Rewinding a transformer

I'm not a transformer expert -- but that sounds like a workable solution. You'll have more inter-winding capacitance, which may or may not be a problem.

Make sure that a high winding resistance wasn't an intentional part of the design -- many wall-wart type transformers are intentionally inefficient to limit short-circuit current without having to pay for extra components.

You can test for core loss issues by running the thing with the secondary open -- generally with transformers the core loss is a function of the primary voltage, with copper losses being a function of the currents. If it only gets hot when power is flowing then chances are the windings are the culprit.

If you're careful to get the polarities correct and to match the turns counts, you can add windings in parallel with those existing. Only connect them in parallel after verifying that voltage and polarity are almost identical.

You can determine saturation margin by increasing primary voltage with secondary unloaded. As you approach saturation, primary current will rise rapidly. This is best observed with a current probe, but you can observe the voltage across a small series resistor. For a toroid, abrupt current spikes will occur near voltage zero crossings.

You can also drive either winding with a signal generator, lowering drive frequency until you observe saturation characteristics as described above. Your operating margin in frequency is approx. the same as the operating margin in voltage. For example, if you can go down to

The inductance will not significantly change due to larger wire, you can use the same number of turns as you have now with heavier wire and it will reduce the copper loss and no change to the core loss.

Does the xformer get hot when the primary is powered but the sec is not loaded? If so, then you have core losses.

The optimum design fills up the core opening with wire and usually has about equal copper and core looses. If you have mostly copper losses and no core losses then you can use heaverier wire and FEWER turns. This will reduce the copper loss and increase the core loss.

Fewer turns = more core loss and less copper loss More turns = less core loss more copper loss heavier wire = less copper loss no change to core loss

Also core loss is constant vs load and is a function of input voltage and frequency Copper loss is a squared function of load current

Again the optimum design fills up the space and has about equal core and copper loss at the desired load.

(This discussion assumes 60 Hz i.e. we are not talking about high frequency effects)

Mark

Thanks all, I will work on it now. Test showed that there was very little loss in the core.

lower

Optimum design for a toroid is a somewhat limited concept.

Losses in a toroid are almost always determined by copper losses IME.

Graham

Correct.

Using more copper works brilliantly. I've used the technique for ages.

Graham

"Mark"@y43g2000cwc.googlegroups.com...

Core losses are much lower than the copper loss.

The core may suddenly saturate, then you *are* in trouble.

This is an old hat theory based on large transformers using high loss core designs.

....... Phil

OK if I have less turns, why does the core saturate easier? Is it that more field from the same voltage applied? If the turn ratio is keep the same then where does one determine on how many turns to start with on the primary? At this point the orginal transformer has 30 turns of #16 on th pri and 150 turns of #22 on the sec. The loss is a little high not bad but runs hot at

"JSF"

Or, just do a simple test and all is revealed.

Now is not too late.

Do you have a Variac available and an AC amp meter?

....... Phil

Yes - in one. The magnetising force increases.

Use the same as you have now.

Just increase the copper gauge. Messing about with the working flux is likely to lead you into trouble.

Graham

The above should have been 150T of #18 not 15T. Its a 1:5 winding ratio.

Yes the applied ac voltage can be vaired from 0 to 7 v .

That should reduce your losses quite a bit.

Graham

The peak core flux is proportional to the voltage per turn times the time it is applied for each half cycle (actually the time integral of the per turn voltage over the half cycle). Adding turns lowers the volts per turn.

Use the number it has now, to keep the core flux swing the same as it is now. Add a proportional bit to both windings to lower the flux swing a little.

The original wire gauges have cross sections of 2583 and 642 circular mils, with a ratio of about 4 to 1. So the primary is producing more heat per volume of copper than the secondary is (the current density is higher). I would probably go up two gauges on the primary and 1 on the secondary, to 14 and 21, with cross sectional areas of 4107 and

On Tue, 2 May 2006 10:30:06 -0500, "JSF" Gave us:

Insulated wires or magnet wire? Also what gauge are they now? Are they on opposite sides of the toroid or stacked?

If insulated, you could switch to SPC (silver plated copper), and increase gauge.

Also, could you have some DC riding in there? That will cause saturation or heating effects.

If magnet wire, you can double the wire up (bifurcated) to effectively double the area of the conductors while keeping the same turns count and ratio. You can do this on both the primary and secondary.

How big is the toroidal core? Also, it could be cracked if it is ferrite construction. That will cause problems like you describe.

On Tue, 2 May 2006 21:44:13 -0500, "JSF" Gave us:

The term is "volts per turn". Another to look up is "ampere/turn"

On Wed, 03 May 2006 00:51:17 -0400, John Popelish Gave us:

Great link!