Isolated Flyback design

REJO -


I think our DPA Switch product family is going to be best suited for your needs. I've put together a very quick/rough design using our PI Expert software that has your rough specifications. If you haven't already, I would highly recommend downloading this software.

-The Traveler

Thank you for your reference schematic my friend , can you tell me on what basis you selected frequency

Most of our product lines have a fixed switching frequency.  Some have the option of two switching frequencies.  Some of our products have internal features that dynamically adjust the switching frequency.  This makes it hard to answer your question.

With the DPA switch starting design I gave you, I just let PI Expert use the default max switching frequency.  This will reduce your transformer size and makes filtering the ripple components on your output easier.

If you can be more specific about your question, I might be able to provide a better response.

-The Traveler

1)In general How can we decide Swithching frequency , any designing procedure ? or just  trial and error method , i heard that switching frequncey can be affect  the core loss and transformer loss

2) Yes Frequency can minimize the core size but how can we determine which type of core can be used, Most of the documents is mentioning frequncy with repect to wattage , in this case also first i chose ef 25/13/7 for 70 khz ( before your refernce document)  but i dont know as the frequncy increase how to determine the core ?   

Looking back at your post, I thought I'd follow up with a bit more information for some of your questions.

There are *many* different factors that come into play when choosing a switching frequency:

1. Conducted and radiated emissions requirements for regulatory agency approval
2. Power delivery requirements
3. Space constraints
4. Switching device parasitics
5. Practical core/bobbin geometry

In general, if you're subject to conducted EMI regulatory requirements, you will likely want to choose a switching frequency below 150kHz.  The reason for this is that it keeps the fundamental component (1st order harmonic) of your switching frequency  out of the EMI scan.  This is the reason you will see 132kHz as a fairly common switching frequency in offline power supplies.

Regarding space constraints and power requirements, the amount of power you can ideally deliver in a flyback power supply setup is constrained by L x I² x Fsw.  Increasing your switching frequency results in being able to push more power through a particular flyback transformer.  This gives you the ability to use smaller cores and bobbins for the same power level.  However, you're correct in that high switching frequencies tend to create larger core losses.  

The other way to increase the amount of power delivered through one particular core/bobbin, is to increase the inductance of the transformer.  You have several different ways to change the inductance.  You can adjust the gap size of the transformer...but you have to keep the gap size practical.  If you use too small of a gap, maintaining the production tolerances of the gap within reasonable limits becomes difficult.  You can also adjust the turns ratio and number of primary turns.  This method comes with another set of trade offs itself.

With copper/winding losses, the problem can become much more complex.  Skin effects, proximity effects, magnetic field effects from the core cap, etc all start coming into the picture.  You might want to increase the number of primary turns in your design, but if you start using too many layers you'll run into problems with proximity effects.  If you're not sizing your wire correctly for the frequency involved, skin effects can significantly impact your power loss budget.  With the core gap, stray magnetic fields (fringing effects) will also tend to increase the AC resistance of the wire directly around the gap.

So there really are a lot of different factors that come into play.  Most of the decisions you make will require trade-offs and compromise.  

In our PI Expert software, transformer sizing is done somewhat automatically based on the following: known switching frequency (given by the product family), known peak current limit (given by the particular IC), core volume/area (determines how hard the core can be pushed before saturating), practical number of winding turns (how many turns can you physically fit onto the bobbin without creating problems with losses, gap sizes, etc), etc.

If you're experimenting with different transformer designs, I would *highly* recommend getting a low-frequency impedance analyzer.  These are an incredible tool for evaluating for the design/configuration of a particular transformer.  

Here is a great article on the subject:

 http://www.ridleyengineering.com/index.php/transformer-measurements.html

-The Traveler

Thank you travellor for your valuble information

from your information  i came to know why 132khz  ,Can i have one more doubt 

From the application note i noticed pi choose EI28 to EER28 for 66KHz and EF25 for 132KHz ( for the wattage 30W to 50W-an47) . What is the relation connecting ferrite size and the frequency ? i know as the frequency increases size with decrease . but how can i determine this core for some particaular frequency etc ?any calculation 

There isn't a specific equation or calculation I can give you for choosing the right core size.  It's actually a fairly complex subject involving trade-offs and compromises between different cores, bobbins, winding configurations, etc.

In general, you'll need a larger core with lower frequencies to process the same amount of power.

-With smaller core cross sectional areas, you will typically need more windings to get the same inductance (this is a simplified assumption).  This gets into the area where compromises start getting made...more turns means more layers (increased proximity/copper losses) or using smaller gauge wire (increased copper losses). 

-Smaller cores tend to have higher core losses. Higher core losses result in higher core temperatures and possibly saturation under the right conditions.

-You also have to start taking into account what are practical gap sizes for the core/turns/current of your inductor.  Could you use a 0.01mm gap? Sure.  Would you ever want to release it for production?  No.

Our PI Expert design software takes these issues into account and does a pretty good job of making recommendations on the core to use.  If there is a specific core you'd like to use, try adding it to the PI Expert part database.  When the software goes through the optimization procedure, you should have the option of including your specific core into the optimization routine.  Or you can have PI Expert use one of the cores that it comes pre-loaded with and then see how closely it matches the specs of the core you'd like to use.

-The Traveler