Broadband disrributed impedance matching

Depends what you mean by "distributed impedance matching".

It's possible to transform impedances over a very wide range of frequencies, by using a slowly-tapered transmission line.

In general, no... not unless you want to confuse things.

A passive filter always has a gain less than 1... it either absorbs and dissipates, or reflects those frequencies that it doesn't pass, and those that it does pass, always have some loss. A filter can be designed in a way such that its output voltage is greater than its input voltage, but only at the expense of the current being reduced by at least that same factor. Or, vice versa - you can get more current out of a filter than you put in, but you'll necessarily get less voltage. Both of these are simply forms of impedance transformation, and no power is added to the signal.

Now, many devices that we informally call "filters" do *contain* an amplifer, and deliver a power gain of >1 at some of their passband frequencies. The two functions may be intertwined (e.g. you may have an op amp, with filtering taking place in the feedback loop, but you'll have to have *some* amplifying element present, if the overall "filter" is going to have a power gain of >1 at any frequency.

It's not the same technique, but non-1:1 transmission line transformers can do wide-band impedance matching.

shows an example of 4:1 impedance matching transmission line transformer.

Matick R.E. "Transmission-line pulse transformers - theory and application" Proceedings of the IEEE vol.56, pages 47-62 (1968) goes a bit further.

Resistors and transformers are all that come to my mind. There should be the equivalent of an optical multi-layer dielectric coating, in the transmission line world. (one issue with multilayer things is that they need time for the various reflections to add up at the output... the step response stinks.)

George H.

Terminators, attenuators, splitters, transformers, and couplers can work over arbitrarily wide bandwidths.

Filters can't have power gain, and always have some amount of loss. A filter can have voltage gain.

Bill gets the prize for the best answer, the OP gets the prize for the most ambiguous question.

The design direction, and accompanying hand-waving argument that I've been given (via various ARRL publications) is that you make a multi- conductor transmission line assembly that satisfies your matching criteria at 1/4 wavelength, and then wind it around (or stuff it through) a core of ferrite or iron powder or whatever. The thing "looks" like a plain old transformer at low frequencies, and like a transmission line structure at high frequencies.

I make transmission-line transformers by winding micro-coax on a high-mu ferrite core. There's only a couple of inches of coax on each one. Risetime is way sub-ns and low frequency pulse response is limited by core saturation. We do 1:1 up to 50 volts and step-up 1:2 to get 100 volt pulses.

Here's a 50 volt pulse.

Speed is limited by the drivers, not the transformers.

There was no consideration of wavelength here. It's just a transformer with very low leakage inductance.

I've never made a transmission line transformer. Is there some reason they are "better" than a normal transformer. You don't have nearly the range of turns ratio's as you do with a regular transformer.

George H.

They're supposed to work better in the higher frequency ranges.

Electrical length is the fundamental property: capacitance and inductance are mere approximations.

But that's alright, because as you say, the input risetime is much longer than the electrical length, and even with a 2:1 impedance mismatch (depending on where that stepup is being done; I forget), that only effectively increases your electrical length by the same amount. So, it's a well designed transformer for the application.

Tim

All transformers are TLTs. It's a fundamental concept, not just an engineering trick!

If all your transformers have had layers with mismatched numbers of turns (and stacks of layers per winding), then the transmission line properties will be rather poor, but you always get the 1st mode behavior. Which gives the LF approximation that is the conventional 1st-order transformer model: Lp, Ls, DCR, LL and Cp.

Simply noting that, if you keep the primary and secondary strands close together throughout their windings, gives you some crude approximation of TLT (as the engineering design concept), and can do very powerful things: like switchmode transformers with fast risetime, minimal turn-on current surge, and approximately zero common mode EMI!

Tim

A rock is a transmission line too, just not a very good one. ;)

Cheers

Phil Hobbs

Direct experience? Articles to refer to? It sounds believable on the surface, but something deeper than hand-waving would be nice.

I just fiddled with it until it worked.

With eqivalent current loss, of course.

Hah, that basically is my experience with RF transformers and matching. (Long ago.. I also had some ringing in an interment stage, And that screwed up how the pulse turned off for an RF nmr thing. (the RF guy at princeton, found the ringing....)

JL, if you ever have a web page of "bad circuits" I've got plenty, (fortunately few went to production.)

George H.

OK I'd need pictures to know how all those went together, Previously we were talking about inductors as transmission lines, I must admit I've only got a low frequency model of an inductor/tranny, All the turns are linked by the same changing magnetic flux and develop an emf on the ends.

Sure, a transformer made with twisted pair looks like a transmission line at high freq.

George H.

I just designed a really good bad circuit. A photodiode cascoded into a microwave MMIC amp. Didn't work. Luckily I had another circuit on the same board, cascode into a fast opamp, and that worked.

I need a web site for all the crazy stuff that I know. I need a web lackey to do the work.

be

Twisted pair is less bulky - particularly if you twist enamelled transforme r wire. It's a bit harder to work out which end of the transmission line is which - I cheated by using green- and red-enamelled wires - and the charac teristic impedance is higher, but the transformers work just as well, up to the high frequency limit set by the length of the transmission line.

These too?