I have often wondered about the choice of diode in the driver bias circuit like D46 below. I happen to look down at my variable power supply and saw a laser diode and I recalled a time when I would drive laser diodes until they gave up the smoke. Figured I would share how I made sense of this diode for others that might want a simple explanation.
[ATTACH=full]66902[/ATTACH]
Have you ever connected a LED to a variable supply without a current limiting resistor? With careful control of the voltage, you can avoid burning out the LED. In fact, you can quite accurately set the current through it if you carefully adjust the voltage to it. They don't just turn on all at once, they come on exponentially. That means small changes in voltage cause large changes in current and vice versa. So, lets say we connect the driver base directly to our variable power supply as if it were the LED and slowly raise the voltage until we get our desired collector current. Write that down and also write down the current from the variable supply biasing the base. A variable resistor between the 8v supply and the base (with no diode) will have way too much range and within a small amount of rotation, the transistor will go from off to fully on and you'll have a hard time stopping at 50mA. What if we found a diode that, at about 10x the base current, had a voltage drop matching what we needed at the base to achieve the desired collector current? Having the diode allows a large change in current to make a small change in voltage, and that allows for a fine adjustment of the bias current. The resistor values are chosen so that the range in current adjustment moves the diode forward voltage drop above and below the voltage we determined we needed for the 50mA or whatever collector current. Think adjustable voltage regulator with a very narrow range. Those laser diodes did not die in vain.
Edit: Temperature compensation just clicked for me while thinking about this. If you fed the transistor base a constant current and recorded its voltage drift with temperature, you could find a diode that, when biased to match the transistor, its Vf drifted the same amount as the transistor, then attaching that diode to the transistor would lower the bias voltage (thus current) as to maintain that constant current as it warms up. I know thinking about BJTs in terms of bias voltage with the emitter grounded is hard to understand, but remember, that 0.6v Vf isn't rock solid, its dependent on current. So thinking in terms of voltage works if your DMM has more than 2 digits.
