Content originally posted in LPCWare by suckfish on Sat Dec 22 16:07:55 MST 2012
Hi,
I've got a problem with SDRAM reads from the CPU getting corrupted while driving a LCD from SDRAM. Any ideas?
My first guess is a timing problem with the memory accesses, but I really don't have much of a clue how to debug this.
This is a custom board with LPC4357 and a 256Mb 16-bit SDRAM. The display is 1024 by 640 TFT, 52.25MHz pixel clock. I get the problem both in 16bpp and 8bpp modes. The CPU and SDRAM are running at 96MHz.
The display output appears rock solid; no glitches.
The M4 CPU is running test code that reads and writes test patterns; it shows no problems without the LCD running, but with the LCD running it starts showing errors.
My first guess was that I was running out of SDRAM bandwidth, but reducing from 16bpp to 8bpp appears to make no difference.
The memory test code is just:
volatile unsigned * const sdram = (v32 *) 0x60000000;
for (int i = 0; i != size; ++i)
sdram = i * 0x02030401;
putchar ('\n');
for (int i = 0; i != size; ++i) {
unsigned e = i * 0x02030401;
unsigned v = sdram;
if (v != e)
printf ("Memtest: Bugger @ %06x expect %08x got %08x\n",
i, e, v);
}
Cheers,
Ralph.
