Changing VDD in STM32H7 from 3.3V to 1.8V on the fly for UHS-1 support for SDMMC?

[PNare]

Greetings everyone,

I'm thinking of a setup where a STM32H7 series chip is with

• a 1.2V regulator powers the core directly bypassing the LDO (REG1)
• another regulator whose output can be switched between 3.3V or 1.8V (default 3.3V on power up) powering the VDD (REG2)

After powering on the STM32H7, checking the SD card is present, it'll send the CMD11 to say the SD card that the data, cmd and clock line will switch to 1.8V. At the same time it'll configure the REG2 to switch output to 1.8V so that the IO lines would be 1.8V.

Would this setup work? This is so that we can find an inexpensive solution. Voltage translators like IP4855CX25 are expensive since they need PCBs with via under balls as its BGA pitch is 0.4mm.

@Community member​ as you're looking into DDR mode with SDMMC, any thoughts on this?

Thanks, Prithvi

[Tesla DeLorean]

I've used eMMC devices in DDR mode, where the translator is not required.

[Tesla DeLorean]

You might be able to fashion with discret level converters, directional for Data, uni-directional for CLK/CMD, and then a feed-back of the CLK (ie out and back loop)

Perhap Qty 4 of these http://www.ti.com/lit/ds/symlink/sn74axc4t245-q1.pdf

I guess one must decide if that would be too large and cumbersome to get the performance boost.

[PNare]

Thanks Clive. Will explore this option. Can you provide link to any resource to understand the feedback needed for the CLK?

Also what do you think of changing VDD?

[PNare]

Hi Clive,

Would using auto bi-directional chips like http://www.ti.com/lit/ds/symlink/lsf0102.pdf work? Also CMD must be bidirectional right? Thanks.

[Tesla DeLorean]

You'd have to experiment.

Yes CMD has a direction control, was skimming the NXP diagram