Transformer Design in AN2794 for Push-Pull Converter

[Nafi-Can]

Hello,

I’m designing a 1 kW converter using a push-pull topology with 24 V input and 400 V output. While doing so, I referred to the transformer design guidelines in your document AN2794. Unfortunately, I believe there are some inconsistencies or errors in a few of the equations. I would appreciate your assistance on the following points:

• The “RMS Input Current” value shown in Table 3 seems to be derived from Equation 9. However, shouldn’t the value be derived from Equations 6 or 8 instead?
• Regarding Equation 32, is it correct to assume that if the calculated Kg value in this equation is greater than the one calculated in Equation 31, then the selected core is suitable for the application? Additionally, I believe the unit in Equation 32 should be cm⁵ instead of cm².
• In Equation 33, does the number of primary turns represent a single winding on the primary side of the push-pull transformer, or the total primary turns? If it refers to a single winding, then the total would be 4 turns (2+2), considering the center-tap.

• Also, in Equation 33, what does T stand for? What is the definition of ΔB? Can you please tell me their values used in calculation of Equation 33.

I would be very grateful to receive answers to these questions. Thank you very much in advance for your support!

Best regards,
Nafi Can

[Peter BENSCH]

Welcome @Nafi-Can, to the community!

The application note AN2794 refers to the discontinued evaluation kit STEVAL-ISV001V1 and can therefore no longer be supported by STMicroelectronics. But perhaps a community member can comment on this?

Regards
/Peter

[Nafi-Can]

Well, I wish someone, like community member, can help but I do not think anyone here will hear my voice.. I am seriously struggling with transformer design in my project. Each solution gives me different results. Therefore, I am not sure which solution, method I should use. When I saw your application note AN2794, I had a little hope, but ..

Thanks.

[Nafi-Can]

Who told you to accept the answer above as a solution? It neither solved my problem nor answered my question? I do not understand why in this community you are workers doing this like in other questions of other people have asked..

[Peter BENSCH]

Your feedback did not reveal any further questions. But no problem, the issue is marked as unresolved again.

[AScha.3]

Hi Nafi,

many "hear" you - but i think, there are not many, with experience in transformer design, so not much response.

I have a little bit - biggest transformer i made was about 60kW , with 20kV output (for plasma gen.), 3-phase 400V input, 25kHz .

You ever made transformers ?  for "middle" frequency ?

You have the needed parts? (ferrite cores, RF-stranded wire, winding machine )

+

is it commercial or private ...?

+

switches: mosfet ? working frequ. ?

[Nafi-Can]

Hi AScha.3,

I have designed transformer several times by using different calculations other than of mentioned document in this question. But the transformers I have designed do not satisfy the requirements when I test. Therefore, I started looking for similar applications that are tested and approved so that I may use same calculations.

The biggest problem is in any article or book, every calculations are giving different results like core, turns etc.

For example, my transformer design for my application, which is targeted 1 kW output at 50 kHz switching frequency by using mosfets as switch, was EE55/28/25 core with 2 turns in each primary side and 46 turns in secondary. But similar core and turns has been used here in this document at 100 kHz ?! In another article titled "Designing an uninterruptible power supply based on the high efficiency push–pull converter" similar core and turns has been used for 20 kHz?!

As a consequence, I am very confused about which calculations I should rely on and how to design the transformer for my circuit.

I do not have ferrite cores, RF-stranded wire, winding machine, I ask transformer producers to produce a transformer with giving data I provide.

My circuit specs:

• Topology: Push-Pull
• Input Voltage (DC): 24 V
• Output Voltage (DC): 400 V
• Maximum Output Current: 2.5 A
• Output Power: 1000 W
• Operating Frequency: 50 kHz
• Purpose: Commercial

Thanks for your interest.

[AScha.3]

Hi,

from your question about 

>In Equation 33, does the number of primary turns represent a single winding on the primary side of the push-pull transformer, or the total primary turns?

+

>Also, in Equation 33, what does T stand for? What is the definition of ΔB? 

Which is very basic knowledge about transformer design and obvious from the  AN2794 ,

i would recommend :

- if you want learn something, buy some parts (cores and wires) and wind some yourself, to see the result and learn.

- if you need a working design "cheap" (= for nothing) try the transformer in the  AN2794 and run it at 100kHz.

- if you need a matching design without the time , to try and learn yourself , use someone to help you: https://www.fiverr.com

[Nafi-Can]

Hi,

The value of T in Equation 33 can be accepted as 10 us, it was a bit absurd to ask this, I accept it.

However, I still have doubts about the value of ΔB. The maximum operating flux density is given as 0.05T in Table 3. If ΔB is accepted as 0.1T due to the push-pull topology and these values ​​are written in the formula, the result is 2.57 x 10^(-4). When I multiplied this value by 10^(4) due to unit conversions, the value of 2.57 was rounded to 2 turns. I am not satisfied with these mathematical operations, but if you do not see a problem with these operations, I will understand that I need to work a little more for this "very basic knowledge" part :).

On the other hand, I still could not understand whether the value of 2 found is for one side of the primary winding or the total for both sides, despite re-examining it.,

Thank you.

[AScha.3 · Accepted solution]

Hi,

well, > I am not satisfied with these mathematical operations ... , me too. :) 

Just for the 2.57 turns: you cannot build 2.57, you have to decide: 2 or 3 turns. Your decision then, whats better...

But from my experience, its also some "feeling" , what to do, and this you (or at least me) get only by doing some working and experimenting ; what happens, if you increase  ΔB , when core heats up, wires get hot...

or wires : changing some heavy wire -> to stranded  wire, then test thinner stranded wire and higher working frequency...

So the mathematical operations are more the "starting point" of a design, to see what size of core needed etc.,

but then comes some optimizing and here the personal experience is helping a lot.

>whether the value of 2 found is for one side

Its a push-pull, so it needs 2 identical windings , each has the number of turns, needed for the transmission factor N you want. So...for 400V / 24V -> N = 16,7 ; -> 17 . With primary 2 T -> 34 T secondary. And 2 primary windings.

And there will be some loss, so depends on what you want: output can be little lower, then ok;

or should be minimum 400V, then need more N, say for 10% possible losses add 10% N : 37  T secondary.

(But some more output voltage under low load then. > There is no free lunch <  )

ok ?