Power Supply Design  623

Equation 22.1 above must be changed from 7 ms to 17 ms. The input capacitors C1, C2 should be the same size as the reservoirs.

Mutual Shutdown Circuitry

The 5532 opamp is in general a tough item, but it has an awkward quirk. If one supply rail is lost and collapses to 0 V, while the other rail remains at the normal voltage, a 5532 can under some circumstances get into an anomalous mode of operation that draws large supply currents, and it ultimately destroys itself by over-heating. Other opamps may suffer the same problem, but I am not currently aware of any. To prevent damage from this cause opamp supplies should be fitted with a mutual shutdown system. This ensures that if one supply rail collapses because of overcurrent, over-temperature, or any other cause, the other rail will be promptly switched off. A simple way to

implement this is shown in Figure 22.7, which also demonstrates how to make a ± supply using only positive regulators. Two separate transformer secondary windings are required; a single centre-tapped secondary cannot be used.

The use of a second positive regulator to produce the negative output rail looks a little strange at first sight, but I can promise you it works. The regulators in this case are the high-current Linear Tech LT1083, which can handle 7.5 amps but is not available in a negative version. That should be more than enough current for even the most complex active crossover . . .

Figure 22.7: A high-current ±17 V power supply with mutual rail shutdown, and using only positive regulators.

624  Power Supply Design

Figure 22.8: A ±24 V power supply using only positive regulators.

The extra circuitry to implement mutual shutdown is very simple; R5, D3, R6 and Q1 and Q2. Because R5 is equal to R6, D3 normally sits at around 0 V in normal operation. If the +17 V rail collapses, Q2 is turned on by R6, and the REF pin of U2 is pulled down to the bottom rail, reducing the output to the reference voltage (1.25 V). This is not completely off, but it is low enough to prevent any damage to opamps.

If the −17 V rail collapses, Q1 is turned on by R5, pulling down the REF pin of U1 in the same way.

Q1 and Q2 do not operate exactly symmetrically, but it is close enough for our purposes.

Note that this circuit can only be used with variable output voltage regulators, because it relies on their low reference voltages to achieve what is effectively switch-off.

Power Supplies for Discrete Circuitry

One of the main reasons for using discrete audio electronics is the possibility of handling larger signals than can be coped with by opamps running off ±17 V rails. The use of ±24 V rails allows a 3 dB increase in headroom, which is probably about the minimum that justifies the extra complications of discrete circuitry.A±24 V supply can be easily implemented with 7824/7924 IC regulators.

Aslightly different approach was used in my first published preamplifier design. [1] This preamp in fact used two LM7824 +24V positive regulators, as shown in Figure22.8, because at the time the LM7924 −24 V regulator had not yet reached the market. This configuration can be very useful in the sort of situation where you have a hundred thousand positive regulators in the stores but no negative regulators.

Note that this configuration does however require two separate transformer secondary windings; it cannot be used with a single centre-tapped secondary.

Reference

[1] Self, D. R. G. “AnAdvanced Preamplifier Design” Wireless World, November 1976

CHAPTER 23

An Active Crossover Design

Here I present an active crossover design which will illustrate a good number of the principles and techniques put forward in this book. A design that demonstrated all of them would be a cumbersome beast, so I have aimed instead to show the most important while also providing a practical and adaptable design which will be of use in the real world.

The design is a generic crossover that can be adapted to a wide range of applications by changing its parameters and its configuration. The crossover frequencies can be changed simply by altering

component values, and circuit blocks for equalisation or time delay can be added or removed, with due care to preserve absolute phase, of course. Since widely varying drive units may be used, I have made no attempt to specify equalisation or integrate the driver response into the filter operation.

The crossover design is primarily aimed at hi-fi applications rather than sound reinforcement. It does not, for example, in its basic form have variable crossover frequencies or balanced outputs, though instructions are given for adding the latter.

Design Principles

The aim of this chapter is not just to provide a complete design but to demonstrate the use of various design principles expounded in the body of this book.

Example Crossover Specification

This is the basic specification for an active crossover that we will use as an example.

625

626  An Active Crossover Design

(As noted in Chapter 13, the path lengths to be compensated for were carefully chosen to give nice round figures for the delay times required.)

The crossover has balanced inputs, but in its basic form the outputs are unbalanced, the assumption being that if a balanced interconnection is required, this will be provided at the power amplifier inputs. The possibility of adding balanced output stages is examined at the end of the chapter. If the overall audio system can be satisfactorily designed with balanced crossover inputs but unbalanced crossover outputs, as opposed to unbalanced crossover inputs but balanced crossover outputs, then the former is clearly more economical, as for a stereo crossover it requires two balanced input stages rather than six balanced output stages.

Low-impedance design is used throughout; the circuit impedances are designed to be as low as possible without causing extra distortion by overloading opamp outputs. This reduces the effect of opamp current noise and resistor Johnson noise but has no effect on opamp voltage noise.

The design assumes that there is no level control between the crossover and power amplifier, or if there is, it is set to maximum and left there; this allows us to use elevated internal levels in the crossover to improve the noise performance, as described in Chapter 17.

No equalisation stages are included in the signal paths.

The block diagram is shown in Figure 23.1.

The Gain Structure

In Chapter 17 I explained how with suitable system design the internal levels of an active crossover could be significantly raised to reduce the effect of circuit noise. Here I have decided to go for an internal level of 3 Vrms, 12 dB higher than the assumed input voltage of 775 mVrms (0 dBu). This

Figure 23.1: Block diagram of 3-way Linkwitz-Riley 4th-order crossover.