Squeal on Older Hard Drives

Tony Duell ard at p850ug1.demon.co.uk
Fri Feb 24 17:38:48 CST 2006


> It was a lot more serious than mere read errors.  The static charge on =
> the
> surface of the disks could build up to thousands of volts in just =
> seconds.
> And its main discharge path was through the heads - they were the =
> closest to
> the charge.  I changed many a head on the early 14 inch drives that had =
> a
> clean little burn hole right through the flying surface.

Ouch!. I am suprised there was enough energy stored to do that.

> 
> The static discharge springs had a metal button that fit on the end of =
> the
> spindle shaft.  It was metal to metal so there was always some wear.  If

Some of the ones I've seen seem to have a graphite composition (a 
bit like the material used for motor brushes) button on them to make contact 
with the spindle. Makes sense, you don't need a very low resistance path 
to ground.

> Very quickly, other schemes can into use that alleviated the need for =
> the
> discharge spring. The most practical was a ferrofludic seal on the top =
> of
> the bearings.  The design goal was a path of around 10K ohm to ground.  =


I thought that all winchesters with the motor outside the HDA had 
ferofluid seals, if only beacuse that was the only practical way to 
prevent dust coming into the HDA round the spindle. But some of those 
drives still have the earting contact.

> 8 inch drive with a discharge spring.  After that, the ferrofludic seal =
> was
> standard.

They were common on older 5.25" drives (heck, I think I've seen a 
SCSI-interfaced drive with one.)

One other thing bothers me. At least one of my drives has the earthing 
contact on the PCB, but that PCB is hinged to the chassis at one end, and 
is designed to be swung away from the chassis for maintenance (so you can 
get a probe on the testpoints). In that position, of course, the brush 
does not contact the spindle. Did the manufacturers just hope there 
wouldn't be a discharge when you were doing tests, or what?

-tony




More information about the cctalk mailing list