groupgets / kimchi-micro Goto Github PK

Type-C port controller PTN5110 (U9) not showing up in i2cdetect. This part should be located at 0x50 on i2c3, but does not show up in a scan:

root@kimchi:~# i2cdetect -y 2
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00:          -- -- -- -- -- -- -- -- -- -- -- -- -- 
10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
40: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
50: -- -- -- -- -- -- -- 57 -- -- -- -- -- -- -- -- 
60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
70: -- -- 72 -- -- -- -- --                         

This issue will stay updated for any hardware bugs found in development of v0.9 board until v1.0 hardware is ready. If you think you've found a bug in the v0.9 hardware, please open a separate issue for discussion, and this document will be updated once that discussion is resolved.

• Problem: 1.8v-capable SDHC (UHS-I and newer) cards won't work in Linux [#2]
  Root cause: microsd SD2 powered by NVCC_SD2, but needs a constant 3.3V
  Workaround (software): add device tree property no-1-8-v
  Proposed v1.0 solution: supply microsd SD2 VDD with VDD_3V3

• Problem: voltage sag on initial power causes pmic to shut down [#3]
  Root cause: excessive inrush current on VDD_3V3 due to aggressive decoupling for PCIe
  Workaround (hardware): don't stuff C125-127. Change reflected in released v0.9 BOM.
  Proposed v1.0 solution: add load switch with inrush-limiting for PCIe 3.3V rail

• Problem: Type-C port will not take host role [#4]
  Root cause: can't communicate with U9 (swapped sda/scl pins)
  Workaround (hardware): cut and jump to pull up resistors R19/R20 (see linked issue)
  Proposed v1.0 solution: fix scl/sda pin swap

• Problem: Advise in BQ27426 notes is not followed [#5]
  Root cause: U12 pin BIN shorts to Vss when battery is used
  Workaround (hardware): cut and jump R51 to configure as BIN pulldown to GND (see linked issue)
  Proposed v1.0 solution: don't include Q1 and configure R51 as pulldown

• Problem: Battery over discharge protection circuit does not work [#6]
  Root cause: disabling discharge with IO on charger did not work as intended
  Workaround: use a pack with integrated protection circuits.
  Proposed v1.0 solution: add a FET to properly disconnect battery on fault

• Problem: Some out-of-spec but common mPCIE cards do not fit [#7]
  Root cause: cards interfere with components on the board
  Workaround (hardware): replace mPCIE connector with 2041262-1
  Proposed v1.0 solution: update BoM

Issue

U12 is a 1.6mm x 1.6mm WLCSP with a 3x3 BGA array underneath. If assembly houses want to make use of x-ray inspection it may be difficult, because there is a large power inductor L8 directly opposed, which casts a large shadow.

This may be a non-issue, but just something I stumbled across when inspecting my board after assembly.

The switch to 1.8V operation for UHS-I mode is undervolting most cards, causing errors in operation.

J10 VDD should be supplied by VDD_3V3.

According to the BQ27426 datasheet:

NOTE: The BIN pin must not be shorted directly to VCC or VSS and any pullup resistor on the BIN pin must be connected only to VDD and not an external voltage rail. If an external thermistor is used for temperature input, the thermistor should be connected between this pin and VSS.

When a pack is used with the board, the BIN FET Q1 is turned on, shorting BIN to VSS

At least one high-power wireless card (Compex WLE1216V5-20) has components that extend beyond the vertical keepout area defined in the mPCIE specification. It will not fit without interfering with components on the board.

Excessive inrush current due to aggressive decoupling on PCIe 3V3 is causing this voltage sag.

Removing these extra decoupling caps allows the pmic to power up properly.

Leaving a battery pack that does not have internal over discharge protection results in a flat pack overnight.