[time-nuts] ThunderBolt question

[John Moran, Scawby Design]

Everyone ...

I must admit to being amazed at the cavalier attitude to impedance matching. I dread to think what a state we would be in if the original telecoms networks were designed with such disregard.

OK, my background is in the old telecoms - land-line stuff where we had a variety of impedances to work with, balanced and unbalanced, but mainly 600 ohms for the audio and line side and 50 or 75 ohms for the internal stuff. But whatever we were working with, designing to match the impedance closely was a critical parameter, and not difficult at all. Regarding connectors, you could mix and match types as long as everything you used was designed to match the same impedance and terminate the cable properly.

When you are designing amplifiers to be flat within 0.1dB over a wide bandwidth, impedance matching matters both for steady-state amplitude settings and ringing caused by the reflections.

There is a whole discipline around transmission lines going back nearly 200 years, for a reason.

OK, TimeNuts tend to be piping single frequencies around the place, but I thought this was a place looking for precision, and playing with low-level signals, and hunting down esoteric artefacts and anomalies. Wading roughshod through transmission theory is at odds with that for me, sorry.

It's not as though designing stuff to have the right input and output impedances is difficult. Nowadays with integrated amplifiers you can just use the brute-force method of hanging a 50 ohm resistor across a 'high-impedance' input and another 50 ohm resistor in series with a 'zero ohm' output impedance. Back in the day we designed them to inherently have the right input and output impedance and so saved throwing lots of signal away.

My humble apologies for the rant ... but I just couldn't believe what I was reading this morning when I opened the mail.

I guess I'll get thrown out for this ...

John