4/19/2013

Simple power supply ripple rejection for battery systems MMBT3904






Many amplifiers, including integrated devices, are afflicted with undesired, implicit input terminals in their power connections. Due to less than perfect power-supply rejection, these implicit terminals admit noise into the signal that the amplifier will magnify. Thus, a clean supply is essential for amplifiers.

But a regulated supply, even one using a low-dropout regulator (LDO), can create problems in a battery-based amplifier design. If the selected regulation voltage is close to the fresh battery voltage, the LDO will quickly drop out as the battery discharges, eliminating the noise protection. If the regulation voltage is far below the fresh battery voltage, the design sacrifices the amplifier’s output voltage swing and wastes power until the battery voltage drops close to the LDO regulation voltage.

Fortunately, most audio amplifiers can readily operate with an unregulated supply because they do not need to provide dc precision. Unregulated supplies are especially useful in battery-powered designs that seek to use the battery fully because the circuits will continue to operate until the battery is quite discharged. The unregulated supply must be clean, however, and of sufficiently low impedance to avoid injecting noise and inducing inter-stage coupling that might cause instability.

A good way to clean an unregulated supply is to use the so-called “ripple-eater.” The ripple-eater’s output voltage tracks slightly below its input voltage and continues to operate as the input drops to a fairly low voltage. The ripple-eater also provides a low output impedance most of the way down.

This ability to track declining input voltage can be very useful when the power source is a non-rechargeable battery. It may not be good, however, for battery systems that can be harmed by deep discharge.

One well-known ripple-eater design that has been popular among audio enthusiasts is available in kit form. That design, though, is relatively complex and operates by bucking the ripple. The simple ripple-eater shown here (Fig. 1) is far simpler and offers about 40 dB of low-power cleaning at audio frequencies (Fig. 2), making it useful for such functions as pre-amplifiers.

The design simply comprises a filter buffered by a follower. A voltage divider (R1 and R2) provides headroom for subsequent regulation. A low-pass filter (R3 and C1) provides a clean voltage that an emitter-follower (Q1 and R4) buffers, adding one diode-drop of offset. A pass transistor (Q3) and feedback amplifier (Q2) provide gain and cancel the diode drop of Q1.

This circuit has sufficient gain to require compensation, which R6 and C2 provide. The voltages at the base of Q1 and at the output are essentially equal, so the transistors form a fed-back follower.

The ratio of R1 and R2 should be chosen to set the ripple-eater’s output far enough below the input so the output is below any troughs in the input voltage. Otherwise the circuit will “drop out” like any other regulator.

If more power is needed than the simple ripple-eater can provide, using a composite transistor in the feedback will provide more gain (Fig. 3). The higher-power version adds Q4 and R7 to increase gain beyond what Q2 alone provides, and the diodes (D1 to D3) increase headroom for the composite transistor’s operation. The result has greater ripple rejection in the audio band (Fig. 4) but loses effectiveness in the high-frequency band.

The high-frequency performance of the simple ripple-eater is another advantage this circuit has over LDO regulation. Many LDOs, especially those with MOSFET pass transistors, exhibit poor power-supply rejection (PSR) at high frequencies.

While the transistors shown are sufficient for audio operation, with high-frequency transistors such as the MMBT3904 and MMBT3908 the circuit of Figure 1 has proven effective where rejection up to a few megahertz was needed. The MBTH10 and MBTH81 can be used for even higher frequency rejection.



1N5819      ATMEGA64           MMBT3904          FR107


1 comment:

  1. Anonymous29/1/20 14:46

    Thanks for an article. Where did you get this schematic from?

    ReplyDelete