Achieving 98% efficiency in a boost converter

Sure. I'd meant to point out that a) components for super-high efficiency are a lot harder to come by at the 20mW power level, and that is partly because b) it's an uncommon requirement: you don't often throw 10x$ in parts to harvest an extra x$ worth of power. Maybe in spacecraft. On Earth you'd just use a 2% larger solar cell.

IOW, I suspect it's a lot easier, with COTS components, to get 98% efficiency in a 20W converter than at the 20mW level.

Fun project, as usual. Thanks Win.

Cheers, James Arthur

The design has been updated and is ready to order.

Here's what my document says, achieving a 98% result: One test condition: 600mV input from a solar cell, 2.4V out at 10mA load = 24mW. Boost converter operation at 250kHz. Critical for success: an RM8 inductor, low-loss winding, measured DCR 0.018 Ω, calculated loss 0.03mW, and AC loss 1.3 Ω, 0.06mW. Best estimated CMOS switch, TS3A24159 powered from Vcc = 3.3V. Ron 0.10 Ω, loss 0.16mW. Calculated switch loss, switching 2.4V with Coss=150pF, 0.21mW.

Continuous total loss 0.46mW = 2% of 24mW (not counting 0.18mW in sense resistors).

I had only one email about a free PCB, but with no followup to my answer. My offer still stands.

My RIS-767 testbed is not meant to be a full-blown boost converter, instead it's meant to evaluate inductors and synchronous switches, under various conditions, by independently setting the switching frequency, the PWM duty cycle, and the load current. Fix two parameters and step a third, plot data points.

The enigma of the long delay to receive your pcb is cleared up at last. By process of elimination the one email that you received belongs to me. Another email was just sent to you with my address in it.

Thank you,

OK, Don, great, you'll get two boards to play with. The file on DropBox has changes (errors fixed), so over-write the old one.

Assembly parts location images added.

Folder includes assembly info:

My order is going in shortly; I need to know how many boards I'll be shipping, so place your (free) orders.

I will take one, if you haven't placed the order yet. I'll email my address.

What's the prize for best efficiency? Do I get shamed if I can't achieve 95%?

OK, two boards for you.

The design has been further updated, more capability, now with 18 flip-chip daughterboards. There's a new 11-page document describing the project, and with five possible simplified build schematics.

I'll be placing PCB orders shortly.

"That file isn?t here anymore Someone might?ve deleted the file or disabled the link"

:-(

That is a very good achievement!

Model and calculations need to be more thorough. This write-up seems to be a comprehensive analysis. I didn't read the whole thing:

Thanks, I hadn't seen that writeup, looks interesting. You're right, I haven't given details of my 98% calcs in my documents, because I'm concentrating on the PCB.

Hmm. See my post this morning, it should work.

Well, the inductor is an achievement, the rest will be calculation / speculation until the testbed arrives.

Thanks, that link works. It hadn't shown up in my news server yet before walking the dogs.

If the MPPT is a required part of the game you'd have to add in the Arduino power consumption. Maybe use a MSP430 instead? Those can be operated on small drops of power.

Of course. But my rule is that none of the controller power consumption counts. Why? We assume with effort you could drive that as far down as you want, but that's not the point. The point is to achieve very efficient power conversion, period. Any wild control scheme you dream up is OK.

Scaling assumptions that approach an order of magnitude are often troublesome. Two orders of magnitude will require more than just serious thought.

RL

Yes, right. And also bench experimenting. Hence my offer for free RIS-767 boards.

Is there a recommended 600mV source, or are input filter components considered adequate impedance stabilization?

RL