What jumpers settings for HDs in IEEE1394 enclosures?

[Bazarhuu]

What are the recommended jumper settings for connecting multiple (I have two) internal HDs to a PC via Firewire? I have used 'Cable Select' with success, and 'slave' with very limited and unpredictable success. Does 'Single or Master' setting make more sense? What about 'Master w/slave'?

I am using two Western Digital 8Mb cache drives, one 200Gb and on 250Gb. I have them in FireXpress Compucable (from newegg) enclosures, which allow for Firewire and USB 2.0 connection to PC, as well as pass-through. I connect these drives to the laptop via one 6pin-4pin Firewire cable. I have to do one at a time, and will probably pick up another cable tomorrow. The laptop is a Sony Vaio R505GL with docking station running XP Home. I have all the relevant XP updates installed. The drives are formatted NTFS with 64K block size, and permanent drive letters (P: and R:, does not conflict with other things I have). I have gotten the failed write delay error now, although never seen it before using the Vaio or getting these enclosures (two weeks ago). I've disabled write caching on both external drives, kept caching on for the internal C: drive. I've used USB 1.1 (don't have 2.0) to get the drives to show up under Disk Management when the Vaio refused to show them using the Firewire connection.

Ideally, I would like to plug one drive in, daisy chain the second, and have the ability to copy large files from one drive to the other (I am using the drives to mainly store and access ~200Gb of raw audio). But I will be satisfied with being able to plug the drives at the same time or individually (Vaio has two 4pin Firewire ports), have a stable system, and not end up corupting the drive. I have moved the drives between laptops, and a Dell 8200 laptop liked them a lot better. I've formatted and reformatted, ended up with corrupt partitions, recovered with "Active File Recovery Pro", lost drive letters, mounted the drives internally to a desktop, etc. Funny tho, I have not lost any date. This all happened in the middle of an East Coast-West Coast move, so I couldn't do too much organzied testing. I'd stick XP Pro on the Vaio instead of Home if I knew the Vaio would not have any problems with it, providing the right drivers are installed.

I have a KEC 1002 2-port Firewire card that gets recognized by the system, although nothing plugged in the card gets recognized. I seemed to have had some of my Firewire problems after plugging in the card and stopping it, so I think I'll retire this one for now.

For anyone who's knowledgeable, I'd love to know the benefits of the different jumper settings. It seems that these settings matter for Firewire, but USB (1.1, have not tried 2.0) ignores them. I'd appreciate any background on this situation, as I've had inconsistent problems that I've solved (for now) without much logic. Planning for a much bigger move in two months, and don't want to lose access to the data on the drives on the road.

Thanks,

Alex

 

[Lord Evermore]

Drives should be set to Master or Single drive when placed in an enclosure. It doesn't matter if you're daisychaining them -- the master/slave setting only applies to drives on a single IDE cable to the same IDE controller chip. Setting it to anything but master could possibly create problems but there's no technical reason for it.

Plugging them into separate ports would give you the best performance, since the drives wouldn't need to compete for bandwidth. As far as how the system sees them, there's no benefit from daisy-chaining (you don't see one drive of the combined sizes or anything like that).

The write delay error seems to be something buggy with XP's detection system. Even with every possibly background program and indexing service of any type disabled, and caching turned off, XP still claims something is accessing my drives so I can't disconnect.

 

[Bazarhuu]

That was much appreciated. Had I known this a week ago, much anxiousness would have been avoided. Will definitely try and see how it goes. Thanks for pointing out the daisy chaining and bandwidth issue. I was not sure until now whether bandwidth was measured per port or per an entire firewire interface. I guess USB would be the same way?

Alex

 

[Lord Evermore]

USB's bandwidth is per root hub when it comes to integrated chipsets. Usually each root hub has support for 2 or 3 ports, so if you have a board with support for 6 ports, and device manager shows 2 root hubs, then 3 ports per hub. I think this would usually be divided up depending on the number of ports on the back panel and the number of headers for an expansion slot bracket. If the board has 4 ports on the IO panel, then each block of 2 would go to a single hub, and then the internal headers would each be the remaining port for each hub. I don't know yet how Intel's 875P with 8 ports is laid out, but I should know in a few days. For each set of 3 ports, the total bandwidth available is 12Mbps or 480Mbps depending on the version.

Some expansion cards allow the full 12Mbps or 480Mbps per port rather than having ports share a root hub. They may show up in Device Manager as one root hub for every port.

As far as I know, Firewire is 400Mbps per port (800 with FW2). This of course means a 3-port controller could saturate the PCI bus.

The reason I think is that USB is a master/slave design, and Firewire is peer-to-peer. The USB hub is dependent on the controller for all transfers, while a Firewire device can act independently. So Firewire doesn't have a "root" hub like USB.