2
\$\begingroup\$

When you look up power supply theory online all of the sample schematics have the switching transistor on the input side of the transformer. Is there a reason for this? What engineering tradeoffs are involved if it is possible to put the switching transistor on the output side of the transformer?

Especially in the case of a power supply converting 100 volts or more to a lower voltage it would seem you could lower the voltage requirements of the transistor by putting it after the transformer. It would also completely isolate the voltage/current regulation circuitry from the high voltage side of the circuit. Additionaly, it would make it dirt simple to make a high current dc to dc power supply for hobby purposes. You could take a microwave transformer, put a power transistor on the transformer output, then connect the transistor output to a rectification diode and smoothing capacitor. Quick, dirty, simple, reasonably dangerous, and cheap.

Since my main question is if you can convert one of the standard power supply types to have the switching transistor on the output side of the transformer I figure this flyback converter example circuit (pictured below) from Wikipedia will be a good circuit to focus the question on.

The main points I would like answered are:

  • Is it generally possible for a power supply to work in practice when designed this way?
  • Are there any engineering tradeoffs this causes?
  • Are there any design pitfalls you must avoid to make it work? (For example, would this design require you to put the switching transistor directly after the transformer or could you put it after the capacitor and rectification diode and still have it work fine - this part may need to be a second question spin-off post though.)

If it is generally not possible in practice to put the switching transistor on the output side of the transformer please explain why. For the purposes of the question I am assuming the input power source is 100v dc. I am also assuming the transformer windings are not a 1:1 turn ratio. This design change would require the transformer to change the voltage.

Edit: I am on mobile so it is not easy for me to edit a proper picture at the moment, but for my question the switch S would be on the secondary side of the transformer. At the node labeled "I subscript D" in the picture below.

flyback converter example circuit taken from wikipedia

\$\endgroup\$
2
  • 3
    \$\begingroup\$ Instead of showing a circuit of what you don't propose (switching the input side) show a circuit of what you propose (switching at the output side). I guess that would clear any misunderstandings. \$\endgroup\$
    – Curd
    Commented Jun 30, 2021 at 7:30
  • \$\begingroup\$ normally I would, but I am on mobile so it is not easy to make that happen. \$\endgroup\$ Commented Jun 30, 2021 at 7:48

3 Answers 3

8
\$\begingroup\$

You always have to have AC input for the transformer to work, doing its job stepping down the voltage. So powering a transformer directly with 100V DC is a non-starter. You have to use a primary switch just to get the transformer to do anything.

Say your supply is 120V AC. You can indeed make a step-down switching supply using a transformer followed by the rectifier to make the raw low voltage DC, and use a switching regulator in the secondary. Before off-the-line switchers became common this was how it was done. It was, and still is, a really easy way to make an AC/DC power supply. It avoids nearly all the hassle of high-voltage components and safety certification, and can use easily-sourced transformers. Is it a good way, given the technology and design tools available to us now? Hard to say.

A long time ago I made an audio product that used exactly that AC/DC approach, as an effort to reduce noise. That product used a large, heavy (and expensive - it was a low-leakage toroidal) transformer to deal with 60Hz. The designer / manufacturer was an audio person who did guitar amps as a business (a famous one at that) and pushed for that approach so I went with it.

Was it better than using a more-typical 30KHz or so used with primary switching? It certainly was heavier and more expensive, and it added some heft to the product making it seem more substantial. Nevertheless I don’t think I’d do it that way again. It’s less efficient, bulkier and costs more than the primary-switch approach.

\$\endgroup\$
6
  • \$\begingroup\$ Ignoring the blatently obvious obvious short circuit condition created that I overlooked, wouldn't pulsing the transistor on the secondary side be enough change in current to cause power to flow from the primary to the secondary side of the transformer? It would be using the transformer like a inductor to store energy though. Basically what a flyback converter does. \$\endgroup\$ Commented Jun 30, 2021 at 8:08
  • 2
    \$\begingroup\$ You can’t ignore the ‘dead short’ condition. At any rate, for energy transfer to occur at all across a transformer there needs to be a varying magnetic field. You’re transferring from primary to secondary, so the variation has to be in the primary. \$\endgroup\$ Commented Jun 30, 2021 at 8:15
  • 2
    \$\begingroup\$ Ah. I assumed that you would have a voltage on the secondary you could use to trigger the energy transfer. Apparently I need to properly teach myself how transformers work fully because I looked it up and that is not the case. \$\endgroup\$ Commented Jun 30, 2021 at 8:29
  • \$\begingroup\$ @JoshNabours There is a voltage - for a few milliseconds when the voltage source is first turned on. After that it decays to zero \$\endgroup\$
    – slebetman
    Commented Jul 1, 2021 at 1:28
  • \$\begingroup\$ @JoshNabours As for a short-circuit. Not quite. It would quickly heat up to a point where resistance increases and then it's no longer a short-circuit. The energy required to maintain the magnetic field also adds to resistance. DC electromagnets are a thing. Also, incandescent light bulbs are a thing. Both start with a short-circuit but both still work and are used fairly regularly. \$\endgroup\$
    – slebetman
    Commented Jul 1, 2021 at 1:32
5
\$\begingroup\$

The main reason why this doesn't work at all is that a transformer doesn't work with DC.

\$\endgroup\$
6
  • \$\begingroup\$ transformers can work with square waves. This is what the switching transistor would make happen at a bare minimum \$\endgroup\$ Commented Jun 30, 2021 at 7:23
  • 4
    \$\begingroup\$ You said "For the purposes of the question I am assuming the input power source is 100v dc.". 100V DC is not a square wave. If you had a square wave at the input, where would it come from without a switching transistor at the input? \$\endgroup\$
    – Curd
    Commented Jun 30, 2021 at 7:26
  • \$\begingroup\$ When the transistor is off the magnetic field would collapse just like in any other switch mode power supply. Transformers don't work because of a specific voltage, they work based off of changes in current flow. \$\endgroup\$ Commented Jun 30, 2021 at 7:35
  • 3
    \$\begingroup\$ And what amount of current would flow through the primary if you connect it to DC? (in theory and in practice) \$\endgroup\$
    – Curd
    Commented Jun 30, 2021 at 7:39
  • 2
    \$\begingroup\$ That's why I asked you to show a circuit diagram of what you proposed not of what you don't propose. I think that would have made it obvious. \$\endgroup\$
    – Curd
    Commented Jun 30, 2021 at 7:48
2
\$\begingroup\$

Is it generally possible for a power supply to work in practice when designed this way?

Yes, you can do this but then you lose the massive benefit of having a much smaller and cheaper high frequency transformer being switched at high frequency (circa 100 kHz) on the primary. It's a no-brainer when you think about it.

Are there any engineering tradeoffs this causes?

Yes, the big trade-off is the transformer because, if you "switched" at the secondary side then, your transformer has to be rated for low frequencies such as 50 Hz and that means much bigger primary magnetization impedance and a wholly bigger and more expensive design.

Are there any design pitfalls you must avoid to make it work? (For example, would this design require you to put the switching transistor directly after the transformer or could you put it after the capacitor and rectification diode and still have it work fine - this part may need to be a second question spin-off post though.)

You just wouldn't do it based on the clear benefits of switching the primary at circa 100 kHz.

it would make it dirt simple to make a high current dc to dc power supply for hobby purposes.

Yes, conceptually simpler but not an economic masterpiece. This is basically how a traditional old-fashioned power supply was designed but, most used linear regulators on the output rather than a buck regulator. It works but, for the same power output it is bigger, heavier, less-power-efficient and more costly.

You could take a microwave transformer, put a power transistor on the transformer output...

Think again; the output of a microwave transformer is thousands of volts.

\$\endgroup\$
3
  • \$\begingroup\$ you would obviously have to rewind the microwave transformer to get a low voltage out of it. I am aware of that. \$\endgroup\$ Commented Jun 30, 2021 at 7:29
  • \$\begingroup\$ I thumbed up your answer because of all the good information you provided, but I think my choice of circuit picture made you misunderstand my question. \$\endgroup\$ Commented Jun 30, 2021 at 8:34
  • \$\begingroup\$ @JoshNabours I disregarded your circuit because it didn't affect my answer. Nothing wrong with the circuit; that's how a modern switching power supply works. The transformer circuit does actually work with DC (although the devil is in the detail). \$\endgroup\$
    – Andy aka
    Commented Jun 30, 2021 at 8:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.