Re: [PLUG] Desktop Support ?

["Keith C. Perry" <kperry@daotechnologies.com>]

That's a good question.  I'm going off of personal experience, my ECE background and circuit design experience.  Hpwever, like with anything else, YMMV  :D

Designing the circuit as you say is not the problem.  Its handing the out of spec situation that is the challenge.  There are only so many conditions you can deal with intra-system (e.g. on a disk controller) before you have to start building out inter-system (e.g. the motherboard, bus controllers, other sensors, etc) and even then, there is still a range of what can be dealt with effectively.

A perfect example of this is understanding why the typical UPS is not a good choice for voltage regulation or power shaping (line conditioning).  Some units do some of that as part of their function but as I'm sure you know, there is no UPS that can **properly** deal with a lightning strike or brown outage (persistent voltage sag).  The reason for that is because the switching logic has a certain range of operation to trigger redirecting to its batteries and on the overage side, the fuse or MOSFET is designed to fail if too much power is detected on main power. This problem exists when you restore power from a blackout condition but its actually more serious for brown restarts because there generally aren't alarms for that so components erode quicker until they are out of spec and its that next power spike that gets through even though the fuse or MOSFET even if it burn out (i.e. enough power for enough time to ionize the air and close the circuit... and keep going).

At that point nothing is guaranteed to protect the system.  Most digital circuits are built to handle unknown states but they still have to be properly detected which means the component that do the detection must in spec.  Current is like water, it takes the path of least resistance, no pun intended :)  With enough power, todays small surface mount electronics are easy to short out or spark across a gap.  Traditional drives have motors and other induction (the importance of that will be hit on later) so they tend to do better than other components but that doesn't mean they always do.  Factor in drive age or whatever the drive was doing at the time and now you have that dynamic situation that can't be modeled.

After years of observing these types of things, I now use a custom power solution to provide UPS, line conditioning and power protection functions.  The component I use to addresses the overage condition is called a "Brickwall" (basically just a large inductor, http://www.brickwall.com/) which can take lightning strikes and then parasitically sink excess power to ground.  Current flows through inductors so there is no switching time.  You can think of them as always being on so, you always have passive inline power protection.  Hospitals use these types of devices because their high frequency equipment much like our typical computers are sensitive to power fluctuations.

Sorry to be so long but my point is that there is always a scenario that can cause failure even if it is normally (but thankfully) rare.  Just in past 3 years, I've lost 2 motherboards and 3 drives but it was from long term UPS / power issues.  With enough "random" events and testing it became obvious what the root cause was.  Last time these types of failures occurred for me was at least 14 years ago so I could be too mad :D

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 
Keith C. Perry, MS E.E. 
Owner, DAO Technologies LLC 
(O) +1.215.525.4165 x2033 
(M) +1.215.432.5167 
www.daotechnologies.com

----- Original Message -----
From: "Rich Freeman" <r-plug@thefreemanclan.net>
To: "Philadelphia Linux User's Group Discussion List" <plug@lists.phillylinux.org>
Sent: Wednesday, April 13, 2016 2:41:05 PM
Subject: Re: [PLUG] Desktop Support ?

On Wed, Apr 13, 2016 at 1:35 PM, Keith C. Perry
<kperry@daotechnologies.com> wrote:
> Things aren't quite that clean- once electrical systems go out of spec (e.g.
> under or overvoltage how that is caused... e.g. source, parasitic, cascading
> events creates a dynamic failure) all bets are off since it is impossible to
> detect and trap multi-system faults unless you have external supervisory
> circuitry in place.

Is there any kind of study on this?  That is certainly not my
expectation for a hard drive, and (not being an EE) I'd think that you
could design a circuit that stopped writing when the voltage supply
dropped.  The drive already has the ability to retract the heads after
a power failure, so it has some kind of capacitor/etc for such things.
If the drive is retracting the heads on a power failure that would
also preclude it from writing anything while those heads aren't over
the drive.

It sounds like somebody just needs to boot a rescue disk and fsck the
filesystem and see what is going on.  If it looks serious I'd probably
clone the drive before attempting repairs.

--
Rich
___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug
___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug