In the mid 90s I decided to throw together what I called a "Bat Listener" - a simple receiver used to convert ultrasonic sound down to the audible range.

Figure 1: The exterior of the ultrasonic receiver, complete with fancy labeling! Click on the image for a larger version.

Two types of circuits:

There are two common ways to convert a higher (ultrasonic) signal to the audible range, whether this is done using analog or DSP (Digital Signal Processing) techniques.

Frequency division

There are several ways to do this, the simplest being the "divider" type which digitally converts ultrasonic frequencies to audible by integer division of the input to a lower frequency.

The problem with this simple approach is that it does not preserve the amplitude (loudness) of the original sound since it must take the input signal, amplify/convert it to a series of logic-level pulses - which loses any amplitude reference - and do a brute-force digital division.  Additionally, if there are multiple signals present, for the most part only the strongest one will be converted down.

Clearly, one cannot "tune" this type of circuit:  A signal at 40 kHz will always be divided down by a fixed integer amount,  Let's say that the circuit digitally divides by 32:  That 40 kHz signal will be at 1.25 kHz.

Additionally, the direct "A-B" frequency differences between ultrasonic signals is lost, instead being "(A-B)/N" where "N" is the number of divisions.  In other words, the relative frequency differences between signals is not preserved.

Heterodyne conversion

The other way to do this is to convert the frequency.  In this technique, two signals - the ultrasonic to be converted - and another generated by the device (the "local" oscillator) are mixed together.  The result is an arithmetic shift in frequency.

The biggest advantages of this method are the fact that that not only are the differences in frequency preserved (e.g. two tones 1 kHz apart at ultrasonic will appear as two tones 1 kHz apart at audio) but the relative amplitudes (loudnesses) of the received signals are preserved as well.

Frequency conversion:

I chose to build a heterodyning receiver to convert the input frequency to a lower one.  This can preserve the amplitude and frequency relationships  - plus it is fully tunable, allowing one to choose the frequency range to convert to audible sounds - and since it is a simple conversion, multiple signals present will also be preserved.

When it comes to frequency conversion, there are two ways:  The simplest - direct conversion - would involve mixing a variable oscillator with the incoming signal and filtering/amplifying the resulting audio.  This has the advantage of being the easiest, and it is the method described in this article:

     April, 2006 QST article, A Home-made Ultrasonic Power Line Arc Detector - link)

While I could have easily built something like this a decade before the above article was published, as I'm sometimes wont to do I decided to make it a bit more complicated, constructing a superheterodyne converter.

While a direct-conversion receive simply mixes an oscillator with the desired signal to cause a frequency conversion, a superheterodyne receiver operates like a conventional AM or FM radio:  The desired signal is first converted to an IF (Intermediate Frequency) - and this IF is then converted to audio.  The advantage of the superheterodyne scheme is that filtering may be applied at the IF to limit the receive bandwidth - and since the IF is fixed, its width remains constant over the tuning range, just like that in a conventional radio/receiver.

Circuit description

Figure 2: Schematic diagram of the superheterodyne ultrasonic receiver. See text for a circuit description. Click on image for a larger version.

As noted above, this circuit is more complicated than it needs to be, so make of it what you will!

VCO:

The heart of the unit is U1, the VCO (Voltage Controlled Oscillator) which uses the venerable CD4046 PLL chip.  Often used for frequency synthesis, we are using (only) the oscillator portion, which provides a linearly-tuned and fairly stable frequency source, adjusted by the voltage applied via R101 (and scaling resistor R102).  The values were chosen to provide an approximate frequency range of 125 to 185 kHz (more on this later) to allow tuning of audio signals from (ostensibly) 0 to about 60 kHz.  The actual tuning range is closer to 115-190 kHz as a bit of extra margin for the frequency range.

The only critical component here is C101 which should be a frequency-stable capacitor.  I used a polystyrene capacitor, but an NP0 (a.k.a. C0G) or silver-mica could be used, instead.  When I reverse-engineered this device, I noted that the marked capacitance value was unreadable, but back-of-the-envelope calculates indicate that a value of "about 150pf" should be in the ballpark.

R103, connected to the "R1" pin of U1, sets the approximate center frequency range while R104, connected to the "R2" pin - sets the lowest frequency - which important, since we want to constrain the tuning to 125-185 kHz.  Additionally, the low end of the tuning range was further refined by R102 on the "ground" side of the tuning potentiometer, which sets the minimum voltage that may be applied to the "VCOIN" pin.

The VCO output, a square wave, is buffered by U2, a hex inverter, and several sections are used to provide both a VCO signal and its inverted version to drive the mixer.

While the 4000 series CMOS chips throughout this receiver will happily run from 3-15 volts, they are operated from a regulated 5 volt supply - mainly to improve frequency stability and to provide a nice, stable voltage for a few other low-level circuits and to provide isolation from the main battery supply which will vary a bit, particularly at higher receive volumes:  This variance, if it gets back into some earlier stages, could cause instability of the receiver in the form of "motorboating" or some other type of feedback.

BFO:

Another circuit is the BFO (Beat Frequency Oscillator) which is used to convert the IF signal back down to audio - both being processes that we'll discuss shortly.  This uses an inexpensive 500 kHz ceramic resonator to form an oscillator using one of the sections of U2C, the signal being buffered by U2B.  This signal is divided-by-two using U3A, one half of a 4013 dual flip-flop - and then divided by two again using U3B, yielding a stable 125 kHz signal.  As with the VCO, two phases of this signal (normal and inverse) are available, this time using the "Q" and "!Q" outputs of the 4013.

Input signal path:

J1, a disconnect-type 3.5mm stereo jack is wired so that an internally-mounted electret "capsule" microphone is connected by default.  This microphone element (M301) is of the "2 wire" type or electret microphone in which a bias voltage is applied to the same pin from which audio is drawn - this voltage being applied via R301 from the 5 volt regulated supply.  The specific make/model of this electret element is unknown as it was selected from a small collection to find the best performer at ultrasonic.

At some point in the future, I'll replace this with a more modern MEMs microphone as described in Another article:  Improving my ultrasonic sniffer for finding power line arcing by using MEMs microphones - link.

The signal from the microphone is applied to U4A which is wired as a unity-gain buffer.  For this, an LM833 is used, an inexpensive, low-noise dual op amp:  An LM358 or many other types may be used here as well - just make sure that it is is fairly low noise:  I'd avoid the use of the LM1458 here as it is quite noisy by comparison!

Section U4B amplifies the signal voltage by 10 (20 dB of power gain) and this signal is applied via R305 to a simple L/C high-pass filter consisting of C303, C304, L301 and L302 the latter two components being inexpensive 18 milliHenry inductors.  Certainly, an R/C-based high-pass filter could have been constructed using U4B, but I chose not to do that for some reason.

Figure 3: Inside the ultrasonic receiver, constructed on prototype board and having been modified several times over the years. Click on the image for a larger version.

In simulation, the C303/C304/L301/L302 filter has a -3dB roll-off of about  23 kHz, it's down by 10dB at about 19.5 kHz, by 20dB at about 16 kHz and by 40 dB at 9 kHz and with the values shown, it's flat to within 1 dB between about 24 and 100 kHz.

The output of the filter is amplified by U5B - and then even more by U5A (which has a bit of roll-off from C307) to yield a whole lot of gain.  It's very possible that I over-did the gain here, but unless the signal source is quite close, there is no clipping observed on the output of U5A.

Its worth noting that a mid-supply voltage is created using R309/R310 to provide a "virtual ground" for the op amps and to maintain stability, it is heavily filtered by C306 and C302, each located near the respective op amp shown on the diagram.

Mixer and band-pass filter:

It is this next section that may seem unfamiliar to some - the use of a CMOS analog switch as a signal mixer.  For this, a CD4066 is used which consists of four separate analog switches.  The filtered and amplified ultrasonic input signal from U5A via C308 is applied to pins 2 and 10 of U6A/U6D.  When the respective signals on the control pins "VCO_A" and "VCO_B" go high, the switches are activated, and because VCO_A and VCO_B are inverts of each other, each of these switches is closed in turn.  The result of this is that the inputted signal is chopped up at the rate of the 125-185 kHz VCO and this produces two mixing products.  

For example, let's assume that there is a 40 kHz signal is present on the input that we wish to hear.  If the VCO is tuned 40 kHz above the 125 kHz IF (again, more on that momentarily) - to a frequency of 165 kHz - the switching action of U6A and U6D produces both the sum (165 + 40 = 205 kHz) and the difference (165 - 40 = 125 kHz).

T301 is a filter/transformer that passes only the 125 kHz signal - the difference signal in this case.  This transformer consists of two separate windings, each resonated using its internal capacitors and the externally-added 820 pF capacitors on each winding (e.g. C309/C310) to "pad" it down to 125 kHz.  This forms a fairly wide (8-10 kHz) filter that rejects signals outside the immediate vicinity of its 125 kHz frequency.  Because this filtering is at a fixed frequency, it does not vary with input tuning which means that its bandwidth is constant over frequency.

Of all of the components in this device, this transformer is unique:  It was originally a 262.5 kHz IF transformer from a 1970s/1980s Philco (Ford) AM-only car radio.  While I could have certainly used the original 262.5 kHz frequency - or even 250 kHz, when I built this I decided to pad it down to 125 kHz using C309/C310  - a frequency that is conveniently 1/4th of the 500 kHz resonator.

It's been so long since I built this, I don't recall why I didn't simply divide the 500 kHz by two and readjust that transformer to 250 kHz.  Practically speaking, I could have also up-converted to 455 kHz and used either transformers or ceramic filters from a modern AM radio as 455 kHz ceramic resonators were certainly available at the time - but I didn't do that.

Each half of T301 has a center tap and to this, a bias voltage is applied via R315 to assure that the voltage on these switches was in the middle of the supply range, away from the protection diodes on the 4066's I/O pins, which could cause clipping/distortion should they be allowed to conduct if the signal voltage got too near the ground or supply rails.  To prevent coupling between the two halves of the transformer via the center tap, R314/C311 was added, the resistor adding isolation with the capacitor bypassing the remainder of the signal.  Practically speaking, being able to adjust the bias voltage was unnecessary as a simple resistive voltage divider to set the bias at 2.5 volts (1/2 the supply voltage) would have been just fine.

On the "other" side of the transformer is the other half of U6 (e.g. U6B/U6C) - this time, clocked from the fixed 125 kHz oscillator.  From this, the signal - previously converted up to 125 kHz is now converted back down to audio.

Post-mixer amp/LPF:

The output of the down-converting mixer is applied to U7B via R316, a 1k resistor and a 0.001uF capacitor, both of which form a simple R/C low-pass filter to attenuate any high-frequency leakage signals from the mixer.  Because the mixing process itself is a bit lossy (about 25% efficient) as is transformer/filter T301, U7B boosts the signal by a factor of 10 (20dB) and then applies it to U7A, which is configured as a variable gain amplifier section.  The output of this is then boosted again by U8, an LM386 which is capable of driving headphones or even a small speaker.

A few comments about the design:

Originally, the circuit lacked U7 at all, but it was added when the gain of U8 (the audio amplifier), by itself, was found to be inadequate.  Since U7 was "patched" into place, this explains the odd gain distribution:  If I were rebuilding this from scratch, I'd certainly not need two post-mixer amplifier sections and I could have likely eliminated one full dual op-amp package.  As it is, I may add a "high/low" gain switch somewhere around U5 to allow reduction of the gain somewhat when in the presence of possibly-high ultrasonic signal levels to prevent clipping prior to the band-pass filter which would surely degrade overall performance.

If I were to build this again I would likely use a 455 kHz IF, instead.  While not as plentiful, 455 kHz ceramic resonators are available to use for the BFO as are either transformer or ceramic-based band-pass filters.  I would also likely reconfigure U4B or U5 to perform the high-pass filter function rather than using harder-to-find inductors.

Again, I built this unit in the mid 1990s and have since lost my original notes, but I do recall that I modified it a few times since, simply tacking changes onto the old circuit rather than completely revising it.

Use as a longwave receiver:

While primarily intended to "hear" ultrasonic sounds such as those produced by bats, insects, leaking pipes, arcing power lines, etc., it is just a longwave radio receiver connected to a microphone:  If one connects a few 10s of feet/meters of wire to to J1 - and provides an Earth/ground reference to its shield connection - one can easily tune in the high-power transmitters used for submarine communications (around 20-30 kHz) plus the WWVB time signal at 60 kHz.  This must, of course, be done away from man-made noise sources such as power lines.

Alternatively, I have used a loop of about 1 foot (25cm) diameter of a dozen or so turns of wire along with a 10uF capacitor in series (to optionally block DC from R301) and been able to hear such signals - even in suburbia - but with this arrangement you'll also likely hear plenty of similar signals from the myriad switching supplies that likely inhabit your house as well!

Final comments:

The reader should be under no illusion that this is an optimized circuit or that I would do it this way again:  It was assembled fairly quickly to suit a need and to test a few random ideas, just to see if they would work.  Will I rebuild it at some point?  I don't know - it works as it should, so I don't plan to re-make something that is currently fit for purpose.

While I've heard very few bats with this - probably due to the deficiencies of the electret microphone at ultrasonic frequencies (which explains the future switch to MEMS-type microphones) - I've used it to find powerline noise (arcs are noisy at ultrasonic) and to test longwave receive antennas.

This page stolen from ka7oei.blogspot.com

[End]

Figure 1: The packaged MEMs microphone, along with the ultrasonic receiver. Click on the image for a larger version.

Years ago - probably 20+ - I constructed a superheterodyne "Bat Listener" to eavesdrop on the goings-on of our winged Chiroptera friends.  (That receiver - the one depicted in Figure 1 - is described HERE.

In retrospect, this device is probably a lot more complicated than it need be as it up-converts from "audio" to a 125 kHz IF, using a modified 262.5 kHz Philco (Ford) car radio IF Can as the filtering element before being converted back down to audio.  This device has a built-in microphone, but it also has a jack for an external microphone, which comes in useful.

This device actually works pretty well for its intended purpose and, in a pinch, can even be used to listen to LF and VLF signals like the time station WWVB at 60 kHz and the powerful transmissions intended for submarines in the 20-40 kHz range if a simple wire is attached to the external microphone input, but I digress.

One of the weak points of this unit has always been the microphone.  To be sure, there exist the 40 kHz ultrasonic transducer modules:  These units used to be common in TV remote controls before the Infrared versions became common and you might still find them in the (now rare-ish) ultrasonic intrusion alarms.  While fairly sensitive, these units do have a "problem":  They are rather sharply resonant around their design frequency - which is typically somewhere around 40 kHz.  In other words, they aren't very good over much of the ultrasonic frequency range above or below 40 kHz.

It would seem that many commercial ultrasonic power mains arc detectors use these things (The MFJ-5008 seems to be an example of one of these) and there have been a few articles on how to make these devices (See the April, 2006 QST article, A Home-made Ultrasonic Power Line Arc Detector - link) but it, too, uses one of these "narrowband" 40 kHz transducers.

While certainly fit for purpose, I was more interested in something that could be used across the ultrasonic spectrum.  When I built my "bat listener" I fitted it with a "condensor" (electret) microphone, rummaging through and trying each of the units that I'd accumulated in my parts box at the time to find the one (make and model unknown) that seemed to be the most sensitive - but compared to a 40 kHz transducer, it was still somewhat "deaf".

This issue has nagged at me for years:  I occasionally break out the "bat listener" to (would you believe) listen for bats and other insects when camping, and it is useful if you have a suspected air leak in a compressed air system - plus it's sometimes just plain interesting to walk around the house and yard to hear what's happening at frequencies beyond human hearing - and it may also be used for finding arcing power lines as the QST article referenced above suggests.

In more recent years, an alternative to the electret microphone has appeared on the scene in the form of the MEMS (MicroElectricroMechanical System) microphone.  This class of devices are literally tiny mechanical devices embodied in silicon structures and they can range from oscillators to accelerometers to exotic tiny motors to (you guessed it) - microphones.  Their small size, which makes them the choice when space is at a premium, as in the case of a phone or web camera, also reduces the mass of the the mechanical portion that responds to variations in air pressure (e.g. sound) which can enable them to respond to frequencies from a few 10s of Hertz to well into the 10s of kHz.

Figure 2: The MEMs microphone mounted and wired up.  The element is mounted "dead bug" by gluing its top side to the circuit board and small (#30) wires connect to the pads. Click on the image for a larger version.

Perusing the data sheets of devices found on the Mouser Electronics web site, I found what seemed to be (one of many) suitable candidates:  The Knowles SPU0410LR5H-QB.  This device, which is a version with an analog output, is about 3mm by 4mm, has a rated frequency response to at least 80 kHz - and it is pretty cheap:  US$ 0.79 each in single quantities at the time of this writing - and, in these days of erratic supply lines, it was available immediately as Mouser reported having more then 30k of them in stock.  

Importantly, this device had its "audio port" on the same side as the wiring - the intention being that it would get its sound through a hole in the circuit board, but this would also make it easier to wire up as described below.

The fact that this is a small, surface-mounted device may seem daunting to the home building - but don't be daunted:  Given the appropriate magnification device (I use a pair of "Geezer Goggles" that I got from Harbor Freight) and a fine-tipped soldering iron, it's perfectly reasonable to solder just a few fine (30 gauge) wires to a device this small.

Figure 3: The completed board, containing the circuit depicted in Figure 4, below.  The board with the microphone is on the left, and the attaching cable is seen in the upper-right. LED1, the one across the microphone element itself, was mounted on the bottom side of the board. Click on the image for a larger version.

First, I cut a small piece of circuit board material to use as a substrate and mounted at a right angle on a larger piece, as shown.  I then took the microphone and "Super Glued" it "dead bug" to the middle of this board (see Figure 2, above) leaving the side with the connections and sound port facing outwards.

With this simple operation, a very tiny part suddenly becomes a larger, easier-to-manage part - albeit with very closely-spaced wire connections.  Being careful with very thin solder not to get any solder or flux in the sound port, I first tinned the connections on the device itself (there are four pads - two grounds, a power and an audio) and then proceeded to use some #30 "wire wrap" wire to make flying lead connections to the device, using a slightly longer section of one to tie the two "grounds" together.  I could have just as easily used some tinned #30 enameled wire, instead, but I tend to keep the Kynar-covered wire wrap wire on-hand for this very purpose.  

With the flying leads and the piece of circuit board as a "breakout" device, I was then free to treat the MEMs microphone as a "normal sized" device and build an interface circuit onto the rest of the board.

In perusing the data sheet, I noted that the power supply voltage rating was 1.5-3.6 volts which was incompatible with the 5 volts of "phantom power" applied by my bat listener to the microphone jack to power a condensor (electret) microphone, but this was easily remedied using the circuit shown below:

Figure 4: The interface circuit used to adapt the MEMs microphone to the existing 5-volt electrect microphone circuit. Click on the image for a larger version.

Circuit description:

This circuit depends on there being power applied via the audio/microphone lead, as is commonly done for computer microphones.  Typically, this is done by biasing the audio line through a resistor (2.2-10k is common) from a 5 volt supply - and that is assumed to have been done here on the device to which this will be connected, as I did on my "bat listener".

DC is decoupled from the audio output of the microphone via C1.  In this circuit, I chose a 0.01uF capacitor as I wanted to reject audible frequencies (<10 kHz) to a reasonable extent - and this means that this capacitor value is way too small if you plan to use it as a "normal" microphone to listen well down into the lower audible range:  Something on the order of 1-10 uF would be appropriate if you do want audio response down to a few 10s or 100s of Hz.

A word of warning:  Do NOT use a ceramic capacitor for C1 as these can be microphonic in their own right.  I used a 0.01uF plastic capacitor (probably polyester) which is neither microphonic or prone to change capacitance wildly with temperature.

Resistor R1 (2.2k shown here, but anything from 2.2k to 4.7k would likely be just fine) decouples the audio from the DC and capacitor C2 removes that audio, providing a "clean" power source for the microphone.

Here, LED2 is used as a voltage limiter:  Being an "old fashioned" green panel indicator LED, its forward voltage is somewhere around 2 volts.  The use of an LED in this manner has the advantage that unlike a Zener, this type type of LED has a very sharp "knee" and practically no leakage current below its forward voltage - and it is much easier to find than a 2-2.5 volt Zener.  It's likely that about any LED would work here - including a more modern Gallium Nitride type (e.g. blue, white, super bright green) but I have not verified that they would properly clamp the voltage in the 1.5-3.6 volt range needed by the microphone.  (And no, there are not any detectable effects on the circuit from light impinging on the LEDs.)

LED1 is present to protect the microphone itself.  When it's plugged in, whatever voltage is present on the audio cable will be dumped into the microphone output as capacitor C1 is charged and it could damage it, particularly if the power source is 5 volts and the microphone's maximum rated voltage is just 3.6 volts.  This LED, which is the same type as LED2, will not normally conduct as the audio output from the microphone typically has a voltage of roughly half that of the supply, so LED1 will be completely "invisible" (in the electrical sense) in normal operation.

Figure 5: A spring, soldered to a wire connecting to the "ground" side of the circuit (also the microphone cable shield) used to make contact with the aluminum tubing. Click on the image for a larger version.

I mounted the board with the microphone in a piece of aluminum tubing that would fit the microphone mount of my parabolic dish (see below) and this not only provides protection for the microphone and circuitry, but also serves as an electrostatic shield, preventing energy - say, from a power line - getting into the circuitry.  To make this effective, the tubing itself is connected to the ground lead (cable shield) by soldering a wire to a metal spring and placing it in the end of the tubing as seen in Figure 5.

To secure things into place, a bit of "hot melt" glue was used, preventing the board from sliding out.  The connection to the receiver was made via a length of PTFE (Teflon) RG-316 coaxial cable - but shielded audio cable would have sufficed:  This cable is firmly attached to the board as seen in Figure 3 as a strain relief. 

The parabola:

While the microphone is sensitive in its own right, its sensitivity can be noiselessly "amplified" many-fold by placing it at the focus of a parabolic dish.  I was fortunate to have obtained a Dan Gibson EPM model P-200 (minus the original microphone element or any electronics, but including the holder) at a swap meet, but the QST article linked above suggests other sources - and I have seen parabolic-based microphones on Amazon - often as semi-serious toys - as well.  Using the holder - the inner diameter of which was the basis for choosing the specific size of the aluminum tubing - the microphone was mounted at the focus of the dish.

Finding this focus can be a bit of a challenge without the proper equipment, so I set up a "test range".  At one end of my back yard I placed a 40 kHz transducer (of the sort noted in the QST article linked above) connected to a function generator set to 40 kHz:  I'm sure that a small speaker would have been sufficient to generate a signal.

Figure 6: The MEMs microphone, mounted in the aluminum tubing, at the focus of the parabolic dish, with attached cable. Click on the image for a larger version.

From across the yard - perhaps 30 feet (10 meters) away, I sighted the emitter through the dish, using its alignment dots and slid the microphone in and out until I had the best combination of the loudest signal, the sharpest aiming, and the "cleanest" pattern.  On this last point, I noted that if I focused too far in our out, the peak of the signal would become "blurry" (e.g. spread out) or, in some cases, I would get two peaks - one on either side of the "real" one, so the object was to have the single, loudest peak possible.  Once this was found, it was marked and a bit of heat-shrink tubing was put over the end of the aluminum tube, corresponding with that mark, to act as a "stop" to set the correct focus depth.

Again, refer to the QST article linked above for additional advice on where to obtain a suitable parabolic reflector, and hints on the mechanical construction.

Does it work?

The answer is yes.  From significant distances, I can hear the acoustic signature of switching power supplies (apparently, many of these have transformers that vibrate at their 30-60 kHz switching frequency) as well as the sounds of insects, and the hissing of the capillary valve of the neighbor's window air conditioner.

Importantly, I was able to verify that a power pole's hardware was, in fact, arcing slightly - although I wasn't able to determine which hardware, exactly was making the racket as it was quiet enough that it became inaudible when I stood far enough away from the (tall!) pole to get a better viewing angle.

When I get the chance, I will replace the capsule electret microphone built into the receiver itself with one of these MEMs units, but that's just one project on a rather long list!

This page stolen from ka7oei.blogspot.com

[End]