On Tuesday, 31 May 2016 14:21:00 UTC+1, The Natural Philosopher wrote:
On 31/05/16 14:00, tabbypurr wrote:
On Monday, 30 May 2016 10:15:32 UTC+1, Rod Speed wrote:

Perfectly possible to design the charger properly so that it is
always completely safe even when something has failed.

Good chargers aren't quite that good. But designing a charger to stay
safe no matter what fails is an impossibly tall order. As anyone with
skills in the relevant area knows. Rodney's ignorant child-like
assumptions are frankly not very constructive.


Lord. I am going to - if not leap - reluctantly step to to the defense
of wodders. Random monkeys and all that.

Oh dear

The way SMPSUs work, is ultimately by having a high voltage bit and a
low voltage bit and separating them physically. They are bridged by
typically a high frequency ferrite cored transformer, and its possible
to do things that isolate mains from LV almost completely. with several
mm of plastic in between.

The only other linkage needed is some kind of negative feedback to
achieve regulation.

You can, in uncritical applications take that off a third winding on the
transformer, or for more precision use things like optical isolators.

What this means is that there is no chance of mains getting onto the LV
side *due to electronic component failure*. You need to somehow breach
the integrity of the physical insulators inherent in air gaps, plastic
transformer bobbins, or optical isolators.

Dropping the thing into a bathtub will do just that of course.

Lets have a reality check now. Any of the following can result in mains appearing on the output:

Transformer insulation failure. In the real world some have the bobbin missing, once in a blue moon one gets wound wrong, or a wire crossing over where it shouldn't etc. Not common with bobbin transformers but it can happen.

Optoisolator failure. Again uncommon, but a big fat voltage spike, which isn't too hard to create, can result in failure here.

Dirt bridging the insulating gaps.

And finally a stray mains wire touching the low v side.

While good design significantly reduces the risk of these, all can and have occurred in the real world.

As will some gross manufacturing defects

they do happen IRL.


That was a one in a million chance. To be so close and yet pass cold
testing, to fail in a hot amp in a hot climate..

I recall watching safety critical components that were failing tests get a pass. Again IRL it does happen.


In the context of USB sockets in bathrooms, they would however certainly
be no worse than an electric shaver socket.

In reality direct mains on output is more common with wallwart size supplies, and the user appliance connected typically provides zero insulation between 5v and user. Mains appliances of course do.


Ex of gross and deliberate assault with a deadly soldering iron a USB
in wall charge point should be mains isolated and completely safe.

Should be? Of course. But it's hard to find much that's completely safe or always as it should be in real life.


NT