Correct HVD SCSI Terminator for tape library

[Matt Liebowitz]

I think the problems I'm having with my TL892 tape library are due to improper termination. I've confirmed that the SCSI card I'm using is compatible and I think I can rule out a mechanical problem with the library. All signs point to a problem with termination.

I am using the terminator that came with the SCSI card, and I think that might be wrong. The only markings on the terminator say:

Differential P/Q Terminator
Amphenol P/N 505260001

Looking up that # shows me the terminator, but not whether it's the correct HVD type. Can anyone confirm for me that it is? And also, does Compaq have a part number for a terminator that's compatible with the TL892?

Thanks in advance..

Matt

[Curtis Ballard]

Almost any time you see "Differential" by itself on a SCSI device it is referring to HVD. Since HVD was commonly called "Differential" when LVD released everything I've seen has been explicitly labeled LVD.

[Matt Liebowitz]

Curtis,

Thanks for the info.. I actually got another HVD terminator and tried it, and the tape library still doesn't work.. I'm simply at a complete loss.

I'm using a HVD device (TL892 MiniLibrary) connected to an HVD SCSI card (Adaptec AHA-2944UW) and I've now tried two HVD terminators. If I set the SCSI ID on the tape library to ANYTHING other than ID 7, it won't show up as a device on the AHA2944. If I set it to ID 7 (which is also the ID of the 2944), it shows up but at ID 11. It will only do this for one device at a time, either the library itself or one of the tape drives.

The reason I was thinking that it was a problem with termination is because I get this behavior whether I have a terminator connected or not.. Do you have ANY idea what might cause this? Sounds like a terminator problem but after two terminators I'm not convinced.

Any help would be greatly appreciated..

Matt

[Curtis Ballard]

That problem sounds like there is a bent pin somewhere. The SCSI negotiation feature only uses 1 data pin. If the library shows up at ID 11 when you set the ID to 7 that implies that data pins 7 and 11 are shorted together so when the library asserts pin 7 pin 11 also asserts or that pin 11 is shorted to ground all the time. For SCSI shorted to ground is the same as always true. As long as everything runs narrow SCSI and no wide parity checking is enabled it is actually possible to limp along with a bent pin on the wide bus so some commands might work.

[Matt Liebowitz]

Curtis,

Where do you think the pin is bent or where the short might be? Is it in the cable connecting the library to the SCSI card? If so, I can just try another cable. Or is this a problem with the library itself?

I've tried turning parity checking off on both the 2944 and the library and that has no effect. Is there another place I can check?

Thanks for the suggestion, this is the first time anyones come up with anything other than "it's probably a bad terminator."

Matt

[Curtis Ballard]

The problem is almost always in a connector somewhere. Remove each connection and examine the pins closely. If a bent pin is found also examing the connector on the device closely. Often a bent pin will damage a connector and just replacing the cable doesn't do anything because the same pin on the new cable gets bent when you plug it into the damaged connector.

If you find a bent pin it is best to replace the cable but if you don't have another you may be able to very carefully straighten the pin and order another cable. Once a SCSI pin is bent it bend easily again and often breaks.

If you find a damaged connector you'll need to use something like a fine tip knife and clean off any damaged plastic that might cause another pin to bend.

Good Luck

[Matt Liebowitz]

Curtis,

I checked the cabling and found no bent pins of any kind.. I did however notice that the cable we're using to go from the SCSI card to the tape library is labeled LVD/SE instead of HVD. I thought I read that if the cable is SE then it will adjust automatically to HVD.. Is that correct? Is this my problem right here?

Thanks so much for your help thus far.

Matt

[CA843380]

Hi,

I find it surprising that the cause of this could be an LVD/SE labelled cable but I guess its possible if it has been wired with LVD/SE in mind which could be the case since the diffsens line is used differently from LVD to HVD.

LVD and HVD are not compatible. LVD stands for low voltage differential and HVD is high voltage differential. LVD and SE (single ended are compatible) as are HVD and SE.

Try a generic SCSI cable (one not labelled LVD/SE or HVD) or a cable labelled SE or HVD and see how you get on.

The other thing that springs to mind about the 11 scsi id compared to an expected 7 is whether you have a mix of 68 way and 50 way connections? HVD SCSI is 50 way narrow and 68 way wide, my assumption is that the TL892 uses a 50 way narrow connection? Your scsi card is a wide HVD card, does it have a 68 way connection? Is you cable a 68way to 50way cable and your terminator on the TL892 a 50 way one? if this is all true that you have only terminated the lower 8 data lines of the bus at the library end, hence the upper 8 are unterminated and this might explain why you are seeing the 11 result.

Maybe you have your scsi card configured to run in wide mode and this with your 68way/50way mix is the issue. Try configuring you scsi card to run in narrow mode so that it doesn't use the upper 8 data bits of the bus.

For me to help further I would need a clear description of exactly what you connections from library to scsi card are, type of cable, type of connector in card/library and number of connections etc as the above ideas are just guesses :)

Cheers,

Dave Dewar

[Matt Liebowitz]

Dave,

Thanks for the info.. I'm ordering a new cable specifically labeled and wired as an HVD cable. I don't think that's going to fix it either, but I'm at a loss for anything else.

Here is a clearer picture and hopefully you can shed some light.. I've got an old Proliant 3000 server that I'm trying to attach a Compaq TL892 tape library to. It comes with a wide SCSI (68 pin) internal card, and an internal tape drive is connected successfully to that.. If I connect the tape library to that it gives an error and doesn't detect it. Whether that's an HVD issue or not, I'm not sure. Because of this, we bought an AHA-2944UW card from Adaptec that supports HVD devices. Both the TL892 and the AHA-2944UW are 68 pin wide SCSI.. There's no 50 way connectors to speak of.

The TL892 has 6 68 pin connectors on it. One for the library, two for each tape drive, and then one at the end to terminate. Whether I connect all of the cables or just the cable for the library (and terminate directly after that), the behavior I get is the same. The only way I can get either the library or ONE of the tape drives to detect is if I set the device to SCSI ID 7. When I do that, it shows up as ID 11 on the card. If I set it to anything else, it detects nothing. If I set both the library AND one of the tape drives to ID 7, it fails going through the scan presumably because it's finding two devices at ID 11.

Does this make any more sense? It still looks to me like the bus isn't being terminated properly despite trying two different terminators because I get this behavior with or without a terminator connected.

Any thoughts or suggestions would be greatly appreciated.. Thanks for your suggestions thus far..

Matt

[CA843380]

Hi Matt,

Ok, I understand things a little better now, sounds like you have individual in/out connectors for both tape drives and library controller making a total of six. Hence you will want to daisy chain them all up, pop a terminator on one end of chain and the cable to scsi card on other end. Everything supports 68 pin wide HVD so you should be all set if it is cabled correctly and you have given each device a unique scsi id.

I have included a link below that details TL892 cabling etc, maybe you already have something like this? manual etc. Can you check you have followed the guidelines in this link please. If it still doesn't work after this, then it does look like a problem with you origianl cable.

Cheers,

Dave Dewar

http://www.phys.uu.nl/DU/Tru64_5.1A/HTML/ARHGWDTE/CHTPDVCS.HTM#sec-tl891-892

[Curtis Ballard]

A cable marked as LVD really shouldn't cause a problem but there could be a difference. Most LVD cables are called multimode and have a lower capacitance than older HVD and SE cables. If there was ever going to be a problem due to the lower capacitance I would expect it to occur on SE, not HVD.

Was this library working previously on another system?

It sounds like you have ruled out the HBA and the terminator. That probably only leaves the cable from the host, the library's cable jumpering the connectors to the controller, or the connector between the libraries controller and cable.

[Matt Liebowitz]

Dave/Curtis,

I reviewed the link that you posted that discusses the correct way to cable the device and it looks consistent with what I've got. We've ordered a new cable that's specifically HVD so hopefully that resolves the problem.

If it doesn't, I'm at a complete loss.. I've never seen this type of behavior in a SCSI device.. I don't know why it's so hung up on ID 11 and it won't detect any device at any other ID.

Thanks for both of your help.. If the new cable doesn't fix it, I'm going to assume that something happened with the library when it was brought over here because it was working in the past.

I'll let you know what happens w/ the new cable.. Thanks again!

Matt

[Matt Liebowitz]

To my surprise, it turns out the cable was the problem all along. Adaptec says their cables all support HVD/LVD/SE, but apparently this one doesn't. I ordered a cable that's specifically made for HVD devices from Granite Digital, and it worked on the first try. Sees all three devices at the correct SCSI IDs, and Windows detects it as a tape library..

Thanks to all who helped and gave suggestions..

Matt

[Curtis Ballard]

Glad to hear you got it working. The old cable is probably damaged and should not be reused. Of all the SCSI signaling types, HVD is the most robust and should be able to use any undamaged SCSI cable.

[Kusuma]

Hi Matt,

The forum encourages authors to assign points to those who have pitched in to solve their technical problems.

To learn more about how to assign points, check out the following link:

http://forums1.itrc.hp.com/service/forums/helptips.do?#28

Good luck and enjoy forum-ing!

Thanks,
Kusuma