Transformer design question

[maudio]

Hi there,

the search function didn't help me out on this one, it only revealed to me that there are not a lot responses on threads regarding this subject...
But I'll give it a try anyway

In the process of designing an esl transformer I run into the following: the available winding area is much too long for the number of primaire turns required.

I see three ways around this:

1. Use much thicker wire than is actually needed so it will fill the entire space
2. Use several smaller wires connected in parallel so it will fill the entire space
3. Evenly space the windings, not filling the entire space but leaving a lot of open space

I wonder which method will result in best performance, in other ways which will give the best trade-off between leakage inductance and internal capacitances.

Method 1 means using wire of around 3 mm thick. Not only is this almost impossible to handle but also will this increase winding height considerably, which will increase leakage inductance. Seems not a good option?

Method 2 is basically creating a 'flat' ribbon of wire made up of, say, 6 parallel wires of 0.5 mm. But can I really consider it to be one big flat wire or will each individual turn contribute to the leakage inductance (which increases with # of turns squared). Since they are in parallel, my guess would be no but I haven't been able to find any documentation on this particulair subject. Good thing about this method is it will result in the lowest possible winding height (which lowers leakage).

Method 3 will result in the lowest primairy to secundairy capacitance, but I can also imagine that coupling will be poorer resulting in higher leakage inductance. Prim to sec capacitance isn't a real problem in esl transformers anyway, it's the secundairy capacitances that cause trouble.

Anyone any ideas about this?

Thanks,

Martin

[jcx]

if pri-sec C isn't too important then interleaving the windings reduces leakage, sandwich the sec beteen two "half" primaries

flat ribbon or paralleled wires are more effeicient from an eddy current loss standpoint - even "bunched winding" - paralleled but but scrambled and twisted is less work and pretty good - sort of transitional home brew litz wire

Its been a while since I looked at transfromer design but having lots of window area left over seemed to only occur when designing very high frequency sw ps xfmrs - are you sure your Lm, core Bsat and low frequency corner requirements are met?

[Calvin]

Hi,

flat Ribbon came to my mind on first thought too

jauu
Calvin

[maudio]

are you sure your Lm, core Bsat and low frequency corner requirements are met?

Yes, that's all covered

I have so much window area because I have been aiming for maximimum coil length, to minimize the amount of wire layers needed. Therefore I use double c-cores on which I put two coil formers, each containing all required prim turns and half the sec turns.

This way I need only 8 layers of secundairy wire (0.2mm), or even 4 if I stack four cores, which is hopefully helping me obtaining low leakage and low capacitance. But the downside is that I have way too much room for the primairy.

Twisting the wires will result in higher copper heigth which is bad for leakage, that's why I decided winding several wires in parallel. It will be tricky to wind but I'll need only 40 turns or so so I think I'll survive that.

Using real ribbon wire (like used in pc's) is a possibility and I agree that it will indeed reduce the winding effort. But using enameled copper will result in higher fill factor and lower primary resistance.
Ideal would be to find insulated flat-crosssection copper wire but I wouldn't know where to find that.

[Capaciti]

Hi maudio,

3mm wire for primary ? How many kilowatts you want to input ?

1mm is more than enough. Consider that 40 windings will not show significant resistance even at smaller wire area. in addition most esl-designs need some resistor in serie to the primary, whcih adds much more resistance than the winding itself.

The double core is the best way to make a stepup. if possible put a second layer primary on top of the secondary and put the primaries in parallel. You can half the size of wire for each primary. the second primary on top reduces leakage significantly and drops it to 1/4. Doing 4 primaries in parallel i would expect 0,6mm wire to be mor than sufficient.


capaciti

[maudio]

Hi Capaciti,

Originally posted by Capaciti
3mm wire for primary ? How many kilowatts you want to input ?
1mm is more than enough.

I know, I know
3mm is just the space I have available for each winding. I know it's ridiculous to use 3 mm wire so that's why I look for alternatives. The question comes down to: Is is better to fill the entire available space with copper (using parallel wires) or can I better leave open space

The double core is the best way to make a stepup. if possible put a second layer primary on top of the secondary and put the primaries in parallel. You can half the size of wire for each primary. the second primary on top reduces leakage significantly and drops it to 1/4. Doing 4 primaries in parallel i would expect 0,6mm wire to be mor than sufficient.

Yes, using interleaving is something I will do as well. With only 4 layers of sec wire I think it's probably best to use the method you propose (prim-sec-prim), it seems rather pointless to use more interleaving sections (will only add capacitance).

By the way, I remember haing read somewhere that it is better (lower leakage) to divide the primairy turns over the sections instead of giving each section all turns needed.
In other words, if I need 40 windings and I decide to make 2 sections on each coilformer it's better to make 2 sections of 20 turns and place them in series instead of making 2 sections of 40 and connecting them in parallel. Anyone who can confirm this?

As for as I understood from some articles I read on leakage inductance modelling, leakage will decrease with factor N when using N sections of interleaving windings. So I am surprised when you write that just adding one layer of primairy will reduce it with factor 4. That's good news indeed!

[Bazukaz]

Hi,
I remember once i disassembled one tube amp transformer. Indeed, it was wound in two sections , separated by ~0.5cm or so.It has had a few layers , i think 4, not really sure.

I observed some other interesting techniques : the insulating paper was made slightly wider than the coil , and there were made many cuts from each side , so it could bend around the corners , without cracking.This prevents the wire to slip on corners.

Best regards,
Lukas.