Taming Noisy Voltage Regulators

The ubiquitous LM78XX voltage regulators are great for general purpose use, but are completely inadequate for low noise applications. I learned this the hard way recently while working on a high gain audio amplifier design; when powered from my bench power supply the circuit worked fine, showing nothing above the noise floor of the ‘scope (about 10mVpp) coming from the audio pre-amplifier section, but when the power supply was run through an LM7810 regulator, noise at the pre-amp output increased to over 120mVpp!

Pre-amplifier noise when powered directly from the bench power supply

Pre-amplifier noise when powered directly from the bench power supply

Pre-amplifier noise when powered through an LM7810

Pre-amplifier noise when powered through an LM7810

The obvious solution of course is to use low-noise voltage regulators to power noise-critical circuitry. If you don’t have any low-noise regulators on hand (which I didn’t), adding a larger filter capacitor across the regulator’s output will help to eliminate regulator noise:

A simple filter capacitor

A simple filter capacitor

The capacitance required to limit the maximum voltage change over a given period of time is a function of the current drawn by the load:

\(
\begin{aligned}
C = \frac{I \times t}{dV} \\
\end{aligned}
\)

Where I is the current through the load, t is time and dV is the maximum acceptable change in voltage.

As the maximum acceptable voltage change is lowered, the required capacitance increases. For low noise circuits, the maximum voltage change that can be tolerated is quite low, and thus impractically large capacitors are often required.

However, as the current through the load is decreased, the required capacitance also decreases. Thus, if you can increase the load impedance across the filter capacitor, then smaller, more practical filter capacitors can be used; this exactly what the “capacitance multiplier” circuit does:

The capacitance multiplier

The capacitance multiplier

This is quite a clever circuit, which is essentially an emitter follower. An emitter follower acts as a voltage source, with the voltage at the emitter mirroring the base voltage, less the .7V diode drop of the base-emitter PN junction. Further, the base current will be less than the emitter current by a factor of Beta, which is typically a nominal value of 100.

If the current seen by the capacitor is 100 times smaller, then the load seen by the capacitor must be 100 times greater, allowing smaller capacitor values to limit the transistor’s base voltage noise to acceptable levels. And, since the emitter voltage mirrors that of the base, the noise of the emitter voltage will be equally reduced.

Another way to look at this is that C1’s capacitance is effectively increased by a factor of Beta; in other words, a 100uF capacitor will act like a 100,000uF capacitor!

There is a voltage drop that results from this circuit (Vout at the emitter will be Vb – .7V), but the noise reduction is excellent:

Pre-amplifier noise when powered through an LM7810 with a capacitance multiplier

Pre-amplifier noise when powered through an LM7810 with a capacitance multiplier

One thought on “Taming Noisy Voltage Regulators

  1. I came to the same conclusion. Capacitance multiplier eliminates the hum the 78xx regulators can’t.

Leave a Reply

Your email address will not be published. Required fields are marked *