I came across a schematic of a computer microphone from Circuit Exchange. This interested me because I have never had any success with using the microphone jack found on the back of desktop PCs. First off, the typical microphone jack seems to be a three conductor, 3.5 mm ‘headphone’ jack, with the tip being the audio signal, the ring connected to +5V through a 2.2k resistor, and the sleeve grounded. Thus it is not stereo, and it is meant to be used with a microphone that has a preamplifier. So I solved these problems by buying a Beringer U-Control for a few tens of dollars. It plugs into a USB port and has stereo.
But what interested me about this circuit is its use of a microphone preamplifier and the use of the computer’s +5V power. For a long time I have wanted to build a remotely powered microphone that uses only two wires, a single twisted pair. Both power and signal are on the same pair, just like a telephone set. In fact, I built a power supply for two telephones, and just used a pair of old telephone sets (see notes at end).
One problem I see with this circuit is it uses three wires. The problem is that most low voltage cables come in two wire unshielded twisted pairs, such as for telephone wiring, or they have wires that are shielded with a piece of foil and a bare ‘drain wire’. Shielded wire is more expensive but not common and unshielded twisted pair is inexpensive and very common.
In this case, one wire is power, so the microphone signal is on the other wire and power and signal return through the ground wire. This means that the audio signal is not balanced, so the cable should be shielded. For a long length of shielded cable this could be expensive.
A solution could be to use two pairs. One pair would be power and ground. The second pair would be the audio signal, balanced relative to ground. The microphone circuit would have a transformer to change the unbalanced primary and change it to a balanced secondary winding of low impedance that is not connected to ground. The other end would use another transformer to change it back to unbalanced. This would allow using unshielded wire, but the transformers are expensive and not easily obtainable. Some designs eliminate the transformers by using transistors to make the signal balanced, but this is still more expensive than a simpler circuit (for more transformer information see notes at end).
My original intercom
At one location where I lived, I built an intercom. It had four locations, and the intercom didn’t have microphones, it used small speakers as microphones. To hook them up, I used a lot of four wire ‘IW’ — inside wire that the telcos used. This cable had four wires and none of them were twisted together to form a pair. I built a preamp that I could connect between the speaker and cable, and I used black for common negative, and yellow, since those two wires were not used for a single line phone.
Some of the longest cables were well over 100 feet or 30 meters long; probably more like 130 to 150 feet. Several cables ran through places where electrical cables were, and underground. I really didn’t have a problem with AC hum – I could hear it faintly. I think the biggest problem was having the speakers pick up low frequencies because the cone was much larger than a microphone. And one time one of the remotes was ‘bugged’ by a cricket that was louder than anything else.
The system was meant to be used for listening for intruders or vandals. I could switch the direction, so I could talk into the speaker and one remote could hear me. But almost all of the time I was just listening. One thing I found was that a speaker a few inches across was too obvious, people could see it and figure out what was going on. Some vandals then tried to damage the speaker.
Back then there was no law about having to get the consent of the person before recording. And of course there were no such things as video surveillance systems. Video would have to use coax cable, which is much larger and more expensive. The coax cable must be fitted with coax connectors, typically crimp-on F Type or Type BNC.
More Recently
Today, things are much different. Higher technology, and more legal complexity. The video surveillance is much better than audio alone. But the audio can still be useful. Because an audio microphone and its cabling is so much cheaper than a video camera and coax, more microphones can be located at more monitoring points. The cabling can be the same as telephone cabling. If two pairs of wires are available then one pair can be used for power, and the other pair for the audio signal. But I had always thought that a single pair should be used for both power and signal, like a telephone.
The idea is to amplify the weak microphone signal at the microphone before it is sent over the long cable. This makes the signal much greater than noise, hum and other interference such as radio stations.
I found several schematics of circuits that used a simple resistor and capacitor to separate the power and audio at both ends. The capacitor blocks the DC power, but passes the audio. The capacitor and resistor are used to keep the audio out of the DC power. There are typically two transistors, the first one is a common emitter which boosts the low voltage microphone signal, and the second is a common collector or emitter follower, to drive the long cable at low impedance.
One other version uses a single transistor to amplify the microphone voltage, and a transformer to match the transistor’s high output impedance to the low cable impedance of a few hundred ohms. But transformers are expensive and can pick up AC hum if they are not shielded. Some of the highest quality microphones use transformers, but in my case I don’t need high quality, or high fidelity audio, and low cost is more important.
The two transistor circuit has a common ground, so it is not balanced. But the circuit is small, only a square inch or 5 cm square, so it has very little difference in its imbalance compared to the overall cable and circuit.
The power is blocked from the audio signal by a capacitor. The power is passed through a resistor and audio is removed by another large capacitor. There is a lot of the power wasted in the filter circuits, but the microphone amp and driver transistors get enough to drive the twisted pair line at high enough level to overcome any noise.
There are other ways of doing this. One was would be to modulate the audio on to a Frequency Modulated carrier. This is more resistant to noise than audio alone. With two separate carriers, the cable could have two separate microphones for stereo.
The audio could be digitized before it is sent over the twisted pair cable. Depending on the cable length, the audio and even video might be sent. But many security cameras are already designed to be powered by the Cat5 datacomm cabling and talk to the network like a computer, so it would be easier to replace the single pair with inexpensive Cat5 datacomm cabling and get high resolution video, and also have high speed datacomm.
One example of reusing a single twisted pair for high speed datacomm is DSL. The telcos use DSL modems to get up to 1.5 megabits per second from the central office to homes, or even higher from a neighborhood concentrator to homes (“at&t Uverse”). This is using a single ‘Cat 0’ twisted pair made for telephone use only.
More Recent Technology
I thought about what has been going on in the last decade with data communication. Most people have heard of WiFi and Bluetooth but few have heard of ZigBee. All are a way of communicating between smart devices so that voice and data can be shared between multiple devices. In this case, radio frequencies are used between smart devices. But since power is needed at the remote end, a cable must be run to it, so it seems logical to use the cable for both power and signals.
The bandwidth will not be that high. A telephone set can use digital compression and get reasonable voice quality with 9600 bps (bits per second). There are modems and line drivers that can go up to a megabit or more over telephone cable, so a few hundred kbps seems reasonable for an unshielded twisted pair of 300 meters or 1000 feet long.
The main thing that is needed is a dedicated computer at each end to implement some kind of protocol to compress, decompress, and do error correction on the data.