improving cross regulation with synchronous rectification

[colin]

Hi,
I have a plethora of power supplies on my project,
250,30,9,5,3.3,1.8,-5
with an input range of 8-15v (ie to be able to use a reasonable range of
wall wart/battery)
at the moment I have 4 switching regulators (ICs with internal switches) and
a few linear ones.

Im looking at synchronous recitification to improve efficiency for the
lowest voltages,
however it looks like this aproach could also be applied to multiple
switcher outputs to provide good enough regulation to get away with just one
switcher for all the lower voltages, by extending or even delaying the on
time of the rectifier to reduce the output votage.

There are a few Maxim SMPS ics wich have drive for external switch and synch
rectifier MOSFETS,
some of these allow feedback from an auxilary output,
although this stops the aux output falling below a certain level it doesnt
seem to offer good regulation.
They also have some synchronous rectification ICs with external drives but
these dont seem to offer regulation at all.

Are there any good solutions with readily available ICs for this aproach ?

Colin =^.^=

 

[legg]

Hi,
I have a plethora of power supplies on my project,
250,30,9,5,3.3,1.8,-5
with an input range of 8-15v (ie to be able to use a reasonable range of
wall wart/battery)
at the moment I have 4 switching regulators (ICs with internal switches) and
a few linear ones.

Im looking at synchronous recitification to improve efficiency for the
lowest voltages,
however it looks like this aproach could also be applied to multiple
switcher outputs to provide good enough regulation to get away with just one
switcher for all the lower voltages, by extending or even delaying the on
time of the rectifier to reduce the output votage.

There are a few Maxim SMPS ics wich have drive for external switch and synch
rectifier MOSFETS,
some of these allow feedback from an auxilary output,
although this stops the aux output falling below a certain level it doesnt
seem to offer good regulation.
They also have some synchronous rectification ICs with external drives but
these dont seem to offer regulation at all.

Are there any good solutions with readily available ICs for this aproach ?

Post-regulating off a winding of the main isolaton transformer,
depending on topology, ends up having to deal with the voltage stress,
lower duty cycle and leakage energy associated with same.

It can be done, but there's seldom a profit generated by the approach,
and it usually means discrete components in series with rectifiers
that are there already - doubling losses. I've only used it where a
single extra output was required as a late addition, at a time when
discrete switching regulators were rare and overpriced.

Integrated controllers have a much easier time dealing with
pre-regulated DC inputs,where the second converter losses need not be
as large as when operating at the stress levels of the actual
pre-regulator.

CS5101 from ON was one of the early parts developed specifically for
the job in single-ended forward converters. I used a simple 3842,
synchronzed by the input waveform being regulated.

RL

 

[legg]

On Sat, 09 Sep 2006 12:16:14 GMT, "colin"

Are there any good solutions with readily available ICs for this aproach ?

Forgot to mention that there are other methods of reducing
cross-regulation. These involve degenerating the main regulation loop,
or introducing a negative impedance in series with the main regulation
loop, to counteract main load effects (only) on main duty cycle.

As this is only one regulator, improving the performance of all
auxiliaries, it can pay off to develop loss-free methods. All that's
left to contend with in auxiliaries, is load regulation.

RL

 

[colin]

Post-regulating off a winding of the main isolaton transformer,
depending on topology, ends up having to deal with the voltage stress,
lower duty cycle and leakage energy associated with same.

It can be done, but there's seldom a profit generated by the approach,
and it usually means discrete components in series with rectifiers
that are there already - doubling losses. I've only used it where a
single extra output was required as a late addition, at a time when
discrete switching regulators were rare and overpriced.

The idea is to use synchronous rectification on both the main and aux
outputs.
this increases efficiency considerably, especially for the low 1.8v.
(1.8v is for a dual 400mhz DDS.)

The regulation comes about when the output needs to be reduced by
controlling the MOSFET that is used as a synchronous rectifier:-
a) not turning the MOSFET on but wich still leaves the body diode in series.
b) allowing the MOSFET to let current flow back from the output into the
transformer, before the main switch is turned on ofc, wich doesnt lose
energy.
the later helps cross regulation even when it is only used on the main
output.

If you look at the MAX1652, MAX1653, MAX1654, MAX1655, they show you the
principle involved.
http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1861

Integrated controllers have a much easier time dealing with
pre-regulated DC inputs,where the second converter losses need not be
as large as when operating at the stress levels of the actual
pre-regulator.

CS5101 from ON was one of the early parts developed specifically for
the job in single-ended forward converters. I used a simple 3842,
synchronzed by the input waveform being regulated.

Yes there are a number of integrated switching ICs wich just need an
inductor and capacitor,
the good ones like lm2594 are quite good but a bit expensive,
the cheap hysteretic ones I dont like much,
they are inclined to screech at light loads.

Actually ive seen some dual output fully integrated switcher ICs with
synchronous rectification such as
http://www.maxim-ic.com/quick_view2.cfm/qv_pk/4595
but unfortunatly the ones with integrated switches have a low voltage input
range.

Might be worth reducing my input voltage range.

Colin =^.^=

 

[legg]

On Sat, 09 Sep 2006 16:21:16 GMT, "colin"

The idea is to use synchronous rectification on both the main and aux
outputs.
this increases efficiency considerably, especially for the low 1.8v.
(1.8v is for a dual 400mhz DDS.)

The regulation comes about when the output needs to be reduced by
controlling the MOSFET that is used as a synchronous rectifier:-
a) not turning the MOSFET on but wich still leaves the body diode in series.
b) allowing the MOSFET to let current flow back from the output into the
transformer, before the main switch is turned on ofc, wich doesnt lose
energy.
the later helps cross regulation even when it is only used on the main
output.

If you look at the MAX1652, MAX1653, MAX1654, MAX1655, they show you the
principle involved.
http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1861

These parts are straight DC-DC regulators. They may employ synchronous
rectification, but they have nothing to do with the reduction of cross
regulation effects seen in an auxiliary rectifier output. If you look
at the data sheet for the CS51001, you will see the switches in series
wth the normal rectifiers, using the output incyctor that would
normally be required in a single-ended forward topology. The same
principle allows it's (very lossy) use in series with flyback
rectifiers. (see St Micro AN511)

The use of active synchronous rectifiers in place of standard
rectifiers has no inherent advantage in reducing cross-regulation
effects, except at light loads where output inductors of standard
rectifiers can become discontinuous.

Yes there are a number of integrated switching ICs wich just need an
inductor and capacitor,
the good ones like lm2594 are quite good but a bit expensive,
the cheap hysteretic ones I dont like much,
they are inclined to screech at light loads.

I fear that I've misunderstood your original posting. If I'd known you
were talking about the uses of DC-DC post-regulators, I would not have
responded. DC-DC post regulators don't give a damn what kind of input
regulation they are correcting, or where it comes from. they are
affected by crossregulation as much as any other load circuit is.

It's only when post-regulators are not used, or when headroom is
critical (ie to reduce linear losses), that the reduction of cross
regulation effects becomes important in multi-output converters.

RL