The six days I checked many possibilities…
Finally, I find it is clock circuit problem.
This is clock sch from v0.5:
This is clock sch from v1.0a:
So do you find the problem?
I moved 27pF cap to other side of the crystal for the new version…I did not even notice that …(when I optimised the layout of the PCB, I changed some parts of schematic, then forgot to change this part back…)
This small careless problem caused the hole board unstable and the wifi not working. Such problem will never happen again.
Why this BUG is so critical?
Normally, hardware is stable and able to afford some problem, for example, if you remove two or three 0.1uF cap, it works normal.
But crystal problem is always a big problem.
If RT5350 is the heart of VoCore, the crystal(clock) is the heart of RT5350. It can not work with a buggy clock.
v0.5 circuit is a standard noise reduction circuit, the high frequent noises are block by the 300ohm, and direct to ground by the 27pF cap. But when I move 27pF cap to another side, this RC noise reduction circuit is broken, so clock noise will be very loud, so some chips are unstable. Wifi signal is 2.4GHz, that is from crystal clock(20MHz) x 120, the noise is even 120 times stronger than the chip, so the wifi is not working at all.
This morning I contact the factory, their repair team is very busy, we have to wait at least one week. And from WuTong, The alpha VoCores had shipped to China Post and I can not make them back to change any hardware, so the alpha contributor have to change this yourself. Really sorry for that problem, I will record a video or write a blog with detail to show how to make that.
For simple, it should be changed like this: