Maker Pro
Maker Pro

Does isolated full bridge converter (DC-DC) not need to use snubber?

Hi everyone, my name is Wilson. I've been trying to design a 300W
12VDC to 380VDC converter lately as part of an inverter project. My
original approach was to use half bridge topology alongside with RC
snubber across primary winding. After I've made the transformer, I
found the energy stored by leakage inductance was unacceptably high. I
measured Lp as 2.5uH. At rated current 25A and switching freq 55kHz,
the average power to be dissipated due to energy stored in Lp will be
1/2 LI^2 * 55k= 43W. To use simple RC snubber will require a really
huge resistor, which is impractical and highly inefficient. My thesis
supervisor told me it's an unavoidable problem of half bridge
converter, and told me to use full bridge converter instead.

So all I want to ask is, is it not necessary to use snubber to
discharge leakage inductance energy in full bridge converter, since
the MOSFET already has body diode built in to freewheel the current
(due to both Lp and Lm) ? Or is my supervisor just setting me up?

thankyou for helping in advance and for looking at my question! ^_^
 
P

Paul Mathews

Jan 1, 1970
0
Hi everyone, my name is Wilson. I've been trying to design a 300W
12VDC to 380VDC converter lately as part of an inverter project. My
original approach was to use half bridge topology alongside with RC
snubber across primary winding. After I've made the transformer, I
found the energy stored by leakage inductance was unacceptably high. I
measured Lp as 2.5uH. At rated current 25A and switching freq 55kHz,
the average power to be dissipated due to energy stored in Lp will be
1/2 LI^2 * 55k= 43W. To use simple RC snubber will require a really
huge resistor, which is impractical and highly inefficient. My thesis
supervisor told me it's an unavoidable problem of half bridge
converter, and told me to use full bridge converter instead.

So all I want to ask is, is it not necessary to use snubber to
discharge leakage inductance energy in full bridge converter, since
the MOSFET already has body diode built in to freewheel the current
(due to both Lp and Lm) ? Or is my supervisor just setting me up?

thankyou for helping in advance and for looking at my question! ^_^

You don't say whether you're using single switch or double switch half
bridge. I recommend the latter for its various advantages, along with
the 2-rectifier approach to steering off-cycle current back to the
primary storage capacitor. You can find lots more info by searching on
a phrase like "2-switch forward converter". Full bridges have their
own set of problems.
Transformer construction determines leakage inductance. Have you done
everything possible to minimize it? It mainly depends on insulation
volume between primary and secondary, but you can greatly reduce it by
using a split primary, also. Choose a wide-window core if possible,
too.
You shouldn't find it necessary to snub the primary switches if you
design the transformer well. However, snubbing the secondary
rectifiers can be very beneficial, since ringing across them as they
switch off is reflected back to the primary. If the optimal snubber
has too high dissipation, check to see if a sub-optimal capacitance
provides good enough reduction of ringing amplitude.
Paul Mathews
 
You don't say whether you're using single switch or double switch half
bridge. I recommend the latter for its various advantages, along with
the 2-rectifier approach to steering off-cycle current back to the
primary storage capacitor. You can find lots more info by searching on
a phrase like "2-switch forward converter". Full bridges have their
own set of problems.
Transformer construction determines leakage inductance. Have you done
everything possible to minimize it? It mainly depends on insulation
volume between primary and secondary, but you can greatly reduce it by
using a split primary, also. Choose a wide-window core if possible,
too.
You shouldn't find it necessary to snub the primary switches if you
design the transformer well. However, snubbing the secondary
rectifiers can be very beneficial, since ringing across them as they
switch off is reflected back to the primary. If the optimal snubber
has too high dissipation, check to see if a sub-optimal capacitance
provides good enough reduction of ringing amplitude.
Paul Mathews

Hi Paul, thanks for answering. Sorry for not having stated clearly. My
original design was to use half bridge two MOSFET push pull converter,
not forward. As for the transformer, I've heard that Lp can be reduced
by doing a P-S-P strructure. However, there is only 6 turns in primary
winding, so it is impractical to split it in to two sections. I think
the small number of turns of Np is the main reason for poor coupling,
which looks unavoidable (Another problem is too low the magnetizing
inductance Lm = 200uH) due to design constraint, such as core
dimension and wire thickness.

I bought a 300W 12V-230VDC pure sine inverter and inside it the
circuit seems to use push pull for DC-DC step up, its transformer uses
the same core as my design (ETD39, about 5cm x 5 cm). Is it possible
that commercial designs use ZVS or ZCS resonant converter to reduce
switching loss, or does this cure the leakage inductance problem?
Sorry I'm pretty novice in power electronics and barely understand the
resonant switching technique.

What are the main problems I'm supposed to see with full bridge
converter? If RC snubbers are to be used, are they connected in
parallel of all 4 switches? And you mentioned the snubber can be
placed on secondary side ( I assume you mean placing it directly
across output winding ). So if designed properly, would it have same
effect as having two snubbers in parallel with the two MOSFETs on
primary side? Once again thanks for you kind help.
 
W

Winfield

Jan 1, 1970
0
w2kw said:
As for the transformer, I've heard that Lp can be reduced
by doing a P-S-P strructure. However, there is only 6 turns
in primary winding, so it is impractical to split it in two

You'd no doubt be better off with a forward than push-pull
transformer configuration, but, sorry the pun, that's not
your primary problem. With respect to the limitation of
having only six turns on the primary, this is easily dealt
with by using multiple paralleled turn sets made from much
smaller wire and interleaving the secondary. Imagine you
were using a wide copper strip for the primary, you could
easily interleave the secondary. But instead of a copper
strip you'll use multiple sets of primary turns, laid down
side by side, and wired up in parallel at the end. I like
to use litz wire in such a scene, because it can be easily
smashed down and smeared into place. Note, this must be
done separately for each primary half. If you meant six
turns CT, then you'd need to redeploy with 4 or 8 turns CT,
so you can interleave each set of 4 or 8 turns with 2 or
4 on each side of the secondary. Or, you can use a S-P-S
winding setup instead. Be careful with your insulation,
a layer of Kapton tape between each interleaved winding.
 
P

Paul Mathews

Jan 1, 1970
0
Hi Paul, thanks for answering. Sorry for not having stated clearly. My
original design was to use half bridge two MOSFET push pull converter,
not forward. As for the transformer, I've heard that Lp can be reduced
by doing a P-S-P strructure. However, there is only 6 turns in primary
winding, so it is impractical to split it in to two sections. I think
the small number of turns of Np is the main reason for poor coupling,
which looks unavoidable (Another problem is too low the magnetizing
inductance Lm = 200uH) due to design constraint, such as core
dimension and wire thickness.

I bought a 300W 12V-230VDC pure sine inverter and inside it the
circuit seems to use push pull for DC-DC step up, its transformer uses
the same core as my design (ETD39, about 5cm x 5 cm). Is it possible
that commercial designs use ZVS or ZCS resonant converter to reduce
switching loss, or does this cure the leakage inductance problem?
Sorry I'm pretty novice in power electronics and barely understand the
resonant switching technique.

What are the main problems I'm supposed to see with full bridge
converter? If RC snubbers are to be used, are they connected in
parallel of all 4 switches? And you mentioned the snubber can be
placed on secondary side ( I assume you mean placing it directly
across output winding ). So if designed properly, would it have same
effect as having two snubbers in parallel with the two MOSFETs on
primary side? Once again thanks for you kind help.- Hide quoted text -

- Show quoted text -

1. I agree completely with Winfield's comments about transformer
construction. Also, ETD39 is an excellent core for minimizing leakage
inductance. An additional advantage of increasing primary turns is
that it increases magnetizing inductance, which has a stabilizing
influence on feedback equivalent to some amount of slope compensation.
2. There should be no need to snub the switches themselves, whether
you're using forward, full-bridge, or push-pull. The rectifiers, which
are much faster switching, are generally what excites the tank
circuits made up of parasitic and stray reactances. Minimize the
strays by proper choice of components (low dynamic capacitance
transistor switches) and layout (minimize leakage inductance, minimize
loop areas and use wide traces in high di/dt loops, use low-profile
components and maximize distances between high dv/dt nodes, etc).
Then, the ringing frequencies will be quite high, so that snubbing
capacitances will be low. When snubbing capacitance is low, power
dissipation in their damping resistors is low. It all plays together.
Books have been written on this subject. One good author: Sanjaya
Maniktala.
3. Resonant and quasi resonant switching techniques can be applied to
any topology, but this adds complexity which may not be worth the
efficiency improvement. I'd start with the simplest control method and
decide if it's good enough. For the same reason, I'd avoid full-
bridge. If you do venture into that territory, be sure to learn the
meaning of the term 'flux walking' and understand that the most
important test conditions for any converter are power-up, power-down,
and brownout.
Paul Mathews
 
T

Tim Williams

Jan 1, 1970
0
The energy stored in the primary is swished back and forth harmlessly for
the most part. As long as primary current (due to inductance) doesn't
exceed saturation current, you're fine. Leakage inductance is the most
important figure.

Consider that the output rectifier and filter, which is choke input,
maintains a roughly constant current due to the filter choke's flyback (at
least one rectifier diode is always conducting). This clamps the secondary
voltage down to roughly zero volts, which likewise keeps the primary
waveform sequestered. The only issue on the primary is leakage inductance,
which will cause a (much smaller) flyback pulse when the transistors switch
off. This can be clamped with diodes in parallel with the transistors (or
by using MOSFETs which have intrinsic reverse diodes, or co-pack BJTs or
IGBTs). The remaining trash (that is, the ringing due to leakage inductance
oscillating with the winding, transistor and diode capacitances) can be
damped with an RC.

Tim
 
H

Heman

Jan 1, 1970
0
Hi Paul, thanks for answering. Sorry for not having stated clearly. My
original design was to use half bridge two MOSFET push pull converter,
not forward. As for the transformer, I've heard that Lp can be reduced
by doing a P-S-P strructure. However, there is only 6 turns in primary
winding, so it is impractical to split it in to two sections. I think
the small number of turns of Np is the main reason for poor coupling,
which looks unavoidable (Another problem is too low the magnetizing
inductance Lm = 200uH) due to design constraint, such as core
dimension and wire thickness.

I bought a 300W 12V-230VDC pure sine inverter and inside it the
circuit seems to use push pull for DC-DC step up, its transformer uses
the same core as my design (ETD39, about 5cm x 5 cm). Is it possible
that commercial designs use ZVS or ZCS resonant converter to reduce
switching loss, or does this cure the leakage inductance problem?
Sorry I'm pretty novice in power electronics and barely understand the
resonant switching technique.

What are the main problems I'm supposed to see with full bridge
converter? If RC snubbers are to be used, are they connected in
parallel of all 4 switches? And you mentioned the snubber can be
placed on secondary side ( I assume you mean placing it directly
across output winding ). So if designed properly, would it have same
effect as having two snubbers in parallel with the two MOSFETs on
primary side? Once again thanks for you kind help.- Hide quoted text -

- Show quoted text -

Hi !
I have some problem in designing the dc-dc converter.
Iam designing a converter with Push pull Boost Topology, 1500W from
27 V to 60V, 90 V outputs. And the 90V output will come in series with
the 60V output circuit, So that the total output will become 150V
during light load (<6A). And full load is 60V 22A. As the load current
increses than 6A, my Second secondary (Booster) will be disconnected.

Here I have two problems, One is to reduction of leakage inductance of
transformer, (I have designed U-U67/27/14 core transformer, switching
frequency of 80kHz, as its winding seeks a bit larger window area
compared to core cross section, window area=8sq.cm).

The second problem is with spike voltages coming in primary winding
(center tapped primary, 2 single winding secondaries folllowed by
bridge rectifier circuits) when the secondary-2 is disconnected from
the circuit and is left open.
 
T

Tesseract via ElectronicsKB.com

Jan 1, 1970
0
Hi everyone, my name is Wilson. I've been trying to design a 300W
12VDC to 380VDC converter lately as part of an inverter project.

In my experience, Wilson, every hard-switched SMPS needs a snubber
*somewhere*. I, too, tried to resist putting snubbers in until I realized
that they don't add to losses, they only shift them out of the semiconductors
(switches or rectifiers, that is). The resonant (ie - soft switched)
topologies are the the only ones that don't need snubbers, and are very much
worth considering. In fact, the benefits of (quasi-) resonant switching are
so great that I barely even consider a hard-switched topology if the power
output is 100W+.

Since you are working with a 12V input, you might want to take apart a couple
of car audio amplifiers as they also have to convert 12V(nom) into higher
voltages and usually at substantial wattages. You can get 'em cheap at any
pawn shop or off of eBay - no sense buying them brand new when you are likely
to destroy it.

That all said, current-mode controlled push-pull is very popular in these
amps, but because you want a much higher voltage output you may want to look
at a current-fed (the output inductor is moved over to the primary in between
the bulk storage cap and the switch(es). I like the push-pull topology for
high power DC-DC converters because of the very simple gate drive (both
switches are "low side") and the penalty of needing to withstand twice the
supply voltage is not a real penalty at all. When current mode control is
used there is no need to worry about flux imbalance.

Leakage inductance is a fact of life; the best way to minimize it is by
interleaving the primary and secondary windings. Also check out "pc board" or
"planar" transformers. These are a bit pricey but they are easy to fabricate
and have very repeatable parasitics. They are most often used in "brick"
converters (as in, "half-brick", "quarter-brick", etc.) running at 200kHz+.

Understand that you *will* be blowing up lots of switches, no matter what, as
you get it all sorted out... ;)
 
M

monkeymiss

Jan 1, 1970
0
Tesseract said:
That all said, current-mode controlled push-pull is very popular in these
amps, but because you want a much higher voltage output you may want to look
at a current-fed (the output inductor is moved over to the primary in between
the bulk storage cap and the switch(es). I like the push-pull topology for
high power DC-DC converters because of the very simple gate drive (both
switches are "low side") and the penalty of needing to withstand twice the
supply voltage is not a real penalty at all. When current mode control is
used there is no need to worry about flux imbalance.

I think you might find that your recipe for quiche is similarly flawed.

Don't ask me, work it out for yourself.

Camilla
 
T

Tesseract via ElectronicsKB.com

Jan 1, 1970
0
monkeymiss said:
That all said, current-mode controlled push-pull is very popular in these
amps, but because you want a much higher voltage output you may want to look
[quoted text clipped - 4 lines]
supply voltage is not a real penalty at all. When current mode control is
used there is no need to worry about flux imbalance.

I think you might find that your recipe for quiche is similarly flawed.

Don't ask me, work it out for yourself.

Camilla

How, exactly, is your comment helpful? It's all well and good to challenge a
comment made by someone, but if you want it to have any sort of impact it's
generally a good idea to provide more substance and less sarcasm.

-Jeff
 
M

monkeymiss via ElectronicsKB.com

Jan 1, 1970
0
Tesseract said:
[quoted text clipped - 7 lines]

How, exactly, is your comment helpful? It's all well and good to challenge a
comment made by someone, but if you want it to have any sort of impact it's
generally a good idea to provide more substance and less sarcasm.

-Jeff

"When current mode control is used there is no need to worry about flux
imbalance."

You were referring to a current fed converter.

You might not have to but if you want your quiche to be better than other
peoples quiche then you might have to pay attention to it.

We don't do sociology, it appears to be an excuse for getting things wrong
without having to think about it.

Have you thought about it? If you take some time you won't look so silly next
time.

Camilla
 
Top