt->SPI_TDR = 0x010e0055;