DDR3 Address Bus Overshoot!

[qammarabbas]

Hi,

While simulating my T1042 processor's address bus with 10 DDR3L memory chips, i found that the driver's(controller) address signal is having overshoot and undershoot of greater than and equal to 400mV. I am using Vtt termination resistor in my design. However, i cannot use series termination on the driver side since it is only recommended for a point to point system. Please let me know what can i do to reduce the overshoot and undershoot? Thank you! 

[Bulat]

Can you provide the circuit being simulated and resulting waveforms?

Regards,

Bulat

[qammarabbas]

Hi Bulat,

Attached is the block diagram of the circuit i am using along with the waveform of an address signal at the driver side. Please find the attached images for your information. Thank you!

[Bulat]

Can you try to change probe location from "at the pin" to "at the die" and run simulations once again?

[qammarabbas]

Yes it gives much better results. Please find the attached image. Overshoot and undershoot is non-existent. But physically how can i take the signal directly from the die when i have a ball package? 

[Bulat]

As I understand your concern was/is the overshoots at the T1042 output pins, right? The last waveform shows that the overshoots are absent at the die, so the silicon is not affected and overvoltage can not damage the device.

And I do not understand your last question "...physically how can i take the signal directly from the die when i have a ball package". What is the task you are trying to solve?

Regards,

Bulat

[qammarabbas]

Please let me know what should be the probe location during simulation? Die or pin?

[Bulat]

Die of course. This is what the device really "sees".

Regards,

Bulat