Narrow band filtering or frequency detection

Hello to everybody,

I am doing a little project in ultrasounds. I am doing a receiver which should deal with simple signals coming from several transducers. The problem is that the different transducers emit in different channels, and I want to be able to pay attention only to the channel I am interested on. Basically, the receiver will be tuned for receiving only one channel.

There are many channels (in Hz): 39.531, 39.688, 39.844, 40.000, 40156, 40.312, 40.469,

So the application it's around 40 kHz. The application itself is very simple. The signal is a pulse of some milliseconds, and you measure the time until the next pulse, and that time is your information. So I should only detect my specific channel... but I am not sure the simpler method to do so. The rest of the electronics and programing I think I know how to do it, but filtering or detecting a signal in such close proximity to the others... I don't know... an analogue filter won't attenuate enough the nearby channels... maybe something could be done with a pll...

any suggestions for detecting a specific frequency?

Thank you very much in advance...

You could count the period versus clock time (stable crystal) with a digital microcontroller, like a digital frequency meter.

Thanks a lot... I'll read more about it... but what would it happend if all the above frequencies were transmitted at the same time at roughly the same amplitude? Can it determine the spectrum and tell the different frequencies appart? I suppose for that it would be better some kind of DFT... As I said, I will read about it, but maybe it can be done easier with a tone detector. I thought those would only work at low frequencies, but actually they work up to 500 kHz. I don't know how they work, but maybe they are the easier solution. I have to keep reading....

Thanks a lot, I appreciate a lot the suggestion.

A PLL would probably have trouble with multiple pitches.

Counting clock tics between zero crossings, or zero crossing during a time interval, would be OK for single tone. Multiple tones will interfere but you may be able to extract pitch from multiple tones from longest, shortest, average.  6 or 7 at once sounds pretty hairy.

I have had some success performing synchronous detection at much lower frequencies by sampling at multiples of the exact period of the note I am reading, but I don't know how well you could parse between 0.3% pitch spacing up at 40kHz.

JR 

Thanks a lot John,

I don't need to work with different frequencies. what I need is to ignore any other frequency but the one I want. The system would be tuned to a specific frequency. So when I said that you could be receiving several of these frequencies, I meant that in the transducer, you might receive these frequencies, and the unit should be able to reject any signal but the appropiate one. So my problem is how to avoid the intereference from the nearby frequencies, since an analogue filter wouldn't have such a narrow band pass (or if so, it would be quite complicated and expensive).

Any suggestion on how to get this? I'll read today about Tone Detection chips... let's see if they can do the job...

About the microcontroler option, if you have several frequencies reaching the transducer at the same time, say 38.9 kHz, 40.0 kHz and 40.16 kHz, would the uC be able to determine whether the 40.0k is presenten or not?... I don't need to know whether there are more frequencies present or not, only if the 40.0k is present, and when it started to be present and when it stopped.

Thank you veru much,

I appreciate a lot your answers....

The microprocessor only does what you program it to... These days they have A/Ds built in, and counters and basic tools, but you need to figure out how to get a useful result.

I guess a FFT with fine enough resolution could work. But the 0.3% spacing of the carriers is not trivial.

JR



 

Hi,

Thanks again...

I've been reading about the LM567 and I really think it is just what I was looking for. The bandwidth can be easily be 0.25% and in the frequencies I have to work with they are about 0.53 to 0.78%... so in theory it is perfect.

I'll try it out this week, if it doesn't get the narrow bandwidth that it is supposed to get I'll be back! .... Thanks, see you around...