<azonenberg>
hoping to write a 1000baseT decode with this in a few weeks
<azonenberg>
or at minimum get some pretty PAM-5 eye patterns
<tnt>
Are those magjack btw ?
<azonenberg>
No
<azonenberg>
Plain old straight through RJ45s then series 0402s that I will load with either caps or 0R's TBD
<azonenberg>
it is incompatible with PoE because the directional couplers are DC short to ground on the pairs
<azonenberg>
if i load 0R's it'd short PoE, so I plan to start testing with caps and see if I get acceptable results
<azonenberg>
that would still not *pass* PoE but it'd fail safely (open circuit, other side gets no power) vs shorting the 48V to ground
<tnt>
Where does your gnd come from ? I'm a bit curious how it's wired tbh. Because eth is 100R balanced right ? So in a ideal world, you'd need only 8 directional coupler, each 100R balanced, but you're using 16 x 50R SE ones.
<azonenberg>
Correct. I was not able to find any 100 ohm balanced ones
<azonenberg>
The directional couplers have a DC ground port on them
<azonenberg>
So that (on all pairs) and the coaxial grounds are all tied together
<azonenberg>
then the RF signal is coupled onto it via some kind of transformer structure
<azonenberg>
its a bunch of coils on some kind of ferrite looking core, but i don't understand the E&M of how the actual coupling works
<tnt>
Well, I'm sort of wondering if it's not the same as in the paper I sent, to be that broadband. That's usually the structure used in VNA AFAIK. And all the coil stuff is just because in the basic resistive structure, the ports need to be floating (different gnd) which is very inconvenient and so with some baluns, you can get rid of that requirements.
<azonenberg>
I'm not sure. It was the most broadband low cost SMT coupler I was able to find