Am 31.01.2014 um 14:07 schrieb Grazvydas Ignotas:
On Fri, Jan 31, 2014 at 11:49 AM, Dr. H. Nikolaus Schaller hns@goldelico.com wrote:
...
Options to solve this are
- replace the eMMC by the Micro-SD (which will have a much lower speed
limit!)
- interface WLAN though SPI (is not optimal for 802.11n) and use the SDIO3
port
- interface WLAN through (internal) HS-USB (likely to draw more power and
needs different chipset/module) and use the SDIO3 port
- interface the Micro-SD slot (or even the second full speed) slot though
USB (needs a probably power hungry SD-USB controller chip like [1])
Both eMMC and internal SD would be a bit overkill IMO, so we probably should choose one of them. The forum posters claim there is not much speed difference but I'm not sure if it's really so in practice, as eMMC has 8 data lines and SD only has 4 and not all MMC interfaces support high speed clocks.
That assumption is definitvely wrong. As you say eMMC has 8 data lines while SD has only 4. But the main factor is the different protocol and clock. So eMMC can go up to 200 MHz clock while AFAIK SD(XC) is limited to 50 or 96 MHz.
Basically I like the idea to have an additional permanently installed SD slot. If we can make it available w/o big problems, I would do it. People will find out what they can do with it... E.g. they could split the rootfs between eMMC and MicroSD, e.g. for permanently storing photos where the high eMMC speed is not that important.
About using SPI, the Jorjin modules ED inquired about don't seem to have SPI mode any more, so probably different modules would be needed (unless Jorjin can offer SPI versions of the same modules).
Another thing that needs to be checked is if pin mux options allow the whatever configuration we want, as lots of things share pins and some peripheral we want to use may use up some pins and make another peripheral unavailable.
I have checked that while trying to get an understanding which interfaces we really have and for the OMAP5 it appears not to be that critical as for the OMAP3. Most pads have a main function and can be used as GPIO and then only have rarely used things (SDIO1 is e.g. shared with JTAG).
So unless I did miss some problem, it will be fine. But this asks for an independent double-check...
BR, NIkolaus