Cheap Firewire Enclosures - BEWARE of Prolific PL-3507 chipset!!

[hourglass]

A few months back, I picked up a cheapo "3.5" Aluminum Firewire/USB2.0 Enclosure" at CompUSA because I needed something on-the-spot.

This was packaged in a CompUSA brand box - and there were tons of them on the shelves.

I managed to destroy my first one in a wiring oversight while installing the bridge board into a more sturdy housing. This whole procedure was eventually proven moot because the bridge board used only supports one drive and I was trying to do a Master/Slave configuration.

Anyhow - the first box I had came with an INITIO chipset. This chipset, while slower than an Oxford911 chipset, worked just fine for the time being.

After I smoked my card, tho, I headed over to CompUSA to pick up a new one since I was still in need of a portable drive solution.

Much to my dismay, I did not immediately notice that this new enclosure had a DIFFERENT chipset.

Everything else was exactly the same - same external box, same cables and power adapter, same terribly translated manual...

But this enclosure used a PROLIFIC PL-3507 chipset.

As I've come to find out in the days SINCE MY DRIVE WAS DESTROYED, the Prolific PL-3507 has a lot of problems.

The primary one being that it lies to the OS on a very basic level.

This device, when connected via firewire, claims it can handle I/O chunks in excess of 128K.

As is true of any IDE device, this is simply NOT POSSIBLE.

So - your operating system queries the device while you are tracking and writing extensive amounts of data and the PL-3507 responds that it can handle a 350K chunk. Or a 512K chunk.

Which the OS gladly hands it.

AND THEN IT HANGS.

Dead. Bang. Gone. If you're on Windows, you'll get a "Delayed Write Failure" error which tells you the data you were trying to write is lost.

What it doesn't mention is that your drive hung-up so ****ing hard that the partition table is corrupted AND the last piece of data which was successfully written to the disk went across RANDOM SECTORS which probably already contained your precious project data. And not just the current project, but everything on the disk.

If you get your disk to come back up (everyone give a shout-out to "Partition Table Doctor") your OS will need to check it. When it does, you'll be told various files are corrupt and are being deleted. You might be told a direcotry is corrupt and it will VANISH. All the data in the directory will become .CHK files under Windows.

It takes about 2 weeks to get anywhere with recovering them (check to see when I last posted about losing files -- I'm nearly finished converting .chk files to .wavs.)

So - if you end up with one of these things, be careful. There are supposedly firmware upgrades which fix the problem, but the manufacturer has removed all support for them from the website.

A private individual has written a patch for Windows which allows the device to operate properly. This requires a manual registry hack as well, which is sort of a good thing, since you can edit the "FriendlyName" of the device to read something more appropriate. My Prolific PL-3507 is now identified as "Horrible ****ing Piece of Shit" when it's connected to any of my computers...

Keep in mind - you don't have to buy one to come across it - someone could bring a firewire drive to your studio and it could have this chipset, so be careful.

From what I can tell, this chipset is used on these CompUSA enclosures (both the 3.5" one and the 5.25" ones) as well as the "Metal Gear Box" enclosures and other related budget boxes which use a single board for Firewire and USB 2.0.

For those interested, the max128k patch which lets this device work properly in Windows is here:

http://member.newsguy.com/~siccos/


For me, I'm glad my nifty Oxford 911 SCSi-retrofit board was waiting by the door when I got home. I thew it into an old 2-bay SCSI tower and it's flying like mad. And was cheaper (with shipping) than the damn PL-3507 Drive Destroyer...

ryan