Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

transformer design for smps

48 views
Skip to first unread message

panfilero

unread,
Mar 8, 2012, 2:01:42 PM3/8/12
to
I have a couple questions, one is about identifying smps topology and
the 2nd is about the 3rd winding in an example I found. I need to
design a smps that can take

Vin: 70V to 105V and output 5V at 1A

I've chosen to use the mic9130 http://www.micrel.com/_PDF/mic9130.pdf

On the front page of the datasheet there is a "Typical Application"
schematic, which is not to far off from what I need to do, it's
Vin: 36V to 72V
Vout: 3.3V @ 4A

So, I think I'll copy that design, but now I have to design a
transformer... from the reading I've done about this, looks like one
of the first things I need to know is the topology of my smps. That's
my question, can anyone tell me the topology of this smps example?
Looks like an isolated synchronous buck converter... or maybe a
forward converter? Also I don't see how the third winding is
supplying the stead voltage to Vcc, according to the datasheet Vcc
operates between 9V -18V (pg 3) confusingly on pg 8 it says Vcc is
typically 8.5V....

much thanks!

Tim Wescott

unread,
Mar 8, 2012, 2:26:51 PM3/8/12
to
An isolated synchronous buck converter is a kind of forward converter,
and that's what this is.

That auxiliary winding on the transformer supplies a DC voltage that's
roughly a diode drop and some resistance loss below 1/4 of the input
voltage. Every time the main transistor turns on, the dotted end of the
auxiliary winding jumps up to (turns ratio) * (input voltage), more or
less. The 200 ohm resistor insures no huge current spikes, the diode
rectifies, and the result is an unregulated power supply to the chip.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

panfilero

unread,
Mar 8, 2012, 7:50:01 PM3/8/12
to
On Mar 8, 1:26 pm, Tim Wescott <t...@seemywebsite.please> wrote:
> On Thu, 08 Mar 2012 11:01:42 -0800, panfilero wrote:
> > I have a couple questions, one is about identifying smps topology and
> > the 2nd is about the 3rd winding in an example I found. I need to design
> > a smps that can take
>
> > Vin: 70V to 105V and output 5V at 1A
>
> > I've chosen to use the mic9130http://www.micrel.com/_PDF/mic9130.pdf
>
> > On the front page of the datasheet there is a "Typical Application"
> > schematic, which is not to far off from what I need to do, it's Vin: 36V
> > to 72V
> > Vout: 3.3V @ 4A
>
> > So, I think I'll copy that design, but now I have to design a
> > transformer... from the reading I've done about this, looks like one of
> > the first things I need to know is the topology of my smps.  That's my
> > question, can anyone tell me the topology of this smps example? Looks
> > like an isolated synchronous buck converter... or maybe a forward
> > converter?  Also I don't see how the third winding is supplying the
> > stead voltage to Vcc, according to the datasheet Vcc operates between 9V
> > -18V (pg 3) confusingly on pg 8 it says Vcc is typically 8.5V....
>
> > much thanks!
>
> An isolated synchronous buck converter is a kind of forward converter,
> and that's what this is.
>
> That auxiliary winding on the transformer supplies a DC voltage that's
> roughly a diode drop and some resistance loss below 1/4 of the input
> voltage.  Every time the main transistor turns on, the dotted end of the
> auxiliary winding jumps up to (turns ratio) * (input voltage), more or
> less.  The 200 ohm resistor insures no huge current spikes, the diode
> rectifies, and the result is an unregulated power supply to the chip.
>
> --
> Tim Wescott
> Control system and signal processing consultingwww.wescottdesign.com

Now, a forward converter usually has some kind of reset winding
right? But, would I be right to think that this circuit doesn't need
the reset winding because there are bidirectional switches that allow
for some current to keep flowing into the secondary, to keep that flux
going and not violate anything? I think that's right... so I think
I'll approach my transformer design then assuming I'm designing for a
forward converter... Vo = nDVin

much thanks, appreciate the help

panfilero

unread,
Mar 9, 2012, 11:33:41 AM3/9/12
to
wait, I just looked at this circuit again, and i thought that in order
to keep the flux flowing through the transformer core that current had
to be able to flow into one of the dots at all times, but when i
analyze this circuit for the switching transistor open, I believe this
would force Si4800DY to turn on and Si4884DY to turn off, which would
break the circuit for the bottom end of our second winding... so I
don't see current able to flow into any dot of this transformer during
the off cycle

Tim Wescott

unread,
Mar 9, 2012, 10:57:00 PM3/9/12
to
I'm not sure what you're thinking. The flux in the core is equal to the
sum of the turns * current for all the windings -- so as far as the core
is concerned, the flux is the same whether there's 1A in a 1-turn
primary, or 0.1A in a 10-turn secondary.

In a forward converter current comes out of the secondary (or
secondaries) proportional to the current going into the primary, for the
most part, except for magnetizing current.

panfilero

unread,
Mar 9, 2012, 11:52:26 PM3/9/12
to
I was just pointing out that there needs to be a path for current to
flow into the transformer in order to support the flux in the core at
all times. When current flows into the dot of a winding it's able to
support the flux. This schematic has an error, during the off portion
of the PWM there is no way for current to enter a dot, no way for
current to support the flux.

Nico Coesel

unread,
Mar 10, 2012, 6:34:37 AM3/10/12
to
panfilero <panf...@gmail.com> wrote:

>On Mar 9, 9:57=A0pm, Tim Wescott <t...@seemywebsite.please> wrote:
>> On Fri, 09 Mar 2012 08:33:41 -0800, panfilero wrote:
>> > wait, I just looked at this circuit again, and i thought that in order
>> > to keep the flux flowing through the transformer core that current had
>> > to be able to flow into one of the dots at all times, but when i analyz=
>e
>> > this circuit for the switching transistor open, I believe this would
>> > force Si4800DY to turn on and Si4884DY to turn off, which would break
>> > the circuit for the bottom end of our second winding... so I don't see
>> > current able to flow into any dot of this transformer during the off
>> > cycle
>>
>> I'm not sure what you're thinking. =A0The flux in the core is equal to th=
>e
>> sum of the turns * current for all the windings -- so as far as the core
>> is concerned, the flux is the same whether there's 1A in a 1-turn
>> primary, or 0.1A in a 10-turn secondary.
>>
>> In a forward converter current comes out of the secondary (or
>> secondaries) proportional to the current going into the primary, for the
>> most part, except for magnetizing current.
>>
>> --
>> Tim Wescott
>> Control system and signal processing consultingwww.wescottdesign.com
>
>I was just pointing out that there needs to be a path for current to
>flow into the transformer in order to support the flux in the core at
>all times. When current flows into the dot of a winding it's able to
>support the flux. This schematic has an error, during the off portion
>of the PWM there is no way for current to enter a dot, no way for
>current to support the flux.

Because it is a flyback. Don't model the 'transformer' like a
transformer but like an inductor! The transformer has an air gap in
which the energy builds up. When the primary FET opens, the inductor
wants to keep the current flowing. The only way is through the
secondary coil.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------

Spehro Pefhany

unread,
Mar 10, 2012, 8:39:08 AM3/10/12
to
On Sat, 10 Mar 2012 11:34:37 GMT, the renowned ni...@puntnl.niks (Nico
Coesel) wrote:

>
>Because it is a flyback. Don't model the 'transformer' like a
>transformer but like an inductor! The transformer has an air gap in
>which the energy builds up. When the primary FET opens, the inductor
>wants to keep the current flowing. The only way is through the
>secondary coil.

It's not a flyback- the relative 'polarity' of the windings is wrong
for that.

It's a resonant-reset forward converter- the 'missing' current flows
are through parasitic capacitances.


Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
sp...@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com

Nico Coesel

unread,
Mar 10, 2012, 12:21:33 PM3/10/12
to
Spehro Pefhany <spef...@interlogDOTyou.knowwhat> wrote:

>On Sat, 10 Mar 2012 11:34:37 GMT, the renowned ni...@puntnl.niks (Nico
>Coesel) wrote:
>
>>
>>Because it is a flyback. Don't model the 'transformer' like a
>>transformer but like an inductor! The transformer has an air gap in
>>which the energy builds up. When the primary FET opens, the inductor
>>wants to keep the current flowing. The only way is through the
>>secondary coil.
>
>It's not a flyback- the relative 'polarity' of the windings is wrong
>for that.
>
>It's a resonant-reset forward converter- the 'missing' current flows
>are through parasitic capacitances.

I think you're right. The extra output inductor should have rung a
bell :-) IMHO its not the easiest design to start with. The OP needs
5W output power. A flyback is the easiest topology to achieve his
goal.

Spehro Pefhany

unread,
Mar 10, 2012, 2:24:11 PM3/10/12
to
On Sat, 10 Mar 2012 17:21:33 GMT, the renowned ni...@puntnl.niks (Nico
Coesel) wrote:

>IMHO its not the easiest design to start with. The OP needs
>5W output power. A flyback is the easiest topology to achieve his
>goal.

If he wants to give it a try, here's a relevant article:-

http://powerelectronics.com/mag/510PET22b.pdf

panfilero

unread,
Mar 10, 2012, 9:21:41 PM3/10/12
to
On Mar 10, 1:24 pm, Spehro Pefhany <speffS...@interlogDOTyou.knowwhat>
wrote:
> On Sat, 10 Mar 2012 17:21:33 GMT, the renowned n...@puntnl.niks (Nico
thanks for the article, I believe it's a forward converter as well...
I've let myself slip into a long back and forth in another forum with
someone who keeps arguing it's a flyback. That person doesn't care
about the inductor at the output

http://forum.allaboutcircuits.com/showthread.php?t=67195

but i've analyzed the circuit, it's a buck derived topology Vo = DVin,
it can't boost

have you guys designed a transformer before? there seems to be a lot
to it

Spehro Pefhany

unread,
Mar 10, 2012, 10:33:42 PM3/10/12
to
On Sat, 10 Mar 2012 18:21:41 -0800 (PST), the renowned panfilero
<panf...@gmail.com> wrote:

>
>thanks for the article, I believe it's a forward converter as well...
>I've let myself slip into a long back and forth in another forum with
>someone who keeps arguing it's a flyback. That person doesn't care
>about the inductor at the output

The dots on the windings tell you immediately by inspection that it's
not a flyback.

>http://forum.allaboutcircuits.com/showthread.php?t=67195
>
>but i've analyzed the circuit, it's a buck derived topology Vo = DVin,
>it can't boost
>
>have you guys designed a transformer before? there seems to be a lot
>to it

Yes, a few (including flybacks), but it's been a while. You can pick
up what you need in an intensive weekend once you have the info in
hand, getting the cores and bobbins usually takes longer. I agree with
Nick Coesel that a (lower frequency) flyback is going to be an easier
way to go.

Nico Coesel

unread,
Mar 11, 2012, 9:19:27 AM3/11/12
to
panfilero <panf...@gmail.com> wrote:

>On Mar 10, 1:24=A0pm, Spehro Pefhany <speffS...@interlogDOTyou.knowwhat>
>wrote:
>> On Sat, 10 Mar 2012 17:21:33 GMT, the renowned n...@puntnl.niks (Nico
>>
>> Coesel) wrote:
>> >IMHO its not the easiest design to start with. The OP needs
>> >5W output power. A flyback is the easiest topology to achieve his
>> >goal.
>>
>> If he wants to give it a try, here's a relevant article:-
>>
>> http://powerelectronics.com/mag/510PET22b.pdf
>>
>> Best regards,
>> Spehro Pefhany
>> --
>> "it's the network..." =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0=
>"The Journey is the reward"
>> sp...@interlog.com =A0 =A0 =A0 =A0 =A0 =A0 Info for manufacturers:http://=
>www.trexon.com
>> Embedded software/hardware/analog =A0Info for designers: =A0http://www.sp=
>eff.com
>
>thanks for the article, I believe it's a forward converter as well...
>I've let myself slip into a long back and forth in another forum with
>someone who keeps arguing it's a flyback. That person doesn't care
>about the inductor at the output
>
>http://forum.allaboutcircuits.com/showthread.php?t=3D67195
>
>but i've analyzed the circuit, it's a buck derived topology Vo =3D DVin,
>it can't boost

IMHO you can think of a forward converter like a buck/boost converter
with the transformer doing a current conversion between input and
output. This helps to figure out the output inductor.

>have you guys designed a transformer before? there seems to be a lot
>to it

I did a couple of months ago. This application note has a very clear
recipe on how to determine the core size and number of windings:
http://www.fairchildsemi.com/an/AN/AN-4134.pdf

I ordered a whole bunch of core size and material types and wound
several prototypes to get a feeling what would be the best solution.

One thing the remember: a big part in the transformer design procedure
is driven by the number of windings a core can hold.

You'll also need a current probe. Otherwise you'll flying blind.

I made a couple of current probes myself using this transformer loaded
with a 5 Ohm resistor which gives me 100mV/A.
http://nl.farnell.com/wuerth-elektronik/749251050/transformer-we-cst-500uh-10a-1/dp/1961657

You'll need to use stranded wire to keep the inductance low. Having
multiple probes allows you to see various currents throughout the
circuit.

Its fun if you have the time.

Anyway, I would use a flyback topology with an off-the shelf
transformer for your project.

panfilero

unread,
Mar 11, 2012, 12:05:19 PM3/11/12
to
On Mar 10, 10:33 pm, Spehro Pefhany
<speffS...@interlogDOTyou.knowwhat> wrote:
> On Sat, 10 Mar 2012 18:21:41 -0800 (PST), the renowned panfilero
>
Thanks for the info. I'm not really looking for easy I'm looking for
efficiency, do you think I would gain efficiency with a flyback
topology? I've been reading that they are not as efficient as other
dc/dc converters, but if they were more efficient than the one I'm
pursuing I'd give them another look. If the efficient one happens to
be easier that of course is nice, but it's not as important as
efficiency.

panfilero

unread,
Mar 11, 2012, 12:07:26 PM3/11/12
to
On Mar 11, 8:19 am, n...@puntnl.niks (Nico Coesel) wrote:
> with a 5 Ohm resistor which gives me 100mV/A.http://nl.farnell.com/wuerth-elektronik/749251050/transformer-we-cst-...
>
> You'll need to use stranded wire to keep the inductance low. Having
> multiple probes allows you to see various currents throughout the
> circuit.
>
> Its fun if you have the time.
>
> Anyway, I would use a flyback topology with an off-the shelf
> transformer for your project.
>
> --
> Failure does not prove something is impossible, failure simply
> indicates you are not using the right tools...
> nico@nctdevpuntnl (punt=.)
> --------------------------------------------------------------

Thanks, yes it does seem like a fun project, do you say you would do a
flyback because it's easier or because it's more efficient?

Nico Coesel

unread,
Mar 11, 2012, 4:35:36 PM3/11/12
to
panfilero <panf...@gmail.com> wrote:

>On Mar 11, 8:19=A0am, n...@puntnl.niks (Nico Coesel) wrote:
>> panfilero <panfil...@gmail.com> wrote:
>> >On Mar 10, 1:24=3DA0pm, Spehro Pefhany <speffS...@interlogDOTyou.knowwha=
>t>
>> >wrote:
>> >> On Sat, 10 Mar 2012 17:21:33 GMT, the renowned n...@puntnl.niks (Nico
>>
>> >> Coesel) wrote:
>> >> >IMHO its not the easiest design to start with. The OP needs
>> >> >5W output power. A flyback is the easiest topology to achieve his
>> >> >goal.
>>
>> I made a couple of current probes myself using this transformer loaded
>> with a 5 Ohm resistor which gives me 100mV/A.http://nl.farnell.com/wuerth=
>-elektronik/749251050/transformer-we-cst-...
>>
>> You'll need to use stranded wire to keep the inductance low. Having
>> multiple probes allows you to see various currents throughout the
>> circuit.
>>
>> Its fun if you have the time.
>>
>> Anyway, I would use a flyback topology with an off-the shelf
>> transformer for your project.
>>
>
>Thanks, yes it does seem like a fun project, do you say you would do a
>flyback because it's easier or because it's more efficient?

Its easier because you can use available parts (very little math
involved) and cheaper due to fewer parts (the transformer is both
transformer and storage inductor). The efficiency doesn't matter much.
5% of 5W is 250mW difference. I'd look into the Topswitch devices I
mentioned earlier. Many transformer/inductor manufacturers have off
the shelf transformers for these chips.

Nico Coesel

unread,
Mar 11, 2012, 4:40:49 PM3/11/12
to
panfilero <panf...@gmail.com> wrote:

>On Mar 10, 10:33=A0pm, Spehro Pefhany
><speffS...@interlogDOTyou.knowwhat> wrote:
>> On Sat, 10 Mar 2012 18:21:41 -0800 (PST), the renowned panfilero
>>
>> <panfil...@gmail.com> wrote:
>>
>> >thanks for the article, I believe it's a forward converter as well...
>> >I've let myself slip into a long back and forth in another forum with
>> >someone who keeps arguing it's a flyback. =A0That person doesn't care
>> >about the inductor at the output
>>
>> The dots on the windings tell you immediately by inspection that it's
>> not a flyback.
>>
>> >http://forum.allaboutcircuits.com/showthread.php?t=3D67195
>>
>> >but i've analyzed the circuit, it's a buck derived topology Vo =3D DVin,
>> >it can't boost
>>
>> >have you guys designed a transformer before? =A0there seems to be a lot
>> >to it
>>
>> Yes, a few (including flybacks), but it's been a while. You can pick
>> up what you need in an intensive weekend once you have the info in
>> hand, getting the cores and bobbins usually takes longer. I agree with
>> Nick Coesel that a (lower frequency) flyback is going to be an easier
>> way to go.
>>
>> Best regards,
>> Spehro Pefhany
>> --
>> "it's the network..." =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0=
>"The Journey is the reward"
>> sp...@interlog.com =A0 =A0 =A0 =A0 =A0 =A0 Info for manufacturers:http://=
>www.trexon.com
>> Embedded software/hardware/analog =A0Info for designers: =A0http://www.sp=
>eff.com
>
>Thanks for the info. I'm not really looking for easy I'm looking for
>efficiency, do you think I would gain efficiency with a flyback
>topology? I've been reading that they are not as efficient as other
>dc/dc converters, but if they were more efficient than the one I'm
>pursuing I'd give them another look. If the efficient one happens to
>be easier that of course is nice, but it's not as important as
>efficiency.

AFAIK a resonant converter (zero voltage switching / zero current
switching) is the most efficient converter of all. Over 95% efficiency
should be possible. But I recall there are some issues with these
converters regarding transient behaviour. Maybe someone came up with a
better topology.

panfilero

unread,
Mar 11, 2012, 4:55:26 PM3/11/12
to
Thanks, I've looked through power integrations stuff, and had no luck
finding parts that can handle my input voltage range 70V-105V

Nico Coesel

unread,
Mar 11, 2012, 5:48:28 PM3/11/12
to
panfilero <panf...@gmail.com> wrote:

>On Mar 11, 3:35=A0pm, n...@puntnl.niks (Nico Coesel) wrote:
>> panfilero <panfil...@gmail.com> wrote:
>> >On Mar 11, 8:19=3DA0am, n...@puntnl.niks (Nico Coesel) wrote:
>> >> panfilero <panfil...@gmail.com> wrote:
>> >> >On Mar 10, 1:24=3D3DA0pm, Spehro Pefhany <speffS...@interlogDOTyou.kn=
>owwha=3D
>> >t>
>> >> >wrote:
>> >> >> On Sat, 10 Mar 2012 17:21:33 GMT, the renowned n...@puntnl.niks (Ni=
>co
>>
>> >> >> Coesel) wrote:
>> >> >> >IMHO its not the easiest design to start with. The OP needs
>> >> >> >5W output power. A flyback is the easiest topology to achieve his
>> >> >> >goal.
>>
>> >> I made a couple of current probes myself using this transformer loaded
>> >> with a 5 Ohm resistor which gives me 100mV/A.http://nl.farnell.com/wue=
>rth=3D
>> >-elektronik/749251050/transformer-we-cst-...
>>
>> >> You'll need to use stranded wire to keep the inductance low. Having
>> >> multiple probes allows you to see various currents throughout the
>> >> circuit.
>>
>> >> Its fun if you have the time.
>>
>> >> Anyway, I would use a flyback topology with an off-the shelf
>> >> transformer for your project.
>>
>> >Thanks, yes it does seem like a fun project, do you say you would do a
>> >flyback because it's easier or because it's more efficient?
>>
>> Its easier because you can use available parts (very little math
>> involved) and cheaper due to fewer parts (the transformer is both
>> transformer and storage inductor). The efficiency doesn't matter much.
>> 5% of 5W is 250mW difference. I'd look into the Topswitch devices I
>> mentioned earlier. Many transformer/inductor manufacturers have off
>> the shelf transformers for these chips.
>>
>
>Thanks, I've looked through power integrations stuff, and had no luck
>finding parts that can handle my input voltage range 70V-105V

You need to dig a little deeper. For instance the minimum operating
voltage for this device is 50V:
http://www.powerint.com/sites/default/files/product-docs/tny274-280.pdf

All their examples and components values are based on off-line
converters but that doesn't mean they can't be used for a DC-DC
converter. Perhaps you should try and contact their technical support.

cbarn24050

unread,
Mar 11, 2012, 8:14:45 PM3/11/12
to
Why not just use one of these or something similar? http://uk.rs-online.com/web/p/embedded-power-supplies/6811556/

panfilero

unread,
Mar 11, 2012, 9:07:54 PM3/11/12
to
On Mar 11, 7:14 pm, cbarn24050 <cbarn24...@aol.com> wrote:
> On Thursday, 8 March 2012 19:01:42 UTC, panfilero  wrote:
> > I have a couple questions, one is about identifying smps topology and
> > the 2nd is about the 3rd winding in an example I found. I need to
> > design a smps that can take
>
> > Vin: 70V to 105V and output 5V at 1A
>
> > I've chosen to use the mic9130http://www.micrel.com/_PDF/mic9130.pdf
>
> > On the front page of the datasheet there is a "Typical Application"
> > schematic, which is not to far off from what I need to do, it's
> > Vin: 36V to 72V
> > Vout: 3.3V @ 4A
>
> > So, I think I'll copy that design, but now I have to design a
> > transformer... from the reading I've done about this, looks like one
> > of the first things I need to know is the topology of my smps.  That's
> > my question, can anyone tell me the topology of this smps example?
> > Looks like an isolated synchronous buck converter... or maybe a
> > forward converter?  Also I don't see how the third winding is
> > supplying the stead voltage to Vcc, according to the datasheet Vcc
> > operates between 9V -18V (pg 3) confusingly on pg 8 it says Vcc is
> > typically 8.5V....
>
> > much thanks!
>
> Why not just use one of these or something similar?http://uk.rs-online.com/web/p/embedded-power-supplies/6811556/

you know i see a lot of stuff like this, and they have the 85Vac -
120Vac input range, which mutiply by sqr root of 2 gives me a Vmin
input 120V, which is too high for my Vmin = 60V need.... but I don't
know, maybe it would work? I'd like it to be within spec of the
datasheet though, within normal operating range
0 new messages