WARNING: Live heatsinks & single pole on/off switches in switching power supplies

[Per Hansson]

This is a warning to inexperienced and experienced techs alike:
The De-facto standard is to have the primary side heatsink live in switching power supplies.
That is the primary side heatsink is referenced to the negative of the bulk filtering capacitor.
Therefore never ever touch the heatsink if the device is plugged in.
And even if it is unplugged let the capacitor discharge first!

Another really serious concern is that the power switch also by De-facto standard is single pole.
This means it only cuts power to the phase or neutral (not both) depending which way you have it plugged in if you have a Schuko plug or another non-polarized plug.
This means that when the device is switched off you will read 230VAC on the primary side heatsink to ground! (Or 115VAC if that is your mains voltage).

In the following picture I show a Fluke 113 measuring the voltage, I'm using this as it has a 3KΩ measurement impedance, so it presents a quite heavy load to the circuit being measured.
This is to allow you to understand that it is a real voltage we are speaking about and not a phantom floating voltage as you can sometimes see in house wiring.
Indeed it is large enough that I can't use the 30mA RCD you also see in the picture: because it will trip!
Then I also use a Fluke 225C Scopemeter to show the actual waveform.

Here is the voltage of the primary heatsink with relation to ground with the unit switched off on the single pole switch:
As you can see we are reading 237VAC (our line voltage)


And here is the voltage waveform of the primary heatsink with relation to ground with the unit switched on:
(As you can see the meters show a voltage of roughly 167VAC but this is a RMS approximation that is not so useful).



The actual DC voltage is of course our rectified AC voltage that is close to 340VDC due to my slightly high line voltage (√2 x 240VAC = 339.4VDC)
I've added some cursors here to better clarify what I mean:


It is also extremely important to point out that the only safe way to measure on the primary side with an oscilloscope is using a differential probe or with a Scopemeter that is not referenced to mains ground.
Because if we wanted to get a useful reading here we would need to clip the negative lead to the primary side heatsink since that is the negative of the primary side.
And I hope you understand now why that would end extremely badly, if not please watch EEVblog #279 - How NOT To Blow Up Your Oscilloscope!

[dicky96]

It is also extremely important to point out that the only safe way to measure on the primary side with an oscilloscope is using a differential probe or with a Scopemeter that is not referenced to mains ground.

Or use an isolation transformer? An excellent post regarding the dangers, nevertheless

[momaka]

I've been repairing PSUs for years now and also been posting some of those repairs here for just as long... yet, it never occurred to me that we should have a warning like this as a sticky thread. But this was a really good idea and also very well written, Per! I'm glad you posted it.

Hopefully this should save us (well, me for sure ) some typing to warn people when they need to poke on the primary side. Now I can just refer them to this thread, instead. Although it should go without saying that doing any measurements on the primary side of a plugged in device is dangerous... indeed I agree that many people don't realize that the primary heatsink is almost always live.

That said, I have also seen the exception a few times where the primary heatsink wasn't connected to anything and just floating. So it's not live on *all* PSUs... but like you said, it is fairly standard practice with most manufacturers for it to be live. So when in doubt... just assume the primary heatsink is live. And even when not in doubt, it's still better to assume that it is live. All in all, it's *ALWAYS* a good idea to avoid touching any parts on the primary side of a PSU with your bare hands.

Moreover, the primary heatsink can be especially dangerous, since it is indeed usually referenced to the negative lead of the primary cap... which means it will have a strong DC component to it. And while AC can penetrate the skin at lower voltages, at least it will usually make muscles on the body twitch. But with DC, there's a danger of muscles contracting one way and not being able to let go. So high voltage DC is something to especially beware of.

Originally Posted by Per Hansson

Another really serious concern is that the power switch also by De-facto standard is single pole.
This means it only cuts power to the phase or neutral (not both) depending which way you have it plugged in if you have a Schuko plug or another non-polarized plug.

Yes.

Moreover, even for countries/places that use polarized plugs - NEVER liberally assume that the wiring in your home was correctly installed. If you would like to use the power switch to turn Off power when working on a device, ALWAYS first verify that the specific plug you are using is correctly polarized. My personal opinion, however, is that one should not depend only on the power switch to disconnect power to the device that is being worked on, as it doesn't take much to accidentally turn On the switch. Unless a situation calls for it to use the power switch, it's always better to physically unplug the device.

And to end on a slightly humorous note here, it should also be noted that the price/quality of a PSU doesn't necessarily mean that the PSU will use a DPST (dual-pole) power switch. For example, here's an old Deer DR-250ATX from the early 2000's that's probably not the best PSU in the world... yet, it has a dual pole switch that disconnects both Live and Neutral. I've also seen many higher-end PSUs with a single pole switch. So again, the build quality of the PSU is not necessarily relevant to the type of switch used on the AC line.

[Dan81]

Originally Posted by momaka;1052522
And to end on a slightly humorous note here, it should also be noted that the price/quality of a PSU doesn't necessarily mean that the PSU will use a DPST (dual-pole) power switch. For example, here's an old [url=https://www.badcaps.net/forum/showpost.php?p=224299&postcount=490

Deer DR-250ATX[/url] from the early 2000's that's probably not the best PSU in the world... yet, it has a dual pole switch that disconnects both Live and Neutral. I've also seen many higher-end PSUs with a single pole switch. So again, the build quality of the PSU is not necessarily relevant to the type of switch used on the AC line.

I've seen those DPST switches in some higher end Allied-built Modecom units as well, though most of the others use the normal single pole switches, with the small note that they heatshrink them unlike the lower (bottom-of-the-barrel) units.

[s675]

Here in the UK most of our circuit breakers at the main board (consumer unit) are single pole. If the incoming supply before the board is reversed, you end up with the situation where everything remains live.

Our wall sockets have switches but a lot of them are also single pole, so with the reversed incoming connection, you can turn off the PSU switch, turn off the switch at the socket, and turn off the circuit breaker, and the PSU will still be at 240v unless its switch is on the neutral (quite common).

Also, I learned this heatsink live situation the hard way yesterday!

[Th3_uN1Qu3]

It's the same here in Romania although the last couple of breakers I bought were double pole. Nothing about it on the box though, you'd have to take it out and look at the diagram printed on it.

[RJARRRPCGP]

Looks like over here in the Americas, having only the neutral switched and the hot, not connected to a pole, can end extremely badly.

It can turn the device that you're working on into a "series-multiplier", causing equipment to get blown out in the building.