SCSI controller problem

[douglasfir]

Hi there! I have an Adaptec 2940w/uw 50/68 pin controller card in an old machine I'm trying to set up, and i'm having a problem getting the drive to even spin up when it connected to the controller. The drive will spin up with the cable connected to it and not the controller. But not when it's connected to the controller. The drive is a seagate st336607lw u320 10k RPM. The cable is an SE/LVD 68 pin with a terminator on one end. Scsi controller is set to defaults. I have enabled termination power with the jumpers on the drive itself. The drive is connected to the last connector before the terminator as it should be. No other scsi devices on the bus. Drive is set to scsi ID 0. Controller ID 7. I've downloaded the pdf for the drive and gone through it extensively, and cannot get the drive to start when connected to the controller. I know the drive will not operate at u320 speeds on this controller, but from what I've read it is compatible. The scsi bios returns an error of "termination not detected" and obviously does not detect the drive. I've used scsi before, and I've never had this problem. I have an identical controller and two other drives and everything behaves the same with the other drives and controller. The only thing I don't have is another cable. The pins are not bent, and there don't appear to be any damaged wires on the cable itself. This is driving me nuts. Please help. What could I be missing? Thanks for your time!

[Curious.George]

Re: SCSI controller problem

Hi there! I have an Adaptec 2940w/uw 50/68 pin controller card in an old machine I'm trying to set up, and i'm having a problem getting the drive to even spin up when it connected to the controller.

The scsi bios returns an error of "termination not detected" and obviously does not detect the drive. I've used scsi before, and I've never had this problem. I have an identical controller and two other drives and everything behaves the same with the other drives and controller.

Are you saying the other drives/controller/cable work properly? Or, exhibit the same sort of failure? Obviously, moving the current disk over to the other controller would be an easy test. Or, moving one of the other drives/cables over to this controller. On the SUT, remove any superfluous SCSI devices so you're POSITIVE there isn't a conflict.

Double-check the "termination" setting in the HBA BIOS. Different Adaptec HBAs treat termination "errors" differently (e.g., some of the adapters will hang if left unterminated) and even different BIOS versions on a single HBA model. IIRC, the termination check is just done by examining the open circuit voltage on the interface (Looking for the Vt equivalent from the termination network).

To simplify things, be sure you're not mixing narrow and wide devices. If there are THREE bus connections on that model cad, you can only use TWO of them at the same time (bur ONE is the safest bet).

There's also a setting for the HBA to send a START_UNIT command to each specific SCSI ID -- make sure your SCSI ID is set for that (else you'll have to wait for the OS to issue the command).

The SCSISelect utility will let you try to issue commands to the drive without having the OS stick its nose into the process.

[Topcat]

Re: SCSI controller problem

That is a very old controller. Make sure it's got the latest firmware on it. What drive specifically? Check the drive jumpers and make sure drive spinup delay is off. Depending on the drive, this may also be called 'start cmd' or 'motor start delay'....or something along those lines. With this off, the drive will spinup when powered up whether its connected to the controller or not. If the drive still fails to spin up, odds are the drive is bad.

[douglasfir]

Re: SCSI controller problem

Thanks for replying! When I move the drives/cable over to the other controller (identical 2490w/uw) the same failure is produced. There are no other SCSI devices on this bus whatsoever, im only connecting one drive at a time. I will check the SCSI bios for the send start unit command. Thanks so much, i'll let you know how it works out. I'm out of ideas otherwise.

[stj]

Re: SCSI controller problem

turn off the drive termination if the cable is terminated.
latest card firmware is 4.something.

what other jumpers are on the drive?

[Curious.George]

Re: SCSI controller problem

turn off the drive termination if the cable is terminated.

There won't be any terminations on the drive -- though there may be a TERMPWR option available. And, maybe even an option to convert between LVD and SE signaling.

As that HBA is designed with the ability to drive from the center of the SCSI bus (both ends of the bus must be terminated), if there are no internal nodes, then the HBA's terminator must be enabled (HIGH and LOW bytes) if there are no internal devices to provide the termination.

Some drives will delay their spinup by a time related to their SCSI ID (staggered spinup to ease demands on power supplies) -- usually something like 6 seconds per SCSI ID (so, if you want a single drive to spin up sooner instead of later, pick a low SCSI ID).

Jumpers are typically only examined on power-up so cycle power for any changes to be noticed.

The fact that the drive DOES spin up suggests it is expecting to be told to do so when connected.

[stj]

Re: SCSI controller problem

well spinup needs to be checked in the card settings.
i was doing some of this yesterday, but not with Adaptec, their bios's are junk and even assign the wrong interupts.

[Curious.George]

Re: SCSI controller problem

well spinup needs to be checked in the card settings.

Drives typically have straps to allow:
- unconditional spinup
- wait for START UNIT
- spinup after delay (usually based on SCSI ID)

A drive configured to unconditionally spin up will usually do so even in the presence of a START UNIT command. OTOH, a drive that is strapped to wait for START UNIT will usually just sit there until/unless it gets one!

I don't, however, know if the behavior will vary depending on whether or not the drive is free-standing (I only spin up drives that are tethered to an HBA). That's the sticky point in the OP's description of symptoms (does the drive know/care that it is free-standing or not?)

Swap drive/cable from "other system" (which is presumably "working") to this system and see if the symptoms change. This isolates issue to HBA.

[When I bring up any sort of HBA (SCSI, FC, SAS, etc.), I always "reset to factory defaults" before getting started. Then, document the changes I make to that starting point. No guarantee that there aren't bugs or options that are unaffected by this -- but, it's a good place to start!]

[Topcat]

Re: SCSI controller problem

well spinup needs to be checked in the card settings.
i was doing some of this yesterday, but not with Adaptec, their bios's are junk and even assign the wrong interupts.

I had a buslogic controller that did this....even after firmware updating. It ended up in the recycler box.

[stj]

Re: SCSI controller problem

i was tempted, but i cant bring myself to scrap my 39160's
or maybe they are 320's - cant even remember now.
they have anoying mini connectors on the back plate.

[Topcat]

Re: SCSI controller problem

i was tempted, but i cant bring myself to scrap my 39160's
or maybe they are 320's - cant even remember now.
they have anoying mini connectors on the back plate.

I've never had a 29x or a 39x ever do anything like that. The only real wonky adaptec controllers I remember were the 2100s raid and the aaa-131u2 raid. They could be very fickle with their setup....and of course had very poor first-release firmwares...

[Curious.George]

Re: SCSI controller problem

i was tempted, but i cant bring myself to scrap my 39160's
or maybe they are 320's - cant even remember now.
they have anoying mini connectors on the back plate.

Dual VHDCI. I use them to connect to my SCSI shelfs.

[Curious.George]

Re: SCSI controller problem

It ended up in the recycler box.

I still keep modest collection of different HBA samples as there are (too!) often times when I need to build a SCSI machine for some purpose (usually related to my Sun kit). Trying to find one of a particular style/make when I need it is just a colossal waste of effort -- easier to throw them all in a box!

I've even got DB50 SCSI cables (along with numerous cables that morph from one connector style to another -- Centronics, DB25, DB50, HD50, HD68, VHDCI, etc.).

But, SCSI RAID controllers just get shit-canned.

[douglasfir]

Re: SCSI controller problem

well, I've tried everything. You cannot update the firmware unless you have an eeprom programmer according to Adaptec.

https://ask.adaptec.com/app/answers/...are-upgrade%3F

All jumper settings are correct as per the drives manual, and the info in this thread. Suppose i'll have to downgrade to a 50 pin drive. that's totally ok for me. But i havent been able to find any. Thanks for helping everyone, I really appreciate it :-)

[Curious.George]

Re: SCSI controller problem

well, I've tried everything. You cannot update the firmware unless you have an eeprom programmer according to Adaptec.

I assume you mean EPROM (one 'E').

All jumper settings are correct as per the drives manual, and the info in this thread. Suppose i'll have to downgrade to a 50 pin drive. that's totally ok for me. But i havent been able to find any.

Did you try connecting the drive to the existing controller and cable in the "other" machine? I.e., to prove to yourself that it's the drive and not the controller (or cable or settings)

<frown> B.C. Probably "paperwork" to ship you anything.

[eccerr0r]

Re: SCSI controller problem

Back in the day, what you sold worked, there is no second chance with a firmware upgrade.

I don't know about SCSI as I don't recall doing this before, but definitely some PATA drives will not spin up if you connect the data cable upside down. Worth doublechecking if nothing else to check. Of course this does not apply to the D-type one way connectors, but the internal SE50s is possible...

I also have to mention that finding a good SCSI68 internal cable is tougher than it seems. I've a few 68-pin wide SCSI cables but they all are wonky and won't work properly except for one -- much worse than the hit rate for PATA cables which they usually just work. YMMV.

[stj]

Re: SCSI controller problem

maybe he has a bad active terminator.

[Curious.George]

Re: SCSI controller problem

I also have to mention that finding a good SCSI68 internal cable is tougher than it seems. I've a few 68-pin wide SCSI cables but they all are wonky and won't work properly except for one -- much worse than the hit rate for PATA cables which they usually just work. YMMV.

I've never encountered a bad internal (or external!) wide SCSI cable that was defective with a failure that wasn't obvious (e.g., bent pins in a connector). IME, machines with SCSI drives see far less "tinkering" with their internals than PATA/SATA machines. And, the folks in a position to do that tinkering tend to be "less consumer-ish" and more likely to understand what they're doing.