torkell | (no subject)

Current Mood: surprised

(no subject)

Plugging in an external USB hard disk apparently made my USB 3G modem disappear for a few moments. Weird.

The hard disk did seem to struggle to spin up - I'm guessing the combination of the spinup current and the draw of the 3G dongle was more than what the laptop was happy to deliver.

Now, in theory, USB devices are supposed to ask the controller how much power is available and draw no more than that, with an upper limit of 500mA. In practice, the rated spinup current on most hard disks (including this one) is much more than that. Isn't it wonderful the way standards get ignored?

Flat | Top-Level Comments Only

Yeah. Western Digital's "Passport" series and Linux is a particularly hilarious combination. If you sniff the bus, you can actually watch the following transaction take place when you plug one in.

WD: I want 300mA
Linux: Uh, no can do, chief. Don't got the power.
WD: Like, whatever, man, you can't tell me what to do.
*WD proceeds to draw about 450mA*
WD: The hell, man? Why you stiflin' ma creativity?

(lather, rinse, repeat until unplugged)

This pleases me not just because it flat ignores the return code that tells it the port can't supply that much power (a perfectly valid response, if an annoying one for a USB HDD), but also because it then proceeds to draw markedly more power than it requested. Feels like a workaround for something, but it sucks.

O.o Another reason to not use their external drives- That's the freaking USB spec they are ignoring...

It's a common theme of USB HDDs - quite often the running current is fine (or only slightly over - I think this drive is rated at 2.8W), but spinup can peak at 1A. And it's not really possible for a HDD to reduce its power consumption and still work. Presumably you could use a small capacitor bank in the enclosure to provide a boost supply for the spinup and keep the overall demand within spec, but I don't think anyone's done that. Early USB HDDs sometimes needed *two* USB plugs to draw enough power from the host!

Heh. My latest ESATA hotplug thingy struggles unsuccessfully to let the drive spin up when it's getting power purely from the wall (the way it smacks the heads against the wall sounds particularly sickening), and when then plugged in to the esata port on the laptop, continues to valiantly die, but then when *also* plugged into the usb (which fortunately doesn't stop esata enumeration in this case), it spins up fine.

I was really excited about esata and firewire 800, but having watched implementations of it out in the wild, I'm harkening back to the good old days of PATA. PATA never dropped the link quite as freely. Never dropped drives out of the array just for the fun of it.

...doesn't spin up when the wall power is connected but the data isn't? To paraphrase Babbage, I cannot rightly conceive of the wrongness of mind that would lead to this way of thinking.

So its external supply cannot provide enough to make it spin up? Then WTF is the point of the external supply given that eSATA doesn't provide power.

I was thinking of getting a couple of Firewire drives to use as external backup for my main system (given that firewire should be massively faster than USB), but the hard part is actually finding the things. Apparently USB is so awesome that everyone uses it (this must be the same definition of "awesome" that makes it almost impossible to find non-glossy non-widescreen screens).

Flat | Top-Level Comments Only