Maker Pro
Maker Pro

Is a 250W Flyback practical?

V

Vladimir Vassilevsky

Jan 1, 1970
0
Hammy said:
I came across this application note from power integrations, that
shows a non-interleaved flyback rated for 250W!

It's in this app note it's 49 pages and the 250W flyback is on page
23.

http://www.datasheetcatalog.org/datasheet/powerint/TOP242-249.pdf

Here is just the schematic.

http://i43.tinypic.com/aoxero.png

The size and the cost of the transformer would probably be impractical.
Also, they claim the efficiency of 85% which is not great, and I also
have some doubts about it.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com
 
M

MooseFET

Jan 1, 1970
0
I came across this application note from power integrations, that
shows a non-interleaved flyback rated for 250W!

It's in this app note it's 49 pages and the 250W flyback is on page
23.

http://www.datasheetcatalog.org/datasheet/powerint/TOP242-249.pdf

Here is just the schematic.

http://i43.tinypic.com/aoxero.png

For modest values of "practical" it is likely ok. The inductor would
need to be a darn good design. The input current is in shortish
pulses so a power factor correction stage would have to go in front of
this stage. The result wouldn't likely end up very effecient.
 
E

Eeyore

Jan 1, 1970
0
Hammy said:
I came across this application note from power integrations, that
shows a non-interleaved flyback rated for 250W!

250W is pushing the edge of practicality for flyback.

It's in this app note it's 49 pages and the 250W flyback is on page
23.

http://www.datasheetcatalog.org/datasheet/powerint/TOP242-249.pdf

Here is just the schematic.

http://i43.tinypic.com/aoxero.png

Power Integrations app notes are generally good but Vladimir has pointed
out a few imperfections.

Graham
 
N

Nico Coesel

Jan 1, 1970
0
Hammy said:
I came across this application note from power integrations, that
shows a non-interleaved flyback rated for 250W!

It's in this app note it's 49 pages and the 250W flyback is on page
23.

http://www.datasheetcatalog.org/datasheet/powerint/TOP242-249.pdf

I have some experience with the 120W version and I must say that 250W
is pushing the limits very very hard. 120W is already on the limits of
a flyback power supply.

You'll need a transformer with an extraordinary low leakage inductance
and you'll lose a lot of power in the snubber circuit. In other words:
I recommend to look into another topology if you need 250W
continuously.
 
E

Eeyore

Jan 1, 1970
0
Joerg said:
No mention of power factor in there? In many areas of the world this is
mandatory for power levels starting around 70W.

Exactly 75W FWIW.

So you couldn't sell it there. Hmm ...

Depends on product category.

Anyhow, while 250W in flyback is certainly possible one has to still
pass EMC and that's no small feat.

Yup.

Graham
 
H

Hammy

Jan 1, 1970
0
The size and the cost of the transformer would probably be impractical.
Also, they claim the efficiency of 85% which is not great, and I also
have some doubts about it.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com


CWS has the coupled inductors ETD44H ; 650uH primary 35uH max
leakage.

XT datasheet ETD44H

http://www.coilws.com/Switch Mode Power/SMPS Tx dwg/TS249-Y03.pdf

I ran the variables through here

http://schmidt-walter.eit.h-da.de/smps_e/spw_smps_e.html

3.2A peak primary and 20A secondary peak.

I do agree a flyback isn't the best choice which is why I had to look
twice at the datasheet. I've only built Flybacks to 70W I've read of
interleaving to extend the power range but never tried it. The reason
I prefer flybacks is because it's simple to get low volume ready made
cores. The 250W mark is pushing it though definitely.
 
K

krw

Jan 1, 1970
0
While I got your attention here, IIRC you are a Kicad user. How did you
get around the fact that it messes up power connections during
annotation? For example when using a LM324 with power pins only on U?A.
Depending on position on the sheet it then places U4C there and orphans
the power wires as well as pins. U4A lands somewhere sans power connection.

We're using OrCad (stop laughing!) and put the power pins on all of
the homogenous parts. I think it's ugly and pretty inflexible so I'm
trying to get our layout guy (also the librarian) to separate the
power pins onto a heterogeneous symbol so I can put the power pins on
a separate page. That'll solve this problem as well as get all the
power supply decoupling off the "logic" pages, while showing it next
to the appropriate device. The reference ID (numbers starting with
Axx00 on each page) will then also show that it's decoupling.
 
E

Eeyore

Jan 1, 1970
0
Joerg said:
While I got your attention here, IIRC you are a Kicad user. How did you
get around the fact that it messes up power connections during
annotation? For example when using a LM324 with power pins only on U?A.
Depending on position on the sheet it then places U4C there and orphans
the power wires as well as pins. U4A lands somewhere sans power connection.

You're mistaken about Kicad. I think I browsed it but my main CAD usage over
the years has been Orcad ( just for schematics - I usually have a seperate
layout guy ). With its demise I'm thinking Logic and Pads. I did some layout
re-work on PADS and found it very intuitive to use. The only other package I've
had moderate use of it a British oddity called Vutrax, plus I've used the
freebie from ExpressPCB which wasn't bad actually. For what I wanted its almost
manual approach was suitable. The only thing that annoyed me was that you
couldn't highlight an entire trace.

The issue of power pins is an oddball. Orcad simply put them on U ( or IC as I
term them ) A and the other packages don't show the power connections.

I don't know what CAD system they use but Soundcraft have a way by which no
individual part has power pins but generates a rectangle for all the parts
therin with the power pins. This is neat since you can keep power and
decoupling to one corner of the sheet.

Random allocation of the power pins must be a real nuisance.

Graham
 
K

krw

Jan 1, 1970
0
Orcad is ok but I crashed the Windows versions way too often, while
Eagle never crashed on me (so far). I also do not like the new princing
methods where they foist the service contract on you and you have to use
those dreaded dongles.

Ours must have a site license; no dongles. In fact, when I did have a
single-user copy, the dongle didn't do anything. I copied the package
to my laptop without problems. Layout needed it but as it turned out,
I never used layout.
DOS-Orcad handled this annotation problem superbly and it always blows
my mind that designers of newer CAD systems don't try to copy software
behavior that worked and that people likes.

Yes, it works but is ugly.
That's what the gEDA guys also suggested because gschem is even worse in
that respect. But it makes a more analog schematic less readable. You've
got to see right then and there in the subcircuit where power comes
from. On my designs the supply power to opamps and logic sections is
often actively switched.

Our schematics are 90% analog and most sheets are entirely analog. I
put "logic" in quotes to differentiate it from ANDs and ORs logic;
meaning the flow (or logic) of the schematic. The power pins and a
billion decoupling caps block that flow. Unlike yours, we rarely
switch power. It's not that we wouldn't like to, but the only way to
tell if power is needed is that the user turned the switch on.
 
N

Nico Coesel

Jan 1, 1970
0
Joerg said:
No mention of power factor in there? In many areas of the world this is
mandatory for power levels starting around 70W. So you couldn't sell it
there. Hmm ...

There is a simple way to get passive PFC with a centre tapped primary
winding and an extra inductor.
 
E

Eeyore

Jan 1, 1970
0
krw said:
We're using OrCad (stop laughing!) and put the power pins on all of
the homogenous parts. I think it's ugly and pretty inflexible so I'm
trying to get our layout guy (also the librarian) to separate the
power pins onto a heterogeneous symbol so I can put the power pins on
a separate page. That'll solve this problem as well as get all the
power supply decoupling off the "logic" pages, while showing it next
to the appropriate device. The reference ID (numbers starting with
Axx00 on each page) will then also show that it's decoupling.

That's exactly the kind of way I'd prefer. Even as it is, I often use the
reference designator CD for decoupling cap as opposed to coupling or other caps.

Graham
 
E

Eeyore

Jan 1, 1970
0
Joerg said:
DOS-Orcad handled this annotation problem superbly and it always blows
my mind that designers of newer CAD systems don't try to copy software
behavior that worked and that people likes.

I used DOS-Orcad for ages. Nice and simple but it does beep at you a lot when
you're working fast !

Graham
 
K

krw

Jan 1, 1970
0
krw said:
krw wrote:
[...]
... I think it's ugly and pretty inflexible so I'm
trying to get our layout guy (also the librarian) to separate the
power pins onto a heterogeneous symbol so I can put the power pins on
a separate page. That'll solve this problem as well as get all the
power supply decoupling off the "logic" pages, while showing it next
to the appropriate device. The reference ID (numbers starting with
Axx00 on each page) will then also show that it's decoupling.

That's what the gEDA guys also suggested because gschem is even worse in
that respect. But it makes a more analog schematic less readable. You've
got to see right then and there in the subcircuit where power comes
from. On my designs the supply power to opamps and logic sections is
often actively switched.

Our schematics are 90% analog and most sheets are entirely analog. I
put "logic" in quotes to differentiate it from ANDs and ORs logic;
meaning the flow (or logic) of the schematic. The power pins and a
billion decoupling caps block that flow. Unlike yours, we rarely
switch power. It's not that we wouldn't like to, but the only way to
tell if power is needed is that the user turned the switch on.


On mine there are often phase shifters consisting of logic chips that
need a super clean supply. Or laser drivers. As a minimum they get an
active capacitor circuit so none of the conducted noise passes through.
Switching happens a lot on battery operated radio circuits. The feature
is called "wake-on-radio" or WOR. The receiver itself keeps on listening
and the rest of the unit is dormant. It is only turned on when a
properly coded RF signal has been detected.

WOR wouldn't give us any advantage. Our radio (mobile belt packs)
wouldn't be turned on unless there was a local base paired with it to
talk to. The base doesn't use enough power (12W) to bother shutting
down for any time there are no bases to talk to. VOX isn't an option
either.
 
K

krw

Jan 1, 1970
0
That's exactly the kind of way I'd prefer. Even as it is, I often use the
reference designator CD for decoupling cap as opposed to coupling or other caps.

That would make the BOM ugly. OrCad already screws with any parts
that have differences in the properties.

Different subject: YOY doesn't Allegro do back-annotation into OrCad?
 
N

Nico Coesel

Jan 1, 1970
0
Joerg said:
That is also an expensive and space-consuming way :)

I doubt that unless I'm missing something. For active PFC you
basically need an extra switcher, inductor and 450V input capacitors
(more expensive). The 400V input voltage eats away headroom for the
flyback switching mosfet and reduces efficiency because the snubber
dissipates more.
 
L

legg

Jan 1, 1970
0
I came across this application note from power integrations, that
shows a non-interleaved flyback rated for 250W!

It's in this app note it's 49 pages and the 250W flyback is on page
23.

http://www.datasheetcatalog.org/datasheet/powerint/TOP242-249.pdf

Here is just the schematic.

http://i43.tinypic.com/aoxero.png

Although there are many reasons why you should be shy of any Power
Integrations app circuit for a 250W flyback, the fact that the output
voltage is 48V, in this instance, is one point in it's favor. The
nearer-unity turns ratio of the higher voltage output assists in
making lower leakage inductance isolation transformers.

You should consult the most recent app note, in any event; yours is
dated 2001 and there has been at least one revision (July'03)since
it's publication for the TOP242-250 series.

RL
 
M

MooseFET

Jan 1, 1970
0
On Apr 18, 12:41 pm, Joerg <[email protected]>
wrote:
[....]
Orcad is ok but I crashed the Windows versions way too often, while
Eagle never crashed on me (so far).

After the Windows version lost me a weeks worth of work by completely
munging the schematics in the project, I went back to using the DOS
version.

If found the Windows version a lot harder to use also. Doing things
like specifying the package for the part was so bad that I had already
given up on the idea of putting that information is non-displayed
attributes in the schematic. Was just going to make a list for the
layout guy.

[....]
DOS-Orcad handled this annotation problem superbly and it always blows
my mind that designers of newer CAD systems don't try to copy software
behavior that worked and that people likes.

What really gets me is that even folks like Orcad didn't stick with
the ideas that worked. I can understand the NIH making another
company not want to copy the ideas.

[....]
That's what the gEDA guys also suggested because gschem is even worse in
that respect. But it makes a more analog schematic less readable. You've
got to see right then and there in the subcircuit where power comes
from. On my designs the supply power to opamps and logic sections is
often actively switched.

I tried gschem and didn't like it because of things like not showing
the power pins. I want the power pins to show even in the logic
parts. You need to be able to see what parts are powered from what
supply. I have never done a design where there was just Vcc and GND
and thats it.
 
M

MooseFET

Jan 1, 1970
0
I used DOS-Orcad for ages. Nice and simple but it does beep at you a lot when
you're working fast !

I wrote a little TSR to make the beep very short.
 
N

Nico Coesel

Jan 1, 1970
0
Joerg said:
All I can tell you from my past designs is that an inductor serious
enough for decent PFC barges into the BOM budget at $3 or more. The
inductor for active PFC is miniscule and usually the grand total comes
to much less.

I used a 100uH inductor rated for the peak primary current. The
rectified AC is fed into centre tap of the primary turn. The input
capacitors are connected to one end of the primary, the switcher to
the other.
 
Top