I recently had to purchase a <used> DLT to replace one which had failed.
The failed model was a DLT2000 (15/30GB) with Centronics connectors. My
SCSI card in the server machine was 68-pin. So I was using an AMP cable
with centronix 50-pin on one end and HD50 on the other.
The new drive is a DLT7000 with 68-pin connections on the back. The
connection on the back of the drive itself is 68-pin as well so I can't swap
out the centronix ribbon connector from one tape to the other so I need
another cable.
In the meantime I took two 68-pin->Centronix cables and hooked them together
using the two centronix connections on the ribbon cable from the old DLT.
I hooked everything up, and the SCSI bios (Adaptec AHA-2940UW card) detects
the drive as "Quantum DLT7000" but hangs at the detection point and goes no
further. I've tried a variety of bios settings for the tape, reducing the
transfer rate, disabling wide negotiation, etc. but can get no success.
The cables are both 1.5m cables so I am sure I'm over the 3m length
recommended as I've also got four internal drives on this bus. Can an
excessively long cable cause the bios to hang like this? It's very
reproducable; I've moved the cables around, cleaned all the connectors, etc.
in order to try to eliminate poor connections as a possible cause.
I've got four ultra-wide 68-pin drives on the internal bus connection of the
card, and just the tape drive on the external card connection. I've got
termination set to 'auto' in the scsi bios but also tried "low off/high on".
I've ordered a new cable; hopefully it's not a problem with the drive
itself.
Dave wrote in news:wjdEi.24716$bO6.14160@edtnps89
> I recently had to purchase a <used> DLT to replace one which had failed.
> The failed model was a DLT2000 (15/30GB) with Centronics connectors.
> My SCSI card in the server machine was 68-pin. So I was using an AMP
> cable with centronix 50-pin on one end and HD50 on the other.
>
> The new drive is a DLT7000 with 68-pin connections on the back. The
> connection on the back of the drive itself is 68-pin as well so I can't swap
> out the centronix ribbon connector from one tape to the other so I need
> another cable.
>
> In the meantime I took two 68-pin->Centronix cables and hooked them together
> using the two centronix connections on the ribbon cable from the old DLT.
>
> I hooked everything up, and the SCSI bios (Adaptec AHA-2940UW card) detects
> the drive as "Quantum DLT7000" but hangs at the detection point and goes no
> further. I've tried a variety of bios settings for the tape,
> reducing the transfer rate,
Set all devices to lower transfer rates.
> disabling wide negotiation,
You should use that.
> etc. but can get no success.
> The cables are both 1.5m cables so I am sure I'm over the 3m length
> recommended as I've also got four internal drives on this bus.
So the limit is 1.5 meters.
> Can an excessively long cable cause the bios to hang like this? It's very
> reproducable; I've moved the cables around, cleaned all the connectors, etc.
> in order to try to eliminate poor connections as a possible cause.
>
> I've got four ultra-wide 68-pin drives on the internal bus connection of the
> card, and just the tape drive on the external card connection. I've got
> termination set to 'auto' in the scsi bios but also tried
> "low off/high on".
You should use that.
> I've ordered a new cable; hopefully it's not a problem with the drive
> itself.
If you have a detachable terminator on the internal cable you can use
the internal cable as if it were an external.
"Dave" <dspear99ca@yahoo.com> wrote in message
news:wjdEi.24716$bO6.14160@edtnps89...
>
> In the meantime I took two 68-pin->Centronix cables and hooked them
> together using the two centronix connections on the ribbon cable from the
> old DLT.
>
Duh. Just occurred to me that if there's 68 pins on the controller, and 68
pins on the drive, then it probably needs 68 pins to operate and that by
losing 18 pins going through 50-pin connectors that's probably my problem...
Dave wrote in news:nmgEi.24747$bO6.12979@edtnps89
> "Dave" <dspear99ca@yahoo.com> wrote in message
> news:wjdEi.24716$bO6.14160@edtnps89...
> >
> > In the meantime I took two 68-pin->Centronix cables and hooked them
> > together using the two centronix connections on the ribbon cable from the
> > old DLT.
> >
> Duh. Just occurred to me that if there's 68 pins on the controller, and 68
> pins on the drive, then it probably needs 68 pins to operate and that by
> losing 18 pins going through 50-pin connectors that's probably my problem...
"Folkert Rienstra" <see_reply-to@myweb.nl> wrote in message
news:46e19d98$0$47108$892e7fe2@authen.yellow.readf reenews.net...
>
> Set all devices to lower transfer rates.
Can you briefly explain why? The only thing that's changed on this system
is the tape drive... with the old DLT2000 I set the transfer rate to 5Mb/s
and disabled wide negotiation, but the internal drives on the other cable I
left at the default setting of 20Mb/S with wide negotiation enabled,
effectively giving me (theoretical maximum of) 40Mb/S. And everything
worked. Is it because of the cable over-length? I'd guess that I might
have less errors with the slower speed and longer cable combo, but could
this hang the bus at boot-up the way I have described?
>
>> disabling wide negotiation,
>
> You should use that.
Are you saying that I need to set the whole bus to narrow? What if I use
the proper cable? Everything would then be 68-pin, I would need no upper
termination from the card, etc. I'd still set the DLT SCSI ID to 5Mb/S but
could then enable wide negotiation, correct?
>
>> etc. but can get no success.
>
>> The cables are both 1.5m cables so I am sure I'm over the 3m length
>> recommended as I've also got four internal drives on this bus.
>
> So the limit is 1.5 meters.
>
>> Can an excessively long cable cause the bios to hang like this? It's
>> very
>> reproducable; I've moved the cables around, cleaned all the connectors,
>> etc.
>> in order to try to eliminate poor connections as a possible cause.
>>
>> I've got four ultra-wide 68-pin drives on the internal bus connection of
>> the
>> card, and just the tape drive on the external card connection. I've got
>> termination set to 'auto' in the scsi bios but also tried
>
>> "low off/high on".
>
> You should use that.
The "high on" gives me termination at the DLT end of the chain because the
DLT is narrow and the terminator cannot provide high termination, correct?
>
>> I've ordered a new cable; hopefully it's not a problem with the drive
>> itself.
>
> If you have a detachable terminator on the internal cable you can use
> the internal cable as if it were an external.
>
The internal cable has an integral (non-removable) terminator. The server
chassis is maxed out for power and especially heat, or I might put the drive
into an empty slot.
"Folkert Rienstra" <see_reply-to@myweb.nl> wrote in message
news:46e19d98$0$47108$892e7fe2@authen.yellow.readf reenews.net...
> Dave wrote in news:wjdEi.24716$bO6.14160@edtnps89
>> reducing the transfer rate,
>
> Set all devices to lower transfer rates.
>
Huh. I set them all to 20 Mb/S (except the DLT which I left at 5Mb/S
narrow) and it worked. 26 Mb/S worked. 32 Mb/S worked. 40Mb/S did not
work. I think my cable is too long. I'll see if I can get through a whole
backup with no SCSI errors, might have to lower it some more still to
accomodate high-throughput situations until my new cable arrives.
In article <nmgEi.24747$bO6.12979@edtnps89>, Dave <dspear99ca@yahoo.com>
writes
>Duh. Just occurred to me that if there's 68 pins on the controller, and 68
>pins on the drive, then it probably needs 68 pins to operate and that by
>losing 18 pins going through 50-pin connectors that's probably my problem...
You've gone from a wide controller to a wide device using two narrow
cables, via a narrow device, which is, umm... suboptimal
And there's the question of termination. If the 68pin - 50pin Centronics
cable you used to connect your DLT2000 has internal high-byte
termination (it should have), and you have an external terminator on the
DLT7000 (you should have, as it's at the end of the bus), you're now
terminating the high byte of the bus three times - twice at the external
end of the bus and again at the other end, probably your controller.
Just get a 60-68 cable and a decent terminator for the DLT7000 and all
will be well. I'd also set your host adaptor config to defaults, unless
you know it has to be configured to suit one of your internal devices.
Mike Tomlinson wrote in news:$izcsTCGQEIHFwiG@jasper.org.uk
> In article <nmgEi.24747$bO6.12979@edtnps89>, Dave <dspear99ca@yahoo.com>
> writes
>
> > Duh. Just occurred to me that if there's 68 pins on the controller, and 68
> > pins on the drive, then it probably needs 68 pins to operate and that by
> > losing 18 pins going through 50-pin connectors that's probably my problem...
>
> You've gone from a wide controller to a wide device using two narrow
> cables, via a narrow device, which is, umm... suboptimal
> And there's the question of termination.
> If the 68pin - 50pin Centronics cable you used
> to connect your DLT2000 has internal high-byte termination
> (it should have),
Oh, and why is that?
It's not a problem with Adaptec cards and not needed on device ends.
> and you have an external terminator on the DLT7000 (you should
> have, as it's at the end of the bus),
> you're now terminating the high byte of the bus three times - twice
> at the external end of the bus
> and again at the other end, probably your controller.
Utter Nonsense. They can't see each other.
>
> Just get a 60-68 cable and a decent terminator for the DLT7000 and
> all will be well. I'd also set your host adaptor config to defaults, unless
> you know it has to be configured to suit one of your internal devices.
Mike Tomlinson wrote in news:c4E0aWAaSeIHFwGd@jasper.org.uk
> In article 4720c831$0$47138$892e7fe2@authen.yellow.readfreene ws.net, Folkert Rienstra writes
>
> > You should feed Bunny more carrots.
> Try taking more Prozac, Folkert.
No thanks, whitnessing the effects that it has on you:
seeingly blind and hearingly deaf.
cable length question 
Linear Mode
