SCSI-320 drives interfering with ATA-133 drives?

[boggsie]

My tower case has two sets of drive cages. In one of the cages I have the boot and program drives. These are 15k rpm SCSI-320 drives. In the other cage is a set of two ATA-133 drives in RAID-1 to hold my data.

The entire PC is setup and the ATA-133 data disks are in the cage, but sitting outside of the computer case. If I run the pc in this manner, the ATA-133 RAID array works like a charm.

However, if I install the array into the case and button it up, the raid controller 'looses' the raid. I have tried this with three separate sets of ATA-133 cables (six in all) and the same result. I even have a set of (supposedly) shielded (braided wire) ATA-133 cables, although I suspect the shielding is more visual than practical ... which is why I tried the other two sets of standard flat (80-oin ATA-133) IDE cables.

I suspect that the interference from the 15k RPM SCSI-320 disks is interfering with the ATA-133 disks. Either that, or the water pumps in the reservoir (water cooled) are throwing off interference.

Is this plausible? Any other thoughts?

 

[mechBgon]

Full system specs?

 

[earthman]

Its totally plausible. What kind of SCSI cable are you using, and how is it routed/bundled?

 

[boggsie]

Good questions. I will follow-up later with detail.

One thing that I forgot, was that I also tried to use the two drives (non-RAID) in the onboard ATA-66 controller. Similar results in that the drives would 'dissappear' from Disk Manager, in mid-test.

A final attempt to be absolutely certain led me to, yet again, moved the two IDE drives (still in the cage) from the server case to the workbench. I didn't disconnect anything. I just pulled the cage from the server case and set it a few inches from the physical machine on the workbench. After a reboot, all of the problems disappeared.

There is some sort of EMI that is having a terrible impact on these IDE drives. It has to be coming from:
a) SCSI-320 drives
b) Koolance water pumps

Again, thanks for the replies. I will respond with details and pictures, later today.

 

[Steve]

I think you may have too many boot ROMs. Try booting to DOS with a floppy and running msd.exe (near bottom of page). If your boot ROMs total more that 128KB (131,072 bytes) then you'll have problems. What kind of video card do you have?

 

[boggsie]

Originally posted by: sm8000
What kind of video card do you have?

Still working on detailed system setup, but this is a quick question to answer.

I have an ATI AIW 9700 Pro AGP and an ATI Radeon 9200 PCI.

The boot ROM thought is interesting. I wonder why (it would appear, anyway) that leaving the IDE drives connected but outside of the case, would make such a difference?

 

[mechBgon]

The aspects I could see being altered by the physical location of the hard drives are:

1) EMI-type stuff

2) Electrical grounding

3) Physical vibration

4) Some sort of physical/mechanical interference (cable gets squashed by presence of extra drive cage, not very likely)

On #1, is your SCSI cable terminated with an LVD terminator just beyond the last drive? The conventional approach is for the card to be at one end of the cable, the terminator at the other end, and then the drives are attached as close to the terminator as possible, working towards the card as you add more drives.

Try testing #2 there, by shutting down the system, leaving the SCSI drives & rack outside the case, but electrically connecting their cage to the case with some wire to establish a uniform ground plane. Fire it up again and see what happens.

For #3, some info on your case might help so we can try to think of a way to damp vibrations. For example, in my Lian-Li, I unriveted the whole HDD rack and mounted it on rubber grommets (not for your reason, however, but for sound damping).

#4, I'm sure you've got that checked already.

 

[Steve]

Originally posted by: boggsie

Originally posted by: sm8000
What kind of video card do you have?

Still working on detailed system setup, but this is a quick question to answer.

I have an ATI AIW 9700 Pro AGP and an ATI Radeon 9200 PCI.

The boot ROM thought is interesting. I wonder why (it would appear, anyway) that leaving the IDE drives connected but outside of the case, would make such a difference?

Well, you do have a lot of boot ROMs - two video cards and three mass storage controllers, plus any other possibilities like bootable USB, bootable LAN, etc. But why the IDE RAID array works outside the case is beyond me.

 

[boggsie]

profile
angle

Iwill MPX2 Dual AMD
2gb PC2100 REG ECC

AGP = ATI AIW Radeon 9700 Pro
PCI1 = 64bit (ATI Radeon 9200)
PCI2 = 32bit (Adaptec 50pin SCSI)
PCI3 = 64bit (LSI 29320-R SCSI)
PCI4 = 32bit (Soundblaster Live! Platinum)
PCI5 = 32bit (SIIG | 10/100 LAN | Firewire | USB 2.0)
PCI6 = 32bit (SIL680 ATA-133 RAID)

Floppy
Plextor Ultraplex 40x Max (SCSI)
Plextor 40/12/40 CDRW (SCSI)
Sanyo BP5 DVD-ROM (SCSI)
Iomega Zip 100 (SCSI)
Seagate Tape Backup (SCSI)

Fujitsu MAS Ultra 320 (boot)
2x Seagate Cheetah Ultra 320 (RAID-0 | Programs)

Coolance parts augmented with other water parts and wedged in for cooling on CPU's and Video

 

[boggsie]

I sold off the 80gb IDE's that I had and returned the 120's.

I am gonna buy a couple of 147gb Seagate 10k.6 drives and put them in RAID-1 and be done with this. I had (in this exact same setup) a couple of Atlas 10k II/III drives and they worked flawlessly. I ran out of room and wanted to try the low-buck IDE route for data storage.

I just wanted some confirmation that it could be EMI, because I couldn't fathom that it was anything else.

I'm still very interested in comments / thoughts on this, though.

Best regards,
-boggsie

 

[Steve]

Even more boot ROMs than I thought. Run the msd.exe diagnostic with the array working and without, see what you get.


BTW the UltraPlex's are 68-pin, right? Can they go on an U160 bus without slowing it down to 40, or do they need their own channel?

 

[boggsie]

Originally posted by: sm8000
Even more boot ROMs than I thought. Run the msd.exe diagnostic with the array working and without, see what you get.


BTW the UltraPlex's are 68-pin, right? Can they go on an U160 bus without slowing it down to 40, or do they need their own channel?

The Ultraplex supposedly came in two variants ... 40x TS Max and 40x TSi Max. The TSi had 68 pins. I owned one a couple of years ago and did not care for the way it did precisely what you hinted at: slowed down the other drives on the SCSI bus with it. I sold it and purchased the 50 pin version of the drive.