AM4378: IBIS Model

Hello Amarnath,

It does not look like your images uploaded properly. Please look at the icons above the text entry area, and you should see a white square with a smaller grey square inside of it and a little circle in the bottom right of the squares. You can use that to insert images.

Regards,

Nick

FYI: the subject owner is on vacation until January 11. If you need assistance earlier than that, let us know and I will see if we can get someone else to comment in the meantime.

Regards,

Nick

Hi Nick,

It would be helpful if any one can support this little earlier as we require the IBIS model to complete our SI simulation

Regards,

M Amarnath

Hello Amarnath,

Ok, I have reassigned the thread for now. Please see my reply above about adding pictures to your post.

Regards,

Nick

Hi Nick,

Pls find images updated as per your suggestion.

 IBIS model (selector) of DQSP and DQSN is not same, both are different. Any particular reason for the different model selector for signals  in same differential pair. Can this be changed in the IBIS model by TI?

Clock pins haven’t been declared as differential pins in IBIS model. Please help us in assigning the same in IBIS model to run simulation

Hi Amarnath,

Thanks for the updated information. I will look into this further and get back to you.

Can you also clarify whether you are looking at the IBIS model sprm636.ibs or sprm636a.ibs? The questions you raised apply to both but sprm636a.ibs contains fixes to some other issues.

regards,

-Shriram

Hi Amarnath,

The buffer definition for strobes defines the P/N separately and should be fine. However I am checking our internal simulations to see what settings were used, as a sanity check.

On a separate note, can you also share the FAE contact at TI whom you are working with?

I will share any files through them, if required, that are too large to attach here.

thanks,

-Shriram

Hi Shriram,

Thank you for the update. We are using sprm636a.ibs model.

I have shared this thread to our TI contact. Please help in sharing the updated IBIS model with the clock signals assigned as differential pair.

Regards,

M Amarnath

Hi Shriram,

Any update on the above IBIS request

Regards,

M Amarnath

Hi Amarnath,

Our simulations team is checking the settings used in our internal simulations and I should be able to provide you an answer shortly.

regards,

-Shriram

Hi Shriram,

The same problem faced by me also. I am colleague of Amarnath.

The DDR3L clock signal(DDR_CK and DDR_CK_N) is not assigned as differential in sprm636a ibis model. Please find the attachment  which I faced during simulation. Please suggest how to proceed or provide the updated mode

Hi,

Can you clarify what tool you are using?

Also are you facing an error or a warning while reading in these models?

As I had mentioned, the IBIS model aligns with our internal SPICE simulations in that both specify the same buffer type for CLK and CMD signals.

I will also check with our team on the specific buffer type used, to further ensure that the IBIS model is accurate.

regards,

-Shriram

HI Shriram,

1. We are using the hyperlynx tool.

Tool: HperLynx SI/PI/Thermal VX2.7_Update2

Model used:

Processor : (AM4378BZDNA100S) , Model: sprm636a

DDR3 IC:(IS43TR16256BL-125KBLI), Model : IS43TR16256BL

2. We are getting "Differential pin mismatches" message

The message observed is like below:

Thanks,

Gunasekar.T

Hi Gunasekar,

Ok. Is the issue you observe in Hyperlynx an error or a warning?

regards,

-Shriram

Hi Shriram,

It showed as "net assignment issue". Please see below image of error.

Thanks,

Gunasekar.T

Sorry - I cannot see the image clearly. Could you pl re-attach?

My question was more with regards to whether you are seeing the simulation abort due to this or whether it is progressing and Hyperlynx just outputs this as information?

regards,

-Shriram

Please find the attachment of error.

Simulation is not able to start due to this differential signal not assigned.

Thanks for the update. I have also confirmed with our simulation team that the exact I/O models specified in the IBIS have been used in our internal simulations. The IBIS model therefore should not be modified. I will check internally if there is anything else that can be done from our side.

regards,

-Shriram