- Shopping Bag ( 0 items )
"Small Signal Audio Design is a unique guide to the design of high-quality circuitry for preamplifiers, mixing consoles, and a host of other signal-processing devices. Learn to use inexpensive and readily available parts to obtain state-of-the-art performance in all the vital parameters of noise, distortion, crosstalk and so on. Focusing mainly on preamplifiers and mixers, this practical handbook gives you an extensive repertoire of circuits that can be put together to make almost any type of audio system." This book is packed full of valuable information, with virtually every page revealing nuggets of specialized knowledge never before published. Essential points of theory bearing on practical performance are lucidly and thoroughly explained, with the mathematics kept to an essential minimum. Douglas' background in design for manufacture ensures he keeps a wary eye on the cost of things.
"Douglas Self, who has been designing both pro and consumer audio products for more than 20 years gave an excellent tutorial on designing balanced interconnections. This talk was part of the Product Design track, so the goal, it seemed, was to expose product designers to the advantages of providing balanced connections (whether the user uses them at not) and that it can be done for very little added cost. He went through several evolutions of the differential amplifier and how to greatly improve common mode rejection for very little parts cost."--MikeRiversAudio.com
An audio signal can be transmitted either as a voltage or as a current. The construction of the universe is such that almost always the voltage mode is more convenient; consider for a moment an output driving more than one input. Connecting a series of high-impedance inputs to a low-impedance output is simply a matter of connecting them in parallel, and if the ratio of the output and input impedances is high there will be negligible variations in level. To drive multiple inputs with a current output it is necessary to have a series of floating current-sensor circuits that can be connected in series. This can be done, as pretty much anything in electronics can be done, but it requires a lot of hardware, and probably introduces performance compromises. The voltage-mode connection is just a matter of wiring.
Obviously, if there's a current, there's a voltage, and vice versa. You can't have one without the other. The distinction is in the output impedance of the transmitting end (low for voltage mode, high for current mode) and in what the receiving end responds to. Typically, but not necessarily, a voltage input has a high impedance; if its input impedance was only 600 U, as used to be the case in very old audio distribution systems, it is still responding to voltage, with the current it draws doing so a side issue, so it is still a voltage amplifier. In the same way, a current input typically, but not necessarily, has a very low input impedance. Current outputs can also present problems when they are not connected to anything. With no terminating impedance, the voltage at the output will be very high, and probably clipping heavily; the distortion is likely to crosstalk into adjacent circuitry. An open-circuit voltage output has no analogous problem.
Current-mode connections are not common. One example is the Krell Current Audio Signal Transmission (CAST) technology, which uses current mode to interconnect units in the Krell product range. While it is not exactly audio, the 4-20 mA current loop format is widely used in instrumentation. The current-mode operation means that voltage drops over long cable runs are ignored, and the zero offset of the current (i.e. 4 mA = zero) makes cable failure easy to detect: if the current suddenly drops to zero, you have a broken cable.
The old DIN interconnection standard was a form of current-mode connection in that it had voltage output via a high output impedance, of 100 k[ohm] or more. The idea was presumably that you could scale the output to a convenient voltage by selecting a suitable input impedance. The drawback was that the high output impedance made the amount of power transferred very small, leading to a poor signal-to-noise ratio. The concept is now wholly obsolete.
At the most basic level, there are four kinds of amplifier, because there are two kinds of signal (voltage and current) and two types of port (input and output). The handy word 'port' glosses over whether the input or output is differential or single-ended. Amplifiers with differential input are very common — such as all op-amps and most power amps — but differential outputs are rare and normally confined to specialized telecoms chips.
The four kinds of amplifier are summarized in Table 1.1.
These are the vast majority of amplifiers. They take a voltage input at a high impedance and yield a voltage output at a low impedance. All conventional op-amps are voltage amplifiers in themselves, but they can be made to perform as any of the four kinds of amplifier by suitable feedback connections. Figure 1.1(a) shows a high-gain voltage amplifier with series voltage feedback. The closed-loop gain is (R1 + R2)/R2.
The name simply means that a voltage input (usually differential) is converted to a current output. It has a transfer ratio A = Iout/Vin, which has dimensions of I/V or conductance, so it is referred to as a transconductance or, less commonly, a transadmittance amplifier. It is possible to make a very simple, though not very linear, voltage-controlled amplifier with transconductance technology: differential-input operational transconductance amplifier (OTA) Integrated circuits (ICs) have an extra pin that gives voltage control of the transconductance, which when used with no negative feedback gives gain control (see Chapter 19 for details). Performance falls well short of that required for quality hi-fi or professional audio. Figure 1.1(b) shows an OTA used without feedback; note the current-source symbol at the output.
These accept a current in, and give a current out. Since, as we have already noted, currentmode operation is rare, there is not often a use for a true current amplifier in the audio business. They should not be confused with current feedback amplifiers (CFAs), which have a voltage output, the 'current' bit referring to the way the feedback is applied in current mode. The bipolar transistor is sometimes described as a current amplifier, but it is nothing of the kind. Current may flow in the base circuit but this is just an unwanted side-effect. It is the voltage on the base that actually controls the transistor.
A transimpedance amplifier accepts a current in (usually single-ended) and gives a voltage out. It is sometimes called an I-V converter. It has a transfer ratio A = Vout/Iin, which has dimensions of V/I or resistance. That is why it is referred to as a transimpedance or transresistance amplifier. Transimpedance amplifiers are usually made by applying shunt voltage feedback to a high-gain voltage amplifier. An important use is as virtual-earth summing amplifiers in mixing consoles (see Chapter 17). The voltage-amplifier stage (VAS) in most power amplifiers is a transimpedance amplifier. They are used for I-V conversion when interfacing to digital-to-analog converters (DACs) with current outputs (see Chapter 21). Transimpedance amplifiers are sometimes incorrectly described as 'current amplifiers'.
Figure 1.1(c) shows a high-gain voltage amplifier transformed into a transimpedance amplifier by adding the shunt voltage feedback resistor R1. The transimpedance gain is simply the value of R1, though it is normally expressed in V/mA rather than ohms.
Negative feedback is one of the most useful and omnipresent concepts in electronics. It can be used to control gain, to reduce distortion and improve frequency response, and to set input and output impedances, and one feedback connection can do all these things at the same time. Negative feedback comes in four basic modes, as in the four basic kinds of amplifier. It can be taken from the output in two different ways (voltage or current feedback) and applied to the amplifier input in two different ways (series or shunt). Hence there are four combinations.
However, unless you're making something exotic like an audio constant-current source, the feedback is always taken as a voltage from the output, leaving us with just two feedback types, series and shunt, both of which are extensively used in audio. When series feedback is applied to a high-gain voltage amplifier, as in Figure 1.1(a), the following statements are true:
Negative feedback reduces voltage gain. Negative feedback increases gain stability. Negative feedback increases bandwidth. Negative feedback increases amplifier input impedance. Negative feedback reduces amplifier output impedance. Negative feedback reduces distortion. Negative feedback does not directly alter the signal-to-noise ratio.
If shunt feedback is applied to a voltage amplifier to make a transimpedance amplifier, as in Figure 1.1(c), all the above statements are still true, except since we have applied shunt rather than series negative feedback, the input impedance is reduced.
The basic feedback relationship is Equation 1.1, which is dealt with at length in any number of textbooks, but it is of such fundamental importance that I feel obliged to include it here. The open-loop gain of the amplifier is A, and β is the feedback fraction, such that if in Figure 1.1(a) R1 is 2 kΩ and R2 is 1 kΩ, β is 1/3 . If A is very high, you don't even need to know it; the 1 on the bottom becomes negligible, and the As on the top and bottom cancel out, leaving us with a gain of almost exactly 3.
Vout/Vin = A/1 + Aβ (Equation 1.1)
Negative feedback can, however, do much more than stabilizing gain. Anything unwanted occurring in the amplifier, be it distortion or DC drift, or any of the other ills that electronics is prone to, is also reduced by the negative feedback factor (NFB factor for short). This is equal to:
NFB factor = 1/1 + Aβ (Equation 1.2)
What negative feedback cannot do is improve the noise performance. When we apply feedback the gain drops, and the noise drops by the same factor, leaving the signal-to-noise ratio the same. Negative feedback and the way it reduces distortion is explained in much more detail in one of my other books .
Nominal Signal Levels
The absolute level of noise in a circuit is not of great significance in itself — what counts is how much greater the signal is than the noise — in other words the signal-to-noise ratio. An important step in any design is the determination of the optimal signal level at each point in the circuit. Obviously a real audio signal, as opposed to a test sine wave, continuously varies in amplitude, and the signal level chosen is purely a nominal level. One must steer a course between two evils:
If the signal level is too low, it will be contaminated unduly by noise. If the signal level is too high, there is a risk it will clip and introduce severe distortion.
You will note that the first evil is a certainty, while the second is more of a statistical risk. The consequences of either must be considered when choosing a level, and the wider the gap between them the greater the dynamic range. If the best possible signal-to-noise is required in a studio recording, then the internal level must be high, and if there is an unexpected overload you can always do another take. In live situations it will often be preferable to sacrifice some noise performance to give less risk of clipping. The internal signal levels of mixing consoles are examined in detail in Chapter 12.
If you seek to increase the dynamic range, you can either increase the maximum signal level or lower the noise floor. The maximum signal levels in op-amp-based equipment are set by the voltage capabilities of the op-amps used, and this usually means a maximum signal level of about 10 Vrms or +22 dBu. Discrete transistor technology removes the absolute limit on supply voltage, and allows the voltage swing to be at least doubled before the supply rail voltages get inconveniently high. For example, ±40 V rails are quite practical for small-signal transistors and permit a theoretical voltage swing of 28 Vrms or +31 dBu. However, in view of the complications of designing your own discrete circuitry, and the greater space and power it requires, those nine extra decibels of headroom are dearly bought.
There are some very basic rules for putting together an effective gain structure in a piece of equipment. Like many rules, they are subject to modification or compromise when you get into a tight corner. Breaking them reduces the dynamic range of the circuitry, either by worsening the noise or restricting the headroom; whether this is significant depends on the overall structure of the system and what level of performance you are aiming at. Three simple rules are:
1. Don't amplify then attenuate.
2. Don't attenuate then amplify.
3. The signal should be raised to the nominal internal level as soon as possible to minimize contamination with circuit noise.
Amplification Then Attenuation
Put baldly it sounds too silly to contemplate, but it is easy to thoughtlessly add a bit of gain to make up for a loss somewhere else, and immediately a few decibels of precious and irretrievable headroom are gone for good. This assumes that each stage has the same power rails and hence the same clipping point, which is usually the case in op-amp circuitry.
Figure 1.2(a) shows a system with a gain control designed to keep 10 dB of gain in hand. In other words, the expectation is that the control will spend most of its working life set somewhere around its '0 dB' position where it introduces 10 dB of attenuation, as is typically the case for a fader on a mixer. To maintain the nominal signal level at 0 dBu we need 10 dB of gain, and a fl10 dB amplifier (Stage 2) has been inserted just before the gain control. This is not a good decision. This amplifier will clip 10 dB before any other stage in the system, and introduces what one might call a headroom bottleneck.
There are exceptions. The moving-coil phono head-amp described in Chapter 8 appears to flagrantly break this rule, as it always works at maximum gain even when this is not required. But when considered in conjunction with the following RIAA stage, which also has considerable gain, it makes perfect sense, for the stage gains are configured so that the second stage always clips first, and there is actually no loss of headroom.
Attenuation Then Amplification
In Figure 1.2(b) the amplifier is now after the gain control, and noise performance rather than headroom suffers. If the signal is attenuated, any active device will inescapably add noise in restoring the level. Any conventional gain-control block has to address this issue. If we once more require a gain variable from +10 dB to off, i.e. -∞ dB, as would be typical for a fader or volume control, then usually the potentiometer is placed before the gain stage as in Figure 1.2(b) because as a rule some loss in noise performance is more acceptable than a permanent 10 dB reduction in system headroom. If there are options for the amplifier stages in terms of a noise/cost trade-off (such as using the 5532 versus a TL072) and you can only afford one low-noise stage, then it should be Stage 2.
If all stages have the same noise performance this configuration is 10 dB noisier than the previous version when gain is set to 0 dB.
Raising the Input Signal to the Nominal Level
Getting the incoming signal up to the nominal internal level in one jump is always preferable as it gives the best noise performance. Sometimes it has to be done in two amplifier stages; typical examples are microphone preamps with wide gain ranges and phono preamps that insist on performing the RIAA equalization in several goes. (These are explored in their respective chapters.) In these cases the noise contribution of the second stage may no longer be negligible.
Consider a signal path which has an input of -10 dBu and a nominal level of 0 dBu. The first version has an input amplifier with 10 dB of gain followed by two unity-gain circuit blocks, A and B. All circuit blocks are assumed to introduce noise at -100 dBu. The noise output for the first version is -89.2 dBu. Now take a second version of the signal path that has an input amplifier with 5 dB of gain, followed by block A, another amplifier with 5 dB of gain, then block B. The noise output is now -87.5 dB, 1.7 dB worse, due to the extra amplification of the noise from block A. There is also more hardware, and the second version is clearly an inferior design.
Active Gain Controls
The previous section should not be taken to imply that noise performance must always be sacrificed when a gain control is included in the signal path. This is not so. If we move beyond the idea of a fixed-gain block, and recognize that the amount of gain present can be varied, then less gain when the maximum is not required will reduce the noise generated. For volume-control purposes it is essential that the gain can be reduced to near-zero, though it is not necessary for it to be as firmly 'off' as the faders or sends of a mixer.
Excerpted from Small Signal Audio Design by Douglas Self Copyright © 2010 by Douglas Self. Excerpted by permission of Focal Press. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
CHAPTER 1: SYSTEM ARCHITECTURES
CHAPTER 1part 2: SPECIFICATION REQUIREMENTS
CHAPTER 2: COMPONENTS
CHAPTER 3: BASIC TECHNIQUES OF NEGATIVE FEEDBACK & LOW NOISE
CHAPTER 3 part 2: DESIGN WITH DISCRETE TRANSISTORS
CHAPTER 4: DESIGN WITH OPAMPS
CHAPTER 5: SIGNAL SWITCHING
CHAPTER 6: LINE INPUTS & OUTPUTS
CHAPTER 7: MOVING-MAGNET DISC INPUTS
CHAPTER 8: MOVING-COIL DISC INPUTS
CHAPTER 9: MICROPHONE INPUTS
CHAPTER 10: TONE CONTROLS & FILTERS
CHAPTER 11: FILTERS
CHAPTER 12: VOLUME-CONTROL & BALANCE
CHAPTER 13: MIXER SUBSYSTEMS
CHAPTER 14: ELECTRONIC CROSSOVERS
CHAPTER 15: LEVEL CONTROL & SPECIAL PROCESSING CIRCUITS
CHAPTER 16: METERING, MUTING & RELAY CONTROL
CHAPTER 17: POWER SUPPLIES
CHAPTER 18: ANALYSIS OF THE MRP 11 PREAMPLIFIER
CHAPTER 19: ANALYSIS OF THE 200 DELTA MIXER