 |
| Figure 1: Front view MFJ-5008 parabolic dish with integrated microphone and receiver (located on the back side). Click on the image for a larger version |
The MFJ-5008 Parabolic Ultrasonic receiver
Note:
Since the MFJ-5008 is no longer being sold, other than finding this unit or similar on the used market is the only option. A future posting in this blog will show how to construct a similar unit using readily-available kits and parts.
The MFJ-5008 was marketed primarily for detecting arcing on failing power line hardware, but there are other reasons why you might use such a device:
- Listen to Bats' echolocation. The "clicks" emitted by bats are well above human hearing.
- Listen to other animals and insects. Other animals and insects also emit ultrasonic sounds - both for echolocation and communication.
- Find high pressure leaks. Leaks in high-pressure systems (water, gas, engines, compressors) often make a lot of noise at these frequencies.
- Locate switching power supplies. These devices often make noise due to magnetostriction of devices (transformers, coils.)
As I find this topic to be interesting, I've written about the detection of ultrasonic signals on two previous occasions in this blog:
- Improving my ultrasonic sniffer for finding power line arcing by using MEMs microphones - Link
- An ultrasonic superheterodyne receive converter (e.g. "Bat Listener") - Link
While there are several devices out there that you can buy to enable listening at these frequencies, the landscape has changed in the past few years when it comes to how one might do this on a budget:
- In years past, the MFJ-5008 was available - its primary purpose being to locate and identify arcing on power lines and related infrastructure. As MFJ is no longer in business, this devices is available only on the used market.
- Some "bat listeners" have used electret microphones. These inexpensive capsule microphones - while having good response across the human hearing range - lose sensitivity rapidly above this, limiting their usefulness above 20-30 kHz.
- Many "bat listeners" have used ceramic transducers. Most often found for the 40 kHz range (and some were made at lower frequencies) these can be fairly sensitive. Their frequency range is quite limited and they are only usable within a few kHz above and below their design frequency at best. As different types of ultrasonic noise sources tend to occur at various frequencies, being able to detect such energy at various points across the spectrum can improve the usability of the device.
- MEMs-based microphones have become cheap and available. These devices - based on microscopic elements - can operate over a frequency range from a few 10s of Hz to over 100 kHz making the excellent replacements for the (increasingly hard-to-find) ceramic transducers. Having a wide frequency range allows the user to tune to the peak frequency of the noise source rather than being limited to the immediate vicinity of 40 kHz.
* * * * *
How the MFJ-5008 works
Made by (the now defunct) MFJ Enterprises, this includes a 18" (46cm) diameter vacuum-formed plastic parabolic dish with a 40 kHz ceramic transducer at its focus. Mounted on the back of the dish is a direct-conversion receiver that is centered at about 40 kHz that converts energy around this frequency to the audible range. As can be seen in Figure 1 there is a bar across the front in which the ceramic transducer is mounted - but it also has holes that - along with one located behind it in the plastic dish - form a crude sighting system that works quite well to determine from where detected noises might be emanating.
If one disassembles the electronics of the MFJ-5008 they will discover a small circuit board with rather common components - namely a 555 timer used as the oscillator, an LM386 audio amplifier to drive the headphones and a few common transistors to amplify and convert the ultrasonic signals to audible. There is a "tuning" control on board consisting of a 10k trimmer potentiometer, but it is not accessible from the outside - and it has a range of about 38-48 kHz: A slight modification will be necessary to allow us to take advantage of the wider frequency response of the MEMS microphone.
Consider the schematic of the MFJ-5008, below:
 |
| Figure 2: Schematic of the MFJ-5008 ultrasonic receiver. The circuitry is straightforward - a simple, run-of-the-mill direct-conversion design that is very similar to the one described in the April, 2006 QST article. Changes to C2/C8 and the added inductor are noted on the diagram. Click on the image for a larger version. |
If we compare the above schematic with that from the April, 2006 QST article, A Home-made Ultrasonic Power Line Arc Detector - link) we see some very striking similarities: Both use a 555 timer for the local oscillator, both use a series of bipolar transistors for signal amplification, and both use a single JFET for the frequency conversion mixer. There are some differences, but these are pretty much superficial when you consider that the same goal is accomplished with the same types of components.
A cursory analysis of the above diagram shows that the first two amplifier stages are coupled with 1uF capacitors allowing the full audio frequency range to pass: This mystified me at first, but in looking at the circuit board and noting some unpopulated parts locations I realized that there may have been plans to allow this circuit to be used at audio frequencies - and, perhaps, have a switch to select audible or ultrasonic ranges as well.
For the original 40 kHz ceramic transducer, this wide frequency range isn't a problem, but for a MEMS microphone - which can hear equally well over a 100Hz through at least 60 kHz range, this would be: As the mixer (Q3) is just single-ended, it will happily amplify the original input as well as do a frequency conversion meaning that you are likely to hear audio-frequency "bleedthrough" on the audio output - and indeed, when I retrofitted it with a MEMS microphone (to be described shortly) I did.
 |
| Figure 3: Picture of the MFJ-5008 with location of the various various components and board locations involved in the modifications annotated. Click on the image for a larger version. |
Adding "proper" high-pass filtering to the MFJ-5008
The only sort of "high-pass" filtering present are capacitors C10 and C11 which are conspicuous by their being in series: Why use two capacitors rather than just a single 180pF capacitor? The answer lies on the circuit board where there are unpopulated locations marked "L1" and "L2" (see Figure 2) which correspond with an (uninstalled) pair of inductors between the junction of C10 and C11 and ground.
To make the unit much less sensitive to audio frequency - and to make it compatible with a MEMs microphone, several changes should be made:
- Change C2 and C8 to 0.01uF (e.g. 10nF) capacitors. This will prevent the first two amplifier stages from being overloaded by audio frequencies and go a long ways in prevent "bleedthrough".
- Install inductance at the positions of L1 and L2. I suspect that two inductors were in mind when they designed the board as high-inductance, surface-mount devices are comparatively rare and expensive, so they could use a pair of lower-value coils to get the desired value. See the footnotes on the bottom of this blog for suggested inductors.
Connect a 4.7k resistor between the center pin of the RCA connector (to the microphone) and the "V+" pad near the un-populated switch. This inserts a current-limited 9 volt supply on the microphone lead.
Figure 4:
Apparently designed to be used in several ways, the MFJ-
5008's board has several unused parts locations, including
positions for an inductor that could be used for improved
high-pass filtering as shown here.
Click on the image for a larger version.
The amount of inductance to install at L1 and L2 isn't too critical, but finding such components may be awkward - but the total amount of inductance to use may be anything between 27mH (that's milliHenries!) and 68mH with 47mH being optimal - a relatively huge amount for an SMD device. In perusing my collection of inductors, I found a through-hole 27mH inductor that I tacked into place, securing it with glue: Note that it gets soldered across the two pads of L1 and L2 closest to the socketed 555 IC as Figure 4 depicts.
When modifying the MFJ-5008, the MEMS microphone was fitted first and it became clear that audio-frequency energy sailed right through the system, significantly reducing its efficacy at the detection of ultrasonic energy. It is my opinion that both the changing of C2 and C8 to 0.01uF capacitors and the addition of the inductor are necessary modifications for good performance.
Note:
If you don't have a suitable inductor for the above modification, the receiver will still
work, but you will hear a bit of audible frequency bleedthrough: In a
location with high ambient noise, this may be a problem, but in an
otherwise quiet location, it probably won't be an issue: Changing C2 and C8 do a reasonable job of reducing audio-frequency response and should be considered to be mandatory if you use a MEMS microphone.
In other words, if you don't install the inductor, don't let that stop you from making the modification to the MFJ-5008 and using it with a MEMS microphone - just be aware of the audio frequency "bleedthrough" issue.
Extending the tuning range of the MFJ-5008
 |
| Figure 5: Potentiometer R10 - originally 10k - was replaced with a 50k miniature potentiometer to allow tuning. A 4.7k resistor paralleling R22 can be seen in this fuzzy photo. Click on the image for a larger version. |
 |
| Figure 6: The modified MFJ-5008 with the (barely visible) tuning knob sticking out on the left. The blue label indicates the approximate tuning frequency. Click on the image for a larger version. |
The final result of the modification can be seen in Figure 5: The cut-off shaft of the potentiometer protrudes slightly out of the left side of the enclosure and there is a label depicting the approximate frequency of the oscillator (and the center of the converted range) with respect to the adjustment of the potentiometer and its white paint mark.
What potentiometer to use?
To fit in the location of the original 10k trimmer, one needs to use a small potentiometer: A suitably small potentiometer is the Bourns 3310C-001-503L which is available from DigiKey HERE and from Mouser Electronics HERE. With a bit of care, it can be mounted to the board and the case modified to allow the shaft to protrude out the side - but it would be a good idea to use something (e.g. "hot melt" glue) to make its mount more rigid. If you are creative, a larger potentiometer might be usable, attached with flying leads, but if it's metal, be sure to connect its body to the V- (battery negative) to minimize noise pick-up.
Note:
If you don't make the (highly recommended!) "tuning" modification, the MEMS microphone is still useful in that its sensitivity extends over a wide frequency range: You may be able to adjust the original potentiometer (which can be adjusted between 35 and 48 kHz) to a frequency that is more suited for the types of noises that you are seeking.
Using a MEMS microphone
Note:
In this section, I refer to a "homebrew" MEMS microphone carrier board - but there are "breakout" boards available that are already assembled: This next section describes how either a "breakout" board or a homebrew board like this may be mounted in the focus of the dish.
 |
| Figure 7: The original 40 kHz ceramic transducer and carrier board (top) and the homebrew version with the MEMS microphone (bottom) both mounted using the pairs of screws on stand-offs in the front bracket. Click on the image for a larger version. |
* * * * *
For the specific MFJ-5008 depicted in this article I used an already-prepared MEMS microphone module: This was described in a previous article linked HERE. This circuit was designed to accept a wide range of voltages (3.5-10) to be imposed onto the same conductor as the audio, making it easy to interface on a single cable as we did here.
In the MFJ-5008, there is an aluminum "U" channel across the front in which the ceramic transducer is mounted and its location places it at the focus of the parabolic dish. What this means is that when we replace this device with something else - a MEMS microphone in this case - it must not only be located at the same axial position (left, right, up, down) as the original, but the sensing element must also be at the same distance from the surface of the dish.
Behind the nesting cover (accessible via the removal of four screws) there is a circuit board mounted on two stand-offs and the focus of this dish is precisely midway between the two. Removing this and peering inside the original ceramic transducer, you can see the element located inside, recessed slightly from the front grille: The distance of that element from the circuit board is that which should be replicated with the replacement microphone.
 |
| Figure 8: Homebrew carrier board with MEMS microphone installed, facing the surface of the dish. The microphone's "sound hole" is centered precisely between the two mounting screws. Click on the image for a larger version. |
Using a Sparkfun MEMS "breakout" board
Soldering a tiny microphone module successfully to a circuit board requires a bit of skill - but there are "breakout" boards that already have the microphone and some of the needed components already on them - and one of these is available from SparkFun (the "BOB-19389") for about US$9.00 at the time of writing - information here:
https://www.sparkfun.com/sparkfun-analog-mems-microphone-breakout-sph8878lr5h-1.html
This breakout board contains both a microphone and an operational amplifier and here are their respective data sheets:
While it is possible to order from SparkFun directly, I ordered it via Amazon for the same price - plus shipping was "free".
As originally designed the SparkFun board "sort of" works for ultrasonic detection, but there are a few circuit elements that require attention before we use it. Consider the schematic, below:
 |
| Figure 9: Diagram of the SparkFun BOB-19389 MEMS breakout board. As can be seen, there's nothing special about this design: A microphone coupled to a single op- amp section - but there's a problem with this circuit in our application: The high gain set by R4 is unnecessary and it and C3 reduce the useful frequency response to less than 30 kHz. Click on the image for a larger version. |
The implementation of this breakout board is nothing special - and it's worth noting that even without the gain of the op-amp, the MEMS microphone itself would have a suitable amount of drive for the MFJ-5008.
As part of our circuit analysis, I will call the reader's attention to R4 and C3 (300k and 27pF, respectively) which form a simple low-pass filter - but these components, along with the limited unity-gain bandwidth product of this op amp being 1 MHz - conspire to cause the frequency response to roll off rather dramatically above 15-20 kHz or so: It will still detect lower-frequency ultrasonic signals, but sensitivity is reduced while the signals that we don't want (e.g. audio-range frequencies) are not attenuated - and even if the frequency response was flat into the ultrasonic range, it would have way too much gain for our application, anyway!
 |
| Figure 10: A close-up of the SparkFun BOB-19389 MEMS microphone break-out board. The location of C3 - now replaced by a resistor. Not also that the "sound hole" of the microphone is on the bottom of the board. Click on the image for a larger version. |
The "fix" is to replace C3 with a resistor. For the MFJ-5008 I would suggest using a 10k resistor in this location and by lowering the gain, the op amp's bandwidth product isn't going to get in the way of the needed frequency response. While it doesn't really matter if one removes the capacitor or not when using a 10k resistor (the -3dB point for a 10k resistor and 27pF capacitor is somewhere north of 500 kHz) it's pretty easy to remove just the capacitor and replace it with the resistor if you have SMD parts on hand. If you have only through-hole parts, it should be possible to tack a 1/4 or 1/8 watt 10k resistor across them. (Note: I used the MEMS board in Figure 10 for a different project which is why there's a 47k resistor at the position of C3: A 10k is appropriate for the MFJ-5008.)
The other issue is that of the voltage range of the breakout board's components. In testing, the board worked "OK" at just 1.8 volts - below the "official" specifications of the the Op Amp - but it worked "better" in the specified 2.3-3.6 volt range. In the modification for the MFJ-5008 described above, the addition of the 4.7k resistor across the "audio in" phono plug put the full 9 volts battery voltage (minus resistive drop) on this line so we need to do two things to make this work:
- Limit the voltage to the 2.3-3.6 volt range.
- Combine split the audio signal from the voltage at the microphone breakout board.
Fortunately, this is quite easy, requiring just a small number of components and the following diagram shows:
 |
| Figure 11: Powering the SparkFun MEMS break-out board from the audio cable with DC bias on it as depicted in the MFJ-5008 modifications, above. Capacitor C1 blocks the DC from the Op Amp, resistor R1 isolates the audio and DC lines while LED1 is used as a voltage shunt to limit the voltage to somewhere between 2.3 and 3.6 volts: An ordinary white or blue LED is perfect for this as they are readily available and provide a voltage in the middle of this range. Click on the image for a larger version. |
Note: I could have simply run a separate DC line from the circuit board to the detector, but this would have still required regulating the voltage down to the voltage needed for the MEMS device: Putting DC on the signal line is easy to do and it requires only a few, inexpensive components.
Capacitor C1 has two functions: Block the DC from the "Audio Out" terminal and to offer a bit of a high-pass frequency response to filter audio-range energy. Resistor R1 extracts the voltage from the "DC + Audio" line and sends it to the "VCC" terminal on the breakout board and across this, the LED acts as a voltage limiter. As noted in the diagram above, one can use a blue or white LED as the voltage limiter: These will "turn on" at between 2.8 and 3.2 volts which is right in the range that we need. Alternatively, if you have some "old fashioned" red LEDs that operate from about 1.7-1.8 volts, two of these in series will do the job.
It is recommended that you use the "diode test" function of an volt-ohm meter to verify the turn-on voltage of your LEDs and to make sure that they are connected correctly. If you have a variable-voltage bench power supply, connect it across the two leads and, starting out at less than 2 volts, slowly increase it while measuring the voltage across the "GND" and "VCC" connections: The voltage should limit in the 2.3-3.6 volt range and you should see the LED(s) dimly illuminate. In testing I haven't found that light falling on the LED causes any effects in the audio, but if you are, for some reason, worried about that, feel free to cover the LED with black paint, put it in some black heat-shrink tube or shield it from light in some other way. (Note that in the MFJ-5008, the carrier board is contained within the "C" channel aluminum pieces and mostly shielded from light, anyway.)
These three components may be mounted on a piece of prototype board to function as the "carrier" board of the same type shown in Figures 7 and 8. Note that the "sound hole" on the breakout board is on the "back" (non-component) side of the circuit board and that it is NOT in the center of the board and take this into account when you are mounting it to the "carrier" board.
Final words on the MFJ-5008 modifications
The above modifications should allow the MFJ-5008 to work over a wider variety of frequencies to allow optimum detection of energy from electric arcs, high-pressure gas leaks, bats, insects, switch-mode power supplies and many other things.
Prior to modification, a "test range" was set up in my back yard: A 40 kHz transducer was driven with a sweep/function generator (an old Wavetek Model 180) and the output set at its lowest-possible setting. From about 33 feet (10 meters) away the "warble" from the swept output was easily audible - but not particularly strong.
After the modification, the subjective impression was that the sensitivity was equal or better than the original 40 kHz ceramic transducer - but a quick walk around the house revealed the ringing presence of several switch-mode power supplies, each producing low-level noises of their own due to magnetostriction of components within - something that was totally inaudible prior to the modification, made possible only by the broad-range response of the MEMS microphone and the added ability to tune the center frequency.
* * * * *
This page stolen from ka7oei.blogspot.com
[END]
