Toroid Power Transformer with Leakage Inductance

[edanderson]

leakage inductance comes from physically separating the field spaces of the primary and secondary windings.  i'm not an ancient elder, but i think the only thing i can think of that would fit your criteria has already been mentioned by mark:

...using each half of the core for respectively Pri and Sec windings...

when you had success with an EI transformer, did it use a two chambered bobbin?  if so, that type of separation is most analogous to a toroid wound in two sectors -- all primary on one side, all secondary on the other (different gauges), looking similar to a common mode inductor.

the leakage inductance would act similar to a load, so the turns would have to be adjusted to still achieve the desired output voltage, etc.

while toroids are generally more efficient than EI core designs due to better utilization (the winding covers the entire magnetic path of the core, and there is no gap), you would essentially be reducing efficiency by increasing the leakage inductance.  not saying you couldn't make a toroid that would work better than an EI design of the same core size, but there's no free lunch -- you'd still have to use a larger core and/or more turns than a stock low leakage toroid transformer.

in the end, PRR is probably right that your best bet would be to use a higher VA and higher secondary voltage EI core transformer.  unless you're willing to make significant qtys, a power toroid wound this way will be prohibitively expensive due to the unusual setup.  and the improvement may end up being marginal anyway.

my best guess, hope that helps...

ed

 

[CJ]

ok the weird dual core alloy toroid is really a regular core nested inside a core that has been cut into 2 semi circles thus creating a gap.
so the gapped core is on the outside of the regular core and is then epoxied over.

this was for for inverter designs that use square waves, and thus saturate easier.

the gapped core takes over when levels get high, the regular core handles the low level stuff, this whole scheme bends the BH loop over,

maybe this type of core would create the leakage you are looking for,

they used to be sold by Magnetics.

toroid cores can not be wound all the way because the machine has a part that must pass thru the core when winding.
EI cores can have the entire window filled with copper, which is what you want when designing transformers, at least until now,  ;D

 

[abbey road d enfer]

Any SPICE gurus who can simulate the separate choke problem?

I can do that in a couple of days from now. The basic effects will show but none of the non-linearities of the iron.

 

[ricardo]

Any SPICE gurus who can simulate the separate choke problem?

I can do that in a couple of days from now. The basic effects will show but none of the non-linearities of the iron.

Thanks for this Abbey.

I just want to know the voltage waveform across the choke as in my description above.  Knowing the peak voltage allows the choke to be designed.  We don't need to know the exact non-linearities.  Just get it a suitable distance from saturation.

Maybe simulate without the choke first to get a feel for the charging spikes & ripple.

 

[ricardo]

Should point out that this ISN'T a choke input supply cos the choke is before the rectifiers.  See my diagram earlier in the thread.

If we ignore the DC current (which makes things really difficult with true choke-input), the waveform across the choke is "similar" if the current is discontinuous.

I'm not sure what happens when the choke current becomes continuous which is how a choke-input PSU is meant to work.  Probably muck things up big time.

Please correct me if I'm wrong.

 

[abbey road d enfer]

Should point out that this ISN'T a choke input supply cos the choke is before the rectifiers.  See my diagram earlier in the thread.

Clear; that's how I've put it. I've drawn a +/- 50V DC supply, with DCR typical of a 200VA xfmr. The choke goes from 0 to 1, 3 and 10mH.
Preliminary conclusion is that, as you expected, the spikes current reduce in amplitude when increasing the inductance, but the voltage across choke increases (which is to be expected), and there is a limit where increasing the inductance drops the DC voltage significantly - with 3mH it drops down to 43V and with 10mH it goes down to 38V.
I will continue tomorrow because I want to see the relative influence of choke inductance vs. capacitance of the smoothing caps.
I think, unless other members are really interested, that I should PM you the results.
Also, I really suggest you download LTSpice from the Linear Technology website and I PM you the .asc file so you could experiment yourself with all the possible variations.

 

[CJ]

ok if you want a leaky toroid, just wind it half way.

you might generate more hum since the field is unbalanced,

which is why you wanted the toroid in the first place,

see how cruel transformers can be?

 

[PRR]

> this ISN'T a choke input supply cos the choke is before the rectifiers

Two parts in series, order does not matter.

It's still choke-input.

As Abbey's numbers are showing:

50V DC ....with 3mH it drops down to 43V and with 10mH it goes down to 38V.

"50V DC" suggests 35V AC winding, 1.414 factor. We expect choke-input DCV to go to 0.9 factor, 32V DC. Apparently Abbey's 10mH has got it 2/3rd of the way to full choke-input, at full current. Go to idle, it will still rise to 50V DC.

> influence of choke inductance vs. capacitance of the smoothing caps.

When "caps are cheap", probably none.

When caps are costly, and you try to minimum-size them, you can get into resonance and huge swings at low current. However at today's Aluminum electro prices, we would not even consider such small rail-caps for a class-B audio amp.

 

[ricardo]

> this ISN'T a choke input supply cos the choke is before the rectifiers Two parts in series, order does not matter.
It's still choke-input.

The big difference is there is no DC.  The choke can have much smaller core.
Also, true choke-input works best when there is continuous choke current.  This can't happen here or rather I dunno wat happens when the current is high enough for this.

> influence of choke inductance vs. capacitance of the smoothing caps.When "caps are cheap", probably none.
When caps are costly, and you try to minimum-size them, you can get into resonance and huge swings at low current. However at today's Aluminum electro prices, we would not even consider such small rail-caps for a class-B audio amp.

I'm playing with LTspice for the 1st time.  Thanks for suggesting this Abbey.  It appears for a given capacitor size & load, there is an optimum choke.  Too big, and the undistorted power of the amp will drop.  About 0.5mH is good with 4mF.  This drops the peak charging current by more than 3x

 

[CJ]

mH or H?    2 pi F L for 60 hzand a 10 millihenry choke means

377 * 0.010 H = 3.77 ohms reactance.


if you are dropping 5 volts across 3.77 ohms, you are dragging

5/3.77 = more than an amp of DC load current?

be sure to put a load on the supply otherwise caps charge to peak value which will not represent a real time circuit,

 

[abbey road d enfer]

mH or H?

Definitely millihenries

  2 pi F L for 60 hzand a 10 millihenry choke means 377 * 0.010 H = 3.77 ohms reactance.
if you are dropping 5 volts across 3.77 ohms, you are dragging 5/3.77 = more than an amp of DC load current?
be sure to put a load on the supply otherwise caps charge to peak value which will not represent a real time circuit,

The sim has been done with 2 amps DC.
Voltage drop is higher than your figure because I is not a sine wave, so each Fourier component of current has its own 2pi.F.L impedance.

 

[mr coffee]

I probably qualify as ancient, but definitely no guru on this stuff - "fer sure."

But if I am not missing the boat here, aren't you looking for what we used to call a "swinging choke"? Something that at least moderates surge current in a capacitor input power supply, especially when current demand is low? And costs less than a real deal choke input inductor?

The choke can be toroid (even if "the ancients" didn't use 'em), and like CJ says, cutting part way through the tape wound core, with or without filling the gap up with powder, modifies the B-H curve when the current surges are worst case. I would think that would keep the nasty EMI in check.

If I'm out to lunch here, forgive me.

 

[ricardo]

.. aren't you looking for what we used to call a "swinging choke"?

A "swinging choke" is entirely appropriate for a conventional choke input PSU with choke after the rectifiers.

What I'm proposing has the choke (either separate or as part of the Leakage Inductance) BEFORE the rectifiers.  As it doesn't carry DC, the core can be MUCH smaller than even a "swinging choke" on a genuine choke input PSU.

Thanks to Abbey, I've had a chance to get to grips with LTspice and play with several scenarios.  So far what I've concluded is

• If you want to do this, it is cheapest/best to have a small separate choke which could be wound on a toroid core.  You could have 2 windings on a single core to do both ends of a CT transformer and +/- supplies as in Class B amps
• My original test case, large Class B amplifiers, isn't too bad and only needs a little help to keep nasty EMI in check
• A small 0.5mH inductor is useful for most cases and doesn't degrade max power before clipping.

 

[CJ]

if the inductor is before the rectifiers, does it not just get lumped into the inductance of the transformer secondary?

the freq response would be better with a small choke, the trans pri is only good to a certain point for filtering out HF stuff,

 

[ricardo]

if the inductor is before the rectifiers, does it not just get lumped into the inductance of the transformer secondary?

It gets lumped into the Leakage Inductance as seen from the secondary.  This was the whole point of the thread.  But the transformer gurus tell us such a transformer would be inefficient and loses the whole point of using a toroid.  Hence the small separate choke.

the freq response would be better with a small choke, the trans pri is only good to a certain point for filtering out HF stuff,

Not sure what you mean by that.  We don't put the small choke on the primary side cos it would then be subject to the huge saturation at switch on problems that toroids are prone to.  It still has these on the secondary side but (proportionally) much less.

The choke is to help with EMI generated by the rectifiers rather than to stop stuff getting in.  For stuff coming in, the IEC mains inlets with built in EMI protection are good if installed properly.

The switch on saturation really needs some sort of slow-start on big Class B amps.  I'd appreciate some ideas that don't involve relays.

 

[abbey road d enfer]

The switch on saturation really needs some sort of slow-start on big Class B amps.  I'd appreciate some ideas that don't involve relays.

http://www.tubeampdoctor.com/de/shop_Widerstaende_VDR_Varistoren_Spezielle_Typen/NTC_Widerstand_ersetzt_Fender_C60_11_zur_Einsch_1689