OMAPL138 Reference Design Power-on Sequence (TPS65070, TPS62420+TPS71733, TPS62353+TPS62231+TPS62232+TPS62111)

Inderjit,

Exceeding the 2V limit may electrically stress the chip and lead to damage. 

-Tommy

It seems that none of the three solutions presented on the L138 webpage, nor the Logic PD schematic guarantee that the 3.3V supply will be no more than 2V above 1.8V on power down (not sure if the Logic PD schematic even guarantees this condition on power-up, but I haven't reviewed it in detail).  On power-down, the 3.3V regulator is shut down when the 1.8V supply goes down, but 3.3V is not actively shunted to ground.  If the load on 3.3V is light, it could decay to 0V much more slowly than the 1.8V supply.

Inderjit,

The solutions are being reviewed to better address the 2V over-voltage requirement.  Until the update is published, I think the safest approach would be to use diodes between 3.3V --> 1.8V and diodes between 1.8V --> GND.

-Tommy

Tommy,

I understand the purpose of the diodes to ensure that 3.3V will pull 1.8V along to never exceed the 2V requirement.

However, I have difficulties to see the use case that the diodes between 1.8V and GND will address. Can you describe that scenario?

Thanks and best regards,

Manfred 

Manfred,

If the diode between 3.3V and 1.8V is activated (for example 3.3V supply is powered up, but 1.8V supply is at 0V or floating), the extra current flowing into the 1.8V plane can create overvoltage stress by raising the voltage above 1.8V.

-Tommy

Won't the diodes cause issues when both power supplies are on?  If the diode from 3.3V to 1.8V has the anode on 3.3V and the cathode on 1.8V (forward biased from 3.3V to 1.8V), it will tend to pull the 1.8V supply to 2.6V when both supplies are on.  A 2V zener diode is only rated at 2V at a particular current:  CZRF52C2 (Comchip) is rated at 2V @ 5mA.  At 20mA, the voltage is 2.4V.

Can you provide more details on what you have in mind?

Inderjit,

I'm not an expert with what diodes are available at reasonable prices, but I have heard of applications where three 0.7V diodes are stacked in series to cap voltages at or around 2.1V.  If there are diodes available with 0.67V drops, you could stack three of those.

-Tommy

It's not as simple as that:  a forward-biased diode has a variable voltage drop that depends on forward current, temperature, and some process variation.  For example, DFSL240 (Diodes, Inc), has an average forward voltage of 0.52V @ 1A, and a maxmium of 0.7V @ 2A.  This adds up to 1.56V and 2.1V respectively.  1.56V is just outside the 3.3V - 1.8V = 1.5V range.  Howver, a forward current of 0.5A has a typcial forward voltage drop of just 0.45V.  This adds up to just 1.35V, so the string of three diodes could conduct during normal operations, and cause an overvoltage condition on 1.8V, not to mention draw more current than necessary.

Inderjit,

You make a very good point.  I should probably stop speculating on what the official recommendation will be.

-Tommy

What happened to this thread?  It was moved to PMU, and then deleted, with no real answers.

Inderjit,

There's no official recommendation for dealing with the powerdown overvoltage at the moment.

-Tommy

Hello:  I am just starting a design with the L138.  Has an official recommendation for dealing with this problem (powerdown overvoltage) been created yet?

Thanks

Graem