Slashdot Mirror
Spec Will Cut External Drive Power Cords
Lucas123 writes "The Serial ATA International Organization just revealed that it is well along the way to finishing a specification that would remove separate power cords to external SATA drives or optical disk drives, allowing them to draw power from the host system. The resulting new cable, being called Power Over eSATA, will be compatible with the existing eSATA connector and support the current maximum interface transfer rate of 3Gb/s. The SATA organization expects the new cables to be released later this year to drive makers."
I think the advantage is supposed to be cost and speed. eSATA is faster than USB and Firewire (I think, dunno about the latest Firewire) and requires absolutely _no_ on board logic to work. With this new spec an external eSATA case is literally a metal box with a hole in it, maybe a passthrough connector if they're feeling swanky. They don't even need the transformer anymore. That makes it cheaper than USB and especially firewire.
I read the internet for the articles.
They also snap off and break! The piece of shit plastic slot on the hard drives, snaps off with ease.
The durability of Sata connectors suck.
Of course proper shielding is necessary, but if you're in a position to design a cable which will support throughput of 3Gb/s, you are in a position to supply power in the same package. In practice, crosstalk from other data cables is a much greater problem than interference on the power supply rail. (Disclaimer: I'm an analog EE; I think about this crap.)
Cretin - a powerful and flexible CD reencoder
FireWire is a fairly general-purpose specification, designed so that devices that require a fixed (and quite large) amount of bandwidth can be guaranteed it, and designed with device-to-device communication in mind. Its maximum bandwidth is 400Mbps (unless you count FW800, which I will as soon as I see a device that supports it).
SATA is a storage-device-oriented specification, designed pretty much so that drives can pump data over it as fast as they can read it, with a centralized paradigm and a much higher peak bandwidth at 1.5Gbps (or 3Gbps, but see the note about FW800 above).
Using USB for storage devices is perverted and wrong; it's synchronous, so your practical bandwidth is limited by the length of your cable and the response time of the nodes at either side. On the other hand, a design like that is pretty great for things like user input devices, which is one reason nobody ever talks about making FireWire mice.
So, in summary, SATA is more suitable for disks than FireWire, and USB is dog-slow. Any questions?
-:sigma.SB
WARN
THERE IS ANOTHER SYSTEM
That's ok; USB isn't designed to power heavy loads. 2.5W (5V x 500mA) is enough to spin a hard drive
It's enough to spin up a laptop HD, but not enough for the cheaper and higher storage but larger3.5" desktop type drives.
Thus those drives need supplimental power, which is still annoying.
I'd have been happier with a limit around 12 watts, which is enough to power a 7200RPM HD, though you might need a capaciter to limit current draw during peaks.
12W@12V would be 1 Amp, so you'd only need a marginally thicker cable(or two), and you wouldn't be limited to trickle charging quite as many devices, or needing auxillery power sources for as many items.
You might even be able to operate an energy-efficient(if slow) laptop inkjet printer off of that.
I don't read AC A human right
The spec has allowed 3200Mb/s over fibre for years but I've not seen any consumer products supporting it. The latest version of the spec (just approved) supports 3200Mb/s over the same cables and connectors as existing FireWire 800 systems.
I am TheRaven on Soylent News
Think of every 12V line going into every hard drive in your machine. Now think of every 12V line having to be routed through the motherboard.
It essentially won't happen because it'll make motherboards much more complicated (read: expensive). That said, power-over-SATA shoudl have been in the e-SATA spec from the beginning, glad I didn't hop on the bandwagon earlier.
Moderation Total: -1 Troll, +3 Goat
Two Firewire 800 devices that I use every day:
Lacie external drives http://www.lacie.com/us/products/product.htm?pid=10922
RME Fireface 800 http://www.rme-audio.de/en_products_fireface_800.php
While they don't use the full bandwidth individually, it's nice to be able to chain without worrying about audio/video dropouts.
So does coupling power with data restrict the potential to chain SATA devices in the future when the bandwidth out paces the drives?
7 pins, but just 4 wires going in (unless you got a 3.3V line too which is not necessary on the bulk of SATA drives). The 7 pins make it hot swapable, you can just yank power to a drive, and it doesn't hose the whole thing. I do failure analysis on hard drives all day (its my job), and for me it is about the best part of the SATA spec (since I don't have to reboot machines just to throw a different drive in for testing).
On another note, I'd guess this is also why it took so long to come out with an eSATA power spec that would work, since occasionally it matters whether you pull the SATA cable or the power cable first. Pulling power first is generally the way to go, as sometimes if you pull the SATA cable first the machine will get pissed, and you gotta reboot as your SATA controller will just sit there waiting to hear back from the drive that no longer has a connection.
Prediction: The real iPhone killer is going to be sex robots from Japan. Think about it.
Why are you talking about batteries?
5v is standard for TTL (transistor-transistor logic) digital circuits. 3.3v for more complex chips and 1.8v for low power stuff.
Good luck getting batteries to produce any of those voltages.
You will find all three of them plus 12v in your computer however.
The main reason is that they don't get re-assigned to different devices. With PATA, you know which physical drive will get /dev/hdc every time.
I understand that some of this can be corrected by configuring udev rules.
However, if you are booting off of a linux software RAID1 disk, it is impossible to write an MBR to handle all types of drive failure. If the two MBRs are set to load the kernel from their respective drives, and one fails such that it disappears to the system altogether, if the other one now becomes the first one's device name, the machine can't boot. One the other hand, if the first (normally booted) one fails such that it is still detected but not a member of the RAID, the machine will boot by reading the kernel off of the bad drive; even if that doesn't cause problems, when you swap the bad drive out, the machine will no longer boot.
I have also had problems with using USB and Firewire disks for scripted backups. They were mounted automatically at night, backed up, and then unmounted, and someone physically rotated the backup set during the day. After a few months the device number would change, and then the backups would stop working; in some cases, the wrong disk was mounted and had a backup dumped on it filling it up. This seniario is fixable by the correct udev rules configuration, of course. But, udev should not have to be a kludgy patch on top of a broken-by-design way to hook up things to a computer.
I am boycotting all USB, Firewire, and SATA devices until something better comes along.