[TowerTalk] spark gaps & GDT's for lightning/static protection

[jimlux]

On 4/9/17 4:01 PM, Roger (K8RI) on TT wrote:
> Agreed, lightning is an RF event and the feedpoint is probably the best
> place to start.
>
> My thoughts on the direct path to the amp, were not toward a DC event,
> it's just a lot easier for the residual voltage to traverse. I say
> residual voltage as reducing the potential to around 300 volts, give or
> take means a lot less for the buried coax to reduce.

An air spark gap will get you no lower than 327V.. 500 is more like a 
reasonable goal.  An sealed gas gap can go a bit lower. And of course, 
there's semiconductor transient suppression devices.

  I'd also ground
> the coax shield with a network of ground rods (at least three spaced at
> 120 degrees and at least 6' to 8' spacing) where the coax enters the
> ground under the antenna.
>
> Perhaps a spark gap (with fine points) at the feedpoint followed by a
> gas discharge device where the coax enters the ground, lowering the
> pulse entering the coax to less than several hundred volts and something
> like a polyphaser where the coax enters the shack at the common point
> ground.  The inductance and capacitance of the 100' of coax should
> greatly reduce that voltage before it gets to the polyphaser/rig.
>
Actually, the coax won't reduce the voltage - it would depend on the 
source and load impedances, and if it's "matched" then it's just the RF 
or DC loss.  1dB would be, what, 12%...

Before the transient suppressor "fires", your coax is probably looking 
into a matched load (or close to it)




> IOW, tackle it in steps, or stages instead of one brute force method.
>

Definitely. Clamp early and often.

Think about this, too: is it energy or voltage that's the thing you're 
trying to deal with - some devices are sensitive to overvoltage, others 
are more sensitive to impulse energy, so something that clamps the 
voltage might not actually reduce the energy, it would just make a short 
pulse into a long one.



For semiconductor devices, a significant over voltage (e.g. above the 
supply voltage of the part) typically results in some diode becoming 
forward biased, and the damage is actually from excess current flowing 
through the circuit.  The traditional way on logic gates and such that 
have high input impedances is to put a series resistor to limit the 
current.

For instance, if you have a balanced line receiver (like for RS422) you 
terminate the twisted pair into 100 ohms (so the line is matched and you 
can send data at high speeds, then have 10k -20k in series into the 
megohm input impedance of the line receiver. 100V on the input through 
the 10k is only 10 mA of fault current, which typically is something 
that the on-chip devices can handle, particularly if it's short duration 
(milliseconds).