I am winding the transformer for my smps, and need a little help.
I know everything about splitting the primary, but need to know something: I wound 25 turns in first layer, then secondary. The problem is that I need 60 turns for primary, so in the last layer i need 35 turns. Point is that primary isn't halved in two equal parts. Is this OK ?
SMPS is low power ( ~120W ), half-bridge topology, core ETD39.
Thanks.
I know everything about splitting the primary, but need to know something: I wound 25 turns in first layer, then secondary. The problem is that I need 60 turns for primary, so in the last layer i need 35 turns. Point is that primary isn't halved in two equal parts. Is this OK ?
SMPS is low power ( ~120W ), half-bridge topology, core ETD39.
Thanks.
Why the imbalance? Was 25T all that could fit in one layer? But, since you ask, I'm not aware of any adverse operation with uneven windings for the primary.
In order to wind a balance winding, bifiller, trifiller method may employed. bifiller means use two same guage copper wire parallel the two wires and start to wind at the starting point, then after the required turns, then ending. If two different colour of copper used, that would be more easily to define coil 1, and coil 2.
Trifiller use three wires, same way as above.
Trifiller use three wires, same way as above.
Hi
Every time I saw split primary was like so: 50% of turns, all sec, 50% of pri. This can be one continuous wire or wires (bifilar,...), or split into two, like where first layer stops, is soldered to second half outside of core (may be on its pins). So what this means is that length of outer 50% of winding will be longer, since winnding is more away from core...
Every time I saw split primary was like so: 50% of turns, all sec, 50% of pri. This can be one continuous wire or wires (bifilar,...), or split into two, like where first layer stops, is soldered to second half outside of core (may be on its pins). So what this means is that length of outer 50% of winding will be longer, since winnding is more away from core...
Yeah, but I'm wondering why he's going with 25T (inside half of primary) + 35T (outside half of primary). Is this because 25T was all he could fit in one layer at a time?
m-tech, I am curious: Can you maybe kick the switching frequency up a bit to accomodate a 50T primary? This could solve your winding area problem. Just a thought.
Luka- You are right on seeing the "center tap", if you will, of the primary on one of the bobbin's pins- this is common, mainly in AT & ATX PSUs.
This gives me an idea
- Trying a center-tap push-pull drive for the primary, using this pinned-out center-tap. Let's see..... for a typical primary of 38T (19T + 19T), at ~160V across the primary, this would be ~+80V applied to each half of the "ct" primary. Hmmm....... Maybe an idea for someone DIYing a DC-DC SMPS operating off ~65-90VDC. HMMM...... 
m-tech, I am curious: Can you maybe kick the switching frequency up a bit to accomodate a 50T primary? This could solve your winding area problem. Just a thought.
Luka- You are right on seeing the "center tap", if you will, of the primary on one of the bobbin's pins- this is common, mainly in AT & ATX PSUs.
This gives me an idea
- Trying a center-tap push-pull drive for the primary, using this pinned-out center-tap. Let's see..... for a typical primary of 38T (19T + 19T), at ~160V across the primary, this would be ~+80V applied to each half of the "ct" primary. Hmmm....... Maybe an idea for someone DIYing a DC-DC SMPS operating off ~65-90VDC. HMMM...... N-Channel said:
Because I made mistake in calculations, and don't want to rewind it again. And, yes, only 25 turns fits in one layer.
N-Channel said:
I don't have problem with winding area.
How high a turn to turn voltage can enamelled wire withstand?
Single layer gives no real issue. Multiple layers are a completely different ballgame.
What if one enamel layer develops a crack or is scratched, can the adjacent enamel provide sufficient insulation?
Single layer gives no real issue. Multiple layers are a completely different ballgame.
What if one enamel layer develops a crack or is scratched, can the adjacent enamel provide sufficient insulation?
m-tech said:
m-tech- I understand now...... is it possible to just go with 50T (25T + 25T)?
luka said:
Luka- Yes, if the core material he chose will allow it. Usually, shouldn't be a problem.
AndrewT said:
Andrew- Most wire should have the dielectric strength of its enamel coating called out in the specs.
[
I don't want to remove 10 turns, because this will increase primary current and dissipation in switching transistors.
I will go with 35 turns.
I will go with 35 turns.
Just a note on the wiring. To get good packing, you can use many strand of fine enameled wire twisted together (i use a drill for that)...also known as litz wire...and here it is not an advertising fad but actually serves a purpose. It helps battle the skin effect by having a lot more copper surface area. The only pain is stripping the enamel off all the fine wire at the ends.
It will work but the forces are not going to cancel completely and your leakage inductance will not be as low as it could be. The windings should be evenly split for optimal performance. I think it will be fine to see if it works, but you should rewind it if you want to improve it.
Check out this TI pdf for great info on winding arrangements...
Eddy current losses in transformer windings.
Check out this TI pdf for great info on winding arrangements...
Eddy current losses in transformer windings.
Hi
Was for me too, but I have found something called Lavacol, it is paint and enamel remover, which works on wood and metal. Great stuff....
luka said:Hi
Was for me too, but I have found something called Lavacol, it is paint and enamel remover, which works on wood and metal. Great stuff....
Lavacol, huh? Need to get me some of this stuff. Sick and tired of scraping last 3/8" manually with exacto.
Hi
One by one wire was my old way, now 20 at one time is no problem, this Lavacom is made here in Slovenia, I tried to find it on net, had no luck so far.
One by one wire was my old way, now 20 at one time is no problem, this Lavacom is made here in Slovenia, I tried to find it on net, had no luck so far.
carvinguy said:
With 60 turns in primary , leakage inductance is ~12uH. Is this OK for 50KHz operating frequency ?
m-tech said:
That depends on your primary voltage, which I don't think you've mentioned yet. I don't know the line voltage in Belgrade...
since v=L*di/dt
at 120VAC ~ 170 VDC rectified, which is twice transformer input voltage (85VDC)
you can get ~2.9 * 10^6 A/s through the leakage ~ 2.9 A/uS of slope...
at f=50KHz, Ts=20uS, half period is 10uS
you probably wont be pulling more than 2A Ip at 120W, if my head math is right. So it'll take less than a microsecond to ramp to that, which is probably OK. With higher line input voltage, it will be faster.
anybody correct me if i'm wrong
- Status
- Not open for further replies.