I have new custom boards designed for DM6467T but stuffed with DM6467. I got two boards working just fine several months ago. I can't get working the third board from the same build.
I have programmed UBL and U-Boot to NAND flash, and I'm trying to connect the console at 115200,8,N,1. On the oscilloscope, the bit timing is supposed to be 8.68us/bit. However, I'm measuring 7.75us/bit. I haven't analyzed the bit pattern, and I don't have a 129000 baud device, but the bit pattern looks like it's good data, just at the wrong speed. I don't have the original working boards around any more, so I can't compare anything to them. I can, however, compare to the DM6467T EVM that I have. Indeed, I measure 8.6us/bit and my TeraTerm program sees the nice U-Boot output.
So how am I possibly getting 7.75us/bit? First, I can't make sense out of the ratio between than and 8.68us. It doesn't relate to any combination of 27 and 33, like the crystal MHz involved.
Note that I am using a TMS320DM6467ZTU7, the same processor I used last time. In both cases, I am programming with dm6467t_495_ubl.bin and u-boot-dm6467t-evm.bin. I realize that these are intended for the DM6467T, not the DM6467. But it worked twice before.
I haven't looked at the U-Boot source code, and I haven't found a document detailing how the DaVinci UART works. I do have decades of UART and RS-232 experience in general, however.
What can possibly be wrong to cause this funny baud rate? Also, can someone give me a quick link to the correct source for this version of U-Boot? Where's a spru*.pdf that covers the UART?
Thanks very much,
Helmut