For some applications, yes. There are switched capacitor filters for example. Though today IQ mixing to baseband and using DSP often makes sense. This is not practically answerable without details of the application, also part shopping questions are off topic. So edit to focus on the problem, not the chip.

Heterodyning by itself cannot extract specific frequencies;it just shifts the spectrum. That process can make it easier to filter because it can lower the bandpass requirements (lower Q)..Phase locked loops are useful as well as cross correlation techniques. Lock in amplifiers are also used to extract specific frequencies from an input. You need to do some research on these topics to see how they work.

I thought heterodyning allows to select easily the frequency of interest, shifting the spectrum to a lower fixed frequency (so that you have only to extract the amplitude of the low frequency). This would make sense for an IC that allows frequency selection.

If you heterodyne then the frequency you want to select is another frequency. I really do think you need to explain what it is you are trying to accomplish because, at the moment it sounds like an XY question. For instance, are you trying to measure the amplitude of one frequency amongst many and possibly noise. If so then say so.

considering such thing as spectrum analyzer exists, it's definitely possible. How narrow do you want it to be and how much/little do you care about neighboring frequencies? You can have a massive bandpass filter that will kill everything far from desired frequency and seriously attenuate the target frequency too, that's the plan (we care about relative amplitutde - frequency we care vs other freq.). Then you amplify the signal, which will almost solely consist of target frequency. This requires some trial and error of course

Andy. You are right, but I think not every question should be "application directed" In my case, it is obvious that I can do with a correctly designed filter, even if this is cumbersome. I simply think an IC that would allow something like that (possibly with some additional components) may simplify many applications. Extracting exactly a 100Hz requency from the 50Hz main noise could be only one of them.

@Chriss stratton. Why not answer the question with your "Switched-Capacitor Filter" (analog.com/media/en/technical-documentation/application-not‌​es/…). This is very close to answer the question.

@MikeTeX because, as explained, as even after the edit the question lacks sufficient detail to be answerable, including to know if switched capacitor filters are applicable. Unanswerable questions get closed, not answered. There are many different technologies on the market, but no way to know which one would fit your woefully underspecified neeed.

@MikeTeX Look up "superheterodyne" and "product detection" and "MC1496", to start.

@Chriss stratton. I respect your opinion, based on two erroneous assumptions: the first is that it is forbidden to ask a question not directed toward a particular application. I have no idea why too many persons in this forum think so. The second assumption is that it is not answerable. You have in your comment proved that it is (perhaps the best answer in fact). Of course, I haven't detailed an application, so you cannot and have not to detail a particular solution. Giving the keywords "switching capacitor" and explaining slightly what they do would have been a perfect answer.

There used to be DTMF decoder IC's. Not sure if they are still around. The reason we ask for more details is because it is frustrating to invest time and effort into a potential answer and then have the original poster say "that won't work for me because I need to restrict the power to 20 uW because I need to run for 6 months off of a CR2023 cell."

"I see no reason to restrict the question to a special range of frequencies ..." The answer to this is that the technique will depend on the frequency range. Many of the contributors guessed incorrectly that you were looking at radio frequency signals. Their solutions probably won't work at < 1 kHz (the range you have in your edited question).

Believe it or not, I was ALSO looking at radio frequency signals (I project to build a poor man "frequency response analyser", from 10 Hz to several MHz).

Techniques that work at 10kHz will not work at 400 MHz at all. A mixer designed to work at audio will be a fairly conventional analog integrated circuit. A mixer designed to work at 400 MHz will be based on a few special transistors and/or diodes. It is a completely different realm of design. The wavelength of a 10kHz signal is 30km. Wavelength at 400MHz is 0.75 meters. So it should be no surprise that different techniques might be used or required.

@mkeith So what? why shouldn't I use different techniques? regarding your previous comment, what you say could be true in another forum. But the building of this forum is such that no discussion or review is possible: you have one question and several answers, that's all. The OP is responsible to choose the answer and to make it work. Normally, comments should be used only to ask details about something, not to discuss or criticize.

My previous comment is based on my actual experience reading and answering questions in this forum. Despite your contention that I am stating an untruth, it very often happens that vague questions (such as yours) get answered, and then further details of the OP's desired application emerge which make some of the answers not valid. This is something I have personally witnessed many times. So, often, we press people to be more specific in their questions.

OK, I understand. Nevertheless I think it is sometimes interesting to know what technologies exist to solve a generic problem. A seemingly underspecified question could be answered simply by general guidelines and by providing key words. This is often all what is needed.

That is a fair comment. I have heard that EESE has a bit of a reputation for being rude compared to other forums. I don't know if it is true. But this is food for thought. There may be a better way to address questions that initially seem vague. As you say, some people may be happy with general answers, and in that case they maybe don't need to clarify their question at all. But on the other hand, sometimes people want specific answers but have not provided enough detail.

Normally one would want to extract a signal from a cluster of frequencies, such as picking out one channel out of thousands on cable TV.

@mkeith. Yes. Actually, I've got the answers I expected. I could simply delete this thread, but what a pity for the answers there: there are really good things, maybe useful for other persons. My problem is now to find an edit of the question that will save the answers of this thread.