Man-wai Chang said:
What most likely kill the hub? Could it be repaired in general?
It's a Samsung PUH-7000NB.
There is a schematic of a four port hub here. It uses a single large chip,
and a few little chips for various things. This one is self-powered
or bus powered, and a small relay is used to switch between sources.
The relay and input power jack, are on the lower left.
http://replay.waybackmachine.org/20080906233617/http://www.smsc.com/main/tools/usb/usb20h04evb.pdf
There are a couple regulator chips next to that. One makes 1.8V, the
other 3.3V, to power the core logic on the main chip.
The main chip is the tall rectangular column in the center of the
schematic. It has some sensitive analog power inputs, and those
are filtered with a ferrite bead. The chip has drivers for status
LEDs, provided on a per-port basis.
The 24C02 is a small EEPROM. If I was manufacturing the hub, I'd use
my own private registered VID and PID, to give my product a unique
electronic identity. Not that it matters.
On the far right, are shown five jacks. But they're not identical, so the
schematic is being less than honest. The top one, is the upstream port,
which connects to the computer. The bottom four, are downstream ports, used
by peripherals.
The two MIC2026 at the top, are "current policemen". They check the current
flow on the downstream ports. These chips are duals, meaning they monitor two
ports each. These turn off the power, if more than 700mA is drawn from the port
(i.e. a bit higher than the 500mA limit). The peripherals would only be allowed
to draw 500mA, if the device was self powered with a wall adapter.
http://www.micrel.com/_PDF/mic2026.pdf
If a MIC2026 failed, then two ports would fail, and two would continue to function.
If either of the two regulator chips fails, the whole box is dead.
If the 24C02 is corrupted, it doesn't appear to be protected by a checksum.
That might make the device behave strangely, depending on what particular
bits are twiddled. There is no particular reason for even bothering with
the expense of that chip, so it's optional.
If I had to guess as to the nature of the problem, I would suspect power.
And power issues can be verified with a multimeter (whereas other issues
in a laboratory, would require an oscilloscope to make progress).
You would start by opening the box, for a look. Identify the component
part numbers and work from there. There is nothing dangerous in there,
because the logic is 5V or lower.
Oh, I forgot one other component. There is a 24MHz quartz crystal near
the bottom corner of the main chip rectangle. If you drop the hub on the
floor, there is a danger of breaking the crystal. Without something to
make a 24MHz oscillator with, the logic inside the chip cannot operate.
And they're not something you can visually inspect.
Paul
