Synchronous rectifier

[[email protected]]

Hi, I am working on a integrated synchronous rectifier, in which a
comparator sense the Vds of a MOSFET to control the turning on and off
of the MOSFET. So the comparator is a zero-crossing detector. In the
design of the comparator, the offset is well controled to enhance the
efficiency. But in very light load/zero load, there is some
oscillation. Thus my question is: is anyone of you once meet this
instability problem and could anyone give some advice on this?

Thanks...
Eric

 

[kell]

Hi, I am working on a integrated synchronous rectifier, in which a
comparator sense the Vds of a MOSFET to control the turning on and off
of the MOSFET. So the comparator is a zero-crossing detector. In the
design of the comparator, the offset is well controled to enhance the
efficiency. But in very light load/zero load, there is some
oscillation. Thus my question is: is anyone of you once meet this
instability problem and could anyone give some advice on this?

Thanks...
Eric

I don't have much experience with synchronous rectification, but lack
of experience never stopped me from shooting my mouth.
Get some hysteresis by using a little positive feedback on the
comparator, maybe.
What's this for, a SMPS?

 

[Harry Dellamano]

Hi, I am working on a integrated synchronous rectifier, in which a
comparator sense the Vds of a MOSFET to control the turning on and off
of the MOSFET. So the comparator is a zero-crossing detector. In the
design of the comparator, the offset is well controled to enhance the
efficiency. But in very light load/zero load, there is some
oscillation. Thus my question is: is anyone of you once meet this
instability problem and could anyone give some advice on this?

Thanks...
Eric

The bane of synchronous rectifiers is dead time which occurs at light
load. Select a topology that contains no dead time at light loads.
cheers,
Harry

 

[Genome]

Hi, I am working on a integrated synchronous rectifier, in which a
comparator sense the Vds of a MOSFET to control the turning on and off
of the MOSFET. So the comparator is a zero-crossing detector. In the
design of the comparator, the offset is well controled to enhance the
efficiency. But in very light load/zero load, there is some
oscillation. Thus my question is: is anyone of you once meet this
instability problem and could anyone give some advice on this?

Thanks...
Eric

Yes, it am lot complicated. Wiv synchronous rectification, if your converter
am allow it, then you am end up with continuous current in the output
inductor even at light/zero load currents. It am mean your inductor current
am also go negative and you am have possibility of returning power from
output to input, what am called 'regen'. If your converter am not allow for
this then it am get upset about things and overvoltage some of its bits.
This am special problem wiv off line stuff which am have transformerer and
leakage inductorance stuff lieing about.

I am before watch people design such stuff and they am see weirdo stuff
happen then am explane why plus think am cool new feature Then get hot
and blow up :-(

If you am not have this problem then it am should work but problem might be
elsewhere. You am rely on drain of catch mosfet to fall to zero and it am
driven by current in inductor. If current in inductor am not enuff to driv
drain to zero then do not happen so am go 'burp'.

I reccomend you am get LTspice and do model to prod about wiv what am going
on. I am have done this myself and am learned lot about it wivout thing
going 'BANG'.

Unfort you am got Gmail accunt plus post from Gugle.

Sigh

DNA

 

[Ken Smith]

I don't have much experience with synchronous rectification, but lack
of experience never stopped me from shooting my mouth.
Get some hysteresis by using a little positive feedback on the
comparator, maybe.
What's this for, a SMPS?

Or...

If the oscillation happens right at the point where the thing is supposed
to turn off, you can include a latch to hold it off.

 

[[email protected]]

Thank you for your reply!

I have said that the comparator is a zero-crossing detector. So it is
used to compare Vds with zero, but in fact there are some offset
voltage in the input of the comparator. Thus the comparator point is
actually +/- several mili volts. If the comparator point is positive
several milivolts, there will be a large delay when turning off the
MOSFET; a negative several milivolts will cause oscillation under
light/zero load. Now you may understand it is very hard to have a
hysteresis for such a comparator and in fact a hyteresis here is of no
use.

This is for a VRM.

 

[[email protected]]

Yes, I once think about adding some latch circuits, but the latch
signal seems must come from the control signal of the main switch in a
buck converter. I hope my chip can work alone, so a external latch
signal is not a good solution. If there are any other solutions...

 

[kell]

Thank you for your reply!

I have said that the comparator is a zero-crossing detector. So it is
used to compare Vds with zero, but in fact there are some offset
voltage in the input of the comparator. Thus the comparator point is
actually +/- several mili volts. If the comparator point is positive
several milivolts, there will be a large delay when turning off the
MOSFET; a negative several milivolts will cause oscillation under
light/zero load. Now you may understand it is very hard to have a
hysteresis for such a comparator and in fact a hyteresis here is of no
use.

This is for a VRM.

this article has some ideas
http://www.edn.com/article/CA216166.html

 

[[email protected]]

To Kell,

Thanks for your reply and time.

I read this article just several days ago. My integration solution is
mentioned in it. And the new proposal to use a sense winding is not
suit for my solution. I want to do a full integration solution. It
seems that it is very hard to do like this...

 

[U.S.]

Thus spake Genome:

Yes, it am lot complicated.

Simple courtesy precludes me from telling exactly what I think of your making
fun of someone who in sincerely attempting to communicate. Just because his
use of the English language isn't first class (and since when is this a US/UK
forum?) you see fit to parody him poke fun at him.

And one wonders why the US has a PR problem around the world?

 

[Terry Given]

I think you missed the point. If you look at other posts by Genome you'll
note an apparently intentional, yet highly consistent use of bad language.
Since I don't have a damned idea what he thinks he's doing, I'd recommend
just quietly placing him in the killfile...

Tim

besides, hes a pom.

Cheers
Terry