Re: Need help with zip drive

[glyn@cs.ualberta.ca (Devin Reade)]

In article <199710241000393074021@slip129-37-106-92.wa.us.ibm.net>,
B.J. Major <bjblackbear@ibm.net> wrote:
>> In article <19971021135719262185@slip129-37-106-177.wa.us.ibm.net>,
>> B.J. Major <bjblackbear@ibm.net> wrote:
>> 
>> >And yes, like any other
>> >SCSI item, its power MUST be on before powering up [any] computer
>> >system.
>
>Devin Reade <glyn@cs.ualberta.ca> wrote:
>> This is not quite accurate; some machines work quite well when their
>> SCSI devices are powered up at the same time as the motherboard.  Take
>> the stock Sun machines for example.  It's really a matter of timing;
>> as long as the SCSI devices are ready to respond when the controller
>> queries them, it shouldn't be a problem.
>
>But if you powered them on FIRST, you wouldn't have to be concerned
>about whether they were going to be ready or not.

Sure, no dispute there.  It was your blanket generalization, and now
your insistence that it will fry components, with which I disagree.

>> Even on my GS I don't have a problem with simultanious power-up.  
>(snip)
>
>Your luck has not run out yet, but I wouldn't push it.  There is
>*always* a possibility that SCSI devices won't be recognized if they are
>not fully powered on before a cpu. From what I've read, you run the risk
>of either frying out the motherboard on the cpu (with the sudden power
>surge of all the devices running through the already-powered-on cpu at
>once) OR ruining your SCSI port/connection, period, if you do not power
>them on before the cpu. 

You're talking about two different things here.  Sure, the devices may
not be recognised if they're not powered on first.  This does not have
anything to do with frying motherboards or other components.

Unless you are physically attaching/detaching cables when one or more
devices is powered up, you run _no_ risk of "surges".  When a motherboard
or a peripheral does a power-up, there may be a short period of time
when the bus is in an undefined state; signals cannot be interpreted as
meaningful.  This is when the digital signal fluctuates between its
two (three, for high-Z logic) states.  It is, however, well within the
voltage limits of the h/w.

You could also look at it this way:  The motherboard/controller and 
the drive logic are, in the scheme of things, all the same parts.  They
are compatible families of digital circuits.  If the mb/controller
could be fried by turning it on before the peripherals, then the 
peripherals could be fried by turning the peripherals on before the
motherboard.  Electricity runs in both directions.

As far as your manuals are concerned, if they do in fact say "you will
fry your motherboard if you power up your motherboard and peripherals
at the same time", then they are wrong.  Keep in mind that such manuals
are written mainly for non-techies.  Futhermore, manuals tend to be
edited *by* non-techies.  Consequently, they often contain well-intended
but misleading information.  This would not be the first time manuals
contained false information.

>The risk is analogous to plugging in a keyboard
>or mouse while the cpu is powered on--although in the SCSI case, there
>is *much* more AC involved.

They are not analogous.  Making physical connections is not the same
thing as merely doing a powerup.  I agree that for *most* (not all)
computing systems, making physical connections when anything is powered
up is a bad idea and can cause fried components.  (An example of the 
exception are disk arrays that use "hot-standby" and "hot-swap"
disks; when a disk problem is detected, it's contents are transferred
to the standby disk, then the bad drive can be removed and replaced
while the array is still powered and in use.  This is used to achieve
zero downtime.)

BTW, there is *no* AC directly involved with a SCSI bus.  The signal
lines are all either GND, TERMPWR, or signal, the latter two of which
are between 0 and 5.25 VDC.

Devin
--
	Devin Reade	gdr@myrias.com