If you are interested in the design files for the RMT1 or the PMT1G, please contact me at info@mediamatech.com.

Regards,

Pete

1. SAM-BA over JTAG (SAM-ICE) is working, with applets written for initializing both LP DDR SDRAM on both DDR0 and EBI.  NAND and DataFlash proven.  ES replacement boot ROM loaded on DataFlash, so now SAM-BA over USB is working.
2. AT91 Bootstrap (1.14) modified to initialize both LP DDR.  So I can now launch U-Boot...
3. Latest U-Boot (not yet released) has native support for AT91SAM9G45, so I started with that as a base.  Added PMT1G support code, and we are now booting into U-Boot!  Ethernet is working under U-Boot.
   There appears to be some stability issues, in which U-Boot prompt becomes unresponsive after ~10 minutes, unless something is actively running (e.g. memory test).  Might have to roll back to Atmel's 1.3.4 demo, just to check.
4. Starting from 2.6.30, maxim.org.za AT91 patches, and Nicolas Ferre's latest patch set, I have added PMT1G board support to linux.  Can now boot through to the linux prompt, using my own root filesystem in NAND (JFFS2).  But after several minutes, a spontaneous cache error occurs.  Still investigating.

And now for a picture of the board, loaded thus far:

DC jack on the lower left, ethernet coming in on the middle-top, and USB miniB cable on lower-right is for console support (on-board SiLabs CP2102 USB/UART).  Console is the DBGU serial interface; in linux, mapped to /dev/ttyS0.

Next up is to analyze some of the U-Boot and Linux stability issues.  And then onto verifying LCD and AC97 codec support.

Stay tuned for more news as things progress...

Thus far, loaded all power supply components and all voltages measure nominal (phew!), on one prototype thus far.  Next will be all circuitry for the SAM9G45 to reach basic start-up (e.g. filtered supplies, clocks/crystals, etc) -- enough to get SAM-BA over JTAG going.

I will post a photo of the prototype soon.  Stay tuned!

Finally got around to getting a LQ043T1DG01 panel.  Beautiful looking output, with relatively high contrast and brightness. The touch panel appears (at least subjectively) to be more sensitive and responsive than the Hantouch panels commonly used.  Looks like it will be a popular "high-end" option.

In terms of the touch pin-out, it turned out it was not a necessity yo change the pin-out (flip Y axis) after all.  It is more an issue of perspective and use.  Oh well, at least tslib is capable of translating any axis flips, and thus compensates for this change simply by running ts_calibrate.

In terms of a new linux release and root filesystems, still waiting for everything to percolate up into mainline -- or I would even settle for just a AVR32 GIT tag when 2.6.31 is finally out.

PCBs and kit at the contract manufacturer.  Expect to have more RMT1 boards available within one week!

After some feedback from the AVR32 kernel list, the patches for RMT1 should get through soon, and should make it in time for 2.6.31.

PMT1x:

Should have the mobile DDR parts soon.  Lead-times are stretching out these days, as vendors are being extremely careful not to over produce.

Looking at the kernel GIT, and the SAM9G45 and -EK board setup code is (finally) next in line.  Again, should make it in time for 2.6.31.  I can soon start test kernel builds, and move onto the board setup code for PMT1G.

On the u-boot front, v2009-08-rc* looks to have the SAM9G45 support in place already.  So it looks like 2009-08 is the place to start the u-boot port for PMT1G.

Kernel patches were submitted, and now I am waiting for them to be accepted and a new tagged AVR32 kernel in GIT.  Then I can publish a new kernel based on 2.6.31*.  RMT1 is support is almost completely in the kernel mainline!  Along with that release will be an updated root filesystem.

In recent days, there was a sudden spike in RMT1-A orders -- more in the last two days than in the last two months!  Not sure what the reason is.  But now I am completely out of stock.  A new batch will be built next week.  This time, I fixed the pin-out of J13, and will load that connector so the Sharp LQ043T1DG01 is supported right away.

Sorry for the delay -- the new RMT1 batch should be available within 2 weeks.

On the PMT1x front, I received the blank PCBs.  Now just waiting on some BGA SDRAMs to arrive, before I can get the first protos made.

There is now a DRAFT PMT1G Brochure (variant with SAM9G45).

I have converted FA1 (FPGA board) into what is now PMTF1, likely the first expansion card for PMT1x.

Preliminary top-side "picture":

And bottom-side:

The PMTF1 PCB is 2.2" x 1.7", and is designed to mate with the PMT1x's EBI Expansion port (16-bit data) over a Samtec FSI connector.  It also has a standard 2x20 0.1" footprint (only 8-bit data) for simple mounting on other boards.  This footprint also matches the Intronix LA1034 USB Logic Analyzer's cable -- very useful for debugging, as it is much less expensive than ChipScope (which does not come with the free Xilinx WebPack tools).

As with FA1, there is SPI DataFlash that holds the FPGA configuration image.  The FPGA takes control of the DataFlash to configure itself right after power-up.  After that, the DataFlash will be placed on the SPI bus, allowing the host CPU to access it.  So in linux, it could be mounted as a typical MTD device, and thus one could easily re-program the bitstream without requiring any special drivers.

The project is now called "PMT1x", to reflect the possible variants:
PMT1G:  based on AT91SAM9G45 (first product)
PMT1M:  based on AT91SAM9M10 (TBD)

Then there will be sub-options, based on BoM customization requests (minimum quantities would apply):  e.g. amount of memory, with/without ethernet, etcetc.

Here is a "picture" of the first-prototype topside:

Except for the USB debug console and the (optional) DC power jack, the main user connections are along the bottom-edge.  Also, the battery cable connections are on the right-edge, as they are intended to be kept internal to the enclosure.

And the bottomside:

Right now I need to finally tackle this elusive SPI/ADS7846 driver bug that has been keeping RMT1 board support from transitioning to 2.6.31!

Then working will begin in earnest on the Atmel SAM-BA bootloader support, U-Boot, and the the Linux board setup (based on the 9G45-EK board).
And remember the FA1 FPGA accelerator?  I am working on revamping it to work with the PMTx board, via the EBI expansion header.  The new board will be a bit larger and will also accomodate a standard 2x20 0.1" header to allow it to be used practically anywhere.  Stay tuned...

No word yet on the 9M10 release schedule.  At least the first PMT10 prototypes will be made with the 9G45.  I am hoping the PMT10 release has the 9M10, but if that is looking too far out (or prohibitively expensive), than I will release with the 9G45.

Still waiting on 9G45 samples.  Not proceeding with the PCB until I get confirmation.  Hopefully soon!

The first General I/O card (e.g. PMTC1) will have dual USB host (built-in USB hub), and options for a bluetooth module, RS-232 DTE (DB-9M), and CAN interface.  The design and layout are completed; so just a matter of waiting until PMT10 is proven.  Or perhaps a better idea is to make a simple PCB with prototyping area, etc.

I am also in the midst of defining the first of the EBC (EBI Bus Extension) cards:  an FPGA video output board, targetting full HD (1080p) decoding applications, with DVI output.  It will likely be based on a Xilinx Spartan-6 FPGA, depending on scheduling, etc.  More details to come.

So for PMT10, I decided to add footprints for connectors to support the DG01.

For now, default stuffing will be for the lower-cost Formike -- but we shall see what the market wants.

Any thoughts about this anyone?

FYI:  it appears that mini-box.com is working on a 9G45 board.  "ituner" (presumably someone from there) seems to be advertising it everywhere he/she can -- including AVRfreaks!  Looks like quite an interesting board.

To correct any possible misperceptions out there, please note that PMT10 is intended to be used as a standalone developmnent board, as well as an OEM-quality board that can be directly integrated into production systems, or even on its own (e.g. intelligent control terminal, etc).  I am hoping the design choices I have made will make it amicable to both kinds of need.

Meanwhile, still waiting on 9G45 samples to get going with the first prototype...