I have a medical device that beeps to alert me that something needs attention. Problem is- I can't hear the beep. (The beep is the same frequency as my tinnitus.)
So, I am wondering about building an audio preamp with a band-pass filter to detect when the input "hears" the tones then publish an MQTT message that my Home Assistant will then send a message to all of the Echo devices in the house.
Any thoughts?
If you put a microphone directly on the beeper, other sounds will be weak in comparison. No filter should be needed. A piezo disk might work well, and may not even need an amplifier.
I thought of that, but the source of the beep is on my body. So the detector needs to be somewhere else in the room.
Detecting a weak tone from somewhere else in a room can be quite a difficult problem. Our ears and brain are very effective at filtering out background noise, so we don't tend to realize that microphones hear EVERYTHING.
But it is a great learning project! Maxim makes some excellent switched capacitor band pass filter chips. MAX7490 Dual Universal Switched-Capacitor Filters | Analog Devices
Seems like a wearable medical device. Have you asked the provider for a device with a different tone? Doubt you are the first to have the problem.
Their "solution" is to stop using their device.
At any rate, I ordered a sample of the MAX7490. I'll solder it to a breakout board so that I can experiment on the bench. I've never worked with an active filter before. I can use the Arduino for the clk, so it's still an Arduino-centric project.
Can you make the device beep on command?
No, it's one time use, so I plan to cut one open to get the piezo beeper out for testing.
It occurred to me that one of the engineers designing the device is a ham because the different alerts beep different Morse letters. "A" for alert, "L" for low, "EEEEEEEEEEEEEE" for expired (change the pod). This is the one that drives my wife up the wall. So making the Arduino decode the Morse code would let me send different voice messages to the Echo devices.
I needed an excuse to buy another oscilloscope. Any recommendations?
Interesting! They are likely using an active piezo. I can't hear them, either.
My wife wore an alarm/sensor for a week to monitor her hear rate. She was supposed to write down the time when it beeped. I never did hear the beep but she did.
I certainly can hear the smoke alarms and that is likely why the don't use active piezo beepers.
Since it is a medical device can you start by defining the physical maximum parameters ?
How small it must be LxWxH ?
What you need is an active filter Notch-Pass filter with a logic signal output.
SMD can be very small, but the circuit design is a bit tricky and not something I have done
I have built Fourth Order Band Stop Active Filters that worked but not band pass.
A band pass is just a high pass and a low pass summed with the cutoff of the former and
latter being the same. (the low pass starts where the high pass ends). The spectrum looks like a high pass on the left and a low pass on the right. The problem is making a very narrow notch pass is not straight forward and I currently don't know how to do that. If you can get the schematic
you can send it to a pcb vendor to have it made using very small SMD parts. I have a microscope
and a soldering station with super small tips so technically , if I had the schematic and the small
parts I might be able to make a prototype but when you get down to 46 guage wire and 0201
resistors , even your breathing makes your hands move. I would never attempt to build one myself. If you googled enough and found a schematic for an active Notch Pass filter you could
simulate it in LTSpice before sending it to a board house.
In a nutshell, what you are asking for is a one trick pony. A graphic equalizer with only one
band instead of the usual 10. As you probably know , one of the features of an equalizer is
you can either suppress or amplify certain bands (obviously , a 10 band equalizer like the one
I built for a front end for a color organ, has 10 specific fixed center frequencies set by component
values so if an audio sound falls between 20 Hz and 20,000 Hz, it will likely fall into one of the
10 bands. Since , as yet , you have yet to post the center frequency of the beep, there is nothing
we can do. Finding that out is your first step. Go to a music store and ask the guy behind the
counter to demonstrate one of their electronic note finders that displays the frequency of a
sound played close to the microphone of the device. If the guy is a musician he likely can tell
you just by hearing it. How else could he tune a piano or guitar ?
Find out the frequency and then look for a circuit that detect and amplify that frequency.
Walter Jung's Audio IC Applications (page 147-151) (This is the edition I used) has the schematic for a graphic equalizer (which is how I built mine, which a certified audio engineer tested and declared 'Studio Quality'.
The "Infinitely Variable Equalizer on page 158 sounds like it would do the trick since it has two
pots, (Q, and frequency). If you turn the Q all the way up and dial the frequency to the one I
told you to find the output can be converted to a logic signal several different ways, (comparator,
CMOS SCHMITT TRIGGER
CD40106B
I have a couple of ideas from the replies here. One using the Maxim 7490 and the other an NE567 tone decoder. Since the tone beeps in Morse code, I also found a circuit and Arduino sketch that decodes received CW.
Tone Decoder is probably the simplest cheapest solution. It should work without any issues because
the beep is probably very close to one of the tones on a touch tone telephone. The OP should use
a touch tone phone to compare to the beep and report which key it is.
I need an excuse to buy a new oscilloscope. (Recommendations, anyone?). I sold my last one because it weighed almost 40 pounds.
I'll just record the beep then play it back while watching the oscilloscope. I may be able to get a tone element from one of the discards to generate a test tone. The 567 will get the first breadboard since I have a few in my workshop.
Pretty standard for medical devices I fancy. The anaesthetic machine behind me as I am sitting here, persistently alarms "di-di-dit di-dit" when the monitor leads are disconnected = "SI" meaning "signal loss". 
I think that was the last sound I heard when I had my Endoscopy at Kaiser
I spent months in a hospital a few years ago and I was hearing beeps in my sleep. In fact, when I was finally home, I couldn't sleep because it was so quiet.
So, no recommendations for an oscilloscope?
When I moved from San Fernando Valley SoCal (USA) (where you hear sirens and gun shots all night) in to Santa Barbara where you can hear a pin drop , I had the same problem.
I hadn't seen the Instek before. I've been looking at Rigol's, Hantek and Siligent.
I asked for suggestions here a few months ago and have narrowed my search. My budget went up because I sold my Tektronix scope for more than I had expected.
The top of my list right now is the RIGOL DS1104Z Plus, or for the $500 budget I could get the Rigol DS1202Z-E plus a signal generator.
There's some sample code in the post below for doing this sort of thing with a microphone module and a digital filter/detector. Sadly the link to the filter coefficient web link is no longer active, but if you knew the frequency of interest, there are other digital filter design tools.
https://forum.arduino.cc/t/walkie-talkies-to-control-dslr-camera-via-arduino/607420/17?u=mrmark
I have a Hantek 2 channel digital oscilloscope purchased a year or two back. It has a sort of chintzy feel and display compared to the commercial grade Tektronix stuff I've used, but I've found it more than adequate for home hobby stuff. I'd spring for the 4 channel version if I were to do it over again.