Help designing power supply

[docmur]

Hello

I'm trying to design a good, low noise, 120V -> 3.3V power supply.   I'm really stuck on where to start, I've never done much power supply design and very little high voltage AC - DC design.  I would prefer a small foot print. 

My first design was a bridge rectifier which turned the 120Vac -> 120Vdc and then use a step down dc-dc to get my 3.3 vdc.

Anyone know any good guides?

Thanks

Andrew

[benjamin545]

id start with one of these



http://www.amazon.com/Tripp-Lite-Isolation-Transformer-Outlet/dp/B00006HPFH/ref=sr_1_1?ie=UTF8&qid=1399507739&sr=8-1&keywords=isolation+transformer

then one of these

http://en.wikipedia.org/wiki/Bomb_suit

then start simple with a linear regulator design. just do AC->fuse->transformer->bridge rectifier->lm317 circuit.

once you have done that you can try some simple smps circuits.

[SArepairman]

well, I would get a isolation transformer, a differential oscilloscope probe, then this

http://cds.linear.com/docs/en/application-note/an25fa.pdf

buying it would probably be 30000 times easier, or using a transformer.

page 8

if you wanna build it, and sell it, your gonna have to worry about EMI protection, radiated noise, compliance, input protection. It is a pretty in depth design to make a mains high voltage not really isolated switch mode supply, and someone else will need to test it for certification, probably multiple times , unless you have nice equipment and know what you are doing, so you only need to test it once.. if you make your own you will defiantly need a spectrum analyzer and some special "jigs" that interface your circuit to the spectrum analyzer, at the very minimum. you will also probobly need some kind of free space probes to measure radiated non conducted interference.

you cannot get away from sending it out for compliance testing, the best you can do is test it yourself and fix errors that the compliance people will have an issue about

i would just plagiarize something that already exists instead of trying to design one, i think this has quite a few traps for young players.. non trivial.

I want to one day design my own, for designs sake.. it certainty is a interesting challenge that I will take over sodoku puzzles.

[rdl]

I've managed to make a few LM317 and LM78xx based power supplies that work pretty well and I don't use isolation transformers nor do I own a differential oscilloscope probe. Why do you guys think these are necessary?

[SArepairman]

I've managed to make a few LM317 and LM78xx based power supplies that work pretty well and I don't use isolation transformers nor do I own a differential oscilloscope probe. Why do you guys think these are necessary?

because he said 170v rectified SMPS

tansformer > linear or switchmode regulator is MUCH more trivial then MAINS > sm PSU > transformer

the first one you can test on a bench and everything is isolated and low voltage. the second one if there is a short circuit on your PCB it explodes violently, and you are dealing with 170 volts at hundreds of kilohertz or even megahertz. you can easily build a "high explosive radio jammer" with the wrong layout and design choices.

[rdl]

Ah, I missed that part about going straight from AC mains to the rectifier. I'd recommend doing as stated in the second post:

AC->fuse->transformer->bridge rectifier->lm317 circuit

Much easier and safer.

[ejeffrey]

How much current do you need?  Less than an amp or so, use a transformer and LM317.  If you need a few amps use the LM317 with an external pass transistor.  If you need several amps a linear regulator starts to dissipate a lot of heat.  You can use a 12 VAC transformer with a low voltage DC-DC converter.  You will still have a heavy transformer, but it is easier, safe, and more efficient than a line operated SMPS.

Finally if you don't need to DIY from scratch, you can get a PCB mount AC-DC converter that takes 120 VAC and gives you 3.3 volts.  Here is a 700 mA one for $13: http://www.digikey.com/product-detail/en/VSK-S3-3R3U/102-1801-ND/1950341 .  Digikey has a wide selection of low power board mount supplies.  A SMPS is a bit noisy.  If you want small and efficient but low noise, get a 120 VAC->5 VDC supply and use a LDO regulator to drop it down to 3.3.  Not as efficient as the straight SMPS, but better than a transformer supply, and still low noise.

[David Hess]

There is nothing to prevent using an input transformer to isolate and step down the line voltage to something safe and then a small non-isolated switching power supply.  That will have greater efficiency than a linear power supply and done will be pretty low noise.