It is especially troublesome that the GFCI trips only when the laptop is
plugged into the power supply. There is supposed to be substantial
electrical isolation between the so-called "hot" and "cold" grounds
(i.e., between the AC and DC grounds). Unless that isolation (usually an
opto-isolator) is compromised, I can't imagine a mechanism by which the
laptop itself could trip a GFCI, even if it were hanging overboard and
immersed in seawater while connected! Not so for the power supply, of
course.

Larry's theory of a monster-sized common-mode switching noise signal
propagating through the AC lines to trip both GFCI outlets is looking
more attractive. At least that theory supports a mechanism by which
plugging in the laptop causes a GFCI trip: going from no-load to
full-load would cause a change in switching characteristics that might
produce more noise on the line.

20,000 ohms or less of leakage can easily be measured on a DMM and
leaking RF filter capacitors are common. It would be a good idea to
measure leakage with and without the laptop connected to the power supply.

Chuck

w_tom wrote:
All appliances have leakage. Leakage that should be so low
as to not add up to a problem. It is doubtful that laptop is
leaking milliamps on startup. IOW there would be something
else on the circuit leaking so much that just a little laptop
leakage could trip the GFCI.

Unlikely that you will find leakage with the ohm meter.
Try. But eliminating other items from the circuit, then
powering on the laptop will probably provide better
information.

Informative may be a current measurement of that circuit's
ground wire. How much current (in AC milliamps and DC
milliamps) is flowing down that ground wire for various
powered on appliances on that circuit?

First, to have a common mode noise (leakage), the computer
must have separate incoming and outgoing electrical paths.
Incoming is AC electric. What is the outgoing path?

Second, leakage through a resistance is rare. Leakage
occurs more often through reactive devices. That means the
ohm meter will not measure leakage through components whose
conductivity increases with frequency and voltage. IOW these
leaks would appear as high resistance (notice I did not say
impedance) to the meter.

Third, all appliances have leakage. GFCI trip is not just
from one device. Sometimes it is leakage from numerous devices
combined. And yet the meter would test every device and see
no leakage from any of them.

To put numbers to what was posted - appliances typically
leak less than 150 microamps when working normally. Show me
the meter that will measure this 800 Kilohm or 1 Megohm
resistor? Meter will declare infinite resistance (more than
tens of megohms) when leakage says 'impedance' (not
resistance) is lower.

Not mentioned is the brand name and model of that laptop.
Not mentioned is whether that power brick is from the
manufacturer or from a third party. Not mentioned is whether
a useless power strip protector is being used. I am not sure
whether the power brick is two prong or three - another
critical fact.

But this is certain. Whether the laptop starts up or is
powered constantly, the current coming in one wire will always
equal the current going out the other - if hardware is working
properly. For those two currents to be different, then
current must have another (third) path out of the laptop /
power brick system. So where is that third path? And how
much current in that third path.

Notice - without numbers then one can only speculate. Get
numbers. How much current is in the safety ground wire for
the entire GFCI circuit? Does the laptop cause this problem
when it is the only item on that GFCI circuit? Contrary to
what that meter will say - all appliances have some leakage -
which is why the ohm meter cannot report 'real' amount of
leakage.

chuck wrote:
It is especially troublesome that the GFCI trips only when the laptop is
plugged into the power supply. There is supposed to be substantial
electrical isolation between the so-called "hot" and "cold" grounds
(i.e., between the AC and DC grounds). Unless that isolation (usually an
opto-isolator) is compromised, I can't imagine a mechanism by which the
laptop itself could trip a GFCI, even if it were hanging overboard and
immersed in seawater while connected! Not so for the power supply, of
course.

Larry's theory of a monster-sized common-mode switching noise signal
propagating through the AC lines to trip both GFCI outlets is looking
more attractive. At least that theory supports a mechanism by which
plugging in the laptop causes a GFCI trip: going from no-load to
full-load would cause a change in switching characteristics that might
produce more noise on the line.

20,000 ohms or less of leakage can easily be measured on a DMM and
leaking RF filter capacitors are common. It would be a good idea to
measure leakage with and without the laptop connected to the power supply.

Chuck

w_tom wrote:
All appliances have leakage. Leakage that should be so low
as to not add up to a problem. It is doubtful that laptop is
leaking milliamps on startup. IOW there would be something
else on the circuit leaking so much that just a little laptop
leakage could trip the GFCI.

Unlikely that you will find leakage with the ohm meter.
Try. But eliminating other items from the circuit, then
powering on the laptop will probably provide better
information.

Informative may be a current measurement of that circuit's
ground wire. How much current (in AC milliamps and DC
milliamps) is flowing down that ground wire for various
powered on appliances on that circuit?

w_tom wrote:
First, to have a common mode noise (leakage), the computer
must have separate incoming and outgoing electrical paths.
Incoming is AC electric. What is the outgoing path?


Well, that's why I lack enthusiasm for the common mode noise
explanation. But it can and does happen, of course. The question is, as
you suggested, what are the numbers. Simply imagine some capacitance to
ground on the AC lines. This forms the return circuit for the
high-frequency noise that trips the GFCI. The SMPS power bricks are
supposed to contain filtering to prevent this kind of "interference" but
the filter caps might have dried out.


Second, leakage through a resistance is rare. Leakage
occurs more often through reactive devices. That means the
ohm meter will not measure leakage through components whose
conductivity increases with frequency and voltage. IOW these
leaks would appear as high resistance (notice I did not say
impedance) to the meter.


I think the leakage resistance is easier to understand if you consider
that all real-world capacitors and inductors contain some resistance.
You can imagine a real capacitor as a perfect capacitor (no resistance
or inductance) in parallel with a perfect resistor. That resistance is
called series resistance or leakage resistance. Sometimes it is
frequency-sensitive, but that is not what I believe we are seeing here.

The ohmmeter will readily measure this resistance (if it is not too
high) because it is quite real.

Third, all appliances have leakage. GFCI trip is not just
from one device. Sometimes it is leakage from numerous devices
combined. And yet the meter would test every device and see
no leakage from any of them.

Yes, I agree fully. However, it was established in an earlier post that
nothing but the computer was connected to this GFCI receptacle. And that
nothing was connected to the computer.


To put numbers to what was posted - appliances typically
leak less than 150 microamps when working normally. Show me
the meter that will measure this 800 Kilohm or 1 Megohm
resistor? Meter will declare infinite resistance (more than
tens of megohms) when leakage says 'impedance' (not
resistance) is lower.

Not sure what you are saying here. Almost any meter will measure 1
megohm of resistance. I don't believe the leakage causing the problem is
due to a low-reactance condition between the hot wire and the grounding
conductor. I believe the problem may be due to a (relatively) low
resistance between the hot wire and the grounding conductor. I believe
that resistance may be leakage resistance in a capacitor in the power
brick. It is a lot more common, I think, for appliances to develop "low"
resistance leakages to ground and relatively uncommon for them to
develop "low" reactances to ground.


Not mentioned is the brand name and model of that laptop.
Not mentioned is whether that power brick is from the
manufacturer or from a third party. Not mentioned is whether
a useless power strip protector is being used. I am not sure
whether the power brick is two prong or three - another
critical fact.

It appears a 3-prong plug is being used.

But this is certain. Whether the laptop starts up or is
powered constantly, the current coming in one wire will always
equal the current going out the other - if hardware is working
properly. For those two currents to be different, then
current must have another (third) path out of the laptop /
power brick system. So where is that third path? And how
much current in that third path.

Yes. And the only path I can imagine is as described earlier.

Notice - without numbers then one can only speculate. Get
numbers. How much current is in the safety ground wire for
the entire GFCI circuit?

That shouldn't matter. The GFCI will work as well without a ground wire
as with one carrying 20 amps! It is not even connected to the ground
wire. As long as there is no net current in the hot and grounded
conductor pair, the GFCI shouldn't trip.

Does the laptop cause this problem
when it is the only item on that GFCI circuit?

Yes.

Contrary to
what that meter will say - all appliances have some leakage -
which is why the ohm meter cannot report 'real' amount of
leakage.


If we measure a leakage on the power brick of 20,000 ohms, then
regardless of what is happening with other appliances, we have
established that the power brick is capable of tripping the GFCI all by
itself and we need look no further. An alternative to using the ohmmeter
is to use a sensitive clamp-meter (or even an AC ammeter) on the
grounding conductor. Keeping in mind, of course, that these instruments
are not likely to give a meaningful measurement of the high-frequency,
common-mode noise that is the other candidate explanation for the
tripping. An oscilloscope or spectrum analyzer would be more useful.

Since there are no other appliances on that GFCI circuit, then other
appliances are irrelevant to this particular problem.

Hope I haven't added to the confusion.

Chuck

w_tom wrote in :

First, to have a common mode noise (leakage), the computer
must have separate incoming and outgoing electrical paths.
Incoming is AC electric. What is the outgoing path?

The ground in the computer is hooked to the ground in the NMEA bus, the
printer through the printer cable, the computer's own troublesome charger.
Because of any NMEA connections, it's also connected to that AC battery
charger under the quarterberth, which is also hooked to AC ground.

How many paths does it need??


Second, leakage through a resistance is rare. Leakage
occurs more often through reactive devices. That means the
ohm meter will not measure leakage through components whose
conductivity increases with frequency and voltage. IOW these
leaks would appear as high resistance (notice I did not say
impedance) to the meter.

This troublesome computer has a 3-prong grounded power plug, so we may
assume it also has in input double pi line filter, or at least some disc
ceramics in the .01 to .05 uF range between "hot" and neutral and ground.
The ac current differential caused by the input filter's capacitors is more
than enough to cause trips, which is why I wanted him to first plug the
computer into a ground buster to eliminate the connection between the
computer power supply ground and the boat AC ground to isolate this type of
tripping. If the ground buster fixes the problem, he merely leaves it
plugged into the ground buster and goes about his business, occasionally
getting a tingle from the ground on the RS-232C shell, maybe. He'd be
fine.

Then, I was going to have him measure the voltage between the unconnected
ground pin and boat ground to see how hot it was. You can imagine the $24
switching power supply of the bargain laptop has nothing but the finest,
mil-spec line filter parts....totaling, probably, 10 cents, tops....


Third, all appliances have leakage. GFCI trip is not just
from one device. Sometimes it is leakage from numerous devices
combined. And yet the meter would test every device and see
no leakage from any of them.

Maybe nothing else is plugged into this OUTLET GFI. It could serve more
than one outlet from its internal terminals, though. The GFI outlet in my
bathroom services the AC outlet on the side of my house, too.

Notice - without numbers then one can only speculate.

Everything we do on this newsgroup is speculation...an exchange of guesses
and ideas that usually come up with a solution or prod the asking party
into taking a different path to the solution than the one he was taking....

No, NOTHING was plugged into the computer. Certainly no NMEA devices. That
said, I probably lied; I use a wireless mouse that MAY have been plugged in
to a USB port every time.

The laptop is a SONY VGN-A190 with Sony AC brick with a 3 wire plug. In a
year of use, I cannot remember it ever starting up without tripping the
GFI - EXCEPT TODAY!

This is totally unreal. I came up to the boat this afternoon and made sure
everything was turned off and nothing else plugged into the AC - only the
ships battery charger was on, even the DC was off. I fired up the AC
outlets and tuned on the computer, No fault. I turned off the computer,
turned everything on and tried again. Still no fault I tried 4 times and
can't cause it to fault. I'll try again in the morning after it has been
off over night.



"Larry" wrote in message
...
w_tom wrote in :

First, to have a common mode noise (leakage), the computer
must have separate incoming and outgoing electrical paths.
Incoming is AC electric. What is the outgoing path?

The ground in the computer is hooked to the ground in the NMEA bus, the
printer through the printer cable, the computer's own troublesome charger.
Because of any NMEA connections, it's also connected to that AC battery
charger under the quarterberth, which is also hooked to AC ground.

How many paths does it need??


Second, leakage through a resistance is rare. Leakage
occurs more often through reactive devices. That means the
ohm meter will not measure leakage through components whose
conductivity increases with frequency and voltage. IOW these
leaks would appear as high resistance (notice I did not say
impedance) to the meter.

This troublesome computer has a 3-prong grounded power plug, so we may
assume it also has in input double pi line filter, or at least some disc
ceramics in the .01 to .05 uF range between "hot" and neutral and ground.
The ac current differential caused by the input filter's capacitors is
more
than enough to cause trips, which is why I wanted him to first plug the
computer into a ground buster to eliminate the connection between the
computer power supply ground and the boat AC ground to isolate this type
of
tripping. If the ground buster fixes the problem, he merely leaves it
plugged into the ground buster and goes about his business, occasionally
getting a tingle from the ground on the RS-232C shell, maybe. He'd be
fine.

Then, I was going to have him measure the voltage between the unconnected
ground pin and boat ground to see how hot it was. You can imagine the $24
switching power supply of the bargain laptop has nothing but the finest,
mil-spec line filter parts....totaling, probably, 10 cents, tops....


Third, all appliances have leakage. GFCI trip is not just
from one device. Sometimes it is leakage from numerous devices
combined. And yet the meter would test every device and see
no leakage from any of them.

Maybe nothing else is plugged into this OUTLET GFI. It could serve more
than one outlet from its internal terminals, though. The GFI outlet in my
bathroom services the AC outlet on the side of my house, too.

Notice - without numbers then one can only speculate.

Everything we do on this newsgroup is speculation...an exchange of guesses
and ideas that usually come up with a solution or prod the asking party
into taking a different path to the solution than the one he was
taking....

Not unreal. For example, some leakage could be between
wires inside an electric box that become more conductive when
humidity increases. Not conductive enough alone to trip a
GFCI, but conductive enough when combined with leakage from
the laptop.

Also the battery charger was on - but was it charging when
tests were conducted.

All appliances have galvanic isolation. That means (in
theory) low voltage circuits are isolated from AC mains - not
leakage. In reality, even galvanic isolation has microamp
leakage. I have even seen where a low voltage light causes
just enough leakage to create rare and intermittent GFCI trips
only when other appliances were powered just because a
chipmunk chewed into insulation of that low voltage (and
buried) wire.

You have a battery charger on the circuit. How much is it
leaking? The numbers, such as the normal leakage into safety
ground wire where that safety ground wire connects to breaker
box, are so important to those who would answer your posts.

Again, what was humidity then verses today? How long since
the last rain or a last boat hose down? And even what was
powered from the battery when you were testing previously?
Search for alternative circuits for current leakage - and that
even includes the battery charger. GFCI says you have leakage
somewhere. Now all we need do is find that leakage. Its easy
to say - and so damn difficult to put into reality.

Nothing unreal about your problem. Have solved these things
so many times that I have no belief in ghosts. But then
others long since give up before I do. My sympathies for your
frustration and my envy for your challenge.

JohnHH wrote:
No, NOTHING was plugged into the computer. Certainly no NMEA
devices. That said, I probably lied; I use a wireless mouse
that MAY have been plugged in to a USB port every time.

The laptop is a SONY VGN-A190 with Sony AC brick with a 3 wire
plug. In a year of use, I cannot remember it ever starting up
without tripping the GFI - EXCEPT TODAY!

This is totally unreal. I came up to the boat this afternoon
and made sure everything was turned off and nothing else
plugged into the AC - only the ships battery charger was on,
even the DC was off. I fired up the AC outlets and tuned on
the computer, No fault. I turned off the computer, turned
everything on and tried again. Still no fault I tried 4
times and can't cause it to fault. I'll try again in the
morning after it has been off over night.

Thanks for your feed back. I have been getting it to fail again by
unplugging the adapter from the wall, waiting a minute and plugging it back
in. It seems like it happens far more consistantly if any other 110
appliance is running - like the refrigerator or heater or network router,
but not 100% consistant. It's raining cats and dogs today so I am sure I
have experienced it in every weather condition.

By the way, my digital multimeter measures about 40 meg ohms before it goes
to infinity.

I am one of those who is going to give it up. I don't think it is serious
and I'm losing interest. The problem will go away anyway when I replace
the AC-DC converter with a DC-DC one. I'm not really frustrated, I was just
curious. I fully expected someone to say, oh yea, that always happens with
those power bricks.

I'm going to bag this thing for now.

Thanks to all of you who replied, it's been informative.

John




"w_tom" wrote in message
...
Not unreal. For example, some leakage could be between
wires inside an electric box that become more conductive when
humidity increases. Not conductive enough alone to trip a
GFCI, but conductive enough when combined with leakage from
the laptop.

Also the battery charger was on - but was it charging when
tests were conducted.

All appliances have galvanic isolation. That means (in
theory) low voltage circuits are isolated from AC mains - not
leakage. In reality, even galvanic isolation has microamp
leakage. I have even seen where a low voltage light causes
just enough leakage to create rare and intermittent GFCI trips
only when other appliances were powered just because a
chipmunk chewed into insulation of that low voltage (and
buried) wire.

You have a battery charger on the circuit. How much is it
leaking? The numbers, such as the normal leakage into safety
ground wire where that safety ground wire connects to breaker
box, are so important to those who would answer your posts.

Again, what was humidity then verses today? How long since
the last rain or a last boat hose down? And even what was
powered from the battery when you were testing previously?
Search for alternative circuits for current leakage - and that
even includes the battery charger. GFCI says you have leakage
somewhere. Now all we need do is find that leakage. Its easy
to say - and so damn difficult to put into reality.

Nothing unreal about your problem. Have solved these things
so many times that I have no belief in ghosts. But then
others long since give up before I do. My sympathies for your
frustration and my envy for your challenge.

JohnHH wrote:
No, NOTHING was plugged into the computer. Certainly no NMEA
devices. That said, I probably lied; I use a wireless mouse
that MAY have been plugged in to a USB port every time.

The laptop is a SONY VGN-A190 with Sony AC brick with a 3 wire
plug. In a year of use, I cannot remember it ever starting up
without tripping the GFI - EXCEPT TODAY!

This is totally unreal. I came up to the boat this afternoon
and made sure everything was turned off and nothing else
plugged into the AC - only the ships battery charger was on,
even the DC was off. I fired up the AC outlets and tuned on
the computer, No fault. I turned off the computer, turned
everything on and tried again. Still no fault I tried 4
times and can't cause it to fault. I'll try again in the
morning after it has been off over night.

"JohnHH" wrote in
news:1128650796.e0315dfd2c551b142d05eb9a0bd4ebdf@t eranews:

This is totally unreal. I came up to the boat this afternoon and made
sure everything was turned off and nothing else plugged into the AC -
only the ships battery charger was on, even the DC was off. I fired
up the AC outlets and tuned on the computer, No fault. I turned off
the computer, turned everything on and tried again. Still no fault I
tried 4 times and can't cause it to fault. I'll try again in the
morning after it has been off over night.


See there? I knew we could fix it on the newsgroup!...(c;

Ok, next problem to solve.......

--
Larry

Another lie, the power brick is only two wire and the prongs aren't
polarized. Funny how different things look when you look at them. Maybe it
was the Dell that had the three wire plug.

"Larry" wrote in message
...
w_tom wrote in :

First, to have a common mode noise (leakage), the computer
must have separate incoming and outgoing electrical paths.
Incoming is AC electric. What is the outgoing path?

The ground in the computer is hooked to the ground in the NMEA bus, the
printer through the printer cable, the computer's own troublesome charger.
Because of any NMEA connections, it's also connected to that AC battery
charger under the quarterberth, which is also hooked to AC ground.

How many paths does it need??


Second, leakage through a resistance is rare. Leakage
occurs more often through reactive devices. That means the
ohm meter will not measure leakage through components whose
conductivity increases with frequency and voltage. IOW these
leaks would appear as high resistance (notice I did not say
impedance) to the meter.

This troublesome computer has a 3-prong grounded power plug, so we may
assume it also has in input double pi line filter, or at least some disc
ceramics in the .01 to .05 uF range between "hot" and neutral and ground.
The ac current differential caused by the input filter's capacitors is
more
than enough to cause trips, which is why I wanted him to first plug the
computer into a ground buster to eliminate the connection between the
computer power supply ground and the boat AC ground to isolate this type
of
tripping. If the ground buster fixes the problem, he merely leaves it
plugged into the ground buster and goes about his business, occasionally
getting a tingle from the ground on the RS-232C shell, maybe. He'd be
fine.

Then, I was going to have him measure the voltage between the unconnected
ground pin and boat ground to see how hot it was. You can imagine the $24
switching power supply of the bargain laptop has nothing but the finest,
mil-spec line filter parts....totaling, probably, 10 cents, tops....


Third, all appliances have leakage. GFCI trip is not just
from one device. Sometimes it is leakage from numerous devices
combined. And yet the meter would test every device and see
no leakage from any of them.

Maybe nothing else is plugged into this OUTLET GFI. It could serve more
than one outlet from its internal terminals, though. The GFI outlet in my
bathroom services the AC outlet on the side of my house, too.

Notice - without numbers then one can only speculate.

Everything we do on this newsgroup is speculation...an exchange of guesses
and ideas that usually come up with a solution or prod the asking party
into taking a different path to the solution than the one he was
taking....

JohnHH wrote:
Another lie, the power brick is only two wire and the prongs aren't
polarized. Funny how different things look when you look at them. Maybe it
was the Dell that had the three wire plug.

So, progress, eh?

It's easier to see when you use your eyes. It's easier to think
when you don't put garbage into your brain.

So, with a 2 wire cableset, unpolarised, did you try it plugged in
both ways?

Does it do "it" both ways? Is there a great bloody gash and bare
wires hanging out the cableset? Covered with black tape? "Sealed up"
with smunge from years of handling? Smunge is that black goo that
costs so much when spread thinly, hundreds of coats, on "antique"
furniture. It comes from human hands and exhalations, sometimes
called "patina."

If only one way, there is probably a .01 uf "decoupling" capacitor
from charger "chassis" ground to one side of the supply, depending
on which way it's plugged in. It will be leaky, or possibly open.
Replace it with a good one. Don't ask how I know, I've been told I
get too technical. The device might have internal transformer caused
eddy currents or switching transients capable of tunnelling through
the insulation into the ether.

Otherwise, there ain't no way this thing can pop a gfci unless there
is leakage through the electrocutee. That's you. Try it again with
rubber boots on, and rubber gloves. If that solves the problem look
for something like an almost dried puddle of puke somewhere near
where your hand contacts the case or perhaps a salty damp berth
cushion and sweaty underpants, all too close to the wire mesh ssb
"ground" plane or something really strange.

Reroute the power cableset, away from where it usually is. Try
again. Does it happen at only one one outlet / location?

Laugh if you want, but this is the real world talking, and truth is
stranger than friction, as they try to say...

Does this thing have an antistatic plastic case, possibly conductive
enough to allow this leakage? Paint it with a good insulating
varnish, or keep it in a plastic baggie.

Buy a different brand of gfci, maybe a cheaper one will serve better.

Terry K


"Larry" wrote in message
...

w_tom wrote in :


First, to have a common mode noise (leakage), the computer
must have separate incoming and outgoing electrical paths.
Incoming is AC electric. What is the outgoing path?

The ground in the computer is hooked to the ground in the NMEA bus, the
printer through the printer cable, the computer's own troublesome charger.
Because of any NMEA connections, it's also connected to that AC battery
charger under the quarterberth, which is also hooked to AC ground.

How many paths does it need??


Second, leakage through a resistance is rare. Leakage
occurs more often through reactive devices. That means the
ohm meter will not measure leakage through components whose
conductivity increases with frequency and voltage. IOW these
leaks would appear as high resistance (notice I did not say
impedance) to the meter.

This troublesome computer has a 3-prong grounded power plug, so we may
assume it also has in input double pi line filter, or at least some disc
ceramics in the .01 to .05 uF range between "hot" and neutral and ground.
The ac current differential caused by the input filter's capacitors is
more
than enough to cause trips, which is why I wanted him to first plug the
computer into a ground buster to eliminate the connection between the
computer power supply ground and the boat AC ground to isolate this type
of
tripping. If the ground buster fixes the problem, he merely leaves it
plugged into the ground buster and goes about his business, occasionally
getting a tingle from the ground on the RS-232C shell, maybe. He'd be
fine.

Then, I was going to have him measure the voltage between the unconnected
ground pin and boat ground to see how hot it was. You can imagine the $24
switching power supply of the bargain laptop has nothing but the finest,
mil-spec line filter parts....totaling, probably, 10 cents, tops....


Third, all appliances have leakage. GFCI trip is not just
from one device. Sometimes it is leakage from numerous devices
combined. And yet the meter would test every device and see
no leakage from any of them.

Maybe nothing else is plugged into this OUTLET GFI. It could serve more
than one outlet from its internal terminals, though. The GFI outlet in my
bathroom services the AC outlet on the side of my house, too.


Notice - without numbers then one can only speculate.

Everything we do on this newsgroup is speculation...an exchange of guesses
and ideas that usually come up with a solution or prod the asking party
into taking a different path to the solution than the one he was
taking....