Michael Berkowitz writes:

> Mike Paivinen writes:
>
>  [log in to unmask] wrote:
>  : Item Subject: cc:Mail Text
>  :      I have a customer with seven(7) DDS-2 drives.  He has had to have
all
>  :      the drives replaced over the past few months.  The problem he is
>  :      encountering is that, during backups, he is getting what would be
>  :      considered excessive tape cleaning requests.
>
>  : o Don't keep your DDS drives near laser printers.  [I have yet to
discover
>  :   what "near" is.]  HP discovered some DDS drives were contaminated
>  :   with toner.
>  : o Only fill 1 to 2 DDS tapes per device per backup per night.  These
devices
>  :   weren't designed as high-volume, high-end backup devices.
>

-----------------------------------------------------------------------------
>  HP said the same thing to us with regards to paper dust in the computer
>  room.  But what kind of devices are they if they can't live in a raised
floor
>  computer room complete with all sorts of peripherals, and if the DDS
>  drives were not designed as high volume devices, then what is the
>  purpose of the DDS autoloader that is or will soon be available?

To which Dave Gale added:

=======================================

We have had a long history with DDS-1 and DDS-2. Equally we have had
problems with the mechs. On average we replaced a particular mech every 6
months.

This continued until I moved the printers from the computer room to a
separate area (including the laser printer). After about one more
relacement, (about 18 months ago) we haven't replaced a mechanism again.
This is really great after having to take down a machine every four months
to replace the tape mech (I never like to take the 3000s down).

The symptoms were always the same. Operators finding that the backups were
still waiting for another tape (first tape marked as bad) and putting
production behind schedule. This would increase with frequency until we had
to clean the tape mech everyday (and still have problems).

Any more, if we have a recurring bad tape problem, I call in the RC.... but
not lately.

========================================

To which Jeff Kell added:

========================================

We just had an "interesting" one... had a DDS drive which had no
indications of a problem' - it was used in backups and passed VSTORE
without a hitch.  However, when we tried to restore a database on the
development system that was created on that drive, it failed.
Subsequent backup sets failed.  Multiple drives failed to read the
tape.  Doing a DSLINE/:FILE foo;dev=other#tape and restore *foo
worked fine.  Tapes written on other drives wouldn't even AVR (the
process that does the "Volume (unlabelled) mounted on ldev #x) failed
on tapes written on other drives.

The drive had creeped out of alignment, but the only indication was
trying to read it's tapes on another drive.  We had the drive replaced
today, but bear in mind that by "repairing" such drives means that you
can no longer "read" tapes recently written on that drive, so beware.

=========================================

All of which leads to the following inevitable result:

<drum roll please>

A SHORT COURSE ON WHAT'S CAUSING
THESE PROBLEMS -- AND HOW TO REPAIR
THEM YOURSELF (IF YOU WANT TO)

</drum roll>

There tends to be two primary problems with DDS drives: (i) "eating" the
tape, and (ii) and messages that say something like excessive retries, bad
cabling, or a request to insert cleaning tape.

If you're used to working on small, precision machinery, both problems are
relatively easy to correct -- at virtually no cost. You do however want to
work on a clean metal table in a non-carpeted room and take all standard
anti-static precautions, as well as have on-hand a suite of standard minature
electronic repair tools, a good CFC-based degreaser (trichloroflouroethane &
isopropyl alcohol), a high-grade light lubricating oil, and a high-grade
lithium ("white") grease.


Problem 1: Eating tapes

When a DDS tape is inserted into the front of the drive, a sensor detects the
presence of the tape and begins a synchronized procedure, sucking the tape
into the innards of the drive. As the tape descends from its original,
upper-level position, where you inserted it, to its lower-level play
position, the tape cartridge is dropped in a manner so that four items are
inserted into the cartridge on the backside of the tape: two metal fingers, a
capstan pinch roller, and a spindle spool. Once the tape is in its play
position, the fingers move out on tracks, drawing the tape from the cartridge
so as to place the tape's magnetic surface next to both the helican scan head
and the capstan drive spindle. Simultaneously, the capstan pinch roller moves
into position behind the tape, so as to create a "sandwich" consisting of the
pinch roller, tape, and capstan drive spindle. This later arrangement is the
mechanism that propels the tape through the drive -- and it is especially
important to Problem 2: "Excessive cleaning messages."

Any surface that is lubricated will get dirty with time. Indeed, a lubricated
surface is nothing more than a dirt trap. If it were possible to operate
small machinery without lubricants, the reliability of small
electromechanical devices would rise dramatically. The fingers that draw the
tape out ride on two slot-based tracks. These tracks are lubricated and do
get dirty, but they're generally not a primary cause of failures.
Nonetheless, to clean and relubricate these tracks, wipe them clean with dry
Q-tips, being sure that you leave no part of the cotton tip behind, until the
uncovered surfaces become so clean that they sparkle. Then add only a smidgen
(no more) of white lithium grease somewhat adjacent to each of the resting
fingers (the working mechanism will distribute it along the track for you).
If you put down too much grease, clean it up and repeat the procedure. Be
absolutely sure never to let any lubricant, of any sort, touch any of the
rubber surfaces. It will absolutely destroy both the rubber and the
reason-of-being for the rubber surface.

The primary mechanism that causes the drive to "eat" tapes is due to the
spindle spool arm that resides directly adjacent to the capstan pinch roller.
This arm has a spring attached to it so as to quickly draw it and the capstan
pinch roller out of the way when the tape is being retracted from the drive.
When they do not move quickly enough, the tape is left partially in the drive
at the time when the DDS cartridge's lid begins to close -- thus you're left
with a small and irritating mess.

The diagnosis you sometimes hear is that the spring has become "weak", but
that's unfortunately not true. The cause of the problem is that the oil that
lubricates the axle of the spindle arm has become dirty and gummy,
sufficiently so that the little spring can't overcome the resistance of the
gum. The arm should snap (essentially fly) back, once it's released. The
diagnostic character for a gummy axle is that it moves back slowly. As it
progressively gets worse, it will move more and more slowly, until it reaches
that point where it will not have the tape back under the DDS cartridge's lid
in time.

Correcting this problem is a bit of work. The Sony drive has to be removed
from its HP enclosure. The spindle arm axle is mounted on the bottom side of
the drive. Carefully remove the retaining clip and extract the spindle arm.
Clean it and the axle shaft with the degreaser and very lightly relubricate
the shaft with a very good sewing machine oil. Reassemble the spindle arm
assembly, reseating the top-mounted spring, and replace the drive into the HP
enclosure (if you remember how it all went together :-).

Basically, these two simple cleaning procedures should, under almost all
normal circumstances, completely clear the drive from eating tapes.


Problem 2: Excessive cleaning messages

There are two motor-driven devices that are critical to writing data onto the
tape. The first, of course, is the rotating helical scan head. It has four
heads, each with a rather large gap, and each recessed a bit into the surface
of the drum. Because the gaps on the heads are large, it is difficult to find
a magnetic particle big enough to actually bridge the gap and create the
equivalent of a magnetic short circuit. That's not to say that it can't
happen; but it shouldn't be considered the primary or most probable cause of
most failures. Nonetheless, cleaning the recessed heads themselves can be
done with a very lightweight, very soft brush, wetted with head cleaner
solution.

In general, the helical scan head will not be the source of your problems.
The heads themselves never touch the tape and rarely get dirty. Indeed, if
there tends to be any problem with the head, it is that it has had some tape
oxide residue attach itself to either the stationary portion of the tape path
or onto the drum itself. If this should happen, the small imperfection
represented by this oxide will cause the tape to "fly" off of the surface of
the rotating drum -- and thus reduce the amount of magnetic field intensity
that is laid down by the heads. Worse, of course, is the fact that the
retrieved (read) signal drops by the square of the signal decrease caused by
the tape not flying properly.  Cleaning both the stator portion of the tape
path and the rotating drum is normally very straightforward and very easily
accomplished with a Q-tip, wetted with tape head cleaner solution.

It may be surprising, but the primary source of the "cleaning requests" is
due to the fact that either the capstan spindle, or pinch roller, or both,
has gotten dirty -- not the rotating read/write drum. The capstan spindle's
diameter is very small on a DDS drive, in comparison to a standard VCR, yet
it is the mechanism by which the tape is dragged across the helical scan drum
at a (supposedly) very constant speed.

If the capstan pinch roller has become highly polished (and thus hardened) by
residue from the backside of the tapes, or by ambient particulate dirt, then
the pinch roller can no longer elastically sufficiently conform itself to
cause a tape to move at a constant speed. If the pinch roller is inelastic,
slippage is somewhat inevitable -- and the result is a tape moving at an
inconstant speed, dependent on tape tension, drag, humidity and temperature.
The result is technically called "wow" -- and it makes the tapes difficult to
read when interchanged, drive-to-drive, or from one month to another on the
same drive.

Most often however, the more severe problem is a contaminated capstan
spindle. The spindle is a small diameter, highly polished rod-shaped piece of
metal. Because of its small size, if any contaminate should stick to the
spindle, it will change the diameter of the spindle rather significantly.
Should this happen, not only will the tape be moving at an inappropriate
speed, a form of jerky "gearing" action will be induced into the tape's
motion, producing "flutter." The flutter is intolerable, more so than wow.
The flutter prevents the scan head from reliably detecting and tracking the
signal that has been (or is currently being) laid down on the tape.

Cleaning the spindle is easily accomplished, but because it is partially
ensconced in a sheath, you need to be sure that you get at all sides of the
capstan spindle. Similarly, cleaning the capstan pinch roller is easily done.
Use a pair of long-nosed needle-nosed pliars to pull the pinch roller toward
the rear of the case, where there is more open room -- and then just lift up.
The roller is not attached, other than by gravity, to its axle. Without ever
touching the rubber portion of the roller with your fingers, use a clean
emery board to very gently take away the worst of the polish off of the
roller and then thoroughly clean the rubber surface with the CFC-based tape
head cleaner solution. What you should have when you're done is a matte black
surface, perfectly round.

Non-constant motion due to capstan/pinch roller contamination is the most
frequent cause of "too many retries"/"excessive cleaning" requests. It costs
virtually nothing to eliminate these problems, if you wish to do it yourself.
But it is a delicate operation, so don't attempt it unless you are perfectly
comfortable that you can pull it off with ease, grace, elan and aplomb.

Although I personally have none of these qualities, my Scottish/East Texan
heritage makes me so cheap that I find that I am easily irritated at the
thought of paying $500 for a $0.50 cleaning operation, and thus I do it
anyway.

Wirt Atmar