Off-Line Switching Regulator Design

[andyb]

Hi guys and gals - I'm hoping someone could point me in the direction of a book / website / app note about designing off line switching regulators.

I'm not really a beginner - some experience designing low voltage stuff, analog and digital - I've even designed a 12V to 350V flyback converter - but I've never designed a regulator that is "off-line" (connected to mains).

To motivate the problem a bit, say I'd like to build an isolated +/-15V, 500mA supply from an input of 100-240V 50/60Hz AC. Where would I start?

[Zero999]

The standard advice, is don't. Buy a pre-made PSU with the output voltages you need.

In reality this isn't always practical but an 18V laptop power supply and a buck inverter to give the 15V and buck-boost to give -15V would do.

[AndyC_772]

Is this for fun, or a commercial application?

If it's going to be sold, the first place I'd start looking is a copy of IEC 60950, and possibly 55022 as well. They describe the safety and EMC tests your design will likely need to pass, and that information is always best to have right at the start of the process rather than at the end.

Read up on conducted emissions / immunity, dropouts, surges, and power factor; these are the things that are likely to trip you up. Beyond that, the topology of a simple off-line switcher isn't that different to a dc-dc converter.

You'll find an isolating transformer helpful (to protect you), and an active differential probe (to protect your scope).

[andyb]

Thanks for the feedback, I'll have a look at those standards. I've got most of that gear already, need to build some B field probes and an amp for EMI investigations.

For now I'm just curious for personal projects, but you know how it goes..

I've always found it strange, seems like a completely taboo topic in electrical engineering. Tons of resources for really esoteric stuff like RADAR, SDR, FPGAs etc. but there's a big hole in common knowledge for probably the most common piece of electronics - off line regulators.

A laptop brick is fine for personal projects, but sometimes not fine for commercial applications. Looking around at off-the-shelf regulators, it's not always possible to find what you need. Then, looking into building one you hit a brick wall. No information, big warnings about safety, and suggestions to buy rather than build.

Our profession is not one to buy rather than build...

[Ian.M]

Power Integration's TOPswitch families are worth a look.   I haven't studied  the current series but the original TO220 packaged series had a reference design that was simple enough to be within the capabilities of a competent amateur, including winding your own magnetics.

[Wim_L]

I've always found it strange, seems like a completely taboo topic in electrical engineering. Tons of resources for really esoteric stuff like RADAR, SDR, FPGAs etc. but there's a big hole in common knowledge for probably the most common piece of electronics - off line regulators.

A laptop brick is fine for personal projects, but sometimes not fine for commercial applications. Looking around at off-the-shelf regulators, it's not always possible to find what you need. Then, looking into building one you hit a brick wall. No information, big warnings about safety, and suggestions to buy rather than build.

There are books about power supply design, including off-line. Here, in the beginners section of the forum, the warnings do make sense. Someone just getting started in electronics probably shouldn't make an off-line switcher as their first experiment.

[codeboy2k]

Power Integration's TOPswitch families are worth a look.   I haven't studied  the current series but the original TO220 packaged series had a reference design that was simple enough to be within the capabilities of a competent amateur, including winding your own magnetics.

PI's website used to be good.  Now it's garbage.
Their power designer software is still available and still pretty good, gives you a few pages of how to design and wind the magnetics on cores you can actually buy.

[Neilm]

Our profession is not one to buy rather than build...

Personally, I would buy a module every time. I have designed my own but tended to have all sorts of issues with EMC. When I costed my last effort, I realised that the module was significantly cheaper than what I had done. Now I only roll my own if there is some reason why a bought in module won't work (operating altitude, required clearances etc)

[Mechatrommer]

Our profession is not one to buy rather than build...

Personally, I would buy a module every time. I have designed my own but tended to have all sorts of issues with EMC. When I costed my last effort, I realised that the module was significantly cheaper than what I had done. Now I only roll my own if there is some reason why a bought in module won't work (operating altitude, required clearances etc)

it sounded like this electronics forum for hobbiests should never existed. leave all of this to professionals, we? do what we do for a living... not electronics. if we need electronics then just buy it with our savings... now i'm not saying you are wrong, i'm saying the argument is perfectly valid, from different part of the world...

so where's the fun? as most people said... granted by building on our own it can cost much2 more than buying mass produced stuffs. but this thing is not about money alone, some say because its custom to our needs, some say we know our stuff and some say, we gained knowledge, and knowledge is power. what i hate is i buy a stuff and when it broke i dont know where to fix, esp expensive stuffs that we trusted should last for years, this is electronics for the dump era, so... ymmv...

I've always found it strange, seems like a completely taboo topic in electrical engineering. Tons of resources for really esoteric stuff like RADAR, SDR, FPGAs etc. but there's a big hole in common knowledge for probably the most common piece of electronics - off line regulators.

you can find some schematics online but details explanation, characteristic report, hands on experience etc detail stuffs is pretty limited, so i guess if we have to go further we are on our own, maybe the better bet is snap a schematic from the net and then use own imagination, read some theories, slam down some components on perfboard and dont die. the most important part is dont die so you can have the chance telling the world. make ready some tin foil cap, current probe, glove googles etc. make sure you got the circuit properly fused so if anything whats the name of the guy.... lemme check jw notes... yeah if mr murphy happens, you got controlled destructive area, wear google everytime you start a mains. that sort of advice. whats mystery in this subject is the magnetic or inductive force in the transformer thats hard to imagine for me at least, so make sure HV current probe is around and start small... there is one mystery guy keeps posting videos here, rbola i think i'm going to follow his post when i'm in the mood of building this stuff i dont know if he's real or bluff... but the most important thing is... dont die.. good luck!

[Mechatrommer]

SMPS design and construction is well documented, the web is full of examples. What exactly is it you are looking for?

give me a link on full detailing, current probing, performance plot etc about "custom" smps construction, not the streamline ATX PSU... say i want a 10KHz transformer that can give 300W 5-30V adjustable output, that sort of "custom".

[Ian.M]

That would sing like a canary due to magnetostriction.  Such low switching frequencies haven't been common for SMPSUs since the early '70's TVs with self-oscillating thyristor chopper PSUs.

Every modern supply at power levels that can be run from a normal wall socket will be above 25KHz to keep them out of the human hearing range and most of them will be above 50KHz so they don't make dogs howl.

[T3sl4co1l]

You also have important size-efficiency gains going up to 100-400kHz.  Above 150kHz is more of a challenge due to conducted emissions (they practically don't care what happens below 150kHz, but right at and above, the limit suddenly cuts in), so you usually go much higher (300kHz+) to reap more benefits.  Higher frequency circuits take more careful design, so you don't want to jump right into a MHz or something, but it's quite possible to achieve good results up there.

The size of magnetics goes roughly as 1/sqrt(f), which reflects a combination of things: reduced Bmax, increased core losses, higher frequency materials (which are generally more expensive, but you use less), increased copper losses (skin effect goes as 1/sqrt(f) as well) but more impedance per turn (inductive reactance is proportional to frequency), improved heat dissipation (a 2/3 power law from surface area / volume), the list goes on.

Tim

[LukeW]

Basically, SMPS controller silicon, and switching silicon, is the "easy" part - Power Integrations has already been mentioned, and you've got good old TL494, and countless other options.

The harder part, really, is magnetic design, and the design, specification, and manufacturability of custom transformers for off-line-flyback use.

[Mechatrommer]

thanks thats what this forum for, you proved me wrong

[Zero999]

Our profession is not one to buy rather than build...

Personally, I would buy a module every time. I have designed my own but tended to have all sorts of issues with EMC. When I costed my last effort, I realised that the module was significantly cheaper than what I had done. Now I only roll my own if there is some reason why a bought in module won't work (operating altitude, required clearances etc)

it sounded like this electronics forum for hobbiests should never existed. leave all of this to professionals, we? do what we do for a living... not electronics. if we need electronics then just buy it with our savings... now i'm not saying you are wrong, i'm saying the argument is perfectly valid, from different part of the world...

so where's the fun? as most people said... granted by building on our own it can cost much2 more than buying mass produced stuffs. but this thing is not about money alone, some say because its custom to our needs, some say we know our stuff and some say, we gained knowledge, and knowledge is power. what i hate is i buy a stuff and when it broke i dont know where to fix, esp expensive stuffs that we trusted should last for years, this is electronics for the dump era, so... ymmv...

You raise a fair point about designing it for hobby use but the original poster did talk about it being for a commercial project, which if is the case, then it's normally more economical to buy a pre-built power supply, unless it's something very customised or a very large production run.

[Mechatrommer]

well, if its for commercial aka proffesional design and he has to ask in a forum... something is wrong me think...

[AndyC_772]

Why? There's nothing about getting a job in electronics that automatically confers all possible knowledge.

[Mechatrommer]

i mean if you are up to the professionalism, you should be capable of searching info that is already everywhere rather than asking "can it be done?", conduct personal test etc. but i believe thats not what the OP question, he's asking for resources/links/books which is perfectly valid for both professional or unprofessional, so i was talking about somebody else probably myself.