DSP Design

Demo for Tektronix

Christopher Edmonds — Tue, 04 Sep 2007 05:13:59 +0000

I worked all day on getting the board ready for a demo at Tektronix. In the end I managed to write the program, but never got to test it because I found out that the JFFS2 file system does not work read/write it only works read only. In the process of discovering this I also managed to erase the entire contents of the flash file system. I think I know how to fix this, but I did not have time to get everything done. Thus unless there is a miracle tomorrow morning we will not have much of a demo. Hopefully I will be able to get a simple demonstration of the unix command line ready.

Board rev .02

Christopher Edmonds — Tue, 28 Aug 2007 16:32:53 +0000

Board rev .02 is nearly out the door. I have finished doing layout and just have to get a few bugs out that freedfm found. This new rev fixes a couple errors in the original revision and also updates the audio codec to the AC97 compatible AD1981BL.

Codec Change

Christopher Edmonds — Tue, 28 Aug 2007 15:48:58 +0000

Well, it turns out that cheap isn’t always good. We chose to change the audio codec on the board because it required that we generate three external clocks to feed into it. This used an unjustifiable amount of board resources and also would have required us to change the crystal for the main processor which would have made it harder to implement ethernet in the future. For this reason, the UDA1345TS is out and we have settled on an AC97 codec from analog devices. The new part number is AD1981BL. This codec also has mic and headphone preamps built into it.

Kernel Booting … Done

Christopher Edmonds — Fri, 10 Aug 2007 22:25:59 +0000

The kernel boots! We now have the kernel booting out of flash. It then mounts the JFFS2 from the Dataflash chip and proceeds to boot. Not too much can be done with the current image but all of the files are there and we were able to play a quick game of dungeon on the serial terminal. The next big task is to get the audio driver done. There are files in the kernel for the uda1341 chipset which is similar to our uda1345 chipset. Unfortunately this file seems to be a remnant of a driver from the 2.4 kernel and not of the necessary files exist to get it working. Due to this we are starting from one of the ADI chips that has a driver to use the SPORT on the blackfin in I2S mode and then rewriting the parts of the functions that are chip specific. We hope to have a first shot at the driver early next week.

Booting in multiple ways

Christopher Edmonds — Thu, 26 Jul 2007 00:25:21 +0000

It’s been a little while since I last posted but a lot has happened. We have now successfully run u-boot in both UART boot and SPI boot modes. Tim Louden managed to quickly get together a flash driver for our spi chip that allows us to interface it with u-boot. Now the real work begins. The kernel itself. We found that the version of the toolchain that we were using was not compatible with the version of the kernel that we were using so we reverted back to the most recent release of the kernel rather than the svn version. This is currently being worked into the repository and hopefully linux hacking can begin tomorrow. Once we get the kernel booting the next big task is to see if we can get the audio codec to respond.

Memory or Forgetfullness

Christopher Edmonds — Wed, 18 Jul 2007 00:24:20 +0000

We attempted to test the SDRAM today. We were able to test putting different patterns into memory location 0×00000000 and reading them back. We then tried to do a sweep of all of memory, but we were not able to successfully complete it. The SDRAM chip started getting warm during this process but we are not sure why. Tomorrow we will investigate whether there is any bus contention and we will continue working on our memory test algorithm.

Lights!

Christopher Edmonds — Tue, 17 Jul 2007 00:07:03 +0000

Well, we were able to execute code that we wrote on the board. We now have an 8-bit debug interface made out of LEDs. The next issue seems to be loading the bootloader into the SDRAM. The SDRAM is configured but maybe not properly.

Long Time … No Post

Christopher Edmonds — Mon, 16 Jul 2007 14:42:38 +0000

Well, it’s been nearly 10 months since I’ve done anything useful with this blog. Fortunately it has not been 10 months since I’ve worked on the project. The first batch of prototype boards came back in December and I have been steadily making progress towards getting Linux up and running on the board. All of the chips (minus the audio codec) have at least had an initial test. This summer I have three other guys working on the project with me their blogs are at Tim Louden Jace Akerlund and Allen T. Waters .

Layout Time!!

Christopher Edmonds — Wed, 27 Sep 2006 18:18:49 +0000

The exciting part is nearly here. There are only three more geometries to be made and verified and then layout begins. There are going to be a lot of traces to lay. I am shooting for a 4 layer board. After reviewing the dimensions of everything it seems that Advanced Circuits may be able to make boards for us after all. Not much other news than that.

Parts

Christopher Edmonds — Wed, 20 Sep 2006 23:02:25 +0000

Well, progress has been going well on the design even though progress on the blg stopped temporarily. I have finished schematic capture. I am now going through my BOM downloading all of the datasheets, verifying part selection and creating needed geometries. However much fun all of that sounds I am looking forward to the layout process for this board. We have found that the Analog Devices reference board uses a 6mil drill for some of their vias. Our normal circuit board manufacturer, Advanced Circuits, can’t do that small of drill hole in their normal boards. We may have to search for someone else to produce the PCB if we can’t figure out a way around this.