TPS65916: Best way to power TPS65916 from a Li-ion battery?

Hello Alexander,

You may be able to get away with a fixed-3.3V LDO between Li-Ion battery and LDO12_IN. The only concern here is if you need the system to start up with low battery voltage, you'll need to confirm if you can meet the minimum voltage of the SDIO domain on LDO1 output. But in this case, I don't think you need to worry about LDO12_IN being higher than VCCA.

Alternatively, if your max PHY current on the processor is less than 200mA, you could consider powering the PHYs only from LDO3, and leave LDO2 unused. In that case, you could power LDO12_IN from the VDDSHVx supply, since LDO1 is sequenced after VDDSHVx. LDO2 may not work in this condition if it's enabled before its supply voltage is present, but that won't matter if it's unused in this case.

If those options don't work, I think you may need a buck-boost to supply LDO12_IN if you need ~2.8V to ~4.2V supply voltage.

Regards,
Karl

Hello, Karl

Thank you for the reply.
My main goal is having SMPS powered straight from the battery to eliminate the need for having a high-power buck-boost on board and increase efficiency. I would be content with having a relatively low-power buck-boost supplying LDO12_IN.

Do I not need to worry about LDO12_IN being higher than VCCA? Looking at datasheet 4.5 I am concerned with LDO2 being in normal mode with Vout=1.8 and having input voltage above specified limits. Is is safe to use it that way? Or alternatively, is it safe to power it up that way but not using the ouput voltage from LDO2? It is preferable to be able to start up with battery voltage below 3.3V.

I plan to use USB3.0 SuperSpeed and SATA 2.0 interfaces and power estimation tool gives me about 120 mA consumption on 1.8V Analog PHY rail. Is it safe to assume i can power them both from LDO3?

If i understand correctly, VDDSHVx is a separate supply that PMIC does not provide (except for SDIO) and it is sequenced by PHY through a load-switch using a GPIO, is that right? (looking at TPS65916 User' s Guide to PowerAM571x figure 1), is that correct?

(*edited to clarify the use of interfaces)

Or maybe I can use buck to get 2.8V from a battery and use it to suply LDOx_IN (except for LDO4 that makes 3.3 from 5, I would still need a separate boost for that) and not use LDO1 at all and supply VDDSHV8 (SDIO) from the external VDDSHVx source (datasheet allows ramping them up at the same time).

Sounds nice to me, do you see any problems with this solution?

Hello Alexander,

> Do I not need to worry about LDO12_IN being higher than VCCA? Looking at datasheet 4.5 I am concerned with LDO2 being in normal mode with Vout=1.8 and having input voltage above specified limits. Is is safe to use it that way? Or alternatively, is it safe to power it up that way but not using the ouput voltage from LDO2? It is preferable to be able to start up with battery voltage below 3.3V.

[Karl] You do need to make sure LDO12_IN is less than VCCA. What I meant to say is that if you take an LDO from VCCA to LDO12_IN, then it must be less than VCCA due to the LDO dropout. It wouldn't be possible in this case for LDO12_IN to be higher than VCCA. I think this is the same as your proposal, except you suggested a buck instead of LDO.

> I plan to use USB3.0 SuperSpeed and SATA 2.0 interfaces and power estimation tool gives me about 120 mA consumption on 1.8V Analog PHY rail. Is it safe to assume i can power them both from LDO3?

[Karl] Yes, as long as you don't intend to add more PHY usage which would increase the current. LDO3 can supply up to 200mA.

> If i understand correctly, VDDSHVx is a separate supply that PMIC does not provide (except for SDIO) and it is sequenced by PHY through a load-switch using a GPIO, is that right? (looking at TPS65916 User' s Guide to PowerAM571x figure 1), is that correct?

[Karl] Correct, VDDSHVx supplies should be powered by an external switch sequenced by REGEN1.

> Or maybe I can use buck to get 2.8V from a battery and use it to suply LDOx_IN (except for LDO4 that makes 3.3 from 5, I would still need a separate boost for that) and not use LDO1 at all and supply VDDSHV8 (SDIO) from the external VDDSHVx source (datasheet allows ramping them up at the same time).

[Karl] LDO1 is needed if you want to support dual-voltage SD cards. If so, then you will need to set the SDIO supply to 1.8V, while the remaining VDDSHVx supplies stay at 3.3V. So you must have a separate supply for that.
In case you only need 3.3V for SDIO, then you can combine it with other VDDSHVx rails and not use LDO1.

Regards,
Karl

Hello, Karl

Thank you for the clarifications.

I have one last question - is it preferable to use REGEN as "enable" pin on a converter or is it better to use it as a control signal for a load-switch with converter being enabled all the time? I guess in both cases I would need to make sure the time from REGEN going high to voltage appearing on processor pins is within power sequencing limits so it doesn't appear too late. It is tempting to eliminate the need for a separate IC for load-switch functionality but I guess that's a faster option.