AM3358: SI/PI simulation non-monotonic signal

Hi Sreenivasa,

AM3358 ZCZ + H5TQ4G63EFR

It's same at all pin for communication with DDR3.

Thanks.

Hello David Park, 

Thank you for the inputs.

I also see you pressed the resolved button.

David Park said:

AM3358 ZCZ + H5TQ4G63EFR

It's same at all pin for communication with DDR3.

Could you please elaborate the above inputs.

Regards,

Sreenivasa

Hi Sreenivasa,

Could you please request customer to provide additional details regarding the simulation performed including the AM335x device package and the memory interface used?

--> I attached IBIS file of DDR3H5TQ4G63EFR IBIS.zip and used IBIS of AM335x ZCZ IBIS Model(Rev.C)

Could you ask customer to elaborate what dq input means - is this behavior observed on a specific input.

--> DQ input is not specific pin. it's  all pin for communication with DDR3

Hello David Park, 

Thank you and noted.

Is customer seeing the issue on a specific input or all the inputs.

I see that you also pressed the resolved button.

Can you please confirm if the issues is resolved?

Regards,

Sreenivasa

Hi Sreenivasa,

It's all the inputs.

Issue is not closed. resolve click was my mistake.

Thanks.

Hello David Park, 

Thank you for the input.

Let me check internally and update.

Regards,

Sreenivasa

Hello David Park, 

Please refer below inputs i received from the device expert (Paul Eaves)

Any LVCMOS signal will have a mid-supply step during the transitions when measured at any point along the signal path except at the far end.  So their simulation may show this step at the package terminal since the signal is still propagating to the far end, which is the pad on the die.  AM335x is a wire-bond package, so the wires may have enough inductance to create the non-monotonic transition they are observing at the package terminal.

Based on the note, the non-monotonic waveform at the pin should not be a concern.

I have also reached out to the simulation expert who is on vacation.

I will update any additional inputs i receive.

Regards,

Sreenivasa

Hello David Park, 

Please refer below additional inputs i received 

Probing at the die is what needs to be done for read transactions. We do not care about the waveform at the BGA pin for DDR read transactions since it is expected to be distorted, given that the probe point is farther away from the on-die terminations.

The signal quality at the PHY (i.e. on-die) is what matters.

Regards,

Sreenivasa