Stacking Cores

Status
Not open for further replies.
Jump to Latest
#43
So we are deciding whether Sevart's software is wrong.
Surely ESP would not continue to recommend it if it was wrong. I think I saw that about 10years ago on the ESP site.

Let's for the moment assume Sevart is correct.

Using that modeling prediction, let's work out the Primary winding to fit through that 50mm diameter hole and leave room for ~ same volume of secondary winding.

VA is 73.8, efficiency is ~90% input VA is ~82VA and max load Primary current is ~82/230 ~357mAac.
0.12sqmm of wire cross sectional area has a current capacity to satisfy that.
the required diameter is >=~0.4mm or use 0.5mm diameter wire.
So we have 1000T of 0.5mm diam enameled copper wire for the single core Primary.

VA is 295.4, efficiency is ~95%, input VA is ~311VA.
Max Primary current is 1.35Aac.
Wire area is ~0.44sqmm.
Wire diameter is >=0.75mm or use 0.8mm diameter.
So we have 500T of 0.8mm diam enameled copper wire for the double core version Primary.

Using the double core version as the worse case for getting the copper in.
The 50mm diam hole can get ~170T with 3 layers giving about 2.6mm deep Primary winding. The net hole diameter (after 4layers of insulation) for the secondary is ~ 44mm diameter.
 
Last edited:
#44
Andrew Eckhardt said:
I forgot to short the meter until the inrush passed.

AndrewT those numbers look like some kind of inverse square is going on. No wonder we're still talking about this. The whole thing is definitely a linear relationship. Throw out that software? I don't know what's going on with that.

Edit: Either I went cross-eyed or you fixed the numbers. But I still don't see how you're getting 4 times the power with twice the core. It would be pretty interesting if that actually worked out.
Click to expand...

yep, throw out that .exe

there's no provision for the non traditional dimensions.
when you plug in the core area, it assumes that what you might call the form factor is the same.
I.e. double the core area, perimeter increases by 41%.
 
#45
AndrewT said:
So we are deciding whether Sevart's software is wrong.
Surely ESP would not continue to recommend it if it was wrong. I think I saw that about 10years ago on the ESP site.

Let's for the moment assume Sevart is correct.

Using that modeling prediction, let's work out the Primary winding to fit through that 50mm diameter hole and leave room for ~ same volume of secondary winding.

VA is 73.8, efficiency is ~90% input VA is ~82VA and max load Primary current is ~82/230 ~357mAac.
0.12sqmm of wire cross sectional area has a current capacity to satisfy that.
the required diameter is >=~0.4mm or use 0.5mm diameter wire.
So we have 1000T of 0.5mm diam enameled copper wire for the single core Primary.

VA is 295.4, efficiency is ~95%, input VA is ~311VA.
Max Primary current is 1.35Aac.
Wire area is ~0.44sqmm.
Wire diameter is >=0.75mm or use 0.8mm diameter.
So we have 500T of 0.8mm diam enameled copper wire for the double core version Primary.
Click to expand...


slow down man, you can't calculate the efficiency until after you choose the wire, remember, 90% of your losses are resistive.

supposing you have 1000 turns and the length of the wire is 4000 units long.
and you have 500 turns and the length of the wire is 3000 units long but twice as thick.
the ratio in resistances here is 4: 1.5 or 2.66. for the same losses, we can increase the current to 1.63 times the 1000 turn transformer. but the surface area is 1.5 times as much, so we can push that up sqrt 1.5 to 1.22 times 1.63 = 1.98 times the 1000 turn transformer.
 
Last edited:
#47
I think Sevart was written for EI core.
But does that make any difference to the modeling?

I realise that you can't accept this square law I have proposed.
I have seen corroboration from other transformer papers/articles, but can't recall the sources.

But I do have lot's of toroid transformers. They do corroborate something much higher than pro rata.
Shame ai can't measure some cores, All my transformers are wrapped in copper wire.

Till now I cannot recall any source that contradicted my understanding of the relationship of core area to VA capability.
 
Last edited:
#48
AndrewT said:
I seem to have a fundamental misunderstanding of transformer science, if I'm to come to accept your statement that doubling the core area gives a doubling of VA capability.

I often stack c-cores for doubling of power.
Checking my German Vacuumschmelze databook confirms the linear relationship between power and weight (magnetic path length has also some influence).
For instance:
core type SU60b: power 116 W / weight 0,92 kg; magnetic path length 22,6 cm;
core type SU75b: power 297 W / weight 1,94 kg; magnetic path length 28,2 cm;
core type SU90b: power 610 W / weight 3,49 kg; magnetic path length 34,0 cm.
Click to expand...
 
#50
AndrewT said:
I think Sevart was written for EI core.
But does that make any difference to the modeling?
Click to expand...


yes, the ratio between iron and copper is all different.
in addition, many E-I calculators assume that the E-I is a lossless cut.
many aren't, and it makes a big difference.

Edit: i thought i summed it up pretty well in the first post i made to this thread.

you can't see it too much when you just stack two cores together, but if you could lump those two cores together into one of circular cross section, i have no doubt you could get 4x power out.
 
Last edited:
#53
uzernaam said:
I intend to wind my own transformer core for a custom application. I understand that with higher power requirements and higher currents through the transformer windings means that you need a physically larger core to handle all the flux without saturation.

Assuming you had 2 or more toroidal transformer cores, could you stack them to make something that looks less like a donut and more like a sort of "tube" and wind around them all at once to make for a larger core if you can't find one the size you need to begin with?
Click to expand...

what i did with mine was that i wrapped the other torroid into the first one, so that i get a bigger donut.......another way to do it...
 
#60
Tony,

These RDH formulas apply for prehistoric quality cores; you cannot use them for nowadays quality grain oriented cores.
I use grain oriented silicon steel c-cores, strip thickness 0.3 mm, for power supply transformers.
SU48b, SU60b, SU75b and SU90b are c-cores I use, single or stacked (the stacked core is still a "single" core with two coils, one on each leg of the c-core).
These b versions are higher than the a versions, which have the same core profile (magnetic path length).
See here the secondary power rating for these cores, and you will see that the difference in core area for the a and b versions is almost 100% proportional to the difference in power:
 
Status
Not open for further replies.
Top Bottom