How to design flyback transformer

parthiban s

Newbie level 5

how to design flyback transformer, (buck, boost, buck boost).

in continuous mode and discontinuous mode.

treez

AN18 by power integrations shows flyback transformer design.

Easy peasy

How to design a flyback Tx:-

the best flybacks are DCM or discontinuous mode as they make FAR LESS NOISE..! i.e. RFI ( the very best are resonant CrCM - but we're keeping it easy ).

OK so I need 24V at 4A from a 250VDC HVDC bus, for example,

a good choice for the flyback volts is about 80% of Vin, so 200V

the peak energy stored in the Tx (in the air gap mostly) is 0.5 L Ipk^2 this is all (nearly all) released to the output when the mosfet is off so the power is Freq x 0.5LIpk^2.

But, we need to get to the Ipk in 40% of the period, di/dt = V/L, V = 250VDC from above

Lets choose 100kHz (and why not - pretty doable today), Ton is now 4uS, dt in the above.

and power (100W here) x 2 = L Ipk^2 x Freq (100kHz) assuming no losses.

Now the peak energy in the core is also 0.5BHAe.Lg ( from your 1.01 machines paper )

and B=Uo.Ur.H (H = ampere-turns per meter of magnetic path length, often given as NI, but really NI/meter), so now we have Epk = 0.5 Bpk^2 Ae. Lg / ( Uo.Ur)

We assume all the energy is in the gap, Lg, so Utot = Uo = 1.256x10^-6 (no Ur, Ur = 1)

We need to now consider V/L = di/dt and 2 x Power = L Ipk^2 Freq

Solving for L on the LHS equation and putting this into the RHS, yields one and only one solution for near optimum Tx design for this case: 2.P = (Vdt/di) Ipk^2 Freq. (di = Ipk)

We know; P, V, dt, Freq, this solves for Ipk, in this case Ipk = 2P/(V.dt.Freq) = 2.00 amp

& therefore L = 500uH ( check 0.5 L.I^2.Freq = 100W - tick, V/L = di/dt - tick ).

Now we need to pick a core:

WE CAN NOW EQUATE: L Ipk^2 = Bpk^2 Ae Lg / Uo = Power / Freq = Joules (pk)

for a good ferrite (3F3, N87) at 100kHz on a flyback, we conservatively choose 120mT pk

This gives us Ae . Lg = L. Ipk^2. Uo / ( Bpk^2 ) = 174 E-9 = gap x Area...! ( in m^2)

Thus a 1mm gap requires a core area of 174mm^2 or 15mm diameter,

An RM14i core has just this and a good measure of self shielding, and enough room for 100W of copper, pri and sec.

now AL = Ae/lg x Uo ( ignoring the core - which we can ) = 218.5 nH / T^2 ( or N^2 )

So now we know the Pri turns are: 48 for 500uH, and the Sec turns for 200V flyback volts & 25VDC out ( 1 volt for the rectifier ) are 6 turns ...

Voila...! 100W DCM flyback designed in 2 minutes, 2Apk pri current, 500uH, 1mm gap on an RM14i core

Super Moderator
Staff member
* Set duty cycle 50 percent at first. Hardware control circuit can adjust length when running.

* Choose wire gauge for windings based on expected maximum Amperes you wish.

* Turns ratio may need to factor in Ampere ratio as well as voltage ratio.

* Faster switching speed allows transformer to be less bulky and less expensive.

* Core needs saturation rating which is capable of conveying your desired Amperes (flux field).

* Avoid human tendency to extract too much performance from a transformer. Operation becomes uncertain if you try to overdo the step-up ratio, or carry too many Amperes, or choose extremely short or long duty cycle.

treez

here is a basic flyback for you to run in ltspice, the free sim.

chakge the K L1 L1 1 to eg K L1 L2 0.995 so you see it with leakage in the txfmr....you can compare aswell the dcm and ccm.
--- Updated ---

..also attached is the matyhs you will nened to design a flyback
Are you at least aware of this maths....if so, then it is very good news for you

Attachments

• 17 KB Views: 5
Last edited:

treez

What is your spec? (vin, vout, pout)
If you give this then we can see for you.

If you give the spec i will send you an excel sheet shwing the calcs, and a representative sim in free ltspice.

The transformer you can do by emperical means.......pick a core, calc turns and see if your bmax etc is ok...if not, change it to next size up/down etc.

How to pick number of turns in transformer

A question about the actual numbers of turns to use is a very common one.

Turn it round, and tell yourself, you can have as many , or as few turns as you like as long as…

1…..The number of turns for a coil generally fits across the bobbin length (even if you do multiple layers) …you generally get better coupling if you use the full bobbin width.
2…you can get your turns ratio……for example, if you want a turns ratio of 3.5, then you cant pick a single turn primary because you cant have a 3.5 turn secondary.
3….You don’t saturate the core
4….you don’t suffer too high core losses
5…the turns fit on the bobbin space
6…you can terminate a coil to the former pin (eg a 3mm thick wire wont be easily solderable to a small former pin)
7…you dont have too many layers, as if you do , you may get proximity effect problems.
8…you dont get too much winding loss
9….you get leakage inductance reduced to your requirement
10….the turns fit on the bobbin…turns being litz or double coil etc

……………..

So make out a spreadsheet and carry it through….when you see you violate some condition, just change it till it converges to the right solution.

If you don’t know what number of turns to start with…pick any number out of your head and try and work it through…if its not viable your spreadsheet calculations will soon flag this up to you and you will be able to see if you need to reduce or increase it.

parthiban s

Newbie level 5
What is your spec? (vin, vout, pout)
If you give this then we can see for you.

If you give the spec i will send you an excel sheet shwing the calcs, and a representative sim in free ltspice.

The transformer you can do by emperical means.......pick a core, calc turns and see if your bmax etc is ok...if not, change it to next size up/down etc.

How to pick number of turns in transformer

A question about the actual numbers of turns to use is a very common one.

Turn it round, and tell yourself, you can have as many , or as few turns as you like as long as…

1…..The number of turns for a coil generally fits across the bobbin length (even if you do multiple layers) …you generally get better coupling if you use the full bobbin width.
2…you can get your turns ratio……for example, if you want a turns ratio of 3.5, then you cant pick a single turn primary because you cant have a 3.5 turn secondary.
3….You don’t saturate the core
4….you don’t suffer too high core losses
5…the turns fit on the bobbin space
6…you can terminate a coil to the former pin (eg a 3mm thick wire wont be easily solderable to a small former pin)
7…you dont have too many layers, as if you do , you may get proximity effect problems.
8…you dont get too much winding loss
9….you get leakage inductance reduced to your requirement
10….the turns fit on the bobbin…turns being litz or double coil etc

……………..

So make out a spreadsheet and carry it through….when you see you violate some condition, just change it till it converges to the right solution.

If you don’t know what number of turns to start with…pick any number out of your head and try and work it through…if its not viable your spreadsheet calculations will soon flag this up to you and you will be able to see if you need to reduce or increase it.
my spec is :- 12 v input , 8v output 2A current, 5v output 1 A Current

Last edited:

treez

my spec is :- 12 v input , 8v output 2A current, 5v output 1 A Current
Thanks, and i presume you need isolation?
If not then two bucks would probably be easier than a flyback
Specially if their is tight tolerance on the 12v and the 5v

parthiban s

Newbie level 5
yes i need isolation

treez

here is the flyback designed for you...jpeg schem and ltspice sim. LTspice is free
Just convert the .txt to .asc and you can run it.

Do you want me to send you link to the ferroxcube catalog so you can pick a core and start winding some turns?

Do you know the equation to work out how many turns gives inductance L?
And do you know about gapping in flyback txfmr?

Attachments

• 142.1 KB Views: 4
• 7.7 KB Views: 3

parthiban s

Newbie level 5
here is the flyback designed for you...jpeg schem and ltspice sim. LTspice is free
Just convert the .txt to .asc and you can run it.

Do you want me to send you link to the ferroxcube catalog so you can pick a core and start winding some turns?

Do you know the equation to work out how many turns gives inductance L?
And do you know about gapping in flyback txfmr?
Do you know the equation to work out how many turns gives inductance L?

And do you know about gapping in flyback txfmr?

dont know, pls share details

treez

The equation for number of turns to give a certain inductance is in here...
...it involves the "AL" parameter

A flyback transformer must be gapped. This doc explains the gap

Google "why do i need to gap a flyback transformer" to find out why........as long as you know a gap is needed.

This website
has all the maths you need to wind your transformer

Here is the transformer catalog.....from which you can pick one.
--- Updated ---

..pick the PQ2620 core on page 726....choose the AL=250 version of 3C90 ferrite . (its a bit big but less fiddley for you)
Wind 16 turns for primary
wind 10 turns for the 5v secondary
wind an extra 6 turns on top of the 10 turns for the 8v secondary (16 turns total) (as in the schem)

(...or, if you find it easier to have a "non stacked secondary"...wind 16 turns for the 8v output, and a separate 10 turns for the 5v output)

I reduced the switching frequency to 42khz..(otherwise the current slope got too flat)
Attached is the new schem and ltspice
Sandwich the secondaries between two layers of primary....so have the primary in two layers of 8 turns each.....
choose all turns litz wire...eg you can use furukawa tex-elz if you want.
If you cant find litz, then i supoose you just have to make each coil of two (or more) paralleled strands of enamelled copper wire.

Attachments

• 151 KB Views: 2
• 7.7 KB Views: 1
Last edited:

parthiban s

Newbie level 5
give excel data and any one example with formulas

treez

Sir, here is your flyback in excel (basic)
--- Updated ---

Here too is my smps course, there is more on flyback in here

Attachments

• 7.5 KB Views: 2
Last edited:

parthiban s

Newbie level 5

i want to know how to design fly back transformer.

so my focus only on the fly back transformer.

i know input and output ,

for example 12V i/p, 8v 2A , 5v 1A O/P.

So how to select core, bobbin, wire gauge, switching frequency,air gap,

how to calculate primary inductance, 1 Henry inductance how much current it can store.

explain fly back waveform.

treez

What you ask is a common request on this forum, and others....you seek magic....

It sounds like you are seeking the “magic equation”, you shove in your parameters, and voila, its done…..but I don’t believe such is possible….you design a flyback, you then pick a core…..wind it up (in excel) …..notice that its not quite to your liking,….then you re-do it…..and eventually you will nicely converge to the solution of choice……I would say it’s a kind of thing of converging to a solution.

For example, your first core choice may have a Bmax of 0.5T….so you would then need a bigger core……you may then find that all your turns layers don’t fit on….so you pick a different former, or a different core and former………..you may then notice that due to wire diameters, you need 36 turns instead of 37 turns per layer….so you go back and adjust the design as such….......you then pick another core and former, but notice that the former doesnt have enough pins, ...etc etc etc

…so it is this method of “do it then check it, then re-do it..possibly multiple times”…..but its not so bad, because when you do it in excel you can re-use the equation cells….so it’s a good way

But if you are stopping yourself from starting because you want to get it all right first time…….then you will never get started…….just get in there and start doing it……..if you go the wrong way, you will soon see and go back.

parthiban s

Super Moderator
Staff member
explain fly back waveform.
1) In the first half of the cycle, current conducts in the primary side only. Flux field builds.

Secondary winding does not conduct due to the diode blocking current. (Ordinary transformer usage has the secondary generating current while the primary admits current.)

2) The primary is switched off. Then in the second half of the cycle, flux field collapses, causing the secondary to generate current, as permitted by the diode.

parthiban s

Newbie level 5

please may i know above wave form continuous current mode or discontinuous current mode?

here i got fully confused , from where i want to learn,i want able to make fly back transformer. which is beginning,
can i get any learning flow chart.

Akanimo

Hi,

View attachment 162700
please may i know above wave form continuous current mode or discontinuous current mode?

here i got fully confused , from where i want to learn,i want able to make fly back transformer. which is beginning,
can i get any learning flow chart.
This waveform is Vds. It's adviced to use the inductor current waveform to tell whether the converter is operating in CCM or DCM.

One thing you should really take note of is the phases of the flyback transformer windings. If wound improperly, there will be no output and all the primary inductance will act like leakage inductance. There will be too much energy not to be transferred to the secondary. One way to go about winding a flyback transformer is to start each winding at its dotted end and wind all in the same direction.

Just like has been pointed out already, succeed with a functional transformer and converter first and then you can begin optimization afterwards.

In post #1, you talked about buck, boost and buck-boost. With this I think you are also interested in dc-dc converter inductors, so I'd refer you to a book with which you can start from: Fundamentals of Power Electronics by Robert Erickson and Dragan Maksimovic.
--- Updated ---

The book also has a step-by-step procedure for designing a CCM flyback transformer with optimal number of turns and it also explains how to reduce losses due to skin effect by making Rac=Rdc (i.e. by using multiple appropriately-selected, small cross-section, insulated strands of wire for the windings instead of a single wire).

Last edited: