Review any text on power supply design, there will be a chapter devoted to each topology. But they don't discuss the A vs. B trade-offs of the various cases; that is, how to choose which, for any particular application?
I'd like to find a table, where each row represents a specific topology, and columns list the various attributes, or trade-offs, with recommendations for appropriate uses.
Anyone know of such? Or any publicly available articles discussing this topic?
Linear or Switch mode. Or by different topologies do you mean the different types of switch mode supplies. Win's AoE3 has an nice section (chap 9.) I think he posted it free online somewhere.
Mostly I end up buying someone else's power supply.
There's optimism-driven design, where you design the power supply first. Pessimism-driven design, where you design the power supply last. Client-driven design, where you design everything simultaneously because would you hurry up already
Win did post it here: but the Dropbox link no longer works. I have the 109 page PDF somewhere, but can't find it right now.
Laying out one big PCB with analog stuff and FPGAs and all that, a dozen or so power rails, you have to do it all at once. It's tedious to estimate all the currents of all the rails, and it's prudent to overkill the regulators by maybe 2:1.
My latest fave is the TI TPS54302, a cute little SOT23 synchronous switcher. It lets you start with a 24 volt wall-wart and switch down from there, sometimes cascaded switchers.
No worries, if he doesn't have the book he should buy it. GH
I use a spreadsheet to track the power tree requirements. I can immediately see if I have a problem as things progress with the rest of the design. Overkill is good, if you can afford it.
We cascade them all the time. Going directly from 24V to 1V can get problematic. However, cascading switchers can cause EMI headaches. We, almost always, synchronize all switchers. Dueling power supplies can cause issues that aren't seen until you're in production, or worse, in the field.
I have a bunch of these modules based on the TPS6120, they're my go-to for anything that has to run off a couple AA or AAA batteries, or a LiPo cell:
They'll drive several incandescent filaments fine, it won't start up into a big cold filament but PWMing the shutdown pin for a few seconds at a low duty cycle pre-heats it.
That TPS part is not synchronizable - it only has 6 pins - but is radically spread-spectrum.
$5.49 per? You might want to look at AliExpress. You'll probably get modules with similar specs for a tenth of that price - with free shipping if you can wait for a few weeks.
Oh, true enough, I just used that site as an example because there are some specs and links to datasheets there.
There's still one brick-and-mortar electronics components retailer about ten miles down the road from me, a large two story warehouse building operation packed with components floor to ceiling, in a pinch I can always go there they have a wall of various SMPS modules like that for approximately the same price.
By dueling supplies, do you mean when one turns on before the other? Which can then make some downstream IC misbehave. (I hate that problem, sometimes it only shows up with weird AC power cycling.)
George H.
More likely hetrodyning, intermittent synchronzation, of switcher frequency.
Powerup sequencing is often a big deal. That little TPS part has internal soft-start, and one can daisy-chain some outputs into the enables of others for deliberate sequencing.
Without tricks like that, you can fold-back a wall-wart.
Right, power up sequencing. (I can never find the right words.) GH
Right. That's another no-no in our shop, for the same reasons.
No, though I hate it when that happens, too. What I mean is that the two radiate at different frequencies and can beat against each other, causing all sorts of problems. Since it can't be predicted, you don't know when it'll crop up so you really can't test for it, either.
Spread-spectrum is a no-no? Why? It helps pass EMI tests and keeps multiple switchers from heterodyning.
TI must have shaped the spreading spectrum carefully; the final DC output is very quiet.
Great little part.
Spread-spectrum prevents that.
I don't think that synchronizing switchers is beneficial. If they are not SS, and are frequency programmable, I like to keep them well apart.
You bet! It makes the heterodyning problem far worse than without it.
TI makes good stuff. I use their stuff all the time but there are must haves for most applications. One is, must have spread-spectrum - NOT.
No, it doesn't. It makes it worse.
Nope. Exactly the opposite. Keep them all the same, where they're predictable. If EMI has to be fixed, fix it and it stays fixed.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required