Fried ethernet on G5 iMac

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

Jeffrey Goldberg - 23 Aug 2006 04:31 GMT

A few hours ago we came up from a power outage due to a thunderstorm.

My wife's G5 iMac (more than a year old and not under Apple Care)
appears to have lost its ether.

I get an active light on the switch that the ether is connected to, but
no matter how I configure the network preferences I get absolutely no
network connections. Network Diagnostics tell me to reboot the switches
and routers in between, but they are all fine. I've tested cables. And
another machine with the same configuration, works fine where the iMac
doesn't.

As a temporary solution, I've brought back the household wireless
network, so she is connected that way. But I would like to get the
ether fixed/replaced.

How difficult/expensive is that to do?

-j

Signature

Jeffrey Goldberg http://www.goldmark.org/jeff/
I rarely read top-posted, over-quoted or HTML posts
My Reply-To address is valid.

Reply to this Message

John Snow - 24 Aug 2006 03:43 GMT

It's probably a motherboard swap.

> A few hours ago we came up from a power outage due to a thunderstorm.
>
[quoted text clipped - 15 lines]
>
> -j

Reply to this Message

Jeffrey Goldberg - 24 Aug 2006 14:56 GMT

> It's probably a motherboard swap.

That's what I feared. But the solution turned out to be to do a PMU
reset. So all is well now.

Cheers,

-j

Signature

Jeffrey Goldberg http://www.goldmark.org/jeff/
I rarely read top-posted, over-quoted or HTML posts
My Reply-To address is valid.

Reply to this Message

Fred Moore - 24 Aug 2006 17:22 GMT

> > It's probably a motherboard swap.

Yes. The iMac is highly integrated.

> That's what I feared. But the solution turned out to be to do a PMU
> reset. So all is well now.

Hi Jeffery,

Just so I get this right, you're saying that the power surge came in
from the outside through, say, a cable modem then a router and then
zapped the ethernet port on the iMac but nothing in between? That would
be remarkable. Usually, one or more of the intervening devices would
blow before the computer port, stopping the surge if you're lucky. Even
if the cables are okay, there could more likely be a problem inside one
of the other devices in the line.

Before you spring for a motherboard, I'd at least remove the router and
see if you can get a signal straight from the cable modem, though it
could be blown too (and most likely since it's first in the chain) and
hard to identify unless you get the cable company to come and check it
out. Sometimes this is a free service call, especially if there was a
known power outage. I mean, how do you _know_ you have a good signal at
the ethernet output of the cable modem? The way you described this, I
think there's a good chance the problem is _not_ with the computer.

Finally, for the future, you could buy a surge strip with an ethernet
filter. I like TrippLite IsoBars the best. They are a bit pricey, but
much better built and with a better guarantee than the others.

Good luck,

--Fred

Reply to this Message

Jeffrey Goldberg - 24 Aug 2006 20:00 GMT

> Just so I get this right, you're saying that the power surge came in
> from the outside through, say, a cable modem then a router and then
> zapped the ethernet port on the iMac but nothing in between? That would
> be remarkable.

I agree. That is extremely unlikely. I did check all of the other bits
of network equipment and that was fine. Basically, I'd misdiagnosed the
problem. After I posted, my wife commented that the machine wouldn't
sleep any more. Also (after I'd first posted) I'd finally found the
install disks and ran Apple's hardware check. Both the extended and
quick tests came to a halt a bit before 2 and half minutes into the
scan, while testing the logic board.

The thunderstorm did lead to a power interruption, and when the power
came back up, it didn't come back up cleanly. So there was no surge
through the network cable. Resetting the power management unit
(Plugging the machine in to the power while the on/off button is
depressed) seems to have fixed it.

> Before you spring for a motherboard, I'd at least remove the router and
> see if you can get a signal straight from the cable modem,

Not an issue. Other machines in the house can see the network just
fine. Believe me, I tried very hard to find something cheap to replace
that was broken: I used cable testers; I swapped things in and out. It
wasn't just an Internet issue, I couldn't see my own (in house) DHCP
server or router; I checked logs on the DHCP server; I ran ethereal on
the DHCP server to look for anything from the right MAC address. I
manually configured TCP options; I put an iBook on the net where the
iMac goes, and the iBook booted fine from DHCP. I did get the iMac
working over wireless. I really did isolate it to ether on the iMac.

Ether really wasn't working on the iMac, but it turned out to be part of
a larger problem that was cheap and easy to fix after a 50 USD call to
Apple Support. I've now bought AppleCare for a newer machine that is
still eligible.

-j

Signature

Jeffrey Goldberg http://www.goldmark.org/jeff/
I rarely read top-posted, over-quoted or HTML posts
My Reply-To address is valid.

Reply to this Message

w_tom - 25 Aug 2006 10:27 GMT

There is not filter that will stop or absorb such surges. In fact, a
better 'filter' already exists on all Ethernet ports. But then no
such filter exists in Tripplite Isobars.

It is possible for a surge to enter on cable, not damage modem, and
overwhelm (damage) ethernet port. But the analysis is not complete
if the rest of that path to earth is not defined.

There is no reason for surges to enter on cable if the cable was
properly earthed (not any ground- earth ground) as even required by
electrical codes. If cable was not properly earthed to same earthing
electrode used by AC electric and telephone, then you have a serous
complaint with the cable installers who should know better. Meanwhile,
the homeowner takes ultimate responsibility. The homeowner typically
provides single point earthing. To avoid future problems, fix the
defective cable ground where cable wire enters building.

> Just so I get this right, you're saying that the power surge came in
> from the outside through, say, a cable modem then a router and then
[quoted text clipped - 20 lines]
>
> --Fred

Reply to this Message

Fred Moore - 25 Aug 2006 18:22 GMT

> > Just so I get this right, you're saying that the power surge came in
> > from the outside through, say, a cable modem then a router and then
[quoted text clipped - 16 lines]
> > filter. I like TrippLite IsoBars the best. They are a bit pricey, but
> > much better built and with a better guarantee than the others.

> There is not filter that will stop or absorb such surges. In fact, a
> better 'filter' already exists on all Ethernet ports. But then no
> such filter exists in Tripplite Isobars.

Agreed, no surge suppressor will stop a direct or very nearby lightning
strike. What is good about the TrippLite's products is that the
guarantee, unlike most other surge suppressors, actually covers a direct
strike. In other words, yes, your equipment behind the IsoBar is toast,
but TrippLite will reimburse you, unlike most (all?) other surge
suppressor manufacturers.

Since I last checked the TrippLite web site (at least a couple of years
ago), they have improved their suppressor offering. They now offer a
surge suppressor which filters the _cable_ not the ethernet, thus
increasing the equipment protected.

http://www.tripplite.com/products/product.cfm?productID=2814

> It is possible for a surge to enter on cable, not damage modem, and
> overwhelm (damage) ethernet port. But the analysis is not complete
> if the rest of that path to earth is not defined.

Yes, certainly one must have a valid ground for the system. The thing
is, we don't know exactly from where the surge entered the system. We're
assuming (reasonably) that it came in through the cable. It could also
have entered through the powersupply of the router, especially if it was
plugged into a different outlet.

> There is no reason for surges to enter on cable if the cable was
> properly earthed (not any ground- earth ground) as even required by
[quoted text clipped - 4 lines]
> provides single point earthing. To avoid future problems, fix the
> defective cable ground where cable wire enters building.

[the following is said with all politeness, no flame intended] The above
is not strictly true. I have personally seen instances where a well
protected, well grounded building has been hit by lightning; yet a
computer with a surge suppressor (an IsoBar, in fact) was not damaged
while another computer with an inferior suppressor was.

What is crucial is how much energy from a strike is dissipated BEFORE it
gets to an electronic device you want to protect. SOME surge will ALWAYS
get through. As they say with human electrocutions: it's not the
voltage, but the current which kills you. So too with electronic
equipment; electronic equipment is just much more delicate.

A well protected building will have a good earth ground, as you mention;
and ALL wiring electric--electric line, cable, telephone, or
whatever--will be tied to that ground, again, as you mention. But that
is just the start. A typical 'lightning arrestor' for wires entering a
building consists of a ground to the system for the ground wire of the
incoming service (awkwardly stated, but that's the best I can do),
followed by a spark-gap device from the powered wires (e.g.. the center
wire of coax or across all line voltage wires, including neutral, to
ground). The spare gap typically only operates above 1000 volts or more.
Voltage below this threshold doesn't allow current to jump the gap and
dissipate harmlessly. Hence, they continue toward your computer.

This is the reason you still need a good surge suppressor at your
computer! Even the best at-the-device surge suppressors only clamp at
400 volts (rms, using varistors to dissipate the energy); but if most of
the energy has been abated by the time it reaches the surge suppressor,
there isn't enough current left to generate enough power to melt any
delicate semiconductor connections.

Hope I was able to make this clear.

--Fred

Reply to this Message

w_tom - 25 Aug 2006 21:11 GMT

> The above is not strictly true. I have personally seen instances where
> a well protected, well grounded building has been hit by lightning; yet
> a computer with a surge suppressor (an IsoBar, in fact) was not
> damaged while another computer with an inferior suppressor was.

This assumes surges strike all appliances equally. Not at all true.
For example, a TV and VCR adajcent but only one is damaged. Yes,
because one made a better path to earth and was therefore damaged. You
cannot say the Isobar did anything effective because the circuit was
not defined. That is literally everything connected to earth appliance
from the incoming surge path AND the outgoing earthing path.

Furthermore, notice the many other appliances also not on a surge
protector and also not damaged: clock radio, dishwasher, dimmer
switches, smoke detector, etc. In order for your example to be valid,
than all other appliances not on a surge protector must also have been
damaged. Those other undamaged applainces cannot be ignored. And yet
that is exactly what you have done to claim the Isobar was effective.

Anything that will be effective on the computer's power cord should
already be inside the computer. Electronics are not delicate as
implied. For example, Intel specs for a power supply demands that a
supply withstand up to 2000 volts. Even standards of 30+ years ago
demanded electronics withstand upwards of 600 volts without damage.
Again, internal protection can be overwhelmed if a surge does not have
a better path to earth elsewhere.

We layer protection. Pictures of the primary protection system:
http://www.tvtower.com/fpl.html

Protectors and direct connection to household earth ground is the
secondary protection system. Each protection layer is defined by
protection. Protector is not protection. Protection is what a
protector shunts to - earth.

There is nothing magic about Isobar. It also is a shunt mode
protector. What makes one shunt mode protector better than another?
Joules. Joules determines protector life expectancy. What makes each
protector even better during one transient? A shorter and even better
single point ground. We want that ground to accomplish two things:
equipotential and conductivity. What happens if voltage to computer
rised 2000 volts? Does not matter is earthing beneath that building
also rises a few thousand volts. The computer sees no voltage
different even thought the entire building potential rises. That is
the equipotential function - one reason why protectors need single
point earthing.

One would think that cable center conductor needed a surge protector.
However industry professionals note that leakage between center
conductor and shield makes center conductor protector unnecessary.
Sometimes a spark gap is used for even better protection where, for
example, direct lightning strikes to transmitter towers are frequent
and robust. But an incoming residential coax cable only bonded to
single point earthing is sufficient protection. Surge damage on the
coax is more often due to no short bonding of cable to that single
point ground OR grounding a cable to some other earth ground (see
cinergy.com citation below).

Cable is not filtered by the Isobar. A cable filter in a plug-in
protector would only diminish cable signals and still would not be
sufficient for protection. In fact, this degradation of cable signals
is why cable companies say coax surge protector is not required. But
again, quality of the single point earthing determines how effective
that bonding for surge protection will be. Again, each protection
layer is defined by its earthing point. A figure demonstrates good and
bad earthing:
http://www.cinergy.com/surge/ttip08.htm

Meanwhile that plug-in protector guarantee is so chock full of
exemptions as to be irrelevant. The guarantee typically only gets a
surge protector replaced. A typical example of how a warranty is
honored: Newsman on 10 Sept 2002 in the newsgroup alt.video.ptv.tivo
entitled "SONY TiVo SVR-2000"

> I got a Belkin surge protector with phone line protection soley
> for Tivo purposes.
[quoted text clipped - 9 lines]
> "Belkin at it's sole discretion can reject any claim for any
> reason".

> Agreed, no surge suppressor will stop a direct or very nearby lightning
> strike. What is good about the TrippLite's products is that the
[quoted text clipped - 48 lines]
> there isn't enough current left to generate enough power to melt any
> delicate semiconductor connections.

Reply to this Message

bud-- - 26 Aug 2006 07:06 GMT

> > The above is not strictly true. I have personally seen instances where
> > a well protected, well grounded building has been hit by lightning; yet
> > a computer with a surge suppressor (an IsoBar, in fact) was not
> > damaged while another computer with an inferior suppressor was.

> Anything that will be effective on the computer's power cord should
> already be inside the computer. Electronics are not delicate as
[quoted text clipped - 20 lines]
> the equipotential function - one reason why protectors need single
> point earthing.

> Cable is not filtered by the Isobar. A cable filter in a plug-in
> protector would only diminish cable signals and still would not be
> sufficient for protection. In fact, this degradation of cable signals
> is why cable companies say coax surge protector is not required.

The best information I have seen on surge protection is at
http://www.mikeholt.com/files/PDF/LightningGuide_FINALpublishedversion_May051.pdf
- the title is "How to protect your house and its contents from
lightning: IEEE guide for surge protection of equipment connected to AC
power and communication circuits"
- it was published by the IEEE in 2005
- the IEEE is the dominant organization of electrical and electronic
engineers in the US

A second guide is
http://www.nist.gov/public_affairs/practiceguides/surgesfnl.pdf
- this is the "NIST recommended practice guide: Surges Happen!: how to
protect the appliances in your home"
- it is published by the National Institute of Standards and
Technology, the US government agency formerly called the National
Bureau of Standards
- it was published in 2001
- it was written by Francois Martzloff - the NIST guru on surges and
lightning

Both guides were intended for wide distribution to the general public
to explain surges and how to protect against them. The IEEE guide was
targeted at people who have some (not much) technical background. Read
one (or both) to understand surges and protection

With a plug-in surge suppressor, all interconnected devices, like a
computer and printer, need to connect to the same surge protector. As
Fred Moore said, if a computer, has external connections like phone or
LAN, those wires have to run through the surge suppressor for
protection. This type of suppressor is called a surge reference
equalizer (SRE) by the IEEE (also described by the NIST). The idea is
that all wires connected to the device (power, phone, CATV, LAN, ...)
are clamped to a common ground at the SRE - a local single point
ground. The voltage on the wires passing through the SRE are held to a
voltage safe to the connected device. The primary action is clamping,
not filtering or earthing. These devices are designed not to degrade
the signal on cable, LAN, ....

bud--

Reply to this Message

Jeffrey Goldberg - 26 Aug 2006 02:44 GMT

> The thing
> is, we don't know exactly from where the surge entered the system. We're
> assuming (reasonably) that it came in through the cable. It could also
> have entered through the powersupply of the router, especially if it was
> plugged into a different outlet.

As I've tried to make clear in prior follow-up messages, there was no
surge. My initial guess at a surge was completely wrong. Also, I'm
fairly confident that even if there were a surge, it wouldn't come in
through the cable. I'm on FiOS.

The problem was something about irregular power to the iMac itself,
which was fixed by doing a PMU reset.

> What is crucial is how much energy from a strike is dissipated BEFORE it
> gets to an electronic device you want to protect. SOME surge will ALWAYS
> get through [...]

> Hope I was able to make this clear.

Thank you. That was interesting and informative in general, even though
it isn't directly related to what my problem turned out to be.

-j

Signature

Jeffrey Goldberg http://www.goldmark.org/jeff/
I rarely read top-posted, over-quoted or HTML posts
My Reply-To address is valid.

Reply to this Message

Mac Forum / General / Hardware / August 2006

Fried ethernet on G5 iMac