# How to select a DC-DC converter topology?

I've been simulating a switched mode 12V to 170V boost converter, and it has been quite the learning experience. If my observations are correct, power losses in the circuit not only increase as the square of input current (as usual), but also as something around

$$\frac{1}{(1 - duty\,cycle) ^ 4}$$

for a given load, inductor and switch. So... that's obviously a pretty big deal.

I am aware that a switched mode full-bridge with an HF transformer can wrangle really large amounts of power from a rather low voltage source to a rather high-voltage one, given appropriate selection of core, windings, switches and switching scheme. I am also aware of the flyback converter arrangement and various other similar high-frequency converters, such as the Cuk regulator and SEPIC regulator. So on to my questions.

In general, how does a professional decide what DC-DC topology s/he will use for a given application? And, more specifically, is there a clear cut-off in terms of Watts or duty-cycle that professionals generally use to know when they need to get away from a simple boost converter design?

The simple rule of thumb I use is that when you get beyond about 6:1 it's time to compare other options. You're at 14:1, so well beyond that.

You could always use multiple stages or push that duty cycle down further by heroic measures to get higher ratios, but the approach does tend to run out of gas.

You might want to have a look at Analog Devices application note AN-1126 on using Sepic multiplied boost converters. Your application appears to be right about in the wheelhouse for that approach- which maintains the single-switch simplicity of the boost converter, and does not require custom magnetics.

• Thanks for that. I did simulate two and three stages, and it definitely made it possible (i.e. I could ALMOST regulate the load voltage within reasonable parameters), but I was seeing efficiencies of between fifteen and twenty-five percent. It was ugly. What heroic measures might these be, and where could I look them up? – Sean Boddy Apr 7 '14 at 1:59
• @user39962 I've added a link to another approach that might fit. – Spehro Pefhany Apr 7 '14 at 2:01
• @SpehroPefhany AN-1126 is an good app note. They look at alternative topologies: simple boost, charge pump multiplied boost. But for some reason, they didn't mention flyback as an alternative. I wonder why. – Nick Alexeev Apr 7 '14 at 3:05
• @NickAlexeev Is there much difference between a tapped inductor boost topology and a non-isolated flyback? – Spehro Pefhany Apr 7 '14 at 3:15
• @SpehroPefhany Forgive me, but I am just shy of my bachelor's. If I understand correctly, those two topologies are basically the same, correct? One uses a mutually coupled inductor and the other uses a three winding transformer for the same purpose? – Sean Boddy Apr 7 '14 at 3:31

I am the author of the ADI application note. I just now encountered this blog by googling the name of the article/application note.

Of course flyback can work, but that introduces transformer leakage inductance which causes spikes and ringing. This usually needs snubbing; that hurts efficiency. Even with snubbing you will probably require a higher voltage FET. Also flyback requires higher voltage diodes. Ultrafast diodes have recovery issues. Schottky diodes can be a problem. Recently SiC schottky diodes have been introduced and those can work. But this topology allows high efficiency and clean waveforms using inexpensive multiple-sourced parts.

Something is wrong with the low predicted efficiency mentioned above. I actually measured >91%. I have built many different switchers over many years and this circuit behaves better than many.........

Bob Zwicker

• Hi Bob, and welcome to the stack. It's an honor - your note got me my diploma. The efficiency noted in the comments above occurred in the context of multiple boost converters, each with a fairly high duty cycle. This was attempted before I finally settled on a SEPIC converter topology for the project. To be clear, the efficiency problem was noted in a simulation where I was experimenting with multiple stages of standard switch-and-diode boost converters. – Sean Boddy Jun 23 '15 at 4:52
• Thanks, Sean While simulation can be useful, it is critical to get the models right. The biggest gotchas are transformer leakage inductance, diode forward/reverse recovery, and FET capacitance. In practice, flybacks usually produce a voltage spike at FET turnoff which subjects the FET to voltage stress which significantly exceeds simple calculations. Simple boost is cleaner (no transformer) but (I think due to multiple energy discharge paths) this circuit produces waveforms which are cleaner even than simple boost. Waveform photo in app note is simple & real. – Bob Zwicker Jun 23 '15 at 6:15