Short on my metal case (ouch)

On Tuesday, June 23, 2015 at 5:06:13 AM UTC+8, Paul wrote:


I would be very careful where you stick the meter probes.

Never use a meter without glancing at the spec sheet first!

Mine is:

1000 VDC max
750 VAC max

and it has an AC/DC button that you press, to switch from
one kind of measurement to the other. If you measure AC
while on the DC setting, then you're not going to get
the right reading. I would feel relatively safe measuring
a 230 VAC source. I probably would not use the meter on
much more than that (I wouldn't use my meter on any
industrial 440V circuits).

DC is what comes from batteries.

AC is what comes from the wall.

When jamming probes into a high voltage circuit, you want
to be familiar enough with it, to know where any metal
grounds are located. You'd feel a fool, if there was
a blinding flash, and the tip was blown off your meter
probe, when you touched a "hot" conductor and an adjacent
building ground. The breaker in the panel provides
little in the way of personal protection. You could
be in a lot of trouble, and the breaker would just
sit there and not trip. Don't depend on the breaker
being there when it counts. If you make a bad mistake,
you have to deal with it directly, at the point of
application.

I dropped a 120V *male* plug with live voltage
on it once, onto the jaws of a pair of lineman pliers,
and the spark was powerful enough, to take a gouge
out of the pliers. (And in that case, blow a fuse,
and put me in the dark.)

*******

This is a simplified view of your computer.

~AC ------ filter -------+-- Main stage --- rectifier --- 3.3VDC
circuit | Transformer --- rectifier --- 5.0VDC
~AC ------ leaks to gnd | Isolated ATX --- rectifier --- 12.0VDC
| |
| +-- Standby stage --- rectifier --- 5VSB
Safety -------+ Transformer
Ground | Isolated
+---------------------------------+---------- DC Ground
|
+------- Metal chassis

The ATX supply, has transformer isolation between input and
output. This is an attempt to keep you safe from "conducted"
high voltages. The outside of the ATX supply has a sticker
that says "HiPot Tested" and that is proof that when 1100V
or so was applied between the AC plug pins, and any DC output,
no current could flow. That's a test of the transformer
lacquer insulation. So in a sense, there is an "air gap"
between the AC input side and the DC output side. We hope
never to see a "HiPot Failure", where something leaks through.

Notice, that all the grounds are connected together. The
transformer has nothing to do with the ground.

On the front end, the supply should have a three prong plug.
And a connection from the "leaky leg" of the filter
circuit, to the Safety Ground connection. When this happens,
the user will not feel a shock, when touching the chassis.
The leakage current flows back into the Safety Ground. And
since the Safety Ground wire has a 15 amp rating, 0.002 amps
is hardly going to matter.

The supply has three operating states.

Unplugged or switched off at back = Main stage OFF
Standby stage OFF

Switched on at the back = Main stage OFF
Standby stage ON

PS_ON# from motherboard = Main stage ON (fans run)
Standby stage ON

When you finish with the computer at the end of the day,
and select standby or shutdown, the standby stage
continues to run all night long. There will be 5V on the
5VSB lead on the main ATX harness. On modern computers,
a certain number of USB plugs will also have this +5VSB
present. You can charge your Apple iPad from those ports.

Now, I don't know where you measured the "2V", but the
+5VSB, if that's what you were measuring, would be
5.00 volts DC.

*******

OK, now let's go to Rays house, where the outlet doesn't
have a safety ground. Ray attempts to make a measurement
from computer chassis, to a "ground" in the house. An
example of an unreliable ground, is the house cold water
pipe. While a cold water pipe exits the building and
travels through the earth, sometimes there is plastic
pipe somewhere inside the house, that breaks the circuit.
Care must be taken in the house, to make sure a bridging
strap is fastened between external copper pipe, and an
internal copper pipe. In a house using CPVC plastic
plumbing, getting a "ground" this way, isn't possible.

~AC ------ filter -------+-- Main stage --- rectifier --- 3.3VDC
circuit | Transformer --- rectifier --- 5.0VDC
~AC ------ leaks to gnd | Isolated ATX --- rectifier --- 12.0VDC
| |
| +-- Standby stage --- rectifier --- 5VSB
//--+ Transformer
| Isolated
+---------------------------------+---------- DC Ground
|
Copper -------------- meter on AC ------------ +------- Metal chassis
plumbing
Ground (cold water)

I'm only using the copper pipe as a "reference",
so I have something to touch the meter to.

If I do this, I will see some magnitude of AC voltage.

Perhaps this is the AC voltage you were seeing.

Your meter was set to AC, you touched the chassis with
one lead, and the other lead was connected to something
at ground potential.

Yes, you'll see a voltage.

The worst I've ever seen, in a lab, was ~60V between
two computer chassis. One computer chassis, an idiot
*cut* the safety ground in a distribution rail. Paul
touches the computer on that rail, touches a properly
grounded computer, and gets a *shock*. An arm to arm shock.
It took me a while to find the hidden wiring fault.
Some engineer had cut safety ground, in order to do
a floating scope measurement. Which is a *stupid* thing to do.
You don't **** with building wiring... Doh.

Anyway, seeing a small AC voltage between the chassis
of the improperly grounded computer and some ground
reference, would not be abnormal. And it's the maximum
level of current flow, that helps determine how dangerous
it is. For example, if the leakage was microamps, you
might measure 60VAC with the meter, but if you put
a human body in the circuit, the voltage would drop
under load, to an unmeasurable value. The meter
typically is measuing an "unloaded" leakage voltage.
As soon as you put some load on the circuit, you
get some idea whether the leaker is "dangerous" or
"quite weak".

I can get the same effect, when using an oscilloscope
on Ethernet wires. You will find "tens of volts" of
AC noise on Ethernet wires, but the power source in
that case is extremely weak, and easily grounded out.
As soon as you touch it with your finger, there is
no longer tens of volts there.

Now, let's overlay the power supply states, with
the "shock potential".

Unplugged = 0 VAC
switched off at back = (Not sure)

Switched on at the back = Leakage voltage

PS_ON# from motherboard = Leakage voltage

The filter is fairly close to the front of the supply,
and we don't know if the main switch is an SPST or
an SPDT. I assume both Line and Neutral open for safety,
but I'm not sure.

Once you're switched on at the back, then I assume
you can see the leakage voltage on the chassis.

And don't be surprised if the meter says it is
a significant voltage. What you don't know at
the time, is what the leakage current would be.
The ATX power supply, should be similar in spec
on leakage to that "metal box filter" I showed.

*******

The PWR button on the front of the computer, the front
of that button is generally isolated from the back two
contacts. Only premium computers use metal for
everything, and you might get a chassis shock when
touching the power button. But the front of the
button, makes no circuit contact with the back
two wires.

+5VSB -----------+
|
GND +5VSB 2.2K ohm resistor
| | |
D D | |
Motherboard --+------+-----+---- PWR# |--+ Push
---------- GND |--+ Button
| (Momentary)

Now, if you were to touch the PWR# conductor, it's a
low voltage logic circuit. It does have a slight DC
voltage (pulled up to +5VSB). But cannot source much
current.

If you caused some of that AC leakage to flow, it
could slightly modulate the voltage on the PWR# wire.
But the circuit will generally have diode clamps
(the letter "D" being a stand-in for a diode
symbol I was too lazy to draw). Generally, the PWR# wire
cannot go outside the range of the two clamp points.
To cause the diode to blow, might take 5 to 70mA,
depending on where the diode is located.

There really isn't a reason for you to be touching
PWR#, but I expect the circuit would survive the
contact. As the leakage might be set at a couple
milliamps max.

The logic circuit on the motherboard is
latching, so that acts as a filter of sorts.
The only thing that might happen, is perhaps
the four second timer might not trigger properly,
if the amplitude of the noise on that wire was
high enough.

The important thing to note though, is there is
a bit of an "air gap" between the button
front surface, and the circuitry behind it.

*******

I still recommend the usage of a Safety Ground,
because I don't consider a visitor touching
your computer and falling on the floor, to be
a good thing. And an electrician should be
consulted, as to how best to get a safety
ground. There is probably a right way to do it,
and I don't know the answer. I'm not an
electrician, and don't know the national code.

I have had good luck, getting advice from
the building inspector number at City Hall.
The guy on the other end of the line said
"I didn't tell you this, but...". They will
provide anonymous advice, at least in an attempt
to prevent you from doing something really
stupid. Getting an electrician to do the work,
makes it "official". The electrician can only
do things "according to code". In some cities,
you are not allowed to do your own electrical
work, and only an electrician is allowed to do it.
And in some cases, an inspection is required too.

I don't know of a practical way to make the
computer safe, by insulating it. As you're always
touching the chassis, handling wires, USB ports,
and so on. Only the Safety Ground does a good
job of taming the leakage. If you modified the
filter inside the ATX supply, you might notice
a degradation in your TV reception. There can
be enough RF energy coming out of the modified
ATX supply, to cause an ADSL modem to drop sync.
So that option is out too. Modding the ATX supply
might generate a little too much electrical noise.

Paul

Thanks Paul!

Here are my followup questions, given the above. I do have a science background but not the practical experience you have.

Without a ground in the house (they use PVC pipe here, and, as an aside, they don't even shut off the water when doing repairs, since nobody knows where the main water shutoff is, if there even is one, but use PVC cement and make repairs while the water is running full blast, apparently PVC cement cures even in running water, a first for me but I've seen them do it!), and with power cord having only two prongs (hot/cold +/-) and not a ground wire, and the PC power cord connected to a UPS supply, which is connected to ungrounded mains:

1) is it possible that you can switch off the PC 600W Power Supply (which includes the transformer), and still you can have some small voltage at the mobo? As I type this I just checked the metal chassis (outside case) of the system with the PC 600W PS on, and indeed a small shock like a very tiny pin sticking your hand occasionally when you touch the metal chassis. Not painful, just a tiny pin prick and if you have dry hands no such prick. However, it seems to be (or it's my palms getting more sweaty) getting worse day by day.

2) if 1) is not possible, it points to a problem that needs to be fixed; if 1) is possible, then we can move on. ***UPDATE: I will assume it's a problem, since in years past I don't recall it being this bad, so let's move to the step of what I can salvage from the system, whether it's Mobo, components on mobo, and power supply, or just scrap the power supply?***

I just noticed now (just happened) that the BIOS is not booting up (black screen). It could be that the extra current fried or reset the BIOS. Followup questions given the BIOS is not working, only fan and no beeps from the on-board sound speaker (a tiny little speaker that beeps one, two three times etc)

1) The problems seem to have started when I installed a Korean "Ice-man" power supply that is 600 W instead of the normal 400 W PS that I had. I am starting to suspect the "Ice-Man" was refurbished (very common here, as I said in other posts, all the stuff that is rejected in First World countries ends up in southeast Asia, refurbished, and sold "as new or 'just like new'".. Even and especially clothes--LOL I have T-shirts from investment banks in NYC that I wear) and defective

2) should I get new tower case (and switch) and downgrade to a new-ish power supply that is 400 W? (according to http://www.extreme.outervision.com/p...ulatorlite.jsp I only need 34 W of power for the simple 2 drive HDD SATA and CD/DVD with simple graphics card i5 system that I use)?

3) Should I assume the otherwise excellent (I think) ASUS P8H67 Motherboard from around 2011 and upgraded via flash RAM is at fault? I'd like to keep this mobo since the i5 chip goes with it. Can a motherboard develop an electrical fault? I doubt it. I think the Power Supply is defective but I'm open to suggestion.

RL

On Tuesday, June 23, 2015 at 11:45:38 AM UTC+8, RayLopez99 wrote:

3) Should I assume the otherwise excellent (I think) ASUS P8H67 Motherboard from around 2011 and upgraded via flash RAM is at fault? I'd like to keep this mobo since the i5 chip goes with it. Can a motherboard develop an electrical fault? I doubt it. I think the Power Supply is defective but I'm open to suggestion.

RL

Let me clarify that when I say 'power supply is defective' it could well be that when you are using a powerful 600 W PS with transformer, in a 220 V country like the Philippines, with no ground, you are 'playing with fire' in that strange things can happen? It could be that with a smaller PS (say 250 to 400 W) you'd not have this? Or, it could simply be that the PS was indeed defective,that this problem was always occurring (I've had this 600W PS installed by a shady little PC shop in Manila, and they are famous for re-selling old stuff as new (even stuff I donated to them, they jokingly told me they resold it as new)), and I just did not notice it since I don't often touch the case and the PS was not installed by me but by a shop.

Strange stuff that comes from --again--junk in the Third World resold as new. It's such a problem that for my HDD, after several (two) failures within a year of 'new' or 'newish' HDDs, that I recently had a friend send me a brand new HDD from the USA, that I will install myself. I hope the airplanes don't run on 'refurbished' parts here in southeast Asia, but given the travails of Malaysia Airlines, and the fact the two big airlines in the Philippines are banned from traveling to Europe / USA, you can draw your own conclusions. Typically this kind of problem is not limited to one industry but 'endemic'.

RL
RL

RayLopez99 wrote:
On Tuesday, June 23, 2015 at 11:45:38 AM UTC+8, RayLopez99 wrote:

3) Should I assume the otherwise excellent (I think) ASUS P8H67 Motherboard from around 2011 and upgraded via flash RAM is at fault? I'd like to keep this mobo since the i5 chip goes with it. Can a motherboard develop an electrical fault? I doubt it. I think the Power Supply is defective but I'm open to suggestion.

RL

Let me clarify that when I say 'power supply is defective' it could well be that when you are using a powerful 600 W PS with transformer, in a 220 V country like the Philippines, with no ground, you are 'playing with fire' in that strange things can happen? It could be that with a smaller PS (say 250 to 400 W) you'd not have this? Or, it could simply be that the PS was indeed defective,that this problem was always occurring (I've had this 600W PS installed by a shady little PC shop in Manila, and they are famous for re-selling old stuff as new (even stuff I donated to them, they jokingly told me they resold it as new)), and I just did not notice it since I don't often touch the case and the PS was not installed by me but by a shop.

Strange stuff that comes from --again--junk in the Third World resold as new. It's such a problem that for my HDD, after several (two) failures within a year of 'new' or 'newish' HDDs, that I recently had a friend send me a brand new HDD from the USA, that I will install myself. I hope the airplanes don't run on 'refurbished' parts here in southeast Asia, but given the travails of Malaysia Airlines, and the fact the two big airlines in the Philippines are banned from traveling to Europe / USA, you can draw your own conclusions. Typically this kind of problem is not limited to one industry but 'endemic'.

RL
RL


The power supply is relatively oblivious to the leakage
problem. The power circuits are unaware of what the filter
is doing on the front end. The supply is also unlikely to
be damaged by the leakage. The main symptom, is the user
getting a mild shock.

While it is possible you could have a Hipot failure (transformer
insulation), I would think the result would be spectacular if
that happened. Probably Ray would be found on the other side
of the room. You describe a barely detectable leakage.

You can switch back to your previous 400W supply, if it is still
working, as a test. If the actual problem is the 600W supply
isn't putting out proper voltages on the DC outputs, that
could stop things from working. A test with the previous
supply might tell you something.

There's no reason to suspect the Iceman, unless you had
some symptoms along the way. Smoke. Bad smell. Sizzling sound.
Any of these, before the computer would not start, would
be important symptoms to remember. I've had two supplies here
that gave advanced warning, and one of them made a cute little
puff of smoke (at first I thought it was dust being kicked
out of the system by the fan at power up, but it turned out
later that was actually smoke).

Paul

The ohm symbol O

JT



Loren Pechtel wrote:

On Mon, 22 Jun 2015 11:03:29 -0700 (PDT), RayLopez99
wrote:

2) I forget 2. I guess I would ask Loren P how I can use the
"resistance meter" of the multimeter.

It's usually marked with a symbol that looks rather like a horseshoe
with the opening pointing down.

Hmmm I pasted in the ohm symbol but it showed up as an "O"

Let's try this: O alt-234


JT wrote:


The ohm symbol O

JT



Loren Pechtel wrote:

On Mon, 22 Jun 2015 11:03:29 -0700 (PDT), RayLopez99
wrote:

2) I forget 2. I guess I would ask Loren P how I can use the
"resistance meter" of the multimeter.

It's usually marked with a symbol that looks rather like a horseshoe
with the opening pointing down.

Oh well, I tried

JT wrote:

Hmmm I pasted in the ohm symbol but it showed up as an "O"

Let's try this: O alt-234


JT wrote:


The ohm symbol O

JT



Loren Pechtel wrote:

On Mon, 22 Jun 2015 11:03:29 -0700 (PDT), RayLopez99
wrote:

2) I forget 2. I guess I would ask Loren P how I can use the
"resistance meter" of the multimeter.

It's usually marked with a symbol that looks rather like a
horseshoe with the opening pointing down.

On Saturday, June 20, 2015 at 9:52:09 PM UTC+8, RayLopez99 wrote:


Just to close this thread: the mobo / BIOS stopped working after the first post here, I took it to a local shop, and they found the power supply, which I had installed less than a year ago, was defective. They replaced it, my BIOS and mobo worked (amazingly) and everything is back to normal, no jolt when you touch the outside of the tower case, and I have a working On/Off push button on my tower case again.

RL

On Wed, 24 Jun 2015 00:49:59 +0000 (UTC), "JT"
wrote:


The ohm symbol O

That came through here as a plain O. That's why I described it rather
than trying to actually embed it--no knowing exactly what font support
there will be on the other end.


JT



Loren Pechtel wrote:

On Mon, 22 Jun 2015 11:03:29 -0700 (PDT), RayLopez99
wrote:

2) I forget 2. I guess I would ask Loren P how I can use the
"resistance meter" of the multimeter.

It's usually marked with a symbol that looks rather like a horseshoe
with the opening pointing down.

Loren Pechtel wrote:
On Wed, 24 Jun 2015 00:49:59 +0000 (UTC), "JT"
wrote:

The ohm symbol O

That came through here as a plain O. That's why I described it rather
than trying to actually embed it--no knowing exactly what font support
there will be on the other end.

What happens if you just find the symbol in some
text and just paste it ?

1000 Ω

Let's see how that comes out.

Paul

On Sun, 28 Jun 2015 01:18:00 -0400, Paul wrote:

Loren Pechtel wrote:
On Wed, 24 Jun 2015 00:49:59 +0000 (UTC), "JT"
wrote:

The ohm symbol O

That came through here as a plain O. That's why I described it rather
than trying to actually embed it--no knowing exactly what font support
there will be on the other end.

What happens if you just find the symbol in some
text and just paste it ?

1000 ?

Let's see how that comes out.

Paul

It doesn't look quite the same but it's still showing as a circle, not
an Ohm symbol.