I must take issue with John's statement that "impedance matching is easy".
It's easy only in the case of very limited bandwidth and if you are willing to ignore such issues as getting good IMD performance, good noise performance, and good broadband AC stability. The last item is critically important because in many situations an amplifier is driving (or being driven from) a filter whose out-of-band port impedance is nothing remotely like the intended value within the passband, being highly reflective out of band. And the idea of just adding a matched termination across the input of a high- Z buffer automatically adds noise, and is impractical at high frequencies because at such frequencies there is no such thing as a high input impedance except perhaps over a very small fractional bandwidth. These factors (along with others) help to explain why good RF amplifier designers (which I do *not* claim to be) get paid the big bucks. And why, despite their best efforts, good input and output impedance matching of amplifiers remains an elusive goal. Dana On Sat, Jun 6, 2020 at 2:12 AM John Moran, Scawby Design < [email protected]> wrote: > 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 > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
