AM6442: OSPI driver which supports training on SDR mode

Hello Hideaki,

I got this thread assigned late.

I will work on your question.

Please allow me sometime.

Regards,

Vaibhav

Hello Hideaki,

As of today in the latest MCU PLUS SDK release for AM64, we do not have SDR driver written. Please check below screenshot.

You can find this under MCU_PLUS_SDK_INSTALL_PATH/source/drivers/ospi/v0/ospi_phy.c

I hope this helps. I will check with dev team on this and update you in few business days.

Regards,

Vaibhav

Hello Hideaki,

Please check mail, I have provided some updates and asked a few follow up questions.

Regards,

Vaibhav

Reopening to get the conversation from email to e2e.

Current open questions

1. What is the SDK schedule for the OSPI PHY tuning algorithm for SDR mode?

2. Why is the customer wanting to use SDR and not DDR mode? DDR will be faster. @Hideaki-san can you please check with the customer?

Regards

Karan

Hello Hideaki,

Thanks for your patience.

Apologies in delayed responses.

Karan Saxena said:

2. Why is the customer wanting to use SDR and not DDR mode? DDR will be faster. @Hideaki-san can you please check with the customer?

Can you please check this with the customer.

Karan Saxena said:

1. What is the SDK schedule for the OSPI PHY tuning algorithm for SDR mode?

Currently this is under development for NAND devices like AM62A. As far as AM64x is concerned, we do not have the SDR algorithm till date.

An estimated release for this would be SDK version 10.

Regards,

Vaibhav

Hi Vaibhav,

Thank you for your reply and apologies in my delayed responses.

Vaibhav Kumar said:

2. Why is the customer wanting to use SDR and not DDR mode? DDR will be faster. @Hideaki-san can you please check with the customer?

Can you please check this with the customer.

Because there is the errata i2189 regarding OSPI DDR on AM64x.

They got your answer before that they do not need to take any further action if they use OSPI_phyTuneDDR() for ospi_phy.c.

At least from the errata point of view they don't need to do any modification.

However, OSPI_phyTuneDDR() only performs tuning and it seemed that i2189 Workaround 1, 2 and 3 didn’t exist in the source file.

Therefore, they judged that it was impossible to use DDR in this opaque situation. For this reason, they had asked you to provide them with SDR solution.

They want to confirm again with you. Are this errata i2189 Workaround 1, 2 and 3 not needed as long as API OSPI_phyTuneDDR() of a new ospi_phy.c is used ?

If the i2189 errata Workarounds are not needed at all, they can still use the 8D-8D-8D-8D driver provided by TI and only need to create 4D-4D-4D. That’s good for them.

Thanks and regards,
Hideaki

Hi Hideaki,

I have noted your response.

Allow me a day or two to check internally and provide you a response.

Regards,

Vaibhav

Hi Vaibhav,

Thank you for supporting this thread. Do you have any update ?

Regards,
Hideaki

Hi Hideaki,

Apologies in delayed responses.

I am going to clarify your doubts.

Hideaki Matsumoto said:

They want to confirm again with you. Are this errata i2189 Workaround 1, 2 and 3 not needed as long as API OSPI_phyTuneDDR() of a new ospi_phy.c is used ?

 

If the i2189 errata Workar

So, as of today the errata says the following:

In the current/latest SDK ospi_phy_dqs.c file, you will find the API termed as OSPI_phyTuneDDR(), this calls the OSPI_phyFindOTP1().

The find OTP API is the crucial API which takes care of the errata as mentioned and attached in the screenshot.

A detailed description of the flow of finding OTP will be extremely helpful for the customers, please share this link with them for the same: https://software-dl.ti.com/mcu-plus-sdk/esd/AM62X/latest/exports/docs/api_guide_am62x/DRIVERS_OSPI_PAGE.html#autotoc_md522

Hideaki Matsumoto said:

They want to confirm again with you. Are this errata i2189 Workaround 1, 2 and 3 not needed as long as API OSPI_phyTuneDDR() of a new ospi_phy.c is used ?

So to answer your question, yes they can proceed with the mode you have mentioned.

Regards,

Vaibhav