DP83867IS: RGMII Slew Rate

Hi Ross,

Included below is the customer response.  Let me know your thoughts.

I see those differences but I have some issues with what they’re saying.

In this response: They say the difference is due to extra capacitive loading on pins 27 and 28 due to the multiplexing.

In this response: They say the extra capacitive loading is already reflected in the IBIS model. Looks like pins 25 and 26 have more capacitance than 27 and 28 though. Opposite of what they said before. I also would have expected a larger capacitive difference to cause the large effect that I’m seeing.

Here’s an example, I setup a simulation to run through some trace, a resistor, some trace, and then terminate into DP83867. I took a measurement with and without the resistor and Phy connected:

NOTE: I did not take the time to setup all of the coupling regions or to model the probe capacitance so there is a difference between measured and simulated data, this is a behavioral comparison only.

You will see that in both the simulation and measured data, before the resistor and phy are connected, all four data lines have approximately the same edge rates, so this proves they’re being driven the same. Now connect the PHY through a 100 Ohm resistor and re-simulate. We see that despite the TX_Data 2 and 3 starting slightly faster, they end up being slightly slower (this makes sense with what is in the model currently) however, they are all very similar.

Now jump to measured data and Data_0 and 1 are significantly slower than data 2 and 3. And this is just not reflected in the ibis at all it doesn’t seem.

Regards,
Mark

Hi Ross,

Any updates on this request?

Thanks,
Mark