Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Register Log in

Flyback SMPS core loss

C

cupoftea

Advanced Member level 6
Advanced Member level 6
Joined
Jun 13, 2021
Messages
3,057
Helped
62
Reputation
124
Reaction score
139
Trophy points
63
Activity points
15,948
Hi,
We are doing a CCM Flyback at 45W. Vin 48V, Vout 15v, fsw 150kHz.
Transformer is EFD25 Ferroxcube with 3F3 ferrite. Gap is 0.32mm for AL=250.
Turns gives 1:1 (13:13).

Primary current is 3.16A to 5.16A, so that gives a B(pkpk) of 0.105T. The ferroxcube graphs only require us to consider half of that, so we are taking 0.053T and finding our power density is just 200Kw/m^3. Since the core volume is just 3300mm^2, our core loss is just 0.6W. This is great news, but sounds rather low?

EFD25

3F3 ferrite
 
Sort by date Sort by votes
The ferroxcube charts also show that 3C94 or even 3C90 would be fine for this application........wow!...who needs resonant converters to reduce size.....45W hard switching flyback with an EFD25 !!!
 
Upvote 0 Downvote
LOL - 600mW min as a starting point for something the size of a large bug

add in wire losses - and please tell me what the temp tops out at in your enclosed space at full power .....
 
Upvote 0 Downvote
Thanks, pri is two strands of 7/0.1mm TEX-ELZ (Litz).
So is sec.
But sec is two parallel interleaving coils.
i^2R loss is low, so too will be skin effect losses as we use Litz.
i^2R is 0.5W for pri, and 0.7W for sec
....when you think that 600mW could be handled by three 1206 resistors.

The enclosure has big vent holes.
Are you suggesting their kw/m^3 graphs are incorrect?
 
Last edited:
Upvote 0 Downvote
Thanks, you must believe the ferroxcube provided graphs are wrong....admittedly its triangle flux and not sine. And then theres the fringing field..but we are using Litz.
--- Updated ---

Your prediction of 150degC can only be due to the fact that the core flux is triangular….even if I take the full B(pkpk) of 0.105T at 150kHz, then for 3F3, the kW/m^3 is only 150Kw/m^3. With the EFD25 core volume of 3300mm^3, this gives just 0.495W of core loss…easily dissipated by this shallow, skeletal core. The only way your prediction of 150degC could come to fruition is…

1…the Ferroxcube graphs are incorrect.

2….The difference for triangular flux is so much that the ferroxcube graphs (based on sines) are irrelevant
--- Updated ---

The following shows how there is not much difference between sine and triangular excitation, for purposes of calculating core loss.

Ridley Engineering | - [091] Core Loss Modeling - Part III Sinewave Versus Triangle Wave Losses

ridleyengineering.com ridleyengineering.com
--- Updated ---

The following shows how there is not much difference between sine and triangular excitation, for purposes of calculating core loss.

Ridley Engineering | - [091] Core Loss Modeling - Part III Sinewave Versus Triangle Wave Losses

ridleyengineering.com ridleyengineering.com
 
Last edited:
Upvote 0 Downvote
" Your prediction of 150degC can only be due to the fact that the core flux is triangular "

assumptive - and not even close - let us know when it is running
 
Upvote 0 Downvote
cupoftea said:
Thanks, you must believe the ferroxcube provided graphs are wrong....admittedly its triangle flux and not sine. And then theres the fringing field..but we are using Litz.
--- Updated ---

Your prediction of 150degC can only be due to the fact that the core flux is triangular….even if I take the full B(pkpk) of 0.105T at 150kHz, then for 3F3, the kW/m^3 is only 150Kw/m^3. With the EFD25 core volume of 3300mm^3, this gives just 0.495W of core loss…easily dissipated by this shallow, skeletal core. The only way your prediction of 150degC could come to fruition is…

1…the Ferroxcube graphs are incorrect.

2….The difference for triangular flux is so much that the ferroxcube graphs (based on sines) are irrelevant
--- Updated ---

The following shows how there is not much difference between sine and triangular excitation, for purposes of calculating core loss.

Ridley Engineering | - [091] Core Loss Modeling - Part III Sinewave Versus Triangle Wave Losses

ridleyengineering.com ridleyengineering.com
Click to expand...
The flyback case isn't like a symmetric triangle wave; excitation has net DC component, and that flyback event
has a steep slope, probably higher harmonic frequency content. The copper losses are also growing fast
when the drive current nears saturation.

Temperature measurement of a prototype would be prudent, or one should allow a generous margin of core size to avoid nearing saturation.
 
Upvote 1 Downvote
please let us all know how long it takes for the xfmr to reach 150 deg C @ full power

Thanks, i'd rather choose a bigger core, but there is not space or cost approval for that....The core loss calc is genuine, and gives it green light
 
Upvote 0 Downvote
Recently I ran an E25/13/7 offline Flyback at 62W output from 373VAC input for many hours. Also ran it for considerable periods of hours at 42W from 90VAC input.
It was 84:19 Pri = 1.06mH. Frequency was 60kHz.
Bpk and Bpkpk were significantly greater than what it would be in the aforementioned 150kHz EFD25 example.
Also, even though it was run on the prototype PCB in “open air lab ambient”, the transformer was tightly encased in a potted plastic container.
Sec was 0.7mm TEX-E, pri was ECW.

So getting 48W, from an EFD25, which has greater core area than E25, doesn’t sound quite so out-of-this-world.

Though admittedly the frequency is greater, but coilcraft graphs show that this is OK.

 
Last edited:
Upvote 0 Downvote

LaTeX Commands Quick-Menu:

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top