mirror of
https://github.com/brain-hackers/u-boot-brain
synced 2024-07-30 06:43:45 +09:00
f72e48ba4d
This patch adds skew register settings for KSZ9131. It checks first which phy is on the board and then applies the correct skew settings. Skew settings calculation for the KSZ9131: The i.MX6 SoC has an output skew tolerance of -100ps to 900ps. All PCB traces where routed exactly the same length so we can calculate the skew settings without taking the length into consideration. The traces are all length matched. RXC skew (PHY to MAC): - We use the 2ns DLL controlled delay on the PHY - We do not use the skew registers This results in the following values: RXC PHY fixed Delay 2000ps PHY Added Delay 0ps T_setup_R min 2.00ns T_setup_R typ 2.00ns T_setup_R max 2.00ns T_hold_R min 1.60ns T_hold_R typ 2.00ns T_hold_R max 2.40ns That means we are well within RGMII specs. TXC skew (MAC to PHY): - We use the 2ns DLL controlled delay on the PHY - We then subtract ~0.6ns with TXD[0:3] and TXC clock pad skew register in a resulting ~1.4ns delay. This results in the following values under consideration of the tolerances: TXC min TXC typ TXC max MAC min -100ps -100ps -100ps MAC max 900ps 900ps 900ps PHY fixed Delay 2000ps 2000ps 2000ps PHY added Delay -340ps -600ps -859ps T_setup_T min 1.56ns 1.30ns 1.04ns T_setup_T typ 2.06ns 1.80ns 1.54ns T_setup_T max 2.56ns 2.30ns 2.04ns T_hold_T min 1.04ns 1.30ns 1.56ns T_hold_T typ 1.94ns 2.20ns 2.46ns T_hold_T max 2.84ns 3.10ns 3.36ns This shows that T_hold_T min and T_setup_T min times are out of spec for RGMII timing. However the KSZ9131 has a minimal value for this time of 0.8ns which is met under all circumstances. Signed-off-by: Philippe Schenker <philippe.schenker@toradex.com> |
||
|---|---|---|
| .. | ||
| apalis_imx6.c | ||
| do_fuse.c | ||
| Kconfig | ||
| MAINTAINERS | ||
| Makefile | ||
| pf0100_otp.inc | ||
| pf0100.c | ||
| pf0100.h | ||