I have a bunch of T50-2 powdered iron RF toroids I got at fire sale price. Do you think they could be pressed into service as low power flyback transformers for large step-up ratios e.g. 5 volts to 48
From other user's data it looks like they give about 50uH inductance per 100 turns @ 250kHz I'm pretty confident I could wind about 150-200 turns of 28 gauge magnet wire, overlapping, on there in under an hour. I think there'd still be room in the window for a few tens of turns of 22 or 24 gauge solid core primary winding
Can you elaborate on how to multifilar wind a toroid with two windings and say a 10:1 turns ratio? The stuff I'm bringing up via Google search isn't particularly enlightening
Not real sure if RF would be done this way but for a power supply transformer... If 10:1 then the :1 portion is repeated in parallel to get it to run 10 times along side the 1 winding of 10:
Terminate and connect the 10 separate 1 turns up. I would not think this would work good for RF really but throwing it out there.
If you wanted to do it as a non-delay-matched TLT, you'd do 20 strands together, interleaved in such a way that a group of 10 is distributed evenly in the bundle with the other ten. (The easy way to ensure this would be using red-green twist for each pair, then bundling the twists together. The twists won't line up so the bundle will have much more air inside it than a flat bundle.)
Then wire the group of ten in parallel for the primary side, and the other ten in series for the secondary side.
This will give you a really low impedance for the primary (~10 ohms?), which, actually, maybe isn't too terrible (5V 10W is 4A peak). It's awful for the secondary, though (each strand, ~surrounded by primary, will be ~30 ohms, i.e., considerable capacitance once you get out to the end of the winding).
It's not delay matched, because the primary induces voltage on each tenth of the secondary at the same time, but each of those voltages has to propagate through the other nine strands. The step response (assuming correct termination) will be a staircase ramp -- the primary edge is smeared out by each TL segment. At LF, this manifests as LC (of course), which is what you have to contend with in a switcher.
But really, I was thinking you're probably fine with two layers of primary, and a ball of secondary between them. Wire the secondary however you want;
4 strands wired in series is a pretty easy way to do that. (The ball-of-wire method gives a higher impedance and lower bandwidth. The impedance is helpful for the secondary, but not the primary. You'll experience way more leakage inductance this way, but probably still good enough to use.)
Do you recall the late Vladimir Vassilevsky's "Anti-septic converter" that used a third winding to recover flyback energy during the switch off time due to leakage L, that would otherwise be lost to a snubber? I was thinking about experimenting with that topology, seems hard to screw up, the only big downside seems to be that it makes the flyback non-isolated.
If you're willing to go to the effort of adding another winding to deal with leakage, you're much better off spending that effort improving the shitty windings that gave you that leakage in the first place... ;-)
You can't solve the problem by adding a winding, anyway. You can make it a little better maybe, but never completely gone. You are fundamentally limited by the impedance (and thus leakage) between pri and 'reset', which can't be any better than between pri and any other sec, for a given target impedance. (You can make the leakage arbitrarily low, but at the expense of capacitance. You want the characteristic impedance near the nominal switching impedance (Vpk / Ipk) to minimize reactive power.)
Plus, there's a standard solution: the two-switch converter. For flyback, you need Vflyback < Vsupply(min) (which hurts for wide range converters, but is fine otherwise). For forward, you get flux reset for free, and use a PWM hard limit of 50% (usually done with an internal T-flop, e.g. UC3844).
You could still use a 'reset' winding where additional P-S leakage is intentional, e.g. resonant converter. But those are more likely to be half/full bridge, where it again doesn't matter.
It's been a while since I've looked at the schematic I may not be remembering the topology correctly. Maybe it only required a center-tapped secondary? Have to dig it up on my desktop PC somewhere...
IIRC the original context was to improve the performance of an off-the- shelf wall-wart power transformer where the nominal mains primary was used as the secondary and the LV winding as the flyback switch-side primary.
A Laird 35T0501-10H ferrite toroid is only $0.30. That might not be a fire sale, but it's not much. Buy 10 and they're $0.24. Only needs about 4 turns for 50uH. I don't know how this ferrite would work in your circuit.
The problem with strict ferrite toroids is that yeah you don't need that many turns to get a certain inductance but there is no gap, distributed or otherwise, and a flyback transformer that's good for more than milliwatts of output power needs to store energy.
Also I probably wasn't clear enough when I said "low power" like 10-15 watts would be nice. Unfortunately I think that power level with large step-up ratios like 10:1 takes me out of the realm of SMT inductors but I'm not positive.
Ah, well that's just piling bad on top of worse... :)
Split bobbin transformers are as good as common mode chokes... at any kind of frequency, you can get maybe a couple of watts. Main difference being, CMCs don't go much below 10s of kHz, while mains transformers obviously go down to 50Hz, where they're actually usable at, well, rated power.
You could get by with far fewer turns with a higher switching frequency.
The NCP1230 is old and slow. Modern low-voltage parts are faster smaller and cheaper. For example a PowTech PT1301, recommended here by Legg, switches 300mA at 500kHz. I just received 100 pieces, for 11 cents each, plus $7 shipping, from LCSC, an easy-to-use Chinese distributor. To round out my low-cost boost-converter kits, I ordered quantities of 4, 10 and 12mm Sumida inductors, at 13 cents with free shipping, from three AliExpress sellers. Not all values available; they must be selling manufacturer's overstock. Anyway, that's 31 cents per converter for the two critical parts in low-cost boost-converter designs.
"Primary" ten wires wound at same time for 10:1 partly (uniformly) twisted together similar to Litz, wire bunch partly glued (wax?) for integrity to act like one wire. All in parallel; "secondary" single wire
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.