[U-Boot] MPC83xx DDR Controller setup

[Andre Schwarz <andre.schwarz@matrix-vision.de>]

Kim Phillips schrieb:
> On Tue, 26 Aug 2008 15:26:16 +0200
> Andre Schwarz <andre.schwarz@matrix-vision.de> wrote:
>
>   
>> Kim,
>>
>> can you (or anybody else) tell me what the following bit means on MPC834x ?
>> Actually it's set inside the CLK_CNTL register of the DDR controller.
>> It's completely missing in my (latest) datasheet's register definition ...
>>
>> include/mpc83xx.h:#define DDR_SDRAM_CLK_CNTL_SS_EN              0x80000000
>>     
>
> Hi Andre,
>
> The mpc8349erm rev. 1 has it:
>
> SS_EN Source synchronous enable
>       0 Source synchronous is disabled.
>       1 The address and command is sent to the DDR SDRAMs source synchronously. This bit must
>         be set before enabling the DDR controller.
>
> But in rev.3 of the chip (mpc8349EA rm rev. 1), the bit is obsoleted:
>
> DDR_SDRAM_CLK_CNTL[SS_EN] has been removed from the MPC8349EA
> and thus there is no effect of setting or clearing this bit. The device will work in
> source sync mode by default.
>
> Kim
>   

Kim,

thanks a lot - this means that I can safely undefine it for the mvBL-M7.

Actually I've had a little trouble with my M7 board being unstable on
high temperatures above 50?C.
After reviewing the DDR controller setup it now works stable.

Right now it's hard to get DDR-II working *stable* out of the box with
soldered IC and without SPD Prom. Signal integrity is not easy to
achieve without tuning the driver strength and ODT resistor values of
both CPU and DDR. Especially the "hardware compensation" using the 18Ohm
across MDIC0/1 (Bit 0 @ DDRCDR) seems to fail on light load conditions,
i.e. single soldered DDR with low pin capacitance.



regards,
Andre

MATRIX VISION GmbH, Talstra?e 16, DE-71570 Oppenweiler  - Registergericht: Amtsgericht Stuttgart, HRB 271090
Gesch?ftsf?hrer: Gerhard Thullner, Werner Armingeon, Uwe Furtner