Be careful with those teardowns, kids!

[NiHaoMike]

It sounds like something for Mythbusters to test. The largest caps I have seen in a PC power supply were in an old HP (back when they were good stuff), and those were a pair of 330uF, 450V. (The PC I'm using right now has even bigger ones, but it's definitely not stock...) I'm sure they can kill if fully charged, but it would take a few orders of magnitude more energy to burn more than a little spot. Maybe someone (photonicinduction?) can try discharging capacitors of various sizes and voltages into a piece of scrap meat to see what it would take to make a significant amount of it char.

Chances are, I don't think it's likely for the caps to stay charged for long since the standby circuit and voltage dividers would discharge them pretty quickly. (All bets are off if the power supply is bad, however!) The bleeder resistors are not used much nowadays in order to increase the standby efficiency.

I suspect the power supply was plugged in at some point (maybe he wanted to test it?) and maybe he even figured out the green to black trick to start it up. (I knew that at about age 14 or so, but I was smart enough to check high voltage capacitors before working and I especially kept a distance from exposed mains connections.) Then if he touched the primary heatsink, he'll get both a deadly shock from the mains and severe RF burns from the high frequency AC.

EDIT:

It's perfectly plausible that he went to short out the cap with a screwdriver but touched both sides of the cap as he did. The current flowed through his heart for a fraction of a second, killing him. Then the screwdriver made contact with both sides of the cap which covered his fingers in carbon and metal from the vaporized screwdriver tip.

Anyone viewing the body would see he died while working on something electrical and would notice charring on his fingers and assume they're electrical burns.

Then they do the autopsy and notice that it's only a "surface" burn.

[asbokid]

Sheesh! Bit of a morbid thread is this!

"It's being so cheerful as keeps me going!"

[amyk]

Anyone with medical experience care to suggest how much current (and importantly, for how long) it really takes to STOP a heart? Once we've got that the maths should be easy enough.

More than it takes to cause fibrillation, which is what causes death. Defibrillators give a controlled shock to get all the muscles to start beating in sync again, and they put out a lot more energy (few hundred J) than two PSU caps.

[westfw]

[power supply caps might hold a significant charge ESPECIALLY IF THE SUPPLY IS BROKEN AND HAS NO DISCHARGE PATH]

Ah!  Yes, I didn't think of that.   Very good point.

Anyone with medical experience care to suggest how much current (and importantly, for how long) it really takes to STOP a heart? Once we've got that the maths should be easy enough.

Moore ("Electrostativs") said 10J for electrostatic discharges (ie anything high voltage enough that the discharge is "short.")  I'm not sure if mains voltage is quite high enough for that (but it's pretty close.)
The rule of thumb is "20mA" across the body.
You can be unlucky; Pacemakers run years on a relatively small battery, and having one get messed up can surely kill you.  Actual nerve impulses are very tiny.

I've always been more worried about slicing an artery on sharp sheet metal.

[Pentium100]

It just so happens that I am currently repairing two PC power supplies - a Q-tec 350W (cheap) and Coba 400W (a bit better) - both have bad caps (Coba needs new HV (1000uF/200V x2) caps as the current ones get quite warm) and the Q-tec has a bad fan and a shorted diode (probably because of the fan). Now that I replaced the caps (and the diode with a more powerful one, since I could not get the exact part quickly), the Q-tec will probably be better than new. It's still cheaper than buying new PSUs.

[Mediarocker]

and they put out a lot more energy (few hundred J) than two PSU caps.

I call bullshit. Sorry but a typical switched mode PSU primary cap contains enough voltage and current in it to stop your heart. It only takes a couple of milliamps to cause fibrillation.

http://www.physics.ohio-state.edu/~p616/safety/fatal_current.html
http://van.physics.illinois.edu/qa/listing.php?id=6793
http://van.physics.uiuc.edu/qa/listing.php?id=1339

It takes about 100 mA of current to kill you. Even currents under 75mA your breathing can stop completely, which can kill you.
Provided certain conditions are met, an input cap on a PSU has more than enough current to kill you.

A basic US household switching power supply has over 350VDC in it's primary. 350VDC WILL go through your skin which (when dry) has a resistance of 1,000 to 100,000 ohms depending on thickness and if it's burned or open. Burnt skin allows for an open wound which will readily conduct more current. Granted that the internal resistance of the human body varies depending on muscle content and other factors... Considering this was a kid I'm willing to think:

A. He has no calluses on his hands and the skin on his hand is fairly thin.
B. I'm not assuming he has much muscle as most teenagers aren't very musclebound... most certainly those inclined to work on their family computer.

So assuming those factors we'll take his total resistance:

Rtotal = Rskin(in) + Rinternal + Rskin(out)


Perfect condition:
Rskin(in)= 1000ohm (This is the "Input" resistance)
Rinternal= 100ohm (being generous here as more muscle = more resistance)
Rskin(out)= 1000ohm (This is the "output" resistance)

I=V/R so I=350VDC/2300ohm

350VDC/500ohm=.15A

Best case Scenario (Older gentleman with lots of muscle mass and calluses):

Rskin(in)=100,000ohm
Rinternal=1000ohm
Rskin(out)=100,000ohm

340VDC/201,000ohm= 0.001A

Plausible scenario:

Rskin(in)=1,500ohm
Rinternal=400ohm
Rskin(out)=1,500ohm

340VDC/3,400ohm=.1A

Likely scenario:

Rskin(in)=2,000ohm
Rinternal=500ohm
Rskin(out)=2,000ohm

340VDC/4500ohm= .0755 repeating

Enough to stop breathing.

Depending on the person, it's more than enough to kill you.

Regardless we don't have his skin resistance measurements, we don't know if he had any open wounds which would have readily conducted the current, and we don't know anything other than what they released in the news. Saying a cap CAN'T kill you is negligent and voiding respect for electricity.

ANYTHING. ANY CURRENT can kill you. Period. It just depends on all the factors and if the moon is lined up.

People survive lightning strikes... but capacitors can kill you. Just depends if it's your day or not.

[grumpydoc]

Defibrillators give a controlled shock to get all the muscles to start beating in sync again

Sort of - the idea with a defibrillator is to give a massive enough shock to stun all the heart cells, then hopefully the heart's normal pacemaker can take over once more to produce co-ordinated activity. That's why you continue CPR after giving the shock - because it takes a while for electrical activity to recommence.

So looking at the energy supplied by a defib (360J normally) and saying "well a PSU cap is much less than that" isn't valid - the defib is designed to overwhelm.

If you hit the 200ms or so "relative refractory period" from the peak of the T wave to the start of the next cycle you have a very high risk of causing ventricular fibrillation with even a quite small shock.

[HackedFridgeMagnet]

I don't know if this is more accurate than the other report but it says here that the official cause of death was electrocution. Doesn't mention burns.

http://digitallife.today.com/_news/2012/10/09/14321140-teen-electrocuted-while-taking-apart-unplugged-computer

[westfw]

A basic US household switching power supply has over 350VDC in it's primary.

How do you figure?  I get 120V * 1.414 = 170V, which explains the 200V caps I see in most US-only supplies.  (Most computer supplies are "Universal input", designed for up to 240V input.)

[Pentium100]

A basic US household switching power supply has over 350VDC in it's primary.

How do you figure?  I get 120V * 1.414 = 170V, which explains the 200V caps I see in most US-only supplies.  (Most computer supplies are "Universal input", designed for up to 240V input.)

Old PSUs (the ones that allow you to select between 120V and 230V and have no PFC) configure the caps as a voltage doubler when set in 120V mode. In 230V mode the caps are in series so 200V per cap is enough. Regardless of the input voltage, the primary will see the same voltage, about 330V.

PSUs that have active PFC can automatically adjust to the input voltage, new PSUs with passive PFC are usually 230V only (or I guess in the US they would be 120V only but probably still have the voltage doubler).

[SeanB]

Active PFC does mains voltage compensation, so can wotk on 90-260VAC. The capacitor then is normally always charged to 390V if the PFC is working, irrespective of the mains input voltage.

[tom66]

One other possibility is that while it was live, he touched some switching portion of the power supply.
At 300Vp-p at 30-60 kHz hurts a lot more than 50/60 Hz or DC.

[grumpydoc]

At 300Vp-p at 30-60 kHz hurts a lot more than 50/60 Hz or DC.

Interesting, I'm not about to try but I wonder if the "surface effect" is strong enough to avoid passage of current through the heart at these frequencies.

[tom66]

At 300Vp-p at 30-60 kHz hurts a lot more than 50/60 Hz or DC.

Interesting, I'm not about to try but I wonder if the "surface effect" is strong enough to avoid passage of current through the heart at these frequencies.

VOLUNTEERS WANTED FOR STUDY ON SAFETY.
FREE CAKE FOR ALL APPLICANTS. TAKE ONE.



Another thing to note is many switching power supplies have transients on the switch node. Especially flybacks for the standby supply, upwards of ~700V for many designs (hence why you need a good 800V FET and flybacks aren't used over about 50-60W.)

[necroscope]

Could it have possibly have been an old crt monitor and he never discharged the flyback befire trying to pop the anode out with a screwdriver,That could surely Kill and would be an easy mistake for a lot of adults let alone a child.

[N2IXK]

The charge stored on the anode circuit of a CRT monitor isn't going to kill you.  The voltage is quite high (25-30 kV), but the capacitance involved is a few hundred pF at most, and the stored energy is correspondingly low. There is no actual filter capacitor in most cases, just the capacitance between the internal and external conductive coatings on the CRT itself. Yeah, you will get a zap like a really bad carpet shock, or a spark plug wire. The involuntary muscle contraction can cause problems if you jerk your hand away and hit a more dangerous voltage, smash the CRT, or scrape your hand on the cabinet and cut yourself, etc.

The low voltage power supply in a monitor or TV is far more dangerous than the HV supply to the CRT. 150V or so of regulated DC, capable of an amp or so of current, usually derived from a non-isolated "hot chassis" type mains rectifier arrangement. Similar to the primary side of a switchmode power supply.

[necroscope]

The charge stored on the anode circuit of a CRT monitor isn't going to kill you.  The voltage is quite high (25-30 kV), but the capacitance involved is a few hundred pF at most, and the stored energy is correspondingly low. There is no actual filter capacitor in most cases, just the capacitance between the internal and external conductive coatings on the CRT itself. Yeah, you will get a zap like a really bad carpet shock, or a spark plug wire. The involuntary muscle contraction can cause problems if you jerk your hand away and hit a more dangerous voltage, smash the CRT, or scrape your hand on the cabinet and cut yourself, etc.

The low voltage power supply in a monitor or TV is far more dangerous than the HV supply to the CRT. 150V or so of regulated DC, capable of an amp or so of current, usually derived from a non-isolated "hot chassis" type mains rectifier arrangement. Similar to the primary side of a switchmode power supply.

Thanks i never knew that,i have always been told by arcade machine techs the flyback can store a charge big enough to kill you even after being switched off for a day,I guess they were exagerating,still i wouldnt like to cop a hit from one,Still if the young lad had a weak heart maybe a hit like that could be enough to stop the ticker.

[Mediarocker]

A basic US household switching power supply has over 350VDC in it's primary.

How do you figure?  I get 120V * 1.414 = 170V, which explains the 200V caps I see in most US-only supplies.  (Most computer supplies are "Universal input", designed for up to 240V input.)

Old PSUs (the ones that allow you to select between 120V and 230V and have no PFC) configure the caps as a voltage doubler when set in 120V mode. In 230V mode the caps are in series so 200V per cap is enough. Regardless of the input voltage, the primary will see the same voltage, about 330V.

PSUs that have active PFC can automatically adjust to the input voltage, new PSUs with passive PFC are usually 230V only (or I guess in the US they would be 120V only but probably still have the voltage doubler).

Thank you for pointing that out 

Also, There is something else to point out... what if the kid had really sweaty hands? The high salinity of the sweat would conduct current more readily and possibly grant the "perfect" condition I listed. So as I mentioned before, this is most certainly possible.

[grumpydoc]

xkcd's amusing cartoon, posted by tom66

LOL, I liked that.

Lest it not be clear I'm not advocating anyone try it. Also in case anyone is unclear about my own views I'm in the "this shit is dangerous even if you know what you're doing" camp.

But I do just wonder whether those incautious enough to be in the "I've had loads of whacks from SMPS's and I'm still breathing so it can't be all that bad" group might have been saved by a degree of surface effect.

Medical diathermy/electrocautery units rely on the surface effect to avoid killing the patient while allowing local current density high enough to heat tissue. In fact they vaporize tissue when used in cutting mode. Having seen them used countless times as a student I can attest that the patient does not have an immediate cardiac arrest when they are used. They use AFAIK AC in the range 300kHz - a few MHz at (I think) a couple of hundred V. As an aside this is one of the things that makes ECG (or EKG, if you're that way inclined) machines expensive - their inputs must be totally electrically isolated and be able to record microvolt signals but ignore (and be protected against) 10's or maybe 100's of volts of RF current or defibrillator current.

But I digress....

Does anyone know or have a reference on the surface effect in human bodies and at what sort of frequencies it becomes significant. Idle curiosity you understand - no experiments please

[Psi]

It's perfectly plausible that he went to short out the cap with a screwdriver but touched both sides of the cap as he did. The current flowed through his heart for a fraction of a second, killing him. Then the screwdriver made contact with both sides of the cap which covered his fingers in carbon and metal from the vaporized screwdriver tip.

Anyone viewing the body would see he died while working on something electrical and would notice charring on his fingers and assume they're electrical burns.

Then they do the autopsy and notice that it's only a "surface" burn.

There hadn't been any autopsy at the time the article about burns was written.
The whole "burns" thing was the initial report from the scene.

[G7PSK]

I have had a belt from the fly back of a 26 inch colour TV that was on. I was adjusting the picture at the time when my hand got to near the tube connection, the charge threw me across the room but did not leave any burn marks.
Now the output from a magneto on a paraffin tractor that leaves a nice black burn mark but does not kill.
I don't think that it is just voltage or current that is required for lethal effect but duration as well.
I have collected a few shocks from power plant of many KW output but due to never having closed my hand around leads and always wearing insulating foot wear I have managed to survive and become ever more cautious.
The best way I know of to get shock while working with electrical equipment is to be in abject terror of it, you have to be ever vigilant of the risks but fear will lead to as many mistakes as over confidence.
The only way that a capacitor in a piece of computer equipment will kill is if the person has a dodgy heart or it is still connected to the mains and the person grabs something in such a manner that they cannot release it. 

[HackedFridgeMagnet]

I have been just trying to get a handle on (quantify) the dangers of the non Mains parts of the SMPS's.
I looked around for some coroners cases of people dying from sub 400v dc electrical shocks. I couldn't really find details of any, but often the coroners don't have a clue whether the cause of the shock is AC or DC anyway, so they don't state it.
I did find that sub 70v electrocutions in the US were almost non existent.

this is old but gives some details

http://www.google.com.au/url?sa=t&rct=j&q=&esrc=s&source=web&cd=6&ved=0CEMQFjAF&url=http%3A%2F%2Fwww.lookupandlive.org%2Fmaterials%2FNIOSH%2520Electrocution%2520Study%252098-131.pdf&ei=GM53UJupHcmViAfzgIHwCw&usg=AFQjCNEr9JWYKpyF0XmucOJdz7tADDw7IA

[notsob]

Here's two examples of shit going wrong that I know of,

1. A motor mechanic was repairing a tractor and shorted the metal wrist band of his watch across the rear of the starter motor whilst tightening up a nut/bolt with a spanner, the watch band was melted on his wrist in quick time (12V at the starter can deliver several hundred amps). Lesson - remove all metal from your hands/wrists.

2. A TV repair tech, (this was back in the 60's) was working on a B&W TV when his tie contacted the EHT (yes they wore ties long ago), unfortunately the fashion at that time was for gold coloured stripes in ties, and they were metal stripes, he had a lot of small circular burn marks around his neck as a result.

[G7PSK]

I have just been trying to find the regulations regarding electric fences which some years ago when I was involved with them were limited by law to not more than 15 joules output. Now I cant find any regulations but health and safety recommend not more than 5J but units of 30 J are available in the UK but they work in a different way from the old type units the charge is stored and only released when grounded by an animal rather than a constant pulse.But from the old regulations I have always assumed that it took more than 15 joules of energy to kill a normal person which is why the legal limit was there.

[Psi]

I have always assumed that it took more than 15 joules of energy to kill a normal person which is why the legal limit was there.

There is no real "level" because every person is different.
Not just in skin/flesh resistance but with regard to how strong their heart is and how much current it can take before giving up.

There's also the physical shock of the event (as opposed to electrical).
It's possible to scare a person to death so for anyone who's never experienced an electrical shock before the physical shock of getting a serious electrical shock can be pretty extreme.

It would be interesting to see statistics on deaths from electric shocks and look for any correlation between people who had past experience and knew what electric shocks felt like vs people who didnt.