Well, here goes. I did a lot reading in books I have on lasers as well as some digging around on the internet.
You can review the literature in the reference section below. Believe it or not, some old radio books gave me great insight to
transmitting and receiving signals in general. I wanted to create these circuits from scratch, so that meant reading about and
understanding solid-state diode laser characteristics, as well as, PWM techniques. I found dozens of circuits that were complicated
and unwieldy but in my study of electronics it seemed to me it all boils down to voltage and current. Seems easy enough.
For this project, the pulse width signal is created and the pulse width varied by using an audio signal from a source such as an iPhone
and a separate carrier signal. The audio signal is modulated (mixed) with a high-frequency triangle wave as the carrier signal.
The objective is to transmit the audio signal across the room or across the street to an optical receiver.
The transmit circuitry has three stages; an audio amplifier, a triangle wave generator and a modulator for marrying the
audio to the triangle wave to pulse the laser diode.
A full range audio signal is generally accepted to be in the range of 20Hz to 20KHz. The carrier wave is used as the sampler of
the audio signal and I wanted to sample the audio at a much higher rate than the audio. There is a theorem called the Nyquist
Theorem. The Nyquist Theorem, also known as the Sampling Theorem, is a principle used in the digitization of analog signals.
For analog-to-digital conversion (ADC) to result in a faithful reproduction of the signal, samples of the analog waveform must
be taken frequently. How frequently, you ask. The number of samples per second is called the sampling rate or sampling frequency.
Suppose the highest frequency component, in Hertz, for a given analog signal is fmax. According to the Nyquist Theorem,
the sampling rate must be at least 2fmax, or twice the highest analog frequency component.
In our case, 20KHz is fmax. This means I need a 40KHz carrier (sampling) signal.
Believe it or not, 44.1KHz is the most popular sampling frequency used in CD's and other audio sampling devices so 40KHz seems
okay. 44.1KHz is not hard to create but for a cheaper design using "off the shelf" components, something a little less than 44.1KHz
sounds better to me. Get it, "sounds better". This may be a comedy white paper after all.
Anyway, continuing this idea, I know that a piano has a low-end frequency of 28Hz and a high-end frequency of 4,186Hz.
For humans, the most sensitivity is from approximately 300Hz to 5KHz with a particularly sensitive spot around 2 - 4KHz.
So, all this means is I don't need a 40KHz sampling frequency. Anything lower than 40KHz and higher than 10KHz, and convenient,
will work for this experiment. I chose 18KHz simply because of the way a 555-Timer IC works with common capacitors and resistors.
This 18KHz sample frequency will work fine but any audio signal over 9KHz will not get sampled accurately. With my old ears,
I can't hear anything over about 8Khz, anyway!!!
Now, if you are wondering about other sampling technologies other than audio, how about this. For analog-to-digital (ADC) sampling, one may
need a lot more cycles per second to accomplish a good sample. My Rigol DS1054 digital oscilloscope samples at 1 billion times
per second or 1 sample every nanosecond. Assuming the Nyquist theorem was used to design the scope, that 1B samples per second
should be 2 times the frequency of the maximum signal being sampled so the scope can at most, safely, sample a 500MHz signal,
with accuracy. Well, the scope is only a 50MHz scope so the Nyquist sampling only needs to be 100MHz. At 1B samples per second the
scope gets 10X the samples necessary for a good sample.
Let's move on.
The Audio Signal
Now, this part gets tricky. Say, for testing purposes, we have a pure sine wave which will represent our audio
signal like the one shown at the right. This is a look at the amplified audio signal. I tested the output of an iPad and at
mid volume I tested 100mV. I need about 3 to 5 volts for this experiment. That is about a 40X gain in voltage. This is easily
accomplished with an Operational Amplifier. An Op Amp is a readily available 8-pin component that costs about 60 cents.
I have dozens of them in
my parts bin. I chose to test with the Linear Technology LT1002 version. This component is fairly robust with fast rise times.
Any Op Amp will do, really, such as the LM311 or the LM741 all of which I will test in this project.
"Rise time" is the time it takes the Op Amp to go from 0V to Vmax when the amp is triggered. This trigger will be the
audio signal converging with with the triangle wave. The rise time for LM741 -- also called the slew rate -- is 0.7 volts / µs.
So, let me state this differently because it is important and confusing.
My carrier signal is 18.125KHz which is basically a pulse every 55.172µs. If the Op Amp can deliver 0.7V/µs, and I have
55.172µs before the next pulse of the triangle wave, then the Op Amp can go from 0V to 38.6V in the time it takes for one pulse
of the signal. We only need 3.3 volts so this is ample time.
This is another suble reason why I chose a slower sample rate than 44.1KHz. The faster the sample frequency, the faster the rise times needed to get from
0V to whatever volts you need. There are super fast op amps but they are more expensive.
The Triangle Wave Signal
Next, we need the triangle wave, that is, the 18KHz triangle wave we discussed above.
This signal acts similar to a carrier wave in a radio transmission.
The image to the right shows output from the modeling
software I used. I added the dashed lines with Visio to show how I get my reference numbers.
For the mathematically inclined, subtract 0.8ms from 1.6ms to get 0.8ms (800us). I chose those values as the triangle wave in the image
landed nicely on those values. Now count the peaks between the dashed lines and we get 14.5 peaks. And, 0.8ms divided by 14.5
gives us 55.172us which is the period. Using
we get 1/0.000055172 or 18,125Hz (18.125KHz). Nice. I love it when the math
The base calculations for the 555 timer IC show 18.222KHz for the components I used. See below section on Triangle Wave Generation Stage.
I used an NE555 timer integrated circuit (IC) as my signal generator. It will generate triangle and square waves easily.
With just two resistors and a capacitor I can get 18,125Hz. Even though the formulas for the LM555 give us 18,222Hz, 18,125 is about as close
as we can get given tolerances on components.
Combining the Signals
Now comes the analog goodness. I need to combine these two waves, audio, and triangle, using a comparator.
The intent is to use the comparator to trigger when the sinewave hits the rising and falling edges of the triangle wave.
On the positive going swing of the sinewave, the pulses will be quick as shown in the figure to the right.
This is because as the sinewave swings more and more positive it cuts across less and less of the triangle wave.
On the negative swing of the sine wave, the trigger point is longer. This is because, as the sine wave goes more and more
negative it cuts across more of an open area of the triangle wave. Of course,
we do this 18,125 times a second which samples the sinewave 18,125 times per one cycle of the sinewave. Very cool.
I can use this triggering mechanism to generate a pulse whose width is proportional to any point on the sine wave.
All right, so I measured this on the simulator and using a calculator I found 0.000004s (4µs) pulse on the highest
swing of the sinewave and 0.000030s (30µs) on the most negative swing. Very fine slices but it should work.
The idea is to pulse the laser on and off with pulses of 3.3 volts for time values of between 4µs and 30µs.
Keep in mind that 18,125 times a second may seem fast but compared to MHz and GHz signals, this monster is WAY slow.
Now, what does all this look like, modeled of course. The red signal in the figure to the right demonstrates the pulse width signal which is
exactly what will be pulsing the laser. The image shows all three waveforms combined to illustrate the
effect of the sinewave modulated with the triangle wave and triggering a comparator (LT1011).
And, we have a pulse width modulated wave form. I just hope it works on the bench.
There are some other subtle things going on here for the observant reader. The laser wants a pulse from 0 volts to a maximum of
3.3 volts. The sine wave input signal has a negative swing in it. I need to ramp up the entire signal above zero volts, as well as,
keeping it under 3.3 volts so I don't torch the laser diode. All of this craziness is modeled in the diagrams on the right.
Triangle Wave Generation Stage
The circuit shown on the right is my design for the triangle wave generator based upon some books and
web pages and hours of modeling with software. It produces a nice 18KHz triangle wave with the component values presented here.
The 555 Timer is an industry workhorse for generating square waves and triangle waves. With two 10K ohm resistors and a 2.63nf
capacitor we get 18,125 cycles per second. Some calculations:
t=0.693(9951 + 2x10048)(0.00000000263)
f=18,222 = 18.22KHz
Audio Amplifier Circuit Stage
The audio amplifier circuit is illustrated on the right. There are two functional entities on display here; the audio amplifier
and the test signal source for the sinewave. This is a typical non-inverting Op Amp circuit configuration. I use two resistors
as a voltage divider
to adjust the amplitude and this will be the volume control (R3 and R6). I used the LT1002 as the main signal amplifier although
I tried the LM741 and LM311. All these components worked but the LT1002 is a more recent IC and has more precise rise times than
the older 741 and 311.
The rationale behind this stage of the circuit is that I wanted to be able to amplify, if needed, the audio signal coming into
the project. A typical iPad has about a 100mV signal on the 3.5mm audio output connector. Or, a raw guitar signal has about 150mV
signal depending on the pickups. I may not have needed this stage but in the interest of learning I decided to add it anyway.
The audio generator section is a sine wave generator for a 2,000Hz, 0.5vpp sine wave. This makes it easy to test and
visualize what is going on in the circuit. Once everything works with the sine wave, it should work with an audio signal. I can test
various frequencies in the audio spectrum.
The comparator is a Linear Technologies LT1011 fast precision comparator. Remember from above the slew-rate discussion.
The LT1011 has a 150ns rise time. Very fast. It takes the triangle wave and the audio signal
(sinewave as our test case) and depending on when the signals converge, fires an output signal proportional to the time when the
signals cross. See above for more details. The comparator is interesting, as most op amps are, in that once the trigger is fired,
the signal goes to full power at the rails. In our case when the signal triggers, I get 6 volts and when the signal goes to an off state
I get negative 6 volts. I limit the 6 volts with the 217 ohm R7 resistor so we don't give too much current to the laser. The
comparator can also use negative voltage for the bottom half of the rail but our laser needs positive voltages so I injected
the PWM signal into a voltage divider. Modeling this shows 33 milliamps of current which is perfect actually. My laser diode
is 148 ohms indicated by R8. R8 is a placeholder for the laser diode as I have no way of modeling it directly. Ohms Law states
that if one knows the resistance and the voltage, one can find the current.
I = V/R
The receiver circuit was not too difficult. I used a standard LM386 audio amplifier integrated circuit with some support
components. The LM386 is a very popular 8-pin DIP IC. It is a little tricky in that it will oscillate easily when you don't want it.
But, there are literally hundreds of sites and circuits that show useful ways to use this popular IC.
I used a photo-transistor as the optical receiving element and then amplified the signal with the LM386 and output the
amplified signal to a small speaker.
The only challenge I faced is how to power the receiver without any connections to the transmitter. I did not want anyone to think
there was some funny business going on with the tranmission of the signal. I finally just ran two wires (PS and GND) to the receiver circuit from the
main power supply in the tranmitter board.
This LM386 IC is an amazing chip. I have used it for years. It is suitable for battery-powered devices such as radios,
guitar amplifiers, and hobby electronics projects. The IC consists of an 8 pin dual in-line package (DIP-8)
and can output 0.25 to 1 watts of power depending on the model using a 9-volt power supply.
Etching the Circuit Board
Once the breadboard testing is completed, we move on to actual circuit board construction. This part is more mechanical and
chemical than electrical, for sure. I etch my own boards as it is far cheaper than sending the schematics off to a board house
and paying $200 each for a nice board. Call me cheap!!! Especially, if you are a hack like me and might need two or three
trys before you get the board setup right. Very expensive. I can etch own circuit boards for $2.00 or $3.00.
The figure to the right shows the copper traces, top and bottom, for the transmitter -- blue for the bottom traces and red for the top trace.
For these types of hand-made circuits, I should
have known better than to have too many top traces. They require vias (holes with places for a tiny wire to go through) and
they are difficult to troubleshoot and manufacture in a garage setting. On subsequent boards, I will have no top traces or
perhaps just one trace such as a ground plane.
An entire book can be written about etching boards; the methods, examples, fails, etc. For me, this is a process which takes patience.
First, I use Eagle CAD to layout the components and traces. I use a wide trace since I use a Sharpie to draw the traces. Ferric Chloride
just eats metal so the Sharpie trace remains. I use finger nail polish to remove the Sharpie and the copper trace remains. Once the
board is laid out I get it out to my drill press and drill all the component hole. I drill from the bottom up because the drill
leaves a tiny burred edge that snags the Sharpie. I want to avoid that. Once the holes are drilled, I use the Sharpie to connect
all the holes with traces. After that, double check everything and put it in the Ferric Chloride solution for about an hour
- 45 minutes in the summer. I use nail polish removed or acetone to remove the SHarpie ink. Now solder on the components.
In the links below, I have a picture of the completed etched board with the scope attached, signal generator attached and the power applied.
The scope is reading the triangle wave and that is shown in the links below.
Issues and Next Steps
1. The audio was a little crackly and distorted. The song came through and was impressive but no real sound fidelity out of the 3-inch
speaker. I will devise some high-pass filters to take off the treble aspects of the sound. The goal of the project was achieved and that
was to transmit sound. Sound quality is another project.
2. Also, the transmitter PWM signal has some fuzz on it when viewed in the time domain on the scope. Looks like phase jitter but certainly
not bad enough to stop the sound from being reproduced. I believe this is a bypass issue but I need to research it. I may need to move some
of the capacitors closer to the IC's. Believe it or not, there is nano-Henry inductance and pF capacitance in the traces and leads.
This could be causing some unwanted resonance.
Update 11/03/2016: I found that if I add a 0.01uF tantalum capacitor from pin 6 of the audio amp to ground, the phase jitter is gone.
I will add this to the next test.
3. Combine the power sources. Right now I have the wall wart powering the transmitter and my benchtop power supply powering the receiver.
Both use 12.2 volts DC.
Update 11/03/2016: I now use the 12.2 V wall wart to power the entire project.
As of 09/25/2016, I am working on a newer version and only have 3 short top traces. Much better design. See image to the right.
Okay, this was the most complex project to date for me. It involved many technical areas which was a real challenge but I
learned a lot about signal generation, transmit and receive techniques as well as PWM. All these topics are beneficial to an engineer.
I have not tried to go any real distance with the receiver. I tried a couple of feet and that worked fine. The hardest part of
distance is alignment of the beam.
As you probably deduced from the text above, the majority of these types of projects is design and modeling. That seems fair to me.
I would rather model with software for weeks than build a dozen different circuit boards. In this case, I have only constructed one
circuit board and will build the second one which should finish this project.
Below, are links to PDF, graphics and videos for those that want to
look a little further.
These are all great references books. Over the years, these have become my "go to" books for circuit design and troubleshooting.
- Horowitz and Hill. The Art of Electronics. (1989). Cambridge University Press.
- Jeff Hecht. The Laser Guidebook (2nd Ed). (1992). McGraw Hill.
- Jeff Hecht. Understanding Lasers. (3rd Ed). (2008). IEEE Press.
- Edward L. Safford. The Fiber Optics and Laser Handbook. (1984). Tab Books.
- Delton T. Horn. The Laser Experimenter's Handbook. (2nd Ed). (1988). Tab Books.
- Chih-Tang Sah. Fundamentals of Solid State Electronics. (1991). World Scientific.
- DeFrance, J. J. General Electric Circuits. (2nd Ed.) (1076). Pages 132 to 146. Holt, Rinehart & Winston.
- Bob Pease. Troubleshooting Analog Circuits. (1991).
Web Site References