[oclug] tape drive recommendations

[Brad Barnett]

On Wed, 27 Mar 2002 14:37:01 -0500
"Strosberg, Bill" <bill.strosberg at rcpsc.edu> wrote:

> > From: Brad Barnett [mailto:bbarnett at L8R.net]
> > Sent: Wednesday, March 27, 2002 1:52 PM
> > To: oclug at lists.oclug.on.ca
> > Subject: Re: [oclug] tape drive recommendations
> 
> <snippage>
> 
> > No, you don't :P  Plugging in the 5+ or 12+ first is _not_ 
> > going to result
> > in a dead drive instantly.  Are you even aware that you break ground
> > usually to kill flow of power?  I don't see many light bulbs 
> > exploding due
> > to this fact, DC powered or AC.
> 
> Brad:
> 
> Plugging in the +5VDC and +12VDC at the same time in absence of frame
> ground will result in a +7VDC potential difference, as the ground
> reference changes from 0 to +5VDC.  Dumping +7VDC across 12VDC drive

7VDC if you take the reading across those two connectors.  However Bill,
I've yet to see a hard drive that doesn't have a diode after the +5 and
+12 inputs.

In short, how is the power going to flow?  Until you can respond to this
point, it doesn't make much sense for me to continue to respond to the
rest.  Even if it is well laid out it is, and politely put. ;)

> motors, and/or across+5VDC ttl logic can definately result in cumulative
> damage.  Most equipment today is very reliably decoupled and protected,
> therefore you USUALLY can get away with hundreds of cycles of all kinds
> of BAD practices.  The highly integrated devices of today are a far cry
> from the "collective farms" of discrete logic we used ten years ago. 
> This doesn't mean you should be recommending people do this.
> 
> As with any circuit board level situation, most problems (statistically)
> occur during inrush and startup - without reliable ground reference and
> voltage supply, many components can be started in undefined states, with
> spurious signals/states generated.  A lab day with a good logic
> analyzer, bus probes and a variable transformer, combined with a set of
> data reference books will demonstrate the truth of this conclusively.
> 
> This is a situation where "seat of the pants experience" is trumped by
> theory and electrical engineering principles.  I hope you never have the
> (admittedly rare) situation occur where you (or your clients) are proven
> painfully wrong.
> 
> On another note, all the reliable, enterprise quality hot swap raid
> equipment I have had experience with does establish frame reference
> ground well in advance of power supply application - usually through
> external chassis contacts made during the insertion/removal phase.  The
> ground pins in the physical P/S connector are secondary and redundant.
> 
> I'm not disagreeing with or doubting your "practical" experiences, just
> trying to outline the dangers to others using these practices in
> commercial applications.  If the data is critical enough to consider
> no-interruption backup facilties, it is worth spending the money to do
> so right.
> 
> --
> Bill Strosberg
> 
> 
> 
> _______________________________________________
> oclug mailing list
> oclug at lists.oclug.on.ca
> http://www.oclug.on.ca/mailman/listinfo/oclug
>