AnsweredAssumed Answered

i.MX6 USB-OTG Bring Up Design Verification

Question asked by Atilla Mete Turedi on Nov 29, 2016
Latest reply on Dec 21, 2016 by Artur Petukhov

Greetings,

 

We have a custom i.MX6Q board design and our plan is to bring the board up via serial downloader as the first option. Here I am a bit paranoid about the validity of the design so I would like to get some step-by-step confirmation on whether this will work or not?

 

First below is the schematics to the PF0100 boost regualtor output where the USB OTG VBUS is to be powered;

 

 

I am not sure why Sabre SD followed such logic but I tried and did the same where SWBSTIN is gated by SW2 which powers the VDDHIGH rail on the processor. In Sabre SD, it is gated by AUX_3V15 but I do not believe it will make a great difference. SW2 is prior to SWBST on power-up sequence of the PMIC also.

 

Next is the switching of the SWBST output for current limitng purposes, This is done via MIC2545A current-limit high-side switch.

 

 

As you can see, Enable pin of the switch is not pulled high/low but connected directly to i.MX6 E23 which is the designated USB_OTG_PWR signal. SWBST output is switched here to become the USB_OTG_VBUS which is distributed to both i.MX6 OTG_VBUS and the USB connector VBUS as follows;

 

 

And to the i.MX6;

 

 

This is done so that the USB_OTG_VBUS is not always available but only made available when i.MX is the host. USB_OTG_ID is pulled high default at the connector input (Also there is an ESD chip there in the design in case you wonder what U29 was).

 

This sums up the schematic part of the design. For the layout 4 mils signals with 4 mils airgap was used to get approximately 92 ohms differential impedance and approximately 55 ohm single ended impedance. Tracks travel around 3.5 inches from the connector to the processor. From the processor to the connector the length difference between two differential signals are 0.64 mils.

 

At the connector vicinity;

 

 

Path from the connector to the i.MX processor;

 

 

This sums up my design. Now I wonder, when using the serial downloader to get the image into the processor, at what speed does the USB work, does it act as host or slave, who supplies the VBUS? Do you see any logic gaps in this design or errors in the layout?

 

If someone reading this post brought up the card using the serial downloader, please share your thoughts.

 

Thanks in advance,

Mete

Outcomes