delays and errors from SCSI drives...

[spadnos]

Glad it worked.

(must be karma from the BIOS flashing idea )

I tend to disagree with Chucky Cheese on drive placement. The whole point of the terminator is to prevent reflections. If it's working correctly, there are NO reflections on the cable (it looks infinitely long ti an impulse). Of course, the terminator has to be at the end, but once that's done, you can put drives anywhere.

The only thing that could be useful is if the drive powering the terminator (assuming active termination, which I think is required for U160 and U320) is too far away to supply clean power to the terminator. (That seems unlikely, but it's possible (maybe))

Chucky: Can you point me to something on the web suggesting that the drives are placed close to the terminator?

Thanks
- Steve

edit: typo and word rearrangement for clarity

[Chucky Cheese]

SCSI

Spandos

i actually got that info from a tech support person at
Granite Digital while troubleshooting a scsi bus system that i built for myself. he called it the "stub length", and said that it should not exceed 3" and likened it to an echo.

when i beta tested ezscsi 5 i asked this question of some of the other testers and product engineers and almost everyone recomended this.

from personal experience i have built myself 4 scsi based systems over the years and have never had any problems configuring my scsi bus or errors on the disk when holding to this. while it may not be absolutley necessary it is beneficial...trust me ??

cc

[spadnos]

Chucky:
Thanks.

It certainly can't hurt to do that, I was just wondering what the justification is.

I guess if there's a slight mismatch between the terminator and the cable impedance, then you can have a small reflection. It should only depend on the overall cable length though (or the distance from the device driving the bus to the terminated end), not the distance from the last drive.

Ah well - high speed signaling is closer to voodoo than anything else, anyway.

I've been lucky over the years, because I rarely have a cable with many spare connectors, and due to length considerations, I just about always end up with my devices at the far end of the chain anyway.

-Steve

[schmosef]

I've never heard about terminator placement before but I've always wondered what the optimum position was. Thanks for the info Chucky Cheese.

I plan to buy a new hard drive as soon as I get my video problem (discussed in another thread) resolved and will definitely follow your advice.

[VJ]

Well, it is all done, and working... I checked the grown defects list of the Seagate, and it is empty. I couldn't seem to find a way to check it for the IBM/Hitachi.

As for the terminator: on the U320 cable, the terminator is extremely close (<1 cm) to the last usable connector, so indeed that seems to be hinting at keeping the stub as short as possible... (mind you, on my other cable, the terminator is just plugged in to the last connector, which results in a minimum stub-length of approx. 15 cm )


Jörg

[Chucky Cheese]

Congrats !!!

VJ

i really like scsi when it is working good...i have since moved to SATA, and this system has the "FEEL" of U160 but i definitely think the scsi disk are faster overall

anyways...good job!

cc

[VJ]

Re: Congrats !!!

Originally posted by Chucky Cheese
i really like scsi when it is working good...

Same here, I moved to SCSI when I needed to upgrade harddisks on my PII-450: I already had an Adaptec 2940UW (which hosted my Plextor 4220), so the only cost was the drive (Atlas 10K). The speed improvement was enormous: the system became much more responsive (programs launched faster, compiling went much smoother, image editing, ...).
Of course, there are downsides: noise, heat (less of an issue now), ...

Strange though: I had a similar looking bus on another controller (ok, it wasn't even U160, so that might be it), which didn't have these issues. When I moved to the dual Xeon (and thus to an onboard controller), the problems began...
I will mail SuperMicro, as their technician also couldn't see the problem (the controller manual states that TERMPWR is not required, as it should take care of that , but the manual is the same for both a standalone controller as an onboard one...).

i have since moved to SATA, and this system has the "FEEL" of U160 but i definitely think the scsi disk are faster overall

Well, I don't have experience with SATA, only with IDE (and SCSI). With IDE, I personally feel a lot depends on the mainboard's chipset: some chipsets apparently limit the throughput (esp. noticable when you configure a RAID0: we had computers here where the throughput of the RAID was the same as that of one harddisk , we found out later that some VIA chipsets indeed do this)

anyways...good job!

I just had to slide in a jumper... you did most of the work!


Jörg