synchronous buck converter

[SwedishWings]

Perhaps I'm posting this in the wrong place but...

I've been searching for a clever and simple solution to make synchronous MOSFET step down buck converter with discreete components, i.e. i don't want to use a PWM controller chip (i have a mcu that will handle that).

I just can't seem to find any existing designs. Has anyone managed to build one? The problem of course is the break before make feature that a PWM controller offers.

Any suggestions would be most welcome.

Thanks,

Mike

[Bakmeel]

Mike,

Sattellite power systems are often equipped with very simple discrete power converters, so a discrete buck is certainly feasible. In fact, you will see that you will understand a lot better the workings of a PWM controller once you have made one "hard core" ;-) So don't get discouraged and go forward!

Synchronous rectification is in fact very easy. Use the hysteresis levels inside an IR2110 high-low gate driver. For both high and low side inputs, implement a small RC network with a diode antiparallel (see image). For switch-on, the timing will be delayed while the switch-off is immediate. Tuning the RC delay allows to tune the amount of cross-conduction that you will allow.

Does your mcu generate pwm? or do you need to build a discrete PWM stage as well?

Bakmeel

[SwedishWings]

Thanks Bakmeel!

I have a proper design with almost 90% efficience using a diode at the low side. My idea was to find a way replace the schottky with a MOSFET (without a driver chip) to gain some efficiency. A problem is that my input voltage is up to 30V, which limits my options concerning driver chips.

But maybe the only way to go is to use a driver chip and leverage from the hysteresis, just like you suggest. Kind of irritating that i can't find a single discrete solution :/

Have you seen an all-discrete solution?


Thanks,
Mike

[Bakmeel]

I have not seen one personally, but that's more due to the fact that synchronous rectification is not often used in my area. That doesn't mean it isn't feasible.

Have you considered adopting a transformer-based gate driver? You'd need a few tricks to secure the dead-time.

The reason why I like the 2110 is because it has a very good matching in turn-on delay for high and low side (less than a few nanoseconds), and has a not so big difference in turn-on and turn-off delay. That allows you to accurately define the dead-time you need.

How did you plan to implement the high-side gate driver in the first place?

[SwedishWings]

I checked out the 2110, and it looks like a very good device, a bit expensive though.

As for transformers, it's way to bulky and expensive. Part of the reason that I'm looking for alternatives is to get the cost down.

The current design looks like the attached picture. The hi side is driven by two push pull switching transistors. It turns out that this works very well with most MOSFET's that i tried. I get quite fast switching times even with hard to dive MOSFET's.

Perhaps i should state my question differently: Is there any simple discrete dead time circuit working in the nano/micro second domain?

Thanks,
Mike

[Bakmeel]

Do you consider a dual voltage comparator as a discrete circuit?

Point is that the simplest timing circuit should be RC based. But then to get at least some timing accuracy you will need to feed that signal into any kind of hysteretic comparator. This you could do with simple logic devices, or with analogue voltage comparators. But these are always integrated devices. Other than that you'd have to make up something else using a proper reference.

Have a look for discrete voltage comparators.

Cheers,
Bakmeel