On Sunday, April 28, 2019 at 1:19:56 AM UTC-4, wrote:
On Sat, 27 Apr 2019 21:43:45 -0700 (PDT), Its Me
wrote:

On Sunday, April 28, 2019 at 12:31:07 AM UTC-4, Bill wrote:


Why were the MOV to ground, instead of across the lines? Or a bigger value
Varistor to ground?

Because of the twisted pair, the danger isn't usually voltage spike across the pair, but rather the spike potential from the pair to ground. That's what we were trying to protect from. And what protection components on 66 punch blocks from back in the day did as well.

As far as the value, it's a bit of a tightrope. Too low of a value, and it's always firing and causing issues like we experienced. Too big of a value, and you may as well not have any protection on there at all. Even a transformer doesn't protect you, as it has an arc-over value. We thought the 180v parts would be OK, but we didn't realize that the lines would be as dirty as they were.


===

I'm wondering if you couldn't use an isolated op amp with differential
inputs to extract the signal. It would have to be totally isolated
from any ground reference to protect from the high voltage common mode
spikes. It would need a floating power supply of course, with the
output through an opto isolator or some such.

The perfect isolator for those common mode spikes is a transformer. It doesn't have any reference unless you give it one. You can certainly use an op amp to terminate and receive that twisted pair signal (and we did at times), but a large enough spike will take it out. There's no way to completely protect it. Once you start jumping air gaps inside the IC, or exceeding the reverse voltage spec on junctions in there, things start blowing up.