Power Trans Winding Parallel AC or DC?

[sylin]

Hi,

Have just picked up a 1kVA Toroidal power trans that has 4 secondary windings of 2x40VAC for my 2-channel power amp. These secondary windings have identical voltage and amperage ratings, at least it appears so from the spec label. Since I'm building a 2-channel amp I plan on paralleling the windings, two for the left and the other two for the right. My question is how you would parallel them up, would you do it on AC side or DC side? Would you just connect the leads of the windings together (in correct polarity of course) then feed the rectifier bridge, or would you get them through separate rectifier bridges (one per winding) then bring the output of the bridges together at the filter caps?

The amp is only 2x125W into 8 ohm but I 'd like it to have good current output, therefore, an as stiff power supply as available.

Thanks!
Shaoyi

[zigzagflux]

I don't know that it would make a huge difference either way, but my gut tells me you would have a better chance of sharing current through each winding equally by paralleling the windings, and using one rectifier bridge.

Pity we don't have three phase power to our houses; would make for a killer DC supply.

[wine&dine]

my gut tells me you would have a better chance of sharing current through each winding equally by paralleling the windings, and using one rectifier bridge.

Wrong. How do you ensure fair current sharing elsewhere, let's say when paralleling transistors? You add a small resistor in each leg.

If you use two sets of rectifiers, the diodes perform that function; if you parallel the windings, only the winding resistance (<<) does.

[zigzagflux]

Perhaps you could show how a diode drop in each separate winding will ensure current sharing better than winding resistance? I am of the opinion in small transformers such as this, the delta V due to current mismatch is MUCH greater by resistance than it is by a PN junction. One of the features of a diode is a relatively constant voltage drop vs. current. This does very little to ensure current balancing. The winding resistance will have much greater ability to force current sharing than a fixed voltage drop in each diode.

Typical transformers exhibit regulations on the order of 5%-15%. A diode does not produce this natural 'droop' in output voltage as load is increased.

[darkfenriz]

WD is right.

[Electrone]

Originally posted by wine&amp;dine


If you use two sets of rectifiers, the diodes perform that function; if you parallel the windings, only the winding resistance (<<) does.

leakage inductance can help too, depending on the transformer construction.

[Tweeker]

Leakage inductance probably wont help so much in a toroid.

[EWorkshop1708]

It would be fine to parallel the coils at the transformer, I'm sure that it was designed for parallel and series usage.

[sylin]

Thanks to all for your input. I was leaning toward direct paralleling the windings so I did a couple of measurement to the secondary windings about the no-load voltage and DC resistance, just trying to get a rough idea how much difference there are. To my surprise they are in fact very consistent.

To measure the no-load output voltages I paired up two windings in series, head-to-head, so that the meter would see the voltage differential. The reading was a consistent 0.000VAC on my Fluke 77 in any combinations of the winding pair-up.

To measure the resistance I hooked up all 4 windings together in series, and put through a known stable DC current of about 1A, then measure the voltage drop across each winding. They are within a deviation of 4%.

So I guess I'll be fine with paralleling the windings directly, at least with this particular transformer.

[zigzagflux]

For an interesting ponder, consider what happens when you parallel two secondary windings that are a little different either in turns ratio, stray inductance, winding resistance, or combination. Then leave this combined secondary open circuited (no load whatsoever). You will find that there is current flow in the secondary windings!

Why? The secondary terminal voltages are forced to be equal since they are paralleled. If the internal emf-generated voltages are not equal (very difficult to get perfect balance), a current flows to force terminal voltage balance. This current is in the form of var flow, 90 degrees in quadrature to the voltage (with miniscule in-phase current due to winding resistance). So this configuration is such that one winding is supplying vars, with the other winding absorbing vars. Net effect is a self-balancing action that adds a teeny additional current to flow in the windings, while helping maintain balance over and above basic winding resistance corrective behavior. It is a mistake to consider a transformer a perfect turns ratio device with only winding resistance being significant for nonideal corrections.

No different, really, than paralleling two generators; with improper field control, you can actually cause rated current to flow in each generator, even though there is no load connected to them, and essentially little real watt exchange is performed; all vars. There are a number of field control techniques used to minimize this circulating current; fundamentally, you adjust the field to control the internal voltages such that there is balance (net zero circulating current). This method is effective even when paralleling differently rated generators with different characteristics.

Power system design can accomplish a similar correction with tap changers and parallel connected transformers of different size and/or impedance. Basic premise is to reduce circulating current, which guarantees good balance and optimal transformer utilization.

Now, much of this effect is minimized when the paralleled windings are essentially the same and on the same core, such as your transformer, but the principle is the same. Allow this quadrature current to flow, and you have self-correction. So upon further reflection, use those diodes, and this automatic correction is in force only with regards to the winding resistance, so the self correction is less. Keep in mind that the quadrature current is most effective in causing voltage drop across the inductance of the winding, not the winding resistance. Back to the right triangle; vars on the imaginary axis, watts on the real.

originally quoted by zigzagflux
I don't know that it would make a huge difference either way

Perspective is always good. With small transformers like we are dealing with, X/R ratios are surprisingly small, so either method will work just fine, and I can't say for sure which would actually perform better. Would be fun to perform some measurements on various sizes of transformers. My leaning is that the larger the transformer, the more beneficial this var control will be.