AM3352: DDR VTP calibration

The factory team have been notified. They will respond here.

Mark

How did they conclude that VTP calibration is the root cause of the problem? Can you share more details? What DDR type is the customer using? We generally recommend Dynamic VTP (Voltage, Temperature, Process) compensation, but depending on the frequency and temperature variation expected in the system, we can look at one time static process compensation

Regards, Siva

Hi Siva,

They stopped the customer's board inside of u-boot, and utilized the memory write and read commands to poke at registers.  Based on the forum posting they decided to try toggling if VTP was enabled and you could see the noise come and go by setting and clearing a single bit in a single register.  E.g. the noise being present was tied to the "enable" bit in the "vtp_ctrl" register being set to 1 - if this bit was set to 0, the noise went away with no other changes to the system.

Here's the bit in question (page 1515 of SPRUH73P revised March 2017):

(Register at offset 0xE0C inside the system control module - absolute address was 0x44e10e0c.)

The hardware was placed inside a faraday cage with a spectrum analyzer RF near field probe sitting near the processor on the board.

Regards,
Mark

Mark

Thanks for the details. Can you provide more information on the DDR type, frequency and temperature of operation?

Regard,s Siva

Hi Siva,

That answers the question.

Thanks,
Mark