[TriEmbed] OT-ish: TVS best practices?

[Pete Soper]

The question is whether a surge will cooperate and just develop a 
potential on a convenient path. Everything I've seen over the years 
suggests these surges, most especially when induced by nearby lightning 
strikes, are really bad about cooperating. So we read stories about this 
wire vaporized but that one left untouched, etc.

Chip's been using surface mount TVS 
<https://en.wikipedia.org/wiki/Transient-voltage-suppression_diode> 
arrays. From memory I think the answer for his setup is being able to 
clamp between two arbitrary conductors. I think that's six for four 
wires, but (if I recall correctly, and unfortunately I can't look it 
up), that's just one little package.

I think it's a very good thing for you to raise this subject. In our 
part of the world, in my opinion, having long wires strung around sets 
up a "when" question for surges, not an "if".

-Pete

On 10/30/2017 01:57 PM, Brian via TriEmbed wrote:
> Hi Group,
>
> I've been working on a design for zoned HVAC which involves having 
> electronics at the ends of long cables.  As such, good transient 
> voltage suppression becomes a more important concern.
>
> My device will be connected via Cat-5 UTP across which will be 
> supplied +6V, GND, and an RS-485 differential pair.
>
> I'm wondering if anyone in the group has suggestions on the ideal 
> application of TVS elements.  I've seen reference designs for RS-485 
> that have paths to ground for each of the signal lines, but I've also 
> got the +6V line to worry about.
>
> Can I just provide paths from each signal to GND, or do I need to 
> provide paths from every signal to every other signal?  It seems to me 
> that theoretically, all four conductors could be floating at different 
> potentials relative to each other, and clamping to GND wouldn't 
> necessarily protect against a damaging potential between a 485 wire 
> and +6V.
>
> All input greatly appreciated!
> Thanks,
> -Brian
>
> _______________________________________________
> Triangle, NC Embedded Computing mailing list
> TriEmbed at triembed.org
> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
> TriEmbed web site: http://TriEmbed.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20171030/ca7230d0/attachment.htm>