[Discuss] static discharge protection when building a PC

[tmetro+blu at gmail.com (Tom Metro)]

Steve Litt wrote:
> Obviously, use static protection. I'm not going to tell you to wear a
>  wrist strap, because that can put you in danger in a shock
> situation...

Commercially produced anti-static wrist straps have an integral resistor
to limit current flow through them to non-lethal levels in the event
your body comes in contact with a ground-referenced high-voltage source,
like a power line.


Edward Ned Harvey (blu) wrote:
> Normally, ESD is not a static shock lightning bolt; normally it's not
> even something you can feel or see. ... Normally, ESD is very
> subtle, like you just reached across the table and in doing so your
> clothes rubbed on each other and created a static charge in your
> body

True.


> It's also not normally something you notice after you power on the
> device. [A static discharge can] degraded the life of one transistor
> somewhere in some chip.

True. It weakens the insulating layers in the chip.


> So the symptom of *typical* ESD, which most people don't
> notice, is a shorter system life and/or increased flakiness. Only the
> most dramatic and extreme ESD results in a catastrophic failure,
> where some chip is rendered instantly nonfunctional. 

Right.


> It only takes a few volts to exceed the specs of your device
> and cause damage...

However, going back to like the 1980s (when CMOS - the kind of
fabrication that is most susceptible to ESD - was rising in popularity),
integrated circuits are generally produced with built-in ESD protection
via internal resistors. They're not ESD proof, but but it reduces the
chances of damage.


> Don't touch the case every few seconds. Touch the case literally
> constantly (never let your hand leave the case) as long as your other
> hand is touching any of the electronics.

If you are in a production environment or handling very expensive gear,
then using a strap is advisable.

In my opinion a strap is unnecessary for the occasional PC build if you
follow good practices. Discharge any gross static build up by grounding
yourself before toughing your components, and limit your handling to the
edges of the boards or other insulated or grounded components on the
board, like brackets or heat sinks.

The exposed pins on connectors are already ESD protected. The large
metal objects on the board, like heat sinks are grounded on the board,
and even if the board is ungrounded, a discharge to these components is
unlikely to travel through an IC.

As Ned points out, ESD is more than just the big, visible sparks, but
the smaller stuff that you can't see also can't jump the insulating
paths between the exterior edges of the board and the ESD protection of
the board's I/O pins to reach the ICs.

PC components are designed to withstand a certain amount of handling by
untrained consumers and careless assembly labor.

Just don't go sticking your fingers on the chips or solder pads on the
board, and you'll have sufficiently reduced your probability of causing
damage.


> Of course, touching the case does precisely jack shit, if you have
> your motherboard not inside the case, or not grounded to the case.

If the power supply is mounted in the case, and plugged into the a
grounded outlet, then the case is grounded.

If that same power supply is powering the motherboard, then the
motherboard is grounded.

Touching the case will discharge your potential relative to the motherboard.


[Re: running the motherboard external to the case]
> Woah - that's a terrible idea. When you screw the motherboard down to
> the case, you'll see that some of your anchor points are
> nonconductive and ok to anchor using nonconductive materials (little
> plastic snaps) but some of your anchor points are lined with brass or
> aluminum, and must be screwed down to ensure multiple points of
> ground.
> 
> Yes, of course, your motherboard will be grounded by the ground wire
> in the power harness, but something people often don't realize is
> that the speed of light crossing the distance of the CPU is actually
> significant - You've got a CPU doing 3 * 10^9 instructions per
> second, and the speed of light is 3 * 10^8 m/s. Which means light
> travels 10cm in the period of time your CPU does each instruction. So
> if some particular instruction gobbles a bunch of power and dumps it
> onto ground, those electrons haven't even had time to reach the
> ground yet. So you get a noisy ground.

Ummm...this doesn't have a whole lot to do with the speed of light. It
does have to do with current and clock frequency.

A modern motherboard uses a multi-layer design, which will include a
groundplane within the PCB where and if needed. The design should not
rely upon the metal mounting plate, however that should serve to improve
 (reduce) EMF and make it easier for a completed PC to pass FCC testing.

People regularly run motherboards outside of a case. Increased ground
noise and EMF emissions isn't going to impact your ability to qualify
that components are in basic working order before installing them. (I
often assemble systems outside the case for a preliminary check out.)
But you may spot a difference in performance benchmarks if not using a
case or test bench (though I doubt it; it would suggest a flawed
groundplane design on the motherboard).

 -Tom

-- 
Tom Metro
The Perl Shop, Newton, MA, USA
"Predictable On-demand Perl Consulting."
http://www.theperlshop.com/