Coxing Amplifier - Circuit
Battery Charger
The Ni-Cad cells used here should be charged with a constant current set at a tenth of their rated capacity. I used a 9.6v battery with 700mAh capacity, so the charging circuit is set to 70mA. As the battery charges the voltage drop across it increases, but the current flowing through it remains constant.
The circuit works by using the fixed voltage drop across the LED. This also acts as an indicator to show that the supply is present. R1 limits the current through the LED to approximately 20mA, and the voltage drop across the LED is 1.9v. The base-emitter voltage drop of the transistor is 0.6v, and so the voltage across R2 is 1.3v. I wanted a current of 70mA to flow through the battery, through the transistor and through R2. So I set R2 to the value 1.3v / 70mA = 18.5 Ohms, or 20 Ohms as the nearest higher value available.
The collector of the transistor will draw 70mA from the supply, through the battery, so long as supply voltage is at least 1.9v more than the voltage across the battery. With a 12v supply the current will reduce once the voltage across the battery reaches 10.2v. At this point the battery is charged, although the LED will remain on. The diode at the transistor collector is to stop the battery leaking current back through the transistor and through R1, which would otherwise lead to a flat battery after a week or so.
Microphone Pre-Amp
The pinout for the SSM2167 is shown above, with a link to the data sheet. The SSM2167 replaces the SSM2165. The circuit I used was taken directly from the data sheet and is shown below. The 2v fixed voltage across R4 and the electret microphone (on the headset) was achieved using an LED, which also indicated the power was ON. The current through the LED was limited to 20mA with R3, which was chosen assuming the microphone had no resistance. Changes in the microphone impedance will be small and will not affect the voltage across the LED, so the microphone is protected from noise in the supply rail. This also helps prevent feedback and keeps the circuit stable at high frequencies. If electronic oscillation occurs high frequencies may become distorted, this can be reduced by placing capacitors across the supply near the IC (pins 1 and 8) and across the LED. I did not find this was necessary. The IC requires a supply voltage between 5v and 10v and so the 9.6v battery could supply the IC directly.
The action of the compressor is best described using the graph below. It shows that for small input signals the output remains a zero, removing hiss, hum and noise from wind across the microphone. At low input the gain is high and the output is larger than the input. This is the compression region and it gives an extra boost if the cox talks quietly or the microphone moves away from the cox's mouth. At high input the gain is reduced and the output is roughly constant for any input. This is the limiting region and it stops the output distorting if the cox starts shouting (most of the time!) Only the compression ratio (the amount of extra gain for quiet input signals) can be varied on the SSM2165. The SSM2166 is a 14 pin package that is very similar to the SSM2165, but allows the noise gate and limiting region to be set as well. This adds to the cost and complexity slightly, and I found that the SSM2165 worked fine for this application.
Power Amplifier
The pinout for the TDA2003 is shown above, with a link to the data sheet. The circuit I used was taken directly from the data sheet and is shown below. The TDA2003 Op-Amp has a high output current limit of 3.5 Amps, low cross-over distortion and several protection features. It is designed to run from a single supply between 8 and 18v and is primarily intended for car audio systems. The 9.6v battery was sufficient for this and gave a maximum output power of 15 watts.
The output can be safely shorted to ground indefinitely, and if the device overheats it safely reduces the gain. The tab is connected to ground, and so no insulation washers are necessary. The output can drive speakers down to 1.6 Ohms - we measured ours as slightly over 2 Ohms when they were connected in parallel.
The Input signal strength is set using a potential divider. This is used as a volume control although it could be set to a fixed value if a volume control was considered unnecessary. The two Op-Amps are used in a bridge configuration allowing the output differential two swing from 9.6v down to -9.6v - This is double the range of a single Op-Amp, which would need one side of the speaker connected to ground, and give an output of 0 to 9.6v. The double in output signal multiplies the power by 4 up to a totally unnecessary 15 watts (rms). The bridge configuration is also low in component count, with capacitors for input and output decoupling, and a feedback network. This is described in detail on the datasheet.
The output was connected to the speakers using an Amphenol Series 44 circular 5 pin connector, as used by Nielsen Kellerman. The pin-out for this is given here: (The rate sensor pin was left unconnected)
