Synchronous Buck capable of Sinking?

[bitbanger]

Hi Folks -

I have a little "power ORing" conundrum and I'm curious: how do sync buck converter behave when asked to sink current? I'm sure this is largely controller specific so I'm looking at the LTC3621 (integrated switches) in particular.

http://cds.linear.com/docs/en/datasheet/3621fc.pdf

My first guess was that it would try to regulate by keeping the bottom (NMOS?) switch activated, essentially dumping the inductor to ground (effectively "shorting" the regulator output to ground). However in reading the datasheet closer I'm not so sure (because it's partially regulating on inductor current as well as FB voltage). Especially interesting is in the section describing the "burst mode":

"Thus, when the switcher is on at relatively light output loads, FB voltage will rise and cause the ITH voltage to drop. Once the ITH voltage goes below 0.2V, the switcher goes into its sleep mode with both power switches off. The switcher remains in this sleep state until the external load pulls the output voltage below its regulation point."

Does this mean I can safely present a higher voltage at the output of the inductor? For my project I'm looking at the switcher (output variable between 4.5-7.4V) facing a potential fixed 7.5V.

Another thought was that the regulator would "fall out of regulation", and perhaps I can finagle the PGOOD pin with the RUN to shut the regulator down? But then there's startup...hmm.

Thoughts?

Happy wiring,
Jamie

[Siwastaja]

Synchronous buck happily sinks if the controller allows reverse current. Some controllers prevent it.

It works as a synchronous boost, injecting current into the "input", so you need input voltage measurement circuit to shut it down before the voltage rises too far to cause damage.

[chris_leyson]

how do sync buck converter behave when asked to sink current?

The LTC3621 won't sink current, the bottom mosfet switches off when the current through it drops to zero if running in discontinuous mode or it switches off at the end of the switching cycle if running in continuos mode. There is a "reverse comparator" in the control logic that senses the mosfet current. Off hand I don't know of any synchronous buck controllers that will allow true synchronous operation and allow power to flow back from output to input. They're not hard to build though.

[David Hess]

Off hand I don't know of any synchronous buck controllers that will allow true synchronous operation and allow power to flow back from output to input. They're not hard to build though.

Linear Technology has them under the DDR Memory/Bus Termination catagory:

http://www.linear.com/parametric/ddr_memory|bus_termination