# Is most of non-isolated power supply “safe”?

I need a large power supply (a little over 1kW) for a consumer appliance that only has a single on/off switch and some non-conductive handles. An off-the-shelf isolated supplies run at least $60 for a sketchy Chinese supply that only supports either US or EU mains, whereas I could probably produce something similar to the following schematic for <$20 that would support both US and EU mains:

A lot of the cost savings would just be because I can afford to run the supply "dirtier" (higher ripple current) on input bulk capacitor(s). Even the output can be somewhat dirty, but I would plan to clean it up by using "smart" switching on the input to the buck converter.

However, safety is a concern. While from what I could tell from regulations there is no requirement that the power be isolated for my application, I would feel pretty bad (not to mention the financial ramifications) if someone got hurt somehow using a non-isolated design (e.g. got water into it and got shocked, if that's even possible). However, it seems like only a small portion of a non-isolated circuit is actually dangerous. As annotated in the schematic, is it true that mains is only dangerous to the extent that high voltages are involved, and that stepped-down but non-isolated voltages later in the circuit are "safe"?

Also note: I would plan to earth ground the metal case of the device (such that a short to live, and potentially neutral, would blow a fuse). And also, I would definitely have whatever final design I came up with reviewed by some experts; I'm just wondering if this is a plausible direction, given that using an isolated supply would probably make the product economically infeasible.

Edit: I'm going for 48V and actually closer to 1.2kW.

Sorry for not clarifying, but that ground symbol was not meant to mean that the circuit was connected to Earth (which would short the bridge), but rather just something I placed as required by the simulation software. The output side would be disconnected from any of L/N/G prongs, which I guess could cause some common mode voltage swing. However, the metal case would be earth grounded (and isolated from the rest of the circuit). Then if the live or the ground reference of the output side connected to the case the fuse would blow, but I guess if the pre-bridge neutral connected to the case, that may not be detected if there isn't an RCD (or I could add a low-rated fuse on the Earth ground path).

• Is switch 2 supposed to short-circuit the supply? – RJR Jul 28 '20 at 8:21
• That ground connection is going to bite you if the AC supply isn't isolated. – user_1818839 Jul 28 '20 at 8:38
• Can you make you device double insulated? Like your mains water heater. (class 2) – Jeroen3 Jul 28 '20 at 9:22
• Honestly, such a supply for 1kW makes me a bit uneasy. What voltage do you need? If 12V, have you considered computer PSUs? – marcelm Jul 28 '20 at 13:17
• @marcelm Voltage updated in description. If you can find a 1000-1200W 48V supply for ~\$40, then by all means, but I haven't seen anything close to that. – abc Jul 28 '20 at 14:58

No, even if that was only a 5V supply for digital logic, it is hazardous and unsafe. So at least your rectangle of unsafe voltages is too small. It will extend to all components and wires going into the box. That is because what you drew as the 0V or ground symbol, will actually have peaks of minus 325V in EU and half of that in the US during the negative sine wave cycle. So your supply voltage is not safe and it must not come contact with humans and it must not have connections to any other devices external to the box.

• I think you have the simple answer the OP wanted... – user105652 Jul 28 '20 at 7:08
• Okay, good point. Though it's a bit non-intuitive, as the entire output circuit will be swinging high and low relative to mains despite maintaining a constant relative voltage between the rails. If I isolated the entire circuit from the case, and earth grounded the case, AND ensured that any contact between L, N, or the ground symbol in schematic and the case/Earth resulted in a tripped fuse/breaker, then would it be "safe"? I guess this would be considered a Class I appliance? – abc Jul 28 '20 at 15:41

It is not only unsafe as stated by Justme, it will also blow imediately one of the rectifier diodes (one of the two diodes connected to GND on the left side of your rectifier symbol) because it shorts the AC input. The reason is that one of the AC terminals is also grounded which is not shown in your diagram (the other one of those two diodes will never be conducting).

• Good thinking. If that is mains earth ground you are right, it will explode, no part of the output can be tied to earth ground obviously. I only thought it is the common or negative for the output, and it is not the symbol for mains earth ground. – Justme Jul 28 '20 at 7:28
• Depends whether s/he meant mains earth ground or just circuit ground. – user253751 Jul 28 '20 at 10:24
• It is not mains neutral or earth ground, just a reference for simulation. See the edit. – abc Jul 28 '20 at 15:31
• @user253751: I assumed OP meant earth ground (or something similar) because he placed the GND line outside of the dashed "Unsafe/High Voltage" box. If it meant just any arbitrary 0V reference potential that wouldn't make sense. – Curd Jul 28 '20 at 15:31
• @abc: yes, that's exactly what I told you in my answer ;-) – Curd Jul 28 '20 at 15:33

Look at how appliances are designed to be safe. There are no appliances I can think of off the top of my head that use 1kW-ish power and have that power isolated before it is consumed.

Something like a stove may have isolated control circuits but the power circuits will not be isolated from the mains. Similarly with an air conditioner, a washing machine and its VFD motor controller, even a lowly microwave oven or toaster oven.

That means that the low power parts may need an isolated supply and some control circuit isolation and anything connected to the mains needs to be protected against accidental contact, for example by the ceramic heater insulation in a stove element, surrounded by a grounded sheath. If the insulation fails, a circuit breaker or GFCI trips and removes the power.

• Thanks Spehro. The accepted answer technically answers my question, but I found this to be the most useful/practical. – abc Jul 28 '20 at 16:25