In article , "Z Man"
wrote:

Here's my story, you draw your own conclusion. I just installed this
motherboard about a week ago. One of its selling points was onboard
Firewire. I have more Firewire peripherals than most PC users, including an
80GB hard drive, DVDRAM/-R, Canon ZR10 mini-DV cam, and Apple 30GB Ipod. All
these peripherals are external, and I use them on a variety of computers in
my home and office. After installing my new motherboard, I could not get
these peripherals working correctly. The DVD-RAM never worked with this mb.
the Canon worked all the time. The external 80GB hard drive worked for
several days, then stopped working. The Ipod worked for a day or two, then
stopped working.

Here's what I did: I took my Ipod to a local Apple store, and they confirmed
that it was dead. They replaced it under the warranty. The external hard
drive is in a Firewire/USB2 case. It now works with USB2. The DVD-RAM case
is Firewire only, so I will have to discard it (since the problem, I have
tried it on two other computers, both of which previously used this
peripheral without issues). I have since disabled onboard Firewire and
installed a Firewire card (with Lucent chipset, I have lots of computers and
spare parts). My new Ipod now appears to work correctly, and my mini DV cam
works, as it did before. I am theorizing that the camcorder never suffered
because it does probably does NOT get its power from the port; instead, it
gets plugged into an a/c outlet. If that assumption is correct, it is the
only peripheral that was never at risk. The DVDRAM and hard drive also use
external power, but their six pin Firewire cables carry power, as well.

So, what do you think? Could it be that the power output from the onboard
Firewire port is too high, and ruined my equipment? Any other theories or
ideas would be welcomed. Post them right here so we can all join in. As the
saying goes, 'a word to the wise is sufficient'.

One part of your story I'm having trouble with is the death of the Ipod.
I thought the Ipod was battery powered, and recharged itself from the
cable. In the stuff I read about below, I cannot think of a reason for
the Ipod to die, but some other peripheral powering cases can certainly
result in a dead Firewire interface.

To properly answer your question would require a copy of IEEE1394-1995
specification (or any followups the IEEE published). That is a spec you
can buy from the IEEE, so it is unlikely you'll find a copy floating
around on the net. There is also www.1394ta.com, but like any good
trade association, there is no free lunch there either.

The only useful free information I can find is this Texas Instruments
application note:

http://focus.ti.com/lit/an/slla011/slla011.pdf (Galvanic Isolation IEEE1394)

It states some of the power, grounding, and signal isolation issues with
1394. To do a Firewire interface properly, you need:

* Shield termination (DC isolation)
* Cable power isolation (floating power source)
* Signal isolation (DC isolation between systems, between MAC and PHY)

In the case of shield termination, the idea is to avoid "ground loops"
in the grounds being used by the various Firewire devices. The TI document
says to use a capacitor and high value resistor to join the shield of the
cable to ground on the local equipment. The capacitor makes an "AC ground",
so AC currents are grounded, but there is no DC path for ground potential
differences to cause current to flow. The high value resistor doesn't pass
any significant current, and is used to "bleed" ESD. In the case of my
Antec computer case, this principle is violated by the joining of the
USB ground to the Firewire ground, on the case front connector. You could
have a similar violation buried somewhere in your system.

The second issue is cable power isolation. The VP and VG are supposed to
be powered by a transformer isolated power supply. The PC power supply
joins the green "safety ground" to logic ground, so a separate DC-DC
converter would be required to meet the 1394 requirement. I cannot
imagine any Taiwanese manufacturer wasting their money on something
like this. That means each PC has already compromised that requirement.
(That means you should be especially careful doing Firewire networking
between two PCs !!! If you use six pin cables, there had better be
diodes on the +12V coming from the motherboard...)

In terms of the voltage and current level, the TI doc says the VP/VG
power supply can be 40 volts at up to 1.5 amps. An Apple computer happens
to use 25 volts or so, to power Firewire devices. Every IBM PC will be using
the +12V (the same supply used on the processor Vcore cct). So, the PC
doesn't have too high a power output. To see an example of how powering
is handled in a sample Firewire implementation, look at the schematic
on page 10 in this document:

http://www-s.ti.com/sc/psheets/slla153/slla153.pdf

In the schematic, the Firewire design converts VP/VG
using the LM2574HVM-3.3, which is a high voltage rated DC-DC stepdown
converter, for converting up to 40VDC to +3.3 for local logic. So, that
shows that a responsible peripheral design, if it wants to extract power
to run the Firewire interface at all times, would use a high voltage
part to do power conversion, and wouldn't burn out.

http://www.national.com/search/searc...eywords=LM2574

While looking at that schematic page, you can see some other safety
features. There is a diode and fuse, where the design connects to VP.
This allows current to only flow one way, and that same diode and fuse
should be used on Firewire equipped motherboards. If the diode is not
present, nasty things will happen when a +25V computer connects to a
+12V computer.

Looking at the Firewire connectors, you can also see the use of capacitors
for the isolation functions.

The third thing the TI document mentions is isolating the MAC to PHY
interface. Capacitive coupling is supposed to be used to allow signals
to be passed from chip to chip. Since many chips are now integrating
the MAC and PHY function, this feature would be hard to implement, at
least to handle any reasonable voltage difference.

So, given all the compromises that appear to be at hand, what can
you do ? I would suggest the following:

1) Power up the peripheral before connecting it to the computer.
The idea here, is to allow the peripheral to satisfy its power
needs locally, with all circuitry ready to go, before firing up
the computer. If you want to leave the Firewire cable in place,
then power the peripheral first, following by pushing the button
on the front of your PC case. If you put your computer to sleep,
leave the Firewire device power switch in the ON position.
Otherwise, power the Firewire device down and then disconnect
the cable, before the next time your PC is awakened.

2) For extra protection in cases where you are uncertain about
the isolation of the power source, only use 4 pin connectors, not
6 pin connectors. Get a 6 pin to 4 pin conversion cable, as this
will stop the VP/VG cable power from flowing. Then, the powering
situation when connecting a single peripheral to a PC will be a
lot easier to understand. Your Ipod will no longer be recharging
from the cable, but it might not die either.

3) For devices that are powered from the wall, try to plug all the
devices into the same power strip. Use a quality strip that doesn't
make questionable contact with the green safety ground.

Another way for silicon devices to die, happens when you "hot swap"
a device, and the signal contacts are the first to touch the foreign
system. Then, whatever voltage difference exists between devices gets
applied directly to the pins, and that can cause the device to die.
In lower speed interfaces, clamp diodes can be used to limit this
effect. Since Firewire runs at 400Mbs, clamp diodes would degrade the
signal quality, so they cannot be used. In any case, this is not
an issue, as the Firewire connector design is such that the cable
shield touches first, before anything else does.

So far, I haven't been able to find any more useful tech info on
Firewire, so these are just the things I'll be trying when I get
my Firewire enclosures (soon). As I've read about Macintosh users
burning out Firewire interfaces, I think it pays to be extra
careful, especially when using expensive devices like cameras.
I hope someone has better luck finding some good info on Firewire
than I've had.

My initial impression is the Firewire industry is pretty lax.
I bet you won't see any motherboard manufacturers joining the 1394TA.

If these device deaths are happening after using the computer case
front panel connectors, be aware that many cases have miswired
connectors on the front. If using a connector on the front of the
case, verify the pinout with an ohmmeter, before destroying any
expensive peripherals.

HTH,
Paul