Maker Pro
Maker Pro

AC->DC Flyback

S

Sylvain Munaut

Jan 1, 1970
0
Hello,


I've searched the web, but haven't exactly found yet what I'm looking for so maybe someone here has a direct link.


I'm searching on how to design a "simple" flyback smps that would work on 110AC and 220AC (preferably without change), and has
12V, 5V and 3.3V output. The total power would be < 100W, something like
12V - 4A
5V - 4A
3.3V - 5A


Also, I'm probably gonna need a "custom" transformer, any pointer on that ?

Thanks,

Sylvain
 
P

Pooh Bear

Jan 1, 1970
0
Sylvain said:
Hello,

I've searched the web, but haven't exactly found yet what I'm looking for so maybe someone here has a direct link.

I'm searching on how to design a "simple" flyback smps that would work on 110AC and 220AC (preferably without change), and has
12V, 5V and 3.3V output. The total power would be < 100W, something like
12V - 4A
5V - 4A
3.3V - 5A

Also, I'm probably gonna need a "custom" transformer, any pointer on that ?

I *strongly* recommend that you visit Power Integrations website www.powerint.com and check out their very informative application
notes.

They include several fully engineered examples including a multi-rail supply not dissimilar to what you require.

There is ( unusually for a semi maker ) also a great deal of info on designing the magnetics too, including a simple to use
application that'll do most of the calculating for you and an app note discussion of safety related construction issues.

Graham
 
T

Tam/WB2TT

Jan 1, 1970
0
A small PC power supply has got to be cheaper than what you could build one
for.

Tam
 
P

Pooh Bear

Jan 1, 1970
0
Tam/WB2TT said:
A small PC power supply has got to be cheaper than what you could build one
for.

I see what you're saying - but not all PC supplies are 'universal voltage' for
starters. I'm unsure whether the individual rail current demands would be
compatible either.

In fact most PC supplies use forward converter methodology rather than flyback
too.

I reckon that the OP wants to learn about flyback. It's interesting - I learnt
a lot about SMPS design myself this way.

Graham


p.s. 'top-posting' is somewhat frowned upon in Usenet - you're meant to add
your comments *after* the original post, so as to preserve the 'timeline of
thought' as opposed to *hey I know this and it's best imho and I'm gonna tell
you first*
 
R

Rich Grise

Jan 1, 1970
0
p.s. 'top-posting' is somewhat frowned upon in Usenet - you're meant to
add your comments *after* the original post, so as to preserve the
'timeline of thought' as opposed to *hey I know this and it's best imho
and I'm gonna tell you first*

Whenever someone recommend bottom-posting, I wish they'd not neglect
"Snip Mercilessly."

Only quote the relevant parts of the post, that you're responding to.
The rest is clutter, and very off-putting. I've been known to skip
over posts where all I see is a full page of quoted crap that I've
seen already, that's irrelevant to the particular answer.

Cheers!
Rich
 
F

Frithiof Andreas Jensen

Jan 1, 1970
0
I'm searching on how to design a "simple" flyback smps that would work on
110AC and 220AC

There is really nothing simple about that - what do you need it for (If it
is for a product, in almost any case it will be cheaper to buy the thing as
a unit ready-made & tested from f.ex. Toko, If it is for learning I would
choose something a little less deadly to start out with ;-)?

Philips used to have some good design notes.
Also, I'm probably gonna need a "custom" transformer, any pointer on that
?

The voltages you quote are common - you job will be *a lot* easier - and
safer - if you at least choose a commodity transformer and design around
that!!
 
S

Sylvain Munaut

Jan 1, 1970
0
Hi,
A small PC power supply has got to be cheaper than what you could build one
for.

PC supply has some drawbacks :
- More power than needed ( often >= 200W )
- More voltages so would lead to more complexity that needed
- Probably need fans
- I would need to remove it from it's casing to include it ...


Sylvain
 
S

Sylvain Munaut

Jan 1, 1970
0
Pooh said:
I *strongly* recommend that you visit Power Integrations website www.powerint.com and check out their very informative application
notes.

They include several fully engineered examples including a multi-rail supply not dissimilar to what you require.

There is ( unusually for a semi maker ) also a great deal of info on designing the magnetics too, including a simple to use
application that'll do most of the calculating for you and an app note discussion of safety related construction issues.

Thanks for theses pointers, I'll check them out in details.


Sylvain
 
S

Sylvain Munaut

Jan 1, 1970
0
Pooh said:
I reckon that the OP wants to learn about flyback. It's interesting - I learnt
a lot about SMPS design myself this way.

My purpose is dual ;)

First, learn about design of <100W SMPS design. And from what I've read in many place, the
flyback topology would be the more suited for that.

Second, obtain a design that would be manufacturable at low-cost, to integrate in a project.
So It also need to pass safety and emi tests.
I can't hire someone to do the job since I must first present a working/manufacturable prototype
and it's cost to the investors ...


Sylvain
 
S

Sylvain Munaut

Jan 1, 1970
0
Frithiof said:
110AC and 220AC

There is really nothing simple about that - what do you need it for (If it
is for a product, in almost any case it will be cheaper to buy the thing as
a unit ready-made & tested from f.ex. Toko, If it is for learning I would
choose something a little less deadly to start out with ;-)?

It didn't sound that complicated ;) I just looked at the Viper100 from ST
and their design application and that looks like it could fit my needs with
little external components.

But at that level of power, it's working in continuous mode, I'm not sure to
know all the implications of that.

Prebuilt modules could do it for the final product (even if building one myself
still interests me for learning). I've looked at TOKO and didn't find modules ...


?

The voltages you quote are common - you job will be *a lot* easier - and
safer - if you at least choose a commodity transformer and design around
that!!

Yes certainly. But often in the app notes / design software given by pwm
controller manufacturer, they don't give reference for a part but just
number of turn / core / ...

Where could I get standard transformers ? I may not be searching for the good
word ...


Sylvain
 
F

Frithiof Andreas Jensen

Jan 1, 1970
0
It didn't sound that complicated ;) I

It *gets* complicated with alle the UL ... safety issues and ensuring that
the design *really* works reliably under real line conditions - seem to
remember from about 90 V to 320 ;-)
But at that level of power, it's working in continuous mode, I'm not sure to
know all the implications of that.

A bit higher losses in the inductor and switch; regulation will be slower
than discontionous *and* there may be instability if the operation causes
mode switching between contionuous & dicontinuous operation.

Normally one fixes the design to operate in one mode only.
Prebuilt modules could do it for the final product (even if building one myself
still interests me for learning). I've looked at TOKO and didn't find
modules ...

Well they did little half-open switchers a while back - but almost all
products tend to get subsumed into other businesses at random. The best
person to ask is a large supplier - usually, they keep track of "what is
where".
Where could I get standard transformers ? I may not be searching for the good
word ...

Look under "Magnetic Components"

Philips Components used to - maybe that went to FerroxCube?
Siemens - may be Matsushito(?) now.
Toko, Panasonic, Vishay maybe.
 
J

Joerg

Jan 1, 1970
0
Hi Sylvain,
Yes certainly. But often in the app notes / design software given by pwm
controller manufacturer, they don't give reference for a part but just
number of turn / core / ...

Where could I get standard transformers ? I may not be searching for
the good
word ...

You could try companies such as Pulse Engineering, or scope out Digikey
and others to see if they offer a suitable transformer from one of the
Asian manufacturers. But when you have data about which core to use and
the number of windings you might as well make your own for a lot less
money. Usually for less than the cost of a glass of Jupiler beer in the pub.

However, you need to understand isolation requirements for the wires,
protection of the core from dielectric breakdown etc. Cores can be
purchased for little money from ham radio shops. Often they will carry
the Amidon brand.

Regards, Joerg
 
P

Pooh Bear

Jan 1, 1970
0
Sylvain said:
My purpose is dual ;)

First, learn about design of <100W SMPS design. And from what I've read in many place, the
flyback topology would be the more suited for that.

Typically - yes.

Second, obtain a design that would be manufacturable at low-cost, to integrate in a project.
So It also need to pass safety and emi tests.

You'll find that Power Integrations offer good advice on safety and EMI issues too !

I can't hire someone to do the job since I must first present a working/manufacturable prototype
and it's cost to the investors ...

You might want to purchase one of their eval boards to do that quickly.


Graham
 
P

Pooh Bear

Jan 1, 1970
0
Sylvain said:
It didn't sound that complicated ;) I just looked at the Viper100 from ST
and their design application and that looks like it could fit my needs with
little external components.

But at that level of power, it's working in continuous mode, I'm not sure to
know all the implications of that.

Higher working flux in the core > larger losses.

Prebuilt modules could do it for the final product (even if building one myself
still interests me for learning). I've looked at TOKO and didn't find modules ...

OEM modules are great for small quantity manufacture but about 3 times the cost of
the parts if you're building it yourself.


I'm not aware of any 'commodity transformer'. You have to design to suit.

Yes certainly. But often in the app notes / design software given by pwm
controller manufacturer, they don't give reference for a part but just
number of turn / core / ...

Most semiconductor manufacturers don't give much info about the magnetics. I had to
work it out for myself first time I did a flyback design and it would have been much
easier with the CAD tools that exist now.

Where could I get standard transformers ? I may not be searching for the good
word ...

Are you looking for 'preferred cores' ? I suggest that you consider the ETD and EFD
cores for a European design for starters.

I've also found E cores quite popular too. E.g. used E30/15/7 in my latest design (
35W ). Multi-output supplies need more pins on the coil former - this may influence
your choice of core ( it did for me ).

The magnetic design application I recommended from power Integrations includes a
'stock list' of popular cores listed by region ( USA, Europe, Asia ).

Siemens also do a useful magnetic design tool too. Errrr. that's EPCOS now.


Graham
 
G

Genome

Jan 1, 1970
0
Pooh Bear said:
Higher working flux in the core > larger losses.


Graham

Higher working flux in the core does not strictly translate to greater
losses. It's the peak to peak flux excursion and the area enclosed in that
loop that you are interested in.

Operating discontinuous with Bpk of 200mT gives you about the same losses as
operating continuous with Bmin 100mT and Bpk 300mT, the flux excursion is
the same.

Continuous operation introduces the right half plane zero into the loop gain
equations.

DNA
 
J

Joerg

Jan 1, 1970
0
Hi Genome, Hi Sylvain,
Continuous operation introduces the right half plane zero into the loop gain
equations.
And that RHP issue is nothing to be sneezed at, especially if you need a
really quiet output voltage. Discontinuous mode is generally considered
to be easier if this is one of the first switchers you are designing.
The price to be paid for a larger core and switch isn't so steep. Also,
when you need a very wide load range it will go into discontinuous mode
below a certain load anyways.

Regards, Joerg
 
K

Ken Smith

Jan 1, 1970
0
Joerg said:
And that RHP issue is nothing to be sneezed at, especially if you need a
really quiet output voltage. Discontinuous mode is generally considered
to be easier if this is one of the first switchers you are designing.
The price to be paid for a larger core and switch isn't so steep.

Often the price for the "extra" switch for discontinuous mode is zero. In
continuous mode the switch has to turn on with a current flowing in the
inductor. This adds to the switching losses.
 
Top