TDR with multiple shorts

You will see the first short, as a short, and nothing else. There's no voltage downstream of the first short.

TDR is often imagined as a graph of impedance vs distance, but that's just the ideal/imaginary case. In real life, every feature along the line, including normal losses, changes both the outgoing and return steps.

Thank you, John, this answers my question (and kills a weird idea).

Best regards, Piotr

Well, if a full short, that's right. If a "weak" short, ie. it has significant resistance, then you might be able to see beyond it. If you have multiple partial shorts, or multiple impedance changes all spaced equidistant, then you will have a mess that would take a computer to deconvolve, with reflections traveling back and forth between the discontinuities. So, that would severely complicate what you propose.

Jon

Some things can't be deconvolved. Imaging TDRing a coax that has a 60 dB attenuator somewhere along the line. The attenuator is invisible, but sure affects anything that you think you see downstream.

The other issue with deconvolving reflections and such is noise.

The telcom TDRs can detect bridge taps and load coils in copper, somehow.

It would be nice to massage TDR or VNA S11 data to yield a clean impedance vs. distance plot, if the discontinuities aren't too wild. I don't see off-hand how to achieve this, but somebody somewhere must have done it already?

Jeroen Belleman

Geologists have done a lot of work in untangling TDR data.

My old 11802 does just that. The impedance at each cursor dot is displayed.

Maybe, but does the displayed value still make sense after a preceding reflection?

Jeroen Belleman

In general, no. It is useful on a PC board, but it becomes obvious that trace losses and vias and things smear out the resolution as you go farther out the trace.

As John said, the deconvolution isn't unique, because a perfectly-matched attenuator doesn't cause any reflection but does knock down the returns from anything following it.

It's much easier in optical fibre, because Rayleigh scatter produces a continuous return signal from all distances. The slope of that signal tells you the loss per unit length, and any step discontinuities mean that there's a matched loss, e.g. someone from a TLA tapping into it by bending the fibre.

Optical TDR is a cool technology--you can map a whole multitap network, which you can't in coax on account of not having the continuous Rayleigh return.

Cheers

Phil Hobbs

Does anybody male photon counting OTDRs for use in the field? If so, how is the detector cooled?

I considered that ~1980 in conjunction with BT Martlesham Heath, but couldn't find acceptable cooling technology.

I also wondered if there were any coding schemes that could increase range. Given a peak-power limited transmitter, they might allow there to be more power in the fibre at any one time. The best code I found was a Barker code, but with a max length of 13 (IIRC), it wasn't worth the effort.

has a pretty good discussion of the tradeoffs circa 2009.

In the early '80s, working at 1.3 um, you would probably have been stuck with an S-1 photocathode, right? They're famously bad for dark counts even at low temperature--you have to get them down to 77K or thereabouts to get decent sensitivity.

Cheers

Phil Hobbs

Thanks, and yes, the temperature was a bit too low to be practical!

Can the data circuit optical TDRs see things like splitters/prism/taps? How do they show up? If so does the direction you test them matter? Years ago I was talking with some datacenter people about how the tapping warrants work in facilities with lots of dodgy customers. The answer was it's pretty easy as all optical circuits already have passive optical taps installed. If a customer asks or notices, they're told it's for testing, which is legit and true. When the feds showed up the warrant, they just plug into the "test" ports. Even with a nice clean fiber connection, you're still at the mercy of the optics at each end, and somehow those go bad or just wear out. I've never gotten a solid answer as to why this happens either. Does the laser start to fade out? Does the receiver become less sensitive? It didn't seem to be an issue with older stuff like gigabit ethernet. 8Gb, 10Gb, 16Gb? They all seem to fail in weird ways.

Yes.

As a localized loss without a significant reflection.

Yes. Returns from the branches will in general arrive in a different order seen from the other end.

The lasers have a limited lifetime.

Cheers

Phil Hobbs