Do analog matched filters exist beyond the trivial cases?
I have seen some that use surface acoustic waves (SAWs) and some other simple cases, like an integrate and dump.
I understand that many people do this in the digital world, but I am trying to find out, are there analog matched filters for first or second order FIR filters?
Common filters reflect unwanted signals back to a source. To make filters matched, isolators or circulators are used to absorb such signals, not to bother the source.
There are filters of "absorbing" class that do the same but they are lossy in pass-band. Also SAW and similar filters can absorb unwanted signals fr a high loss (16-20 dB typical for SAWs)
I am looking for "matched" filters in the true theoretical sense (https://en.wikipedia.org/wiki/Matched_filter) that they provide the time-reversed waveform of the basis pulse for optimal SNR, not impedance matching.
Obviously the feasibility of making analog matched filters can be related to being able to generate arbitary pulse responses.
In the general case, you'll need true delay elements to represent it exactly. You already mentioned SAW, in principle you can also synthesize arbitrary pulse responses with transmission lines. An approximation with lumped elements should be possible but may require a large number of elements.
Obviously the feasibility of making analog matched filters can be related to being able to generate arbitary pulse responses.
In the general case, you'll need true delay elements to represent it exactly. You already mentioned SAW, in principle you can also synthesize arbitrary pulse responses with transmission lines. An approximation with lumped elements should be possible but may require a large number of elements.
This is what I was thinking, but it was nice to hear it being confirmed. I am somewhat surprised though, because matched filters are taught as dogma to engineers, but without that mention that it can only easily be done in the digital world.
SAWs use their dispersive nature to do matched filtering with pulses. Just out of curiosity though, you mentioned that arbitrary pulse responses can be constructed with transmission lines - can you provide an example in the literature?