Of Mics and Men
An Adventure with Sideband Audio and Home-brewed Microphones
By Tom Morton, W5TOM
As Seen in March 2003 Issue of QST
Cover Photos by Tom Morton
Since the beginning of radio, Hams have been challenging every aspect of the communication art. The modern transceivers leave little to for the user to improve when the covers are removed and the experimenter is left with what is connected to the magic box. In recent years the attention has been directed to SSB transmitted audio. You only have to tune the HF bands to find Hams trying to make their 3 kc bandwidth sound like a broadcast station. Certainly the original SSB sound (300 to 2.5k cycles) can be greatly improved and produce a more pleasing sound. The first attempt was to use a full range studio type mic. Immediately it became evident that the audio was very bassy and needed equalization. Enter the professional audio gear. Although some achieved their goals, this produced great expenditures and frustration with equipment not designed for the 3kc bandwidth and the RF environment. I also chased my tail with the microphone and EQ game, but my thirst for simplicity and curiosity led me to find out what was really happening.
It seemed logical if you used a mic that was flat in response across your pass band, you should get a flat transmit audio. But something was happening to the high and the very low frequencies. There was a steady slope to the spectrum, and the highest frequency was between 10 to 20dB down from the lowest. I suspected that nature of the sideband formation and reconstruction was the culprit. A simple band noise test made with the Hamalyzer, spectrum analyzer software that runs on your computer and receives the audio signal from your radio, through the sound card. This showed that the receiver in SSB mode received all the frequencies within the 3kcs at equal strength. So, I thought it must be the transmitter. I ran a test using a signal generator that put a constant voltage into my mic input for all frequencies in the 3kc pass band. I ran the output of my transceiver through a Bird peak reading Wattmeter and then into a dummy load. To my surprise the output was a constant 100Watts across the whole 3kc. except for the part below 100cps, which was rolled off a few dB. A west coast engineer and frequent QSOer of mine, Rod, KQ6F ran a similar test using a spectrum analyzer and found the same results. He suspected the human voice to be the problem. Since it was not the mic or the transceiver, I reluctantly accepted his theses. Then I was off to the Internet. I discovered an article from Purdue University dealing with the nature of the human voice. Their project was to come up with a mechanical device that could accurately reproduce the human voice. Needless to say our primary modulator is a very complex device and the things it does to the air after it passes over the vocal cords is quite remarkable. It seems that the low frequency percussive consonants and vowels actually produce a vortex leaving the mouth. Since a vortex is the actual movement of a package of air, it has great energy. The lower, waveform frequencies also have higher energy. On the converse, pinching off the airflow with the tongue, teeth, and lips forms the high frequencies. This produces eddy currents in the air that substantially destroy the waveform and the energy striking the mic diaphragm. This justified the test results and the voice was definitely the problem.
Being a firm believer in the ďKISS PrincipalĒ (Keep it Simple Stupid), I wanted a mic that could be used straight into the mic jack with out all the EQ business. I realized that the mic I needed would have to have a high frequency boost of considerable magnitude and a bit of a boost below 150cps. The Astatic D104 seemed to be the only one I had heard that seemed to do that, however, they never seemed to have the necessary low frequencies. At the time I was using a low Q, passive, 2 band, EQ I had built, and I thought that I could tame the D104. A trip to the local CB shop was in order. After making my way past the sleeping dog and the thousand watt amps topped with old pizza boxes, I discovered a D104 head at a bargain price. It included a spec sheet with a frequency response graph. I was amazed to discover that it went flat down from 1kc to 50cps and had a 12dB rise from there to 3kc. The only catch was it had to see a very high impedance. The higher the impedance the better the lows were. The mic was designed by a couple of hams, Chorpening and Woodworth, back in the tube AM days in 1933, and led to their founding of the Astatic company. It was a new concept utilizing the Rochelle crystal as the transducer. In itís original use it saw a high impedance from the grid of a tube and performed very well. Some of the amplified versions out today transform the impedance to the 600 Ohms required by the modern radios, but donít seem to go far enough. Not having a FET with a high impedance input, I liberated one from a cheap electret mic I had on hand in my bottomless junk box. The input impedance was on the order of 100megohms. I put the D104 cartridge straight into the FET just as the electret element had been. From there on, the mic was treated just as an electret would be. It gets its operating voltage from the radio through a 10k resistor and the output is through a 4.7mF capacitor. The source is grounded, hot mic goes to the gate and the Vin resistor and the output capacitor are connected to the drain. This was made into a small package and placed inside the mic enclosure. The electret FETs apparently are self-biasing and give a very linear, low noise response, making them ideally suited for the job. Since the graph for the D104 specified 5megs for the low frequency response, I put a 5meg resistor across the element. The results were very good, but the lows werenít all there. So I kept increasing the parallel resistor value until the best results were had at 50megs. I noticed that the element had padding in direct contact with the diaphragm without which the mic sounded very sharp and tinny. After playing with the padding, I found I could control the amount of highs, and with the parallel resistor, I could control the amount of lows. My futile little brain had a flashÖI could EQ a mic! My desire for simplicity was in view.
Not wanting to destroy the D104 I had working pretty well, I thought of playing with some cheap crystal mics. I had a couple in my bottomless junk box and I ordered two more from Mouser in the 3 to 4 buck range. Realizing that the most important factor in mics is how they look (you donít have to listen to them, but you do have to look at them), I wanted to have a nice housing for my elements. I had a Shure 520D (the Green Bullet) mic that was lousy for our use. I scraped the guts into the trashcan only to discover that the crystal element I had would not fit. I tore one apart and realized how simple they were. I mounted the crystal in a jar lid I had that fit the Green Bullet. I used a small nail for the fulcrum and Reynolds Wrap for the diaphragm. The new cocktail onion jar lid element and the electret FET were packed into the housing with a 40meg shunt and quilt padding on the diaphragm. I was totally amazed how good it sounded. I was expecting a strong tin can sound, but it was smooth and far better than most manufactured elements for Ham radio. I began experimenting with the padding and the tension on the diaphragm. I found that I could tighten the diaphragm by running my fingernail carefully around the edge and could loosen it by running my finger around halfway between the edge and the center. The tighter it was the higher the frequency shift, the looser the lower. More padding restricted the highs and more pressure increased the highs. Now I had complete control of the spectrum response of the mic, even if it was a bit more of an art than a science. The Reynolds Wrap was much thinner than the diaphragms on manufactured mics and seemed to help the very low frequencies. It was more like tuning a Stradivarius, and microphone element manufacture no longer seemed like a black art practiced by big companies with German names.
After careful tuning, I worked a few Hams on 17mtrs and told them I was using a mic element I had made from a jar lid, Reynolds Wrap, and the FET from an electret mic, plugged directly into the radio. No one believed me because it sounded like I was using a studio mic and some EQ gear. Apparently no one had tried home-brewing mic elements since Chorpening and Woodworth. My web site, which has a lot of humor and put-ons didnít help my creditability either. I wanted to see if I could repeat the accomplishment, but I didnít want to disturb the mic I had working so well in the Green Bullet, so I needed a new platform for my cartridges. I didnít want to build one in a tuna can since the way it looks is primary and my creditability was suffering enough as it is. I began to search for something that I could make into an attractive mic. The parts began to come together by applying ďGazinta-Gizonta EngineeringĒ (looking for something that goes into or onto something else) and combing the discount and hardware stores. I must be the only guy that shops Walmart with a dial caliper, because when I was in the kitchen department checking out the exact diameter of the stainless steel measuring cups a security person was sent to see what I was doing. I tried explaining, but it was futile. After realizing that I was neither a threat to national security or the health of man and beast, she departed. I bought the set of measuring cups that had one at the right diameter. At Target, I had a slam dunk with a stainless powdered sugar shaker. I now had the raw materials for the two new platforms featured in this article, the SMC1 and the SS1 (Strainer/Measuring Cup and Sugar Shaker). A little work with a hacksaw, a drill press, and some screws and hardware they were ready for elements.
Now that I had respectable looking housings, I was interested in making the elements from parts easily available. I had everything but the piezo crystal. I tried the piezos from sounders (like those found in smoke alarms), a phono pickup, cigarette lighters, piezo tweeters, and everything Radio Shack had that was piezo active. Even though they all produced reasonable sounding elements, none had the bass needed and found with my element that had the crystal from the cheap mic. It became evident that the crystal in mics was special and different from other piezo active materials. I researched the Rochelle crystal (Sodium Potassium Tartrate) and found a lot of information and some swell photomicrographs, but no possible sources. The mics arrived from Mouser, one crystal and one ceramic. I tried the ceramic piezo, but it was also missing the lows. The crystal from the crystal mic worked great and found itís way into a cherry jar lid and the SMC1 mic. The SS1 received an element made from another old crystal mic from the junk box. Now I had three great sounding mics that solved the problem.
On the air testing with Sam, KT4QW and his Hamalyzer produced the trace found in Fig.1 while using a studio mic straight into the radio. Itís evident that the high frequencies take a 15dB dive. It sounded bassy and lacked brilliance. If the signal had been near the noise level it would have been uncopyable. Fig.2 shows the SS1 trace, which is pretty flat and sounds clean and crisp and has the EQed broadcast sound. I wanted to see just what my mics were doing so I connected one directly to the computer running the Hamalyzer and placed it in front of my stereo playing music. FM stations put out signals with flat EQ across their whole bandwidth and I have a good and balanced setup. Fig.3 shows a full range studio mic; as you can see, it is flat and lacks the necessary boosts. Fig.4 shows the SS1 with its boost on the bottom and the rise to the top. It reflects the traditional ďsmiley faceĒ shape most used by Hams using graphic EQs, i.e. a boost on the bottom and the top. Note that it goes down the tubes after 3kc but that is out of my pass band.
I have learned a lot, had fun, and have three EQed mics that do the job with nothing between the mic and the radio but wire, each with their own character. My rig is a Ten Tec Pegasus controlled by a laptop computer running N4PY software and the Hamalyzer (Hamalyzer.com). I encourage you to try this; itís easier than you think. Donít waste your time with piezos, just buy the crystal mic #25LM024, $4.17, from Mouser (800-346-6873), and use the crystal in the jar lid of your choice.