/diy/ - Do It Yourself

>>1004139
The main problem is that the output voltage will fluctuate with load and the division ratio will fluctuate with input voltage.

You'd get less variation with input voltage if the lower leg of the voltage divider was a 4.3k resistor and a diode (or another emitter-follower). As it stands, you have to tweak the resistance ratios so that the reference voltage is off by Vbe ~= 0.7V. But clearly any resistive divider is going to produce a reference voltage proportional to the input voltage.

You'd get less variation with load current if you reduce the output impedance and/or increase the gain. This is why an op-amp is usually preferable. But op-amps can only supply ~10mA; if that isn't enough, you can use an emitter follower to boost the output current from an op-amp (include the transistor in the feedback loop to preserve the linearity).

Whatever approach you take, you can't avoid the fact that you'll be dissipating half the supply voltage times the ground current (i.e. you're effectively building a linear regulator). Linear rail splitters are designed for the case where "ground" is a reference voltage, not a power rail.