Blog 1: Transforming my understanding of flyback transformers, one hack at a time

RE: Blog 1: Transforming my understanding of flyback transformers, one hack at a time

BigG — Mon, 18 Dec 2023 20:21:17 GMT

[mention:b0bc65b9ecdc4307bd967592f00e340a:e9ed411860ed4f2ba0265705b8793d05] [mention:4f57fc9d538949ad9eb336ddb9469bb8:e9ed411860ed4f2ba0265705b8793d05] you've both provided me with some great explanation that has helped me better understand the mechanics behind switch mode power supply design. While searching online, I came across this reference design by Texas Instruments. It's interesting because it does away with a snubber. It also introduces an inductor in series (L2) on the primary side with a resistor in parallel (R3) - it's shown in their schematic (PDF). This was something I had assumed most would circuit designs would have included as it limits (inrush) current etc, but that has not been the case. So I'm pondering as to the why...

www.ti.com/.../TIDA-00628

RE: Blog 1: Transforming my understanding of flyback transformers, one hack at a time

jc2048 — Mon, 18 Dec 2023 12:01:09 GMT

With a capacitor, the energy storage is potential energy. It persists until there is a path so that the charge (that you did work to separate) can recombine (where you get the energy back). With an pure inductor, it's more like kinetic energy. It builds whilst the current is increasing. Is held whilst the current is constant, and then released when you remove the voltage that's driving the current (the coil generates an emf to keep the current flowing as the field winds down). The energy absolutely has to come out of the magnetic field - if you don't provide a path at all, the inductor will create one, generating a voltage so high that it will breakdown a semiconductor or the air.

So the energy storage is small scale and temporary. In a SMPS, it all adds up because you take the small dollops of energy that come out on each switching cycle and store them on a capacitor to power a circuit for a short period of time. Do it 120,000 times a second and there's a reasonable effect to it.

If we ignore the secondary just for a moment, the primary circuit you've got there is a simple boost circuit - at this point people will tell me it's 'flyback' or 'buck-boost', but it's easier to understand as a boost. The secondary winding is another coil that shares the same magnetic flux as the primary [the core sees to that, as it channels most of the flux in a magnetic circuit within the confines of the material]. That means the secondary will respond to the changes in flux, just as the primary does, except there is a scaling factor if there is a different number of turns. When the switch turns off, the energy comes out of either the primary [as with a simple single-inductor boost circuit] or the secondary [which is probably what we want], depending on where a suitable path lies.

There are two consequences of all that you need to consider.

Firstly, if you have fixed switching without any control and don't have a load on the secondary, it will keep pumping up the output capacitor voltage as more and more dollops of charge end up there. It will limit at some point. Either something will break down, or the primary path will suddenly become available limiting further movement. But if you want an accurate voltage at the ouput you either need active regulation of the whole converter or have a device on the output that can absorb the excess energy whilst limiting the voltage (i.e. zener - hugely wasteful because the zener will have to take the energy if your circuit doesn't want it and everything is running at full tilt all the time).

There are various schemes for control and regulation. PWM is one. A fixed on-time and variable off-time [so not constant frequency] is another. Monitoring the primary current [leading to variable on-time] is another.

Secondly, if you have that arrangement with the 1n4001 ordinary diode and snubber on the primary, I think you'll find that it takes all the energy and you won't achieve what you want at the secondary. You'll probably need a zener there so that the secondary gets first go and the diode/snubber is only cutting off any transients there.

Hopefully I've mostly got that right. I'm sure someone will correct me if it's wrong in any detail.

RE: Blog 1: Transforming my understanding of flyback transformers, one hack at a time

shabaz — Sat, 16 Dec 2023 22:45:41 GMT

At least a smoking transformer is (sometimes) better than a exploding capacitor : )

Most datasheets will vary when it comes to transformers, but in the case of that particular one, the things that would help explain that behavior that you experienced, would be the Isat value (1.4A) and also the Primary Inductance value they mention (25uH) and the DCR (DC resistance) value of 0.55 ohms.

The Isat isn't a maximum current that the transformer will consume, but is the maximum current that should be provided (ideally less). Once the Isat value is allowed to exceed, current shoots up, and can cause the smoke. That occurs because beyond Isat, the inductance decreases, due to properties of ferrite material.

The question is, how can the Isat value be not exceeded, and the answer is by only allowing current to flow in the primary until it approaches that value (it ramps up), and then disconnect.It's visible by monitoring current (e.g. with a sense resistor), it is possible to see the linear ramp for a while, and then there is a knee where the current shoots up. Since you had a manual switch, it was not possible to limit to below Isat by limiting time. The datasheet mentions the transformer is (probably optimal) for 120 kHz, but the other way to see a ballpark value is to see the primary inductance.

Using the formula impedance = 2 * pi * f * L, you'll see that the impedance is super-low (a fraction of an ohm) if f is low. By pressing the switch manually, the f was under 1 Hz. To increase the impedance (and hence reduce the current), f needs to be far higher, since L is fixed (well, fixed is a first approximation) at 25 uH. Since the impedance using that formula was so low, actually the DC resistance of the primary winding was the dominant resistance, at 0.55 ohms, that means that with the 6V supply, 6/0.55 = 10A was going through the coil, which would make it smoke. Unless you're planning to use some IC, you probably will need a MOSFET or other transistor for general flyback stuff, although purely mechanical methods are possible with relays etc.

I think I get what you were aiming for in that particular experiment (I'm guessing, but I think your intent was that when the DC power disappeared, the magnetic field would collapse, causing current to briefly flow in the secondary winding). That for sure is possible but with high power solenoids, which are designed to have higher DC resistance for a (sometimes) sustained current to flow through them (some may not like that either), whereas a flyback transformer isn't, the flyback transformer doesn't expect to hold that magnetic field near-indefinitely, although it can but only if Isat wasn't exceeded in some way (e.g. current limiting in a circuit, but that's not normally how the transformer will be used). Interesting experiements. What software is that, is it the Analog Discovery 'scope you're using?

RE: Blog 1: Transforming my understanding of flyback transformers, one hack at a time

DAB — Sat, 16 Dec 2023 19:07:34 GMT

Nice update.

Yes, there are a lot of little details you need to understand about flyback transformers.