Fixed off time buck controller

Y

Yzordderex

Guest
A friend of mine is designing a buck controller to run an AM radio
modulator. It's basically a buck controller that has an audio input
as a reference.

Looks like he is going to use a interleaved 2 phase design. I got to
thinking -since duty cycle of each controller phase is not going over
50% could he use a variable frequency approach with a fixed off-time?
If he set up right (with lc network near the source of each fet)then
he might be able to get the source voltage to ring back to zero when
it came time for next turn-on.

I am running full bridge resonant converter, and works nice. This is
just slightly different flavor for my friend.

regards,
Bob
N9NEO
 
In article <a69e22f6.0403151231.471f956a@posting.google.com>,
Yzordderex <yzordderrex@verizon.net> wrote:
[...]
50% could he use a variable frequency approach with a fixed off-time?
If he set up right (with lc network near the source of each fet)then
he might be able to get the source voltage to ring back to zero when
it came time for next turn-on.
Click to expand...
A fixed off-time design was fairly common in roll your own switchers. One
LM339 could do all of the needed logic and timing and work as the feedback
amplifier. Normally the off time was made just long enough to be sure
that all of the energy is out of the inductor before the switch is turned
on again.

The ringing frequency in these beasts was usually quite high and didn't
account for much energy.


--
--
kensmith@rahul.net forging knowledge
 
"Yzordderex" <yzordderrex@verizon.net> wrote in message
news:a69e22f6.0403151231.471f956a@posting.google.com...
A friend of mine is designing a buck controller to run an AM radio
modulator. It's basically a buck controller that has an audio input
as a reference.

Looks like he is going to use a interleaved 2 phase design. I got to
thinking -since duty cycle of each controller phase is not going over
50% could he use a variable frequency approach with a fixed off-time?
If he set up right (with lc network near the source of each fet)then
he might be able to get the source voltage to ring back to zero when
it came time for next turn-on.

I am running full bridge resonant converter, and works nice. This is
just slightly different flavor for my friend.

regards,
Bob
N9NEO
You can't have your cake and eat it too! If you want to go 2 phase
Click to expand...
interleaved, you must go fix frequency. As you point out each channel will
have it's own different variable frequency with fixed off time so no phase
lock.

Regards
Harry
 
Looks like he is going to use a interleaved 2 phase design. I got to
thinking -since duty cycle of each controller phase is not going over
50% could he use a variable frequency approach with a fixed off-time?
If he set up right (with lc network near the source of each fet)then
he might be able to get the source voltage to ring back to zero when
it came time for next turn-on.

I am running full bridge resonant converter, and works nice. This is
just slightly different flavor for my friend.

You can't have your cake and eat it too! If you want to go 2 phase
interleaved, you must go fix frequency. As you point out each channel will
have it's own different variable frequency with fixed off time so no phase
lock.
Click to expand...
Depends on the controller. One controller can develop multiple phases
using the same feedback. With a common load, even simple slaving is
possible if symmetry isn't critical. The benefits of interleaving are
still obtainable.

RL
 
"R.Legg" <legg@magma.ca> wrote in message
news:e715b5cc.0403152322.2302d8b8@posting.google.com...
Looks like he is going to use a interleaved 2 phase design. I got to
thinking -since duty cycle of each controller phase is not going over
50% could he use a variable frequency approach with a fixed off-time?
If he set up right (with lc network near the source of each fet)then
he might be able to get the source voltage to ring back to zero when
it came time for next turn-on.

I am running full bridge resonant converter, and works nice. This is
just slightly different flavor for my friend.

You can't have your cake and eat it too! If you want to go 2 phase
interleaved, you must go fix frequency. As you point out each channel
will
have it's own different variable frequency with fixed off time so no
phase
lock.


Depends on the controller. One controller can develop multiple phases
using the same feedback. With a common load, even simple slaving is
possible if symmetry isn't critical. The benefits of interleaving are
still obtainable.

RL
I'm not buying that, you have two inductors that are not matched exactly
Click to expand...
with the same volt seconds across them. The delta current will be very close
but the DC current will not. One loop can be in continuous mode the other in
discontinuous mode. I don't get "if symmetry isn't critical", but it is
critical to get a current balance in both inductors. If both loops operate
in discontinuous mode the DC current component is zero and you are OK.
Continuous current mode, interleaved, fixed off time = fixed on time = fixed
frequency = DC current unbalance.

Regards
Harry
 
"R.Legg" <legg@magma.ca> wrote in message
news:e715b5cc.0403152322.2302d8b8@posting.google.com...
| > > Looks like he is going to use a interleaved 2 phase design. I got
to
| > > thinking -since duty cycle of each controller phase is not going
over
| > > 50% could he use a variable frequency approach with a fixed
off-time?
| > > If he set up right (with lc network near the source of each
fet)then
| > > he might be able to get the source voltage to ring back to zero
when
| > > it came time for next turn-on.
| > >
| > > I am running full bridge resonant converter, and works nice. This
is
| > > just slightly different flavor for my friend.
|
| > You can't have your cake and eat it too! If you want to go 2 phase
| > interleaved, you must go fix frequency. As you point out each
channel will
| > have it's own different variable frequency with fixed off time so no
phase
| > lock.
| >
|
| Depends on the controller. One controller can develop multiple phases
| using the same feedback. With a common load, even simple slaving is
| possible if symmetry isn't critical. The benefits of interleaving are
| still obtainable.
|
| RL

Not if, as suggested, the OP wants true resonant mode control with zero
voltage transitions. Forcing current sharing will force different
frequency operation. Forcing constant frequency operation will lose zero
voltage switching.

DNA

DNA
 
Ok, I will stay out of his design path then. I see the point about
the inductors not sharing so well. My design is king anyway. I just
thought I might be able to help him out. I know there is a company up
the road here Syncor is the name I think - They make these little
bricks that I think run slaved forward converters with the planar
magnetic thing. Cute little units.

73
Bob
N9NEO




"Emoneg" <Emoneg@nothere.com> wrote in message news:<kiH5c.488$Do3.43@newsfe2-gui.server.ntli.net>...
"R.Legg" <legg@magma.ca> wrote in message
news:e715b5cc.0403152322.2302d8b8@posting.google.com...
| > > Looks like he is going to use a interleaved 2 phase design. I got
to
| > > thinking -since duty cycle of each controller phase is not going
over
| > > 50% could he use a variable frequency approach with a fixed
off-time?
| > > If he set up right (with lc network near the source of each
fet)then
| > > he might be able to get the source voltage to ring back to zero
when
| > > it came time for next turn-on.
|
| > > I am running full bridge resonant converter, and works nice. This
is
| > > just slightly different flavor for my friend.

| > You can't have your cake and eat it too! If you want to go 2 phase
| > interleaved, you must go fix frequency. As you point out each
channel will
| > have it's own different variable frequency with fixed off time so no
phase
| > lock.
|
|
| Depends on the controller. One controller can develop multiple phases
| using the same feedback. With a common load, even simple slaving is
| possible if symmetry isn't critical. The benefits of interleaving are
| still obtainable.
|
| RL

Not if, as suggested, the OP wants true resonant mode control with zero
voltage transitions. Forcing current sharing will force different
frequency operation. Forcing constant frequency operation will lose zero
voltage switching.

DNA

DNA
Click to expand...
 
"Emoneg" <Emoneg@nothere.com> wrote in message news:<kiH5c.488$Do3.43@newsfe2-gui.server.ntli.net>...
"R.Legg" <legg@magma.ca> wrote in message
news:e715b5cc.0403152322.2302d8b8@posting.google.com...
| > > Looks like he is going to use a interleaved 2 phase design. I got
to
| > > thinking -since duty cycle of each controller phase is not going
over
| > > 50% could he use a variable frequency approach with a fixed
off-time?
| > > If he set up right (with lc network near the source of each
fet)then
| > > he might be able to get the source voltage to ring back to zero
when
| > > it came time for next turn-on.
|
| > > I am running full bridge resonant converter, and works nice. This
is
| > > just slightly different flavor for my friend.

| > You can't have your cake and eat it too! If you want to go 2 phase
| > interleaved, you must go fix frequency. As you point out each
channel will
| > have it's own different variable frequency with fixed off time so no
phase
| > lock.
|
|
| Depends on the controller. One controller can develop multiple phases
| using the same feedback. With a common load, even simple slaving is
| possible if symmetry isn't critical. The benefits of interleaving are
| still obtainable.
|
| RL

Not if, as suggested, the OP wants true resonant mode control with zero
voltage transitions. Forcing current sharing will force different
frequency operation. Forcing constant frequency operation will lose zero
voltage switching.
Click to expand...
The methods of generating ZVS transitions are beside the point. So is
the absolute accuracy of phase sharing.

I'm not saying that synchronization of topologies with variable
frequency is easy, or even optimal, only that it is practical and
possible. The effects of interleaving on ripple will be comparable, in
effect.

It is found most commonly in hysteretic and critical conduction
circuitry, where frequency is generally load dependant. Methods to
avoid local bifurcations may be required if the control loop isn't
fairly rugged.

I've fiddled with it myself, where the simplicity of the controller
for a single phase was too simple and cheap to chuck out willingly.

"Interleaved Synchronous Buck Regulator with hysteretic voltage
control"
Wei Gu and Issa Batarseh
Pesc'01

http://apec.engr.ucf.edu/publications/conferences/pesc/WeiguPESC%5B1%5D.pdf

"Comparative Analysis of Three Interleaved Boost Power Factor
Corrected Topologies in DCM"
Teodorescu,R., Kjćr,S.B., Munk-Nielsen., Blaabjerg,F., Pedersen,J.K.
Proceedings of IEEE PESC'01

(Direct link now invalid but available from SB Kjaer by writing)
http://www.iet.auc.dk/~sbk/

RL
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top