HDD died in a spectacular way

From: James Willing <jimw_at_agora.rdrop.com>
Date: Sun May 3 11:13:15 1998

At 04:53 PM 5/2/98 -0400, you wrote:
>Anyone ever seen a worse failure?

Well, seen after the fact when I was called in to 'repair' it... And
actually, quite a good thing that no one was there to witness it! (you
will see why in a moment)

Probably over 15 years ago, NorthStar offered a 14 inch hard drive in an
external cabinet for use with the 'Horizon' microcomputers. It was amusing
to watch with the top of the enclosure off as the entire unit was enclosed
in a slightly smoked brown Plexiglass 'bubble'.

The down side was that the sector wheel/transducer/tachometer assembly was
external to the sealed housing and frequently caused "sector not found"
errors as dust collected on it. This required regular (monthly or so)
cleaning which required removal of the HDA from the external enclosure in
order to access the bottom of the assembly. (quickly rectified by our
staff after a couple of calls by using a 'nibbler' tool to add a 2"x3"
opening on the bottom of the external enclosure that we could access the
assembly through)

It was also noted in one of the service bulletins that since this assembly
also served as the tachometer for the spindle drive, that you could tell if
the wheel was becoming dirty by a "surging" sound coming from the unit even
if you did not experience sector errors. Little did we know...

We had one customer who tended to keep his system up 24x7 since he had an
external sales staff that used the system to file orders and he liked to
work from home. (dial-in lines) He also liked to run the HD with the upper
part of the external enclosure removed so that he could show off to
customers and clients just how advanced their operation was. (ignoring our
warnings that this would allow the unit to attract dust and dirt more rapidly)

Well... One Monday morning I get to the shop and we have a number of
messages on the answering machine (in increasing levels of agitation). He
starts off by explaining that Friday evening he started getting frequend
'sector' errors reported from the system. Over the course of the day
Saturday the errors increased and the system response degraded. Sunday
morning the system would not answer a call at all. Sunday evening he got a
call from the Alarm Monitoring company that something had tripped the
offive alarms. When he went in to check the building, he noted that there
"appeared to be a problem with the hard drive" and wanted us out there
first thing to make sure he did not lose any data.

So... a couple of us went out expecting to have to clean out the wheel
assembly (yet again) and perhaps correct a couple of glitched sectors...

When we entered the computer room, it was quite obvious that there was a
bit more than a "problem with the hard drive"!

The room looked like someone had stood in the center of the room with an
M-16, and used it to try to cut the room in half. Clear around the room at
about table top level were pieces of plastic and metal stuck into the wall
boards. The plexiglas 'bubble' from the HDA was gone (obviously shattered)
and the platters had large chunks missing from them. No need to even look
for the head/arm assemblies.

After an extended discussion with the customer, of which most of the time
was spent explaining that we would NOT be able to recover any data from the
drive, we set about collecting the wreckage and installing a new drive and

Once back at the shop, a call to NorthStar brought a visit a few days later
from one of their technical support staff as well as an engineer from the
manufacturer of the HDA (Micropolis if I recall correctly).

The "official" explaination went something like this:

Due to the design of the (rather basic) tachometer circuit in this unit, a
missing sector pulse was intrepreted by the tach as a loss of spindle
speed. (start-up mode mentality, if you are starting the drive and don't
see a pulse, speed up (or start-up) the motor until the required pulse rate
is achieved) This was the cause of the "surging" sound noted in previous
service bulletins.

In "theory" (now apparently proven) if the sector/tach wheel became
sufficiently dirty the tachometer circuit could attempt to keep increasing
spindle speed up to such a point where a failure of the unit might be
induced either by overload failure of the spindle motor or by excess
vibration caused by excess rotational speed.

Apparently, no one ever considered a 'fail-safe' for the tachometer circuit...

While they never did detail just exactly what failed first, whatever it was
apparently caused a head crash severe enough to fracture one or more of the
platters, and from that point it was all downhill...

Without the cover of the external enclosure to contain it, (a metal
enclosure by the way) when things started to come apart there was nothing
to stop it! And the results were indeed quite spectacular.

Made everyone a good deal more serious about keeping those sensor wheels
clean... (and the covers on!)

BTW: NorthStar did replace the drive for us...


The Computer Garage - http://www.rdrop.com/~jimw
Computer Garage Fax - (503) 646-0174
Received on Sun May 03 1998 - 11:13:15 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:31:11 BST