Transformer question

[Watson A.Name - \Watt Sun, the Dark Remover\]

It is not uncommon for the DC resistance of a center tapped winding to
differ on each half. For the same number of turns, one half could have
more length of wire because that part of the winding is further from
the core piece and one turn uses more wire. A better way to test is to
put an AC signal on each half separately and see if the secondary gets
the same output.

---
I disagree. Using EI laminations and winding the entire transformer on
the center leg, it makes nore sense, if only for the sake of economy,
to wind the CT primary bifilar and avoid using more wire than
necessary. After all, an extra ten feet per transformer is about two
miles of wire, 1000 transformers downstream. Putting down both halves
of the primary at the same time also saves time and tends to make both
halves of the primary look more alike than winding one half on top of
the other.

I would think that as far as toob transformers are concerned, the
voltages between the windings would be too great to wind them bifilar.

 

[John Fields]

I would think that as far as toob transformers are concerned, the
voltages between the windings would be too great to wind them bifilar.

 

[John Fields]

I think this is only half right. Winding them bifilar would certainly
make them more "balanced" in both turns and impedance, but the total
amount of wire used would have to be the same.

---
You're right.
---

Think of it this way: I still need the same number of turns in each
winding and the same number of total turns. Winding them bifilar
(assuming the same wire gauge) will still occupy the same amount of
the winding window.

So instead of "an extra ten feet" of wire in the outer winding, we end
up using five extra feet in each winding.

---
I'm not sure I understand what you mean unless you're referring to the
penalty for not having two separate windings.

To find out I set up a hypothetical situation using wire with a
diameter of 0.1", a length of turn of 1" for the first layer and a
window width of 2" in order to get 20 turns per layer, and using
either bifilar winding or one winding on top of the other, for 10
layers the total length of wire was 765.3". For the bifilar case,
each winding's length was 382.65", but for one winding on top of the
other, the shorter winding was 225.58" and the longer winding 539.72"!

Taking the total number of turns per winding (100)and winding them
separately would require twice 225.58", or 451.16". Since either the
bifilar or "conventional" winding would require 765.3" for the same
number of turns, that's a savings of 314.14" of wire!
---

 

[Chuck Harris]

Side by side works nicely too. Wind with a bobbin that has a
partition down the middle.


------+ +------
| |
------+ +------
| |
------+ +------


-Chuck

 

[Jim Adney]

I'm not sure I understand what you mean unless you're referring to the
penalty for not having two separate windings.

Now I'm the one who's confused, because any way we do this, there are
still 2 separate windings, unless you want to count the one on top of
the other as a single center-tapped winding.

To find out I set up a hypothetical situation using wire with a
diameter of 0.1", a length of turn of 1" for the first layer and a
window width of 2" in order to get 20 turns per layer, and using
either bifilar winding or one winding on top of the other, for 10
layers the total length of wire was 765.3". For the bifilar case,
each winding's length was 382.65", but for one winding on top of the
other, the shorter winding was 225.58" and the longer winding 539.72"!

Yes, note that 225.58 + 539.72 = 2 x 382.65. This was what I
predicted.

Taking the total number of turns per winding (100)and winding them
separately would require twice 225.58", or 451.16". Since either the
bifilar or "conventional" winding would require 765.3" for the same
number of turns, that's a savings of 314.14" of wire!

Yes, but the only way you can get this is to suddenly have more room
to do windings, so it's not really a fair comparison.

You'd have to assume that the voltage difference between adjacent ends
of the bifilar winding would be nearly as great as the B+ voltage. I'm
not sure how much I can expect to get out of magnet wire varnish, but
I'm pretty sure that I wouldn't want to count on this if I were making
a bunch of them to go out into the commercial world running on a 400V
B+ line.

Keep in mind also that this voltage difference would be AC, reversing
itself with the output. This is a lot harder on the insulation and is
a sure way to find flaws in the insulation. The ACVrms between wires
at either end would be about .7 x B+. The voltage difference in the
middle would actually be the same. Interesting how that works out....

Sure I could probably make one and show that it worked just fine, but
what would my return rate be when they went out into the field and
people covered the ventilation slots with magazines and dirty clothes,
while playing the stereo at full volume for hours on end. ;-)


-

 

[John Fields]

Now I'm the one who's confused, because any way we do this, there are
still 2 separate windings, unless you want to count the one on top of
the other as a single center-tapped winding.

---
Separate windings as on two separate bobbins, each with the same width
and starting diameter and occupying a separate portion of the core.
---

Yes, note that 225.58 + 539.72 = 2 x 382.65. This was what I
predicted.

---
Yes, I know...
---

Yes, but the only way you can get this is to suddenly have more room
to do windings, so it's not really a fair comparison.

---
It's a perfectly fair comparison and reflects the savings to be
realized with utilizing more of the core. Consider a transformer wound
on EI laminations with one winding on the center leg and two other
windings wound around the outer legs. If the outer windings were
identical, then winding them separately would result in saving about
41% of the wire which would be needed to wind them either stacked or
bifilar.
---

 

[Jim Adney]

Separate windings as on two separate bobbins, each with the same width
and starting diameter and occupying a separate portion of the core.

Sure, you can do that, but you're changing the rules in the middle of
the game: You're giving yourself a larger winding window to work with.
(otherwise you would have slightly less than half as much length to
work with, so each winding would build up faster and taller. The extra
bobbin takes up some of your previous volume, so you come out behind.

Otherwise, separate bobbins is certainly a desirable thing to do, but
you have to allow extra space for it up front.

I could also suggest that I would just start with a longer core so
that I would need fewer layers, and that would save even more wire....

Folks who make transformers for a living know all that, and they
routinely hipot to 5kV winding-to-winding and winding-to-core, and
make stuff that can run at ungodly high temperatures, so I don't think
it's a big deal if it's planned for.

That's a MUCH easier problem than 300-500 AC volts between adjacent
turns.

-

 

[John Fields]

Sure, you can do that, but you're changing the rules in the middle of
the game: You're giving yourself a larger winding window to work with.
(otherwise you would have slightly less than half as much length to
work with, so each winding would build up faster and taller. The extra
bobbin takes up some of your previous volume, so you come out behind.

Otherwise, separate bobbins is certainly a desirable thing to do, but
you have to allow extra space for it up front.

I could also suggest that I would just start with a longer core so
that I would need fewer layers, and that would save even more wire....

---
At the expense of extra core or, for the same volume, a longer
magnetic path (not good). The game is to minimize the total cost of
the transformer while maximizing its performance, and blithely doing
this-and-that isn't likely to converge on a winning "design"
---

That's a MUCH easier problem than 300-500 AC volts between adjacent
turns.

---
Don't be ridiculous. The only problematic part about it is deciding
what insulation to use on the magnet wire, and since there are many
materials which can easily stand off higher voltages, that turns out
to be a non-problem as well.

Consider Kapton. At about 5kV/mil, a 0.0001" layer of it on adjacent
conductors would stand off a 1000V difference between them.

If you want to spout definitive information instead of mere opinion,
first go slogging through the spec's for the different magnet wire
insulation systems, then spout what you find.

In any case, I'm getting bored with this whole thing so if you choose
to reply you'll be talking to yourself.