Offline flyback SMPS with bigger core generally has less transformer loss

[treez]

We have a 240VAC, 30W (30V OUTPUT) offline flyback SMPS which operates in BCM.
It uses a Ferroxcube PQ3220 core and is litz wound.
Our contractor tells us that it can be done with a Ferroxcube PQ2625 core, with exactly the same LP and LS as was used with the PQ3220 core version......
.....We agree with this.

However, he says that doing it with a PQ2625 core instead, would mean lower OVERALL transformer losses, due to the lower volume of core material in the PQ2625.....
We disagree with that…

..Because even though the PQ3220 has greater volume to suffer the changing flux than the PQ2625, the flux swing will be less in the bigger core. Also, we can get thicker litz wound turns in the PQ3220 and so overall the losses will be lower….who do you agree with?

BTW the inductances LP and LS need to be the same whether its done with PQ2625 or PQ3220 because of other circuital reasons.
(but LP does not equal LS, just in case it seems like I implied that there)
The switching frequencies will be the same, whichever core is used.

 

[iop95]

Need to see Bmax difference, that is area difference of cores.
If have same material, will be same loss curve.
First core (bigger) have effective area A1 and volume V1, second core: A2 and V2.
A1=ka*A2; ka=A1/A2 and V1=kv*V2, kv=V1/V2
Flux density: B1=fi/A1 and B2=fi/A2; ratio B1/B2=A2/A1=1/ka
Ploss ~ (B)^x * V, so ratio of power loss for core1/core2 will be: Kr_loss = (1/ka)^x * kv; now, see x in core curve loss at used frequency and calculate Kr_loss; if Kr_loss is bigger than 1, core 1 have bigger loss and vice-versa.

 

[mtwieg]

Keep in mind the number of turns will likely be different, especially if you optimize the gap. It's certainly not impossible for the smaller core to yield less losses, but that requires more info about the design.

 

[schmitt trigger]

As anyone who has designed a transformer, the transformer losses are the sum of copper and core losses.

Usually improving one aspect has a negative effect on the other. Balancing the core and copper losses usually (but not always) yields the best efficiency.
But again, this is a only generic rule of thumb, and depends a lot on the requirements themselves.

For instance..........One may choose deliberately lower copper losses to improve cross regulation across multiple secondary windings.

 

[FvM]

A hand calculation based on core and coil geometry gives a small reduction (about 8 %) of absolute core losses for the larger core if you keep the winding resistance constant.

 

[c_mitra]

However, he says that doing it with a PQ2625 core instead, would mean lower OVERALL transformer losses, due to the lower volume of core material in the PQ2625.....

Not correct.

Everything else being same, a smaller core will have higher losses.

You are correct to say that smaller flux swing will result in smaller losses (may be very small but small none the less!)

But larger core may need more copper and copper losses may increase (perhaps little and I do not know)

Overall result may be only a small gain in performance: is it worth it? Certainly not cost effective.

 

[Easy peasy]

smaller core gives a smaller mean length of turn, but if the gap needs to be larger to keep the flux swing the same, then the turns will go up ( they will go up any way with a smaller core ) - usually a smaller core gives a larger gap - and now if the wires are not fine litz and the winding structure puts the first layer or two near the gap - you will get local hot spot wire heating which will take the Cu losses up.

In my experience - lowest losses are achieved with a slightly oversized core for flyback - the shorter bobbin length that you get with a smaller core will generally give you more leakage as well - unless you go to a 5 layer ( PSPSP ) structure - and more leakage gives more snubber losses.

- - - Updated - - -

just as a footnote - almost all our flybacks are near 100kHz, and we keep the flux to 100mT Pk, this is to allow for crappy ferrite that our clients purchasing officers like to trend to over time - and to give us low intensity fringing fields near the gap to lower wire heating related to this, it means the core losses are low, and we allow for a longer bobbin to give one layer per winding - which gives the lowest losses and prox effect in the wire - and lowest leakage - we have never had a design overheat in the field.