On 1/13/12 12:11 PM, Stan Labinsky Jr. wrote:
> Jim, hello;
>
> Your question about sharp bends caused me to re-investigate that statement.
>
> "http://www.protectiongroup.com/ProtectionTechnologyGroup/media/PTG/WhitePapersandTechnicalNotes/1485-013.pdf?ext=.pdf";
>
>
> is where it came from. I've seen the good result from following this
> group's recommendations.
but the question really is, are their recommendations overkill.
For instance, they talk about skin effect, which is real, and does
increase AC resistance, but in a typical system, AC resistance isn't the
problem, inductance is, and flat strips just don't have that much less
inductance than the same cross section of copper in some other shape
which might be easier to use/cheaper, etc.
For instance, they say:
"The grounding of the plate or panel is very important. A low impedance
path to ground is a necessity and only copper strap should be
considered. Each piece of equipment in the equipment room must be bonded
to the single point ground panel with a low inductance strap."
And had you asked me 2 years ago, I would have said, sure.. But that was
before I got interested in inductance formulae and discovered that strap
isn't much lower inductance. And nobody will question that it's a royal
pain to work with flat strip compared to wire.
Sure, if you are concerned about low impedance, period, as the
combination of resistance and inductance, strap is where it's at.
The thing is.. for lightning (and other transients) (as opposed to, say,
15 MHz RF), the inductive impedance of the conductor dominates, even for
fairly small wires.
Take the 1 microsecond rise time lightning scenario..
Inductance for a wire is about 1 uH/meter. so the voltage rise is
E=1E-6 dI/dt = 1E-6 I peak/trise as an approximation or E = Ipeak for a
trise of 1 microsecond (the 1 microhenry and the 1 microsecond
essentially cancel).
The resistance of 1 meter of AWG 10 wire is about 0.003 ohms. Ohm's law
says E = Ipeak R, so we have E= 0.003 Ipeak.
So the voltage rise due to the inductance is 300 times that due to
resistance. Sure AC resistance is greater than DC resistance, but it's
nowhere near 300 times. Yeah, and sure, a strap with the same cross
sectional area as AWG10 might have 0.8 uH/meter inductance.. Again,
you're not talking about factors of 100 here... The inductance still
dominates.
ALl I could find on bends is:
"No Sharp Bends
Route all ground straps and grounding conductors so they have a gentle
bending radius. Bends sharper than 8-inch [203.2 mm] radius will add
unwanted inductance to the desired ground path. Even for conductors
buried in the ground, try to prevent sharp bends."
Sorry.. The physics for this inductance just aren't there. Let's take
a really extreme example: a 360 degree turn, with 8 inch diameter..
basically a 1 turn coil with 4 inch radius. (call it 10 cm)
The inductance of a loop with 10cm radius and 2.54mm diameter wire
(AWG10) is 0.56 uH. (per equations from Terman) A straight wire 62cm
long will have inductance about .6 uH (using the 1uH/meter rule of thumb..)
That's practically the same.
And for 90 degree bends, the difference is MUCH less, because less of
the flux from one part of the wire couples to another.
A curly multiturn pigtail is bad, no question.
--------
So here's my take.. There's a pile of info out there on how to do
transient protection. A lot of it is derived from copying from
somewhere else, without going back and looking at the underlying theory.
A lot of it is "conservative" in the sense that it's certainly no
worse than other approaches.
Polyphaser doesn't really care if strap or wire is better. And if they
propose using wire to someone who believes (because they read it
somewhere) that you gotta have strap, they won't get the job. If they
propose using strap to someone who knows that strap and wire are
equivalent, and the cost is reasonable, then the buyer probably doesn't
care, and is really basing the decision on something else.
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|