TMDS64GPEVM: Unknow circuit about Kelvin Sensing

Hello Huang Feng,

Here are some additional inputs i received.

It is intended that supply has a private trace from underneath the power pin BGA as well as a second private trace underneath the nearest GND pin BGA of the processor. This way we can use a high speed power rail probe to measure instantaneous supply ripple noise.

Detailed explanation

Sometimes, we need to characterize AC noise on an SoC’s DC power supply.  For example, we may need to look for voltage dips that might bring the supply below the minimum spec’d voltage.  In the past, it has been difficult to make this type of measurement well.  There are different schools of thought on this.  A relatively convenient method is to use a differential probe to measure across a decoupling capacitor that’s placed close to the SoC’s power balls.  However, the de-cap acts to filter the measurement, leading to an optimistic measurement.  A better approach is to probe at vias near the SoC balls, but this is also difficult since vias may not always be conveniently placed for probing.  In the EVM, Kelvin sensing circuits were included that make it easy to use a power measurement probe to measure AC noise on DC supplies.  The supply and gnd nets are routed away from the SoC using private traces, and the test points are also outside of the SoC socket

I do not think a customer would need this in a production board, but if they have space in their prototype board it is useful to prove they are satisfying the voltage recommended operating conditions at the processor pins, which is where it matters.

Regards,

Sreenivasa