AM3354 HS USB Host Issue

Is this customer using Linux?

If so, please confirm the version number.

I am the customer.

We are using QNX on this control.  As I am the h/w engineer on the project I may not be able to answer all your questions regarding the s/w but will be able to get some assistance on Monday.

As to UBOOT my understanding is that the basis is that used on the TI Dev Kit for this processor. 

Thanks for the assistance.

John,

I don't believe any of the reference platforms (ie. evaluation kits) actually use a 24MHz clock.  I checked the schematics on the various kits and all of them used a 25MHz clock.

My suspicion is that the clock tree setup is off (ie. PLLs, etc) likely because the software was based on assuming a 25MHz clock input.  This is just a theory not knowing anything about what modifications were made in the QNX board support package to adjust for the 24MHz input.

Hi John,

Brandon meant to say all of the reference designs use a 24 MHz crystal.  The change would need to be made in U-boot.  Replacing the 25 Mhz crystal with a 24 MHz crystal might be easiest.

#define V_OSCK                          24000000  /* Clock output from T2 */

Thanks,

Mark

Thanks Mark for catching that.  You statement is true that I meant 24MHz on the EVMs.

Earlier I asked about the SW in use as I suspected that the PLLs were not configured to provide the correct clock to the USB module. As Mark mentions here, it might just be easier to use a 24MHz clock source and see if the issue is resolved.

My expectation is that 24MHz would be better (jitter-wise) for USB as a whole given that it requires much less in the way of internal division/multiplication to meet the required input frequency of the module.