When I start one particular PC on my 3 PC network it causes the ADSL modem to
disconnect and to reconnect to the internet. I cannot find what the problem is.
All computers are configured in the same manner. It is only this one particular
machine that does it.

All PCs run Windows XP SP2, have static IPs, etc.

Any ideas on how I can trouble shoot this?


--
Two rules for success: 1. Don't tell everything you know.

Dear Marts,

Is this BigPond? Uninstall all login clients. Might help.

Otherwise check what is running on startup on all PC's and compare
them with the black sheep one. The only way to do that effectively is
to download "Autoruns" from MS and do some homework. Alos that PC may
have a trojan. They can often bypass any firewall.


Hope that helps.

CreateWindow
http://mymessagetaker.com
The while-you-were-out message program you have been looking for!

(E-Mail Removed) wrote...

> Dear Marts,
>
> Is this BigPond? Uninstall all login clients. Might help.

No. Standard ISP.

> Otherwise check what is running on startup on all PC's and compare
> them with the black sheep one. The only way to do that effectively is

I've done this, but can't see anything obvious.

> to download "Autoruns" from MS and do some homework. Alos that PC may

Will look for this "Autoruns" thing.

> have a trojan. They can often bypass any firewall.

AVG hasn't detected anything. But that's not conclusive.

On Oct 29, 6:55 pm, Marts <marts...@yahoo.com.au> wrote:
> When I start one particular PC on my 3 PC network it causes the ADSL modem to
> disconnect and to reconnect to the internet. ...
> Any ideas on how I can trouble shoot this?

Start by providing sufficient information so that other can post
something based in facts; not in speculation. For example, power on
is a maybe two minute procedure. During which second does your fault
occur? When power switch is pressed? When Bios first starts? When
Bios first starts loading Windows? What is each power cord connected
to? How is the network connected? Ethernet? USB? Implied from a
post with too little information is that malware is irrelevant -
should not even be considered. To obtain a better informed to reply,
some 10 or 50 missing facts must be provided. That last post is a
request for wild speculation. Your replies will only be as good as
the information you provide.

w_tom wrote...

> On Oct 29, 6:55 pm, Marts <marts...@yahoo.com.au> wrote:
> > When I start one particular PC on my 3 PC network it causes the ADSL modem to
> > disconnect and to reconnect to the internet. ...
> > Any ideas on how I can trouble shoot this?
>
> Start by providing sufficient information so that other can post
> something based in facts; not in speculation. For example, power on
> is a maybe two minute procedure. During which second does your fault

It occurs as Windows XP SP2 is loading when the "splash screen" I think that
it's called is still being displayed.

PC is connected via standard ethernet cable directly to the ADSL modem's
ethernet switch, in the same way as the other PCs are.

All PCs have Asus motherboards with built in NICs. All are configured
identically, except for individual machine IP addresses, which are static. No
DHCP is enabled

Dunno what the power cord has to do with anything, but it's plugged into the
wall and is on the same phase as the other GPOs in the house.

What other information do you need in order to point me in the right direction
on this?

Thanks

Part of a long list of reasons for a dropped ADSL include ground
loops created by building AC electric. Assumed: you know all machines
have proper safety ground connections (so as to eliminate floating or
ground loop problems).

Time of the event (as Windows loads drivers into NIC) implies
grounds are not a problem. I can only assume all are 100 Mhz Ethernet
connections and that all wires are known properly constructed (a
problem that sometimes occurs when connectors are attached to wires
'in the field'). The suspect NIC works just fine at 100 Mhz (not 10
Mhz)?

Identical systems means all have NICs from the same manufacturer.
We don't know if the problem is Windows or hardware. Diagnostics will
break that unknown into parts to solve it. Find the NIC
manufacturer's diagnostic for that Ethernet chip. This diagnostic
will load and execute without Windows. Now what happens to the ADSL
connection when NIC diagnostic loads and executes?

The diagnostics have one even better feature - the last (optional)
test is a worst case data test. This is a worthy test of network
stability and ethernet hardware. Setup the diagnostic on the other
machine as a diagnostic server (mirror, echo, whatever they call it).
Software is executing hardware without any Windows. Setup the
optional last test on the suspect machine to output data
continuously. The mirror server will echo worst case data patterns
back to the suspect machine via the router. It should have no affect
on the ADSL connection.

If true, then hardware is eliminated as a reason for failure. Now
move on to another suspect - Windows.

Setup the Ethernet post as disabled (from Control Panel, Device
Manager, or Network Connections). Then reboot the machine. Does this
eliminate ADSL interference?

In the meantime, confirm the suspect computer's safety ground
actually exists. Both safety ground inside walls AND safety ground
connection inside machine. If not, then ADSL interference was created
by two problems - a missing safety ground AND a galvanic isolation
failure in the suspect computer's NIC interface. Based upon new
information, I don't expect this ground problem to exist. However, it
is a simple possibility to eliminate.

On Oct 30, 10:23 pm, Marts <marts...@yahoo.com.au> wrote:
> It occurs as Windows XP SP2 is loading when the "splash screen" I think
> that it's called is still being displayed.
>
> PC is connected via standard ethernet cable directly to the ADSL modem's
> ethernet switch, in the same way as the other PCs are.
>
> All PCs have Asus motherboards with built in NICs. All are configured
> identically, except for individual machine IP addresses, which are static. No
> DHCP is enabled
>
> Dunno what the power cord has to do with anything, but it's plugged into the
> wall and is on the same phase as the other GPOs in the house.
>
> What other information do you need in order to point me in the right direction
> on this?

w_tom wrote...

> Part of a long list of reasons for a dropped ADSL include ground
> loops created by building AC electric. Assumed: you know all machines
> have proper safety ground connections (so as to eliminate floating or
> ground loop problems).

Wiring in the house is fine. In fact, it's all new. We're doing up an old house.
The problem existed before we started the wiring upgrade.

Also, I had the wife's computer connected to the network via a 108mbps Netgear
wireless network. Did it then. Then later on, after we wired in the LAN, I
connected her PC to one of the 4 ports on the Netgear wireless router. One of
its other ports was connected back to the modem as at the time I didn't have
enough spare ports wired in.

Then, when I finished upgrading the wired network in the home, her PC was
connected directly into the ADSL modem via one of its ethernet ports. The
problem still remains.

So, have tried different wiring setups, both for power and ethernet. Made no
difference. Have checked software configs, virus scans, hardware configs in
BIOS, etc.. Nothing out of the ordinary.

Both her PC and mine run Asus P5 motherboards with identical NIC chipsets and
configuration.

And the problem occurs AFTER Windows starts loading and BEFORE things like
network connections and logins are established. Unless of course, all this stuff
happens during the first two or three seconds of the "Loading Windows" splash
screen period.

> Time of the event (as Windows loads drivers into NIC) implies
> grounds are not a problem. I can only assume all are 100 Mhz Ethernet

Presumably you mean 100mbits, mot Mhz?

> break that unknown into parts to solve it. Find the NIC
> manufacturer's diagnostic for that Ethernet chip. This diagnostic
> will load and execute without Windows. Now what happens to the ADSL
> connection when NIC diagnostic loads and executes?

This was a painful exercise. The internet to Asus' website is awfully slow, and
it's not the most friendly site to navigate.

And no, I found nothing like what you describe. And how does it load without
Windows? Do you install it onto a boot floppy or something?

Hey, if you can find what you're describing for an Asus P5 DG1 Pro with a
Marvell/Yukon NIC, then I'd appreciate it. I can find drivers, Asus Update
Utility, but that's about it.

> Setup the Ethernet post as disabled (from Control Panel, Device
> Manager, or Network Connections). Then reboot the machine. Does this
> eliminate ADSL interference?

Good one. Didn't think of doing that.

Now, if I discover that when Ethernet Networking is disabled, what then?

Remember, this process wasn't enabled when the Netgear wireless card was
installed in the PC (and yes, I've uninstalled it and ensureed that all drivers
have been removed).

On Nov 2, 11:31 pm, Marts <marts...@yahoo.com.au> wrote:
>> Find the NIC manufacturer's diagnostic for that Ethernet chip.
>> This diagnostic will load and execute without Windows. Now
>> what happens to the ADSL connection when NIC diagnostic
>> loads and executes?
>
> This was a painful exercise. The internet to Asus' website is awfully slow, and
> it's not the most friendly site to navigate.
>
> And no, I found nothing like what you describe. And how does it load without
> Windows? Do you install it onto a boot floppy or something?

Asus will not provide any diagnostics. Find diagnostics for and
from the Ethernet chip manufacturer.

Long before drivers are loaded and when the Windows splash screen is
apparent, the Ethernet chip is being setup. This may be when the
suspect computer is sending something to interfere with ADSL.

Normally nothing in data should cause that ADSL shutoff (see near
last paragraph for more details). But hardware (electrical) currents
might. Ethernet is supposed to have galvanic isolation to make such
electrical currents not possible. Is it comrpomised? To better
understand what is ongoing, we need something that will reproduce that
problem to know what is (could) produce your ADSL shutoff. A worst
case data test for that Ethernet chip would do that.

Diagnostics are so powerful, in part, because they run as all
computers once executed - without any operating system (Windows). The
trick and power of diagnostics is to execute a known good and specific
software without the complications from OS (such as Windows). But only
the better computer manufacturers provide diagnostics. Normally that
would outrage computer experts. But today, so many computer
assemblers don't know basic diagnostic principles. Instead they only
shotgun - keep replacing parts until something works. Too often,
computer assemblers fix problems using the same joke in Home
Improvement - "more power" rather than fix the problem.

In your case, a failure must be seen before it can be solved. What
is (ie identical computers and identical setups) does not exist until
confirmed. Whatever your problem is, the problem is indeed subtle.
To understand it, for example, I would have put an oscilloscope on
that Ethernet port to see what is happening - another diagnostic
tool. But you (and most every computer technician) would not have
that tool nor the experience to immediately see it on the scope.

Hopefully the Ethernet diagnostic echoing data between computers
might make the problem so reproducible or apparent as to then be
identified.

Everything you have done implies no problem generated by hardware.
But your problem is so typical of a hardware anomaly.

Unfortunately, a third 'similar' computer is not available. Another
test would execute that diagnostic data test between two 'good'
computers. Then power on the suspect machine. Or setup computers in
Windows to do larger and repeated file transfers.

It is possible during startup that the suspect computer sends a
collision signal. This collision signal would temporarily stop
diagnostic data. However this signal should not affect ADSL on
another side of the router. Should not. Just another reason why
better hardware provides diagnostics.

w_tom wrote...

> Asus will not provide any diagnostics. Find diagnostics for and
> from the Ethernet chip manufacturer.

Okaaay, I'll see what I can find about "Marvel Yukon".... But I reckon I'm
wasting my time.

> Long before drivers are loaded and when the Windows splash screen is
> apparent, the Ethernet chip is being setup. This may be when the
> suspect computer is sending something to interfere with ADSL.

We seem to be focusing on the ethernet chip. What is swaying my thinking that
it's a Windows issue (and why I posted it here in the first place) is that it
was occuring when the PC was connected wirelessly using a Netgear 108mbps PCI
wireless network card, with a Netgear wireless router at the other end. That in
turn was connected via ethernet through one of its 4 ethernet ports (there is a
5th one for bridge connections to an ADSL modem but it isn't being used)
directly into the modem. The PC was disconnecting the modem even then. I thought
that when I installed ethernet networking in the house that this problem would
be resolved.

The NIC was disabled both in BIOS and via Windows (well, it didn't see the NIC
as it was disabled). But the problem was still there.

The network itself is sound. It's a new installation, done by a professional
using the appropriate equipment to terminate the cables and to test the cable
connections afterwards.

Fact that it was also doing it when I used a patch cable that I run across the
floor from the suspect PC to the modem and the problem still exists tends to
also eliminate cabling as a suspect.

> might. Ethernet is supposed to have galvanic isolation to make such

What's "galvanic isolation"?

> Hopefully the Ethernet diagnostic echoing data between computers
> might make the problem so reproducible or apparent as to then be
> identified.

If it's hardware which I do not think that it is, as I can reproduce the problem
under a range of different configurations. The common denominator here seems to
be Windows itself.

> Unfortunately, a third 'similar' computer is not available. Another

No, it's a laptop that's connected wirelessly (by the aforementioned Netgear
wireless router).


--
Two rules for success: 1. Don't tell everything you know.

The problem based upon what is currently seen does not break down
into a specific suspect AND does not say anything that is definitive.
IOW the diagnostic test was a way of turning the cube 90 degrees to
get a completely new perspective while eliminating the "Windows"
variable from that test. The test would have tested three lowest
level network layers of which only the first layer is hardware. These
are the three layers that Windows is in control when talking on that
LAN - everything on computer side of network.

Unfortunately, it sounds like your ethernet hardware is from a least
responsible source - no diagnostics. IOW everything is fine as long
as no problems result. Look in Device Manger (and other parts of
Windows) to better identify the actual NIC hardware and manufacturer.
But it sound like your ethernet is from some backroom third world
hardware source. Without that other perspective (worst case
diagnostic data test), we have so little to analyze.

You suspect Windows. There are seven layers in a network
interface. Which layer is the problem? It would probably be in the
first four. First three layer problems should be completely masked
(blocked) by the router. I have been assuming you know ADSL is lost
only because the router light says ADSL is lost. IOW a loss of local
network is not being confused with loss of WAN (ADSL)? What you
suggest is a problem in layer four. But your computers have no access
to that layer on the ADSL side. Only he other side of the router
controls that.

That hardware diagnostic (worst case) data test with two or three
computers would have reported so much more - literally turn the
problem into a set of intersecting lines - so that the problem could
be limited to certain known suspects.

What I am not sure now - it is the wife's computer that created
the network crash? The wife's computer creates that crash when
accessing the router via network wire AND via Netgear wireless card in
wife's computer?

Again, that Ethernet diagnostic was for more than testing the
Ethernet chip. It was testing with the Windows variable completely
removed and it was testing lowest level network layers including the
router. If that test between two computers talking only on the same
local network (only talking via a bridge) was crashed by powering the
suspect machine, then we knew something useful. It was about
establishing 'definitively' what is known good and what does actually
crash.

A less useful test to do same would be to setup two computers
pinging each other constantly (PING -t 192.168.x.xxx) and then
learning what happens as the suspect computer boots.

So that you are not confused. All those ethernet ports for a local
network (LAN) are connected only by a bridge (inside the router box).
Any data output by any one computer is seen by all that share the same
bridge. ADSL (other side of router) sees nothing exchanged via the
bridge. ADSL only sees data with an address that must pass through
and be modified by the router. That is an important difference
between a bridge connecting all household computers as opposed to a
router that connects the bridge to a WAN (ADSL on other side of
router).

Data exchanged between computers on same bridge should never pass
through router - not cause ADSL problems. Only data with IP addresses
outside the bridge get modified and setup to ADSL by router. No data
on bridge should cause ADSL to shutdown.

That ADSL connection is controlled completely by router and DSLAM in
the telephone CO. That ADSL connection literally wraps any computer
data into a package and ignores anything inside that package - called
a packet. Any data from one computer would not crash ADSL because the
data is simply 'boxed and shipped'. WAN network (ADSL) only sees the
box - not what is inside the box.

Galvanic isolation is why Ethernet ports are rated to withstand
thousands of volts without damage. It is also why touching speaker
wires from a TV or stereo does not result in an AC mains electric
shock. During computer hardware setup, one network interface could
send an electric current into the router even causing failure on the
other (ADSL) side of the router. But galvanic isolation should
isolate electric (hardware) currents from computer from the router.

Is Windows creating something that would crash both the local
network AND pass through the router to crash ADSL? Nothing like that
should exist in data streams. I would use tools such as the
oscilloscope and network sniffer to confirm this. But that is well
beyond what we can do here.

Can Windows send data through router to shutdown ADSL? Only if a
specific command unique to that router setup page is sent to turn on
and then turn on the ADSL signal, or to order the router into reset.
IOW a startup program (either in the Start>Startup folder or loaded in
Registry to startup on powerup) unique for that router is somehow
loaded in the suspect machine? Well you could boot the suspect
computer with Task Manager open, then monitor every process that loads
and executes. But Windows must load before Task Manager can. Again,
according to how I read previous posts, your problem occurs before
Windows can load any processes. Problem would occur before Task
Manager appears on Windows screen.

One final information source. You have two computers. Do data
lights (on both computer NIC and on router front panel) flash at the
same times when each computer is booted? Does the suspect computer
appear to output more data OR output unique data just as the ADSL
crashes? How does the data light from suspect computer coincide with
when ADSL is lost and same flashing does not occur from other
computer?

What (low level layer) data from Windows can crash a WAN (ADSL) on
the other side of a router? None that I can think of. Nothing in
lower layer data transfers (what Windows controls) should be able to
do that. Hardware currents blowing through conductors that should not
be carrying a current that could do that (due to galvanic isolation).
My hope here is that additional information identifies some 'noun'
that is being misused and therefore implying something that does not
really exist or that does exist and is not being conveyed by the
appropriate verbiage.

On Nov 3, 6:46 pm, Marts <marts...@yahoo.com.au> wrote:
> ...
> Okaaay, I'll see what I can find about "Marvel Yukon".... But I reckon I'm
> wasting my time.
> ...
>
> We seem to be focusing on the ethernet chip. What is swaying my thinking that
> it's a Windows issue (and why I posted it here in the first place) is that it
> was occuring when the PC was connected wirelessly using a Netgear 108mbps PCI
> wireless network card, with a Netgear wireless router at the other end. That in
> turn was connected via ethernet through one of its 4 ethernet ports (there is a
> 5th one for bridge connections to an ADSL modem but it isn't being used)
> directly into the modem. The PC was disconnecting the modem even then. I thought
> that when I installed ethernet networking in the house that this problem would
> be resolved.
>
> The NIC was disabled both in BIOS and via Windows (well, it didn't see the NIC
> as it was disabled). But the problem was still there.
>
> The network itself is sound. It's a new installation, done by a professional
> using the appropriate equipment to terminate the cables and to test the cable
> connections afterwards.
>
> Fact that it was also doing it when I used a patch cable that I run across the
> floor from the suspect PC to the modem and the problem still exists tends to
> also eliminate cabling as a suspect.
> ...
>
> What's "galvanic isolation"?
> ...
>
> If it's hardware which I do not think that it is, as I can reproduce the problem
> under a range of different configurations. The common denominator here seems to
> be Windows itself.
> ...