Conference ssag::ask_ssag

    John,
    
     The switches you see on the I/O module (8 position switch pack) are 
    NOT related to the TLZ09 brick switches.  These are address switches 
    for the SCSI ID.  
    
     Termination is NOT to be supplied by any drive in Storageworks(tm),
    just at the ends of the bus - typically at the initiator and the
    enclosure.  The TLZ09 jumper/switch is supposed to be factory set 
    (Sony) and should not be changed at the SBB build.
    
     The TLZ09-VA is a narrow drive - but uses the 16-bit compatible I/O
    module.  (All SBB's have been 16-bit compatible since at least Feb 1995
    - most were earlier.)  The 8 pos switch indicates that you have one of
    the newer I/O boards, the old "8 bit only" board used a 6 pos switch
    pack.
    
     Unfortunately I can't add anything to what you've tried.  The only
    guess that I have is that the switch on the drive was set wrong - 
    somewhere - but without removing the drive from the SBB, you can't
    tell.
    			Regards,
    				Bill

    I can always return the drive.  
    
    So a TLZ09-VA should work in a BA364 that has wide drives?  I just
    wanted to make sure I wasn't trying something unsupported.
    
    Anybody want to see the drive before I return it?  Whatever is wrong,
    it sure doesn't play along with the other drives in the box.
    
    				-John

    It's supposed to work. 
    
    You've probably set the address switches to overlay another device. Get
    out your docs and put them back to auto-set by the BA364.
    
    There's probably nothing wrong with the drive.
    
    Chris
    

    I didn't touch the swtiches on the back.  They are all factory default.
    With the drive installed, a SHOW DEVICE at powerup shows the tape
    at its proper SCSI address of #5 (as per the switch settings for the
    BA364 slot I'm using).  Its the other devices that starting going
    away...
    
    As for the docs, the TLZ09 manual doesn't explain the switches on the
    back of the SBB.  (Hence my confusion in .0)
    
    				-John

.4> BA364 slot I'm using).  Its the other devices that starting going
.4> away...
    
    So it's the "other devices" you should return ;-)
    
    Did you try to use different slots.
    
    Guenther

    The engineering spec contains documentation of the switches.
    See subsys::lca:[specs.tape.tlz]tlz09_spec_Reva.ps.

    To return to the original problem, from the WRKSYS::ALPHASTATION500
    conference, it sounds like you're seeing the drives disappear 
    intermittently.  The symptoms are a pretty close match to note 76
    there.  Putting the TLZ09 on the bus may affect the electrical
    characteristics of the bus enough to increase the frequency of failure.
    Please check the date code on your QLogic chip, as suggested in 76.22.

    I did some more playing around.  I found that removing any of the 4
    devices from the BA364 box made it work.  Only when all 4 are in
    does the drives disappear.  Could it be that 3 RZ28D-VWs and 1
    TLZ09-VA draw more power than the BA364 can provide?  Could a low
    voltage situation cause conditions such as this?
    
    				-John
    
    P.S.  I'll check out note #76 to see if that is applicable.
    

    re: .4
    
     The TLZ09 manual WON'T have SBB information - it's specifically for
    the brick that's inside the SBB.
    
     However, you would have received EK-SBB35-UG with your TLZ09-VA.  This
    document has lots of info about SBBs in general.  Amongst other things,
    it explains what the SBB switches do.
    
    				Bill

    Ah, I missed that info in the SBB User's Guide.
    
    So, how can I compute the power draw of the 3 disks/tape to compare
    it with the BA364's power supply.  The TLZ00 manual talks about power
    for the drive, but I'd guess the SBB draws a little more given it has
    to power the 16-bit personality card, etc.  Same for the disks, I
    suppose.  Or is the BA364's power supply so powerful that I'm wasting
    my time and should be looking for something else.  
    
    BTW, I haven't cracked the case yet to check for the date on the
    Qlogic chip as described in the ALPHATSTATION500 conference (#76).
    I'll try that this afternoon.
    
    It has been suggested to me that the stub lengths of the 4 devices
    is making the bus slightly longer than 3M.  That makes sense, but
    the documenation seems to imply that adding/removing SBBs from the
    box doesn't change the length of the bus.  Is that a slight mistruth?
    
    				-John

    John,
    
     The SBB doesn't pull significantly more power than the drive itself. 
    There is NO personality card in a -VA, there is an I/O module that
    simply routes power and signals from the backplane to the drive.  [Since
    the I/O module is not 0 resistance there is a slight power loss there, 
    but it's in milliwatts!]  [The personality card is a BA356 part, not an
    SBB part.]
    
     Sorry I can't address the BA364 design, it's not mine.
    
     Adding drives to a bus does NOT change the bus length - but can change 
    the bus characteristics.  Since the TLZ09 uses the I/O module, while
    disk drives use the I/O Flex, the impedance may be different.  (Node
    spacing remanins the same, but the stub lengths are probably
    different.)  I'm not a SCSI bus expert, so don't know what this impact
    would be.
    
    				Bill

    Just to follow up, I got a BN31G-0E cable (0.5M) and everything seems
    to work now.  This got my bus length from 2.75M down to 2.25M.  It is
    a tight fit to stretch the cable from the back of the 500/500 over to
    the back of the BA364 (sitting to the right of the system box).
    Boy I sure can't want for the UltraSCSI and DOC chips to hit the
    streets.  These ever shrinking cables are a pain in the ...
    
    				-John

    Spoke too soon.  Just 27 hours after installing the new short cable
    and using all 3 disk drives and the tape drive, I got a couple of
    "power on reset"s in the error log.  Something still isn't right.
    I'll check a few power connectors, SCSI chips, and then work this
    back through field service.
    
    				-John