[Ti] Firewire Port Frying External Hard Disks

Jerry Krinock jerry at ieee.org
Sat Jun 17 15:25:10 PDT 2006


After burning up several external Firewire hard disk enclosures in the last
couple months, I've finally found a possible pattern: I believe it happens
whenever I connect one of them to our old 867 MHz Ti Powerbook G4.  Upon
connecting an external drive to this old Powerbook, I hear a snapping sound
inside of the external drive case, smell some faint smoke, and when I take
the drive case apart I often see that the case of the Firewire interface IC
has a crack or a hole it it, and this seems to be the source of the smell.

It happens using either a 4-pin cable (signals only) or 6-pin cable (signals
+ 30VDC unregulated power).  It happens whether or not I provide external
power to the external firewire drive.

According to the official IEEE 1394 pinout, pins 1 and 2 are supposed to be
one differential signal pair, and pins 3 and 4 are another differential
signal pair.  The outer shell is ground.

When I measure the DC voltage presented by the suspect Powerbook, pins 1 and
2 are both at +3.3 VDC while pins 3 and 4 show no voltage source.  On a good
Powerbook, pins 1 and 2 are at +2.3 VDC instead of +3.3 VDC.

So, possibly the firewire driver inside the powerbook is shorted to +3.3V
and is frying whatever I connect to it.  But most chips can handle at least
0.5V more than their VCC, which looks like it's probably +3.3V for Firewire.
Also, the powerbook does not seem to mind if short one of these +3.3V
sources to ground.  If I use a resistor of 10-100 ohms, it initially shows a
source resistance of 100 ohms or so, but then after a few seconds the
voltage reduces to 0 as though a capacitor is discharing.  So, it's not like
a direct short to a powerful supply, and I'm surprised that I'd be frying
external devices so consistently, unless there is an initial current
transient which I cannot see using my Radio Shack DVM.

Anyone with any knowledge or experience in these matters is encouraged to
add to this story!

Jerry Krinock 




More information about the Titanium mailing list