Using Anedio D1 DAC with Logitech Squeezebox Touch
The home computer is increasingly becoming the hub of audio playback and storage. With the availability of large-capacity hard drives at affordable prices, the entire CD collection can be ripped and stored on a hard drive, and any song in the collection can be easily accessed through a media player with just a few clicks. The power and convenience of having all the music collection in one place is very appealing. In this rapidly-evolving era of computer audio, we at Anedio are continuing to experiment with various cost-effective solutions for high-performance digital playback.
For the computer-based audiophile system to be viable, we believe the following factors are important:
- Performance: We would like superior sonic performance that can make the Red Book CD sound better and reveal the sonic potential of the high-resolution 24-bit, 88.2K/96KHz format.
- Accessibility: We would like the digital archive of music to be easily accessible anywhere in the house.
- Cost: We are passionate about keeping the cost of the whole system reasonable so that more people can enjoy the highest quality music.
One obvious approach is to connect a computer to an external DAC through a USB or SPDIF cable, and use a media player program to send the audio stream. This direct approach works reliably; however, without a remote control, the computer may not be the most convenient device for playing music in a living room. Some audiophiles may object to the fan noise emitting from the computer and may want to place it in another room. Others may not want to have their computer tethered to their audio system through a USB or SPDIF cable.
Logitech Squeezebox Touch: Wireless Music Server
Another cost-effective approach is to use a wireless music server, such as the Logitech Squeezebox Touch, together with the Anedio D1 DAC. Priced at only $299, the Squeezebox Touch opens up a window of possibility for those searching for cost-effective, high-performance playback. It is quite flexible in the way it connects to various sources of digital music. It can stream music from a wireless network, Ethernet, or directly from a USB drive or SD card attached to it. Perhaps, the most interesting possibility for the greatest accessibility is the wireless.
Most importantly for audiophiles, it offers three indispensable capabilities:
- Bit Transparency: The Squeezebox Touch passes the audio data through transparently without modifying it in any way.
- High Resolution and Sample Rates: It supports 24-bit resolution and 88.2K/96KHz sample rates.
- Built-in FLAC Decoder: Its supported formats include FLAC, the most widely-accepted lossless compression format for high-performance audio.
These three features enable the Squeezebox Touch to be a high-fidelity digital transport, passing the audio data transparently to an external high-performance DAC.
In the wireless configuration, the Squeezebox Touch receives the digital audio stream from the wireless network already existing at home. The digital music files may be stored on a computer located in another room or on a network drive. The Squeezebox Touch serves as a wireless pass-through device that receives the digital audio stream and passes it on to an external DAC, as well as a music player that can be controlled via its touch screen or remote control. Even though its built-in DAC is good in their own right, especially considering the low price of such feature-rich equipment, its performance can be improved substantially by feeding the digital output to the D1 DAC through the SPDIF interface.
Measurements
Our tests confirmed that the wireless streaming through the Squeezebox Touch indeed preserved bit transparency for both 16 and 24 bits and for 44.1K, 88.2K, and 96KHz sample rates. All test signals were in the FLAC and WAV formats residing on a notebook computer. The process of wireless streaming through Squeezebox Touch is quite robust, as dropouts occur very rarely. We believe that the wireless connection is sufficiently reliable for high-performance home audio. In our test setup, the wireless router was about 30 feet away, with the signal strength in the Squeezebox diagnostics showing between 60% and 80%.
Shown in Fig. 4 is a jitter measurement of the Squeezebox Touch analog output with an 11.025KHz, 24-bit signal sampled at 44.1KHz. To expose as much frequency-coherent jitter as possible, we used very long FFTs, 1024K points, averaged 4 times. The multitude of small peaks that spread symmetrically around the signal are most likely caused by jitter. Most of the jitter-related peaks are within 1.1KHz from the signal frequency. Taking the RMS sum of the peaks between 50Hz and 1.1KHz from the signal, the jitter is estimated to be roughly 50 ps rms or 300 ps peak-to-peak.
It is surprising that this jitter pattern is consistent whether the data is streamed wireless or taken directly from an SD card or a USB drive. One might expect that jitter would be worse through wireless connection, but the fact that it is consistent regardless of the type of interface suggests that there is intelligent buffering taking place internally for the wireless stream.
When the same signal is fed into the D1 DAC, the whole FFT spectrum is cleaned up drastically, demonstrating the effectiveness of the multi-stage jitter reduction scheme for all types of digital input stream. The multitude of small peaks have completely disappeared; the only remaining jitter sidebands are below -150dB, +/- 1KHz from the signal. Also, the overall noise level is at least 20 dB lower, revealing the full capacity of high-resolution 24-bit music.
In sum, the combination of a wireless music server, such as the Logitech Squeezebox Touch, and the Anedio D1-DAC makes it possible to enjoy highest quality music anywhere in the house at a reasonable price. The bit-transparent wireless connection opens up the possibility of accessing the computer archive of music any place within the reach of the wireless network. The Anedio D1 DAC provides jitter-free conversion to the analog domain that will satisfy even the most discerning ears.
Multi-Stage Jitter Reduction
Jitter is an undesirable artifact that is unique to digital audio. In the digital domain, music is represented by a discrete series of samples spaced at a regular time interval. To convert these discrete samples into the analog domain, they must be recreated at regular, precisely-spaced time intervals. This is where the timing error, called "jitter," seeps in. The distortion caused by jitter is non-harmonically related to the signal and sounds unnatural, fatiguing, and harsh. Our goal is to diminish this timing error to a level where it can no longer be perceived.
Various solutions exist for reducing jitter. Some of them work quite well in taming the amount of jitter, but in the process, they may generate other digital artifacts that are undesirable and ultimately degrade sonic quality.
In designing a jitter-reduction circuit, we have two main goals in mind. The first goal is to preserve the original samples in their bit-perfect condition as far as it is possible so that the jitter-reduction process does not introduce digital artifacts. The second goal is to make it work consistently over a wide range of digital interfaces -- USB, SPDIF, and Toslink -- so that the listener may enjoy natural sound regardless of the source of music. In order to achieve such goals, we have devised a multi-stage jitter-reduction strategy that progressively lowers jitter until it becomes insignificant.
Jitter Reduction Stage 1
At the first stage of jitter reduction are high-performance digital transformers for rejecting high-frequency common-mode noise from digital sources. These transformers, made by Scientific Conversion, are optimized specifically for ultra low primary to secondary capacitance (0.5 pF), the most critical parameter for common-mode rejection. Thus, the jitter due to the common-mode noise is eliminated right at the input, even before it reaches the digital receiver circuit.
In addition, the digital transformers completely isolate the noisy grounds of a computer, CD, DVD, Blu-ray player, or music server, cutting off the ground loops and keeping the DAC ground clean.
Jitter Reduction Stage 2
The remaining jitter from the incoming digital data is reduced further through the 2nd stage, implemented using the Wolfson WM8805 digital interface receiver. Often at this stage, a sample-rate converter is used to reduce jitter. However, a conventional sample-rate converter modifies every sample based on a constantly-changing estimate of the input sample rate, inevitably introducing audible artifacts. By contrast, the WM8805 relies on an elastic buffer that absorbs timing errors without modifying the original samples in any way.
In a nutshell, the incoming samples, with their jittery clock, are temporarily stored in the elastic buffer. As they fill up the elastic buffer, they are sequentially read out, controlled by a precision clock from a crystal oscillator. The rate of readout is determined by a PLL (Phase-Locked Loop), which tracks the difference between the incoming clock frequency and the reference clock frequency and makes an adjustment to ensure that the elastic buffer does not overflow.
The chief advantage of this approach is that the audio samples remain bit-perfect, completely unmodified. The only change is the timing information, now referenced to a clean crystal oscillator. The WM8805 can tolerate a very wide range of incoming jitter, and almost all it is absorbed by the elastic buffer, leaving only a small amount of residue (50 ps rms) to the last stage.
Jitter Reduction Stage 3
The final refinement is performed by the ES9018 DAC, with its unique sample-rate conversion technology (patented by ESS Technology, US patent 7330138). It transports the original samples to a completely new clock domain, run by an ultra low phase noise oscillator. The new clock domain is at a much higher frequency (80 MHz in the case of the D1 DAC). Initially, the audio data is oversampled to the new domain simply by duplicating the samples, as indicated by the repeated dots in the diagram. Because the two sample frequencies are unrelated, the sample times of the two clocks will not align exactly. Note that almost all the samples of the output are simple replicas of the input. Only at the point where the input sample changes does the timing mismatch become an issue.
As shown in the figure zoomed into a transition point, the output sample clock falls slightly after the input sample clock, which causes a portion of the signal to be missing (the shaded area). It is this missing area that must be corrected. The key to ESS Technology's innovation lies in transforming this sampling-time mismatch into a single intermediate sample that corrects for the timing error. At the transition point, it generates a new intermediate sample in such a way to match the shaded area caused by the timing mismatch. Thus, the time error is transformed into an equivalent digital level.
To calculate the shaded area, the position of the input clock relative to the reference clock must be determined accurately. This critical task is performed by a DPLL (Digital Phase-Locked Loop), which locks onto the input clock and compares it to the precision low phrase noise reference. The whole process is extremely accurate, achieving errors less than -175 dB.
Benefits
In summary, Anedio's multi-stage jitter reduction circuit accomplishes the two main goals mentioned at the beginning. It keeps the bit-perfect condition until the last stage, where it performs the fine-level jitter correction with extreme accuracy. This assures that whatever the listener hears is as close to the original as it can get. In addition, the wide range of jitter correction can effectively deal with for all types of digital inputs -- USB, SPDIF, and Toslink, enabling the listener to enjoy consistently natural sound regardless of the music source.
Measuring Jitter
We have intentionally left our jitter specification blank. It's not because jitter is unimportant. On the contrary, we care deeply about jitter because it produces non-harmonic distortion, which is the reason they are perceived as fatiguing, metallic, and harsh. However, to measure it reliably down to the picosecond level (one trillionth of a second) remains exceedingly difficult, and even if it could be done, a single number is inadequate unless its frequency-domain behavior is also understood. So rather than attempting to present a singe definitive number, we limit ourselves to presenting certain characteristics of jitter that seem relevant to auditory perception.
Sometimes, the jitter of the master clock is presented as a performance metric of a DAC. But it merely represents a lower bound, not the actual sampling jitter of the DAC. What we're really after is the sampling jitter, measured at the analog output of the DAC, which is what ultimately matters to the sonic quality.
Jitter Sidebands
One direct way of measuring jitter is simply to monitor the DAC output with an ADC. If the jitter is sufficiently small, which is usually the case, then it would manifest itself in the frequency domain as double sidebands, centered around the signal. From the amplitudes of these sidebands and the frequency of the signal, we can then estimate the amount of timing jitter.
To illustrate how this jitter spectrum might look like in the frequency domain, we generated a 1 KHz, 50 ps rms (141 ps peak-to-peak) sinusoidal jitter in software, modulating the sample time of an 11.025 KHz signal (sampled at 44.1KHz, 24 bits). This simulated signal was then fed into the Audio Precision APx525 analyzer for FFT. Since the signal is from a numerically-generated simulation, we do not see the noise floor, which is well below the lowest limit of the plot. But we do see the telltale sign of jitter -- the double sideband centered around the main signal, which in this example is at +/- 1 KHz and -112 dB from the signal. As evident in this plot, the sidebands are not harmonically related to the signal, which is the reason they are perceived unnatural and objectionable.
Some Assumptions
In actual measurements, as we search for jitter on the order of picoseconds, we run into the limitation of measuring instruments. It is not easy to tell what portion of jitter is from the DAC under test or the ADC of the measuring instrument. Fortunately, the jitter of an ADC is usually lower than that of a DAC. The ADC does not need to go through the process of recovering the clock from the incoming data. Its timing-critical conversion stage can be driven almost directly from a pristine clock source without suffering much degradation in quality. As we will see, it turns out that the ADC of the APx525 performs exceptionally well in this regard, and we can assume that its jitter is usually less than that of the DAC under evaluation.
We also assume that, for the most part, the DAC jitter is dominated by frequency-coherent jitter than random jitter. As we have seen already, frequency-coherent jitter will produce discrete tones centered around the signal in the frequency spectrum. By contrast, random jitter will produce a flat frequency spectrum, which is indistinguishable from white noise in the FFT analysis. It is generally recognized that the human ear is more perceptive to the presence of discrete tones than white noise. So we consider only that which exhibits frequency-coherent behaviors.
Another assumption we make is that low-frequency jitter (below about 100 Hz) is masked by the signal and is below the threshold of hearing. In the frequency domain, the low-frequency jitter sidebands congregate near the main tone, and if sufficiently low in frequency, it appears as skirts around the main tone. Research by Julian Dunn indicates that jitter below 100 Hz is at least 40dB less audible than jitter above 500 Hz. This is consistent with well-known research on auditory masking, which has demonstrated that the presence of a large tone masks smaller tones at nearby frequencies.
With the above assumptions, we used the ADC of the APx525 analyzer to sample the analog output of the D1 DAC, being mindful of the fact that what we are measuring is not merely the DAC but the DAC plus the ADC. To increase the sensitive of measurement, we use a high-frequency signal, 11.025 KHz, a quarter of the sample rate, 44.1KHz. And to expose as much frequency-coherent jitter as possible, we used very long FFTs, 1024K points, averaged 4 times.
Moreover, to test if the jitter is sensitive to data patterns, we varied the frequency and magnitude of the signal over hundreds of points, in effect, varying the pattern of 1's and 0's in the data. The results show that changing data patterns do not alter the jitter spectrum of the D1 DAC in any noticeable way.
USB Interface Jitter Measurement (16 bits)
We have a keen interest in measuring the jitter of the USB interface because its level of jitter is typically much worse than the SPDIF interface. If left untreated, the degradation in sonic quality is readily audible. So we wanted to see if our multi-stage jitter reduction circuit would be effective on the USB interface. The FFT plot of the D1 DAC output (16-bit, 11.025 KHz signal) shows that there are no jitter sidebands visible above the noise floor, around -135 dB, which means that if they exist, they are buried under the 16-bit quantization noise. If there were a single jitter-induced tone at -135dB, the corresponding jitter would be 4 ps rms.
Here is the same measurement, zoomed in along the frequency axis. Notice the slight spreading around the signal, about +/-20Hz. This spreading is a characteristic sign of low-frequency jitter. As mentioned above, such a low frequency jitter is masked by the presence of the strong nearby tone and is well below the threshold of audibility.
SPDIF Interface Jitter Measurement (24 bits)
We've already mentioned that these measurements are ambiguous because they not only represent the D1 DAC's performance but also that of the ADC of the APx525. Before proceeding further, it would be illuminating to digress for a moment to see the loopback performance of the Audio Precision APx525 analyzer. In a loopback setup, the analyzer essentially measures itself, with its DAC output connected to its ADC input. With 24-bit data, the noise floor is low enough to reveal numerous discrete tones. Although these discrete tones are quite low in amplitude, all below -132 dB, their behavior is complex. We see several discrete tones that appear to be jitter sidebands (positioned symmetrically around the signal), but also many others that are not jitter-related. Based on this plot alone, it would be impossible to separate the contributions of the DAC from those of the ADC.
Finally, we look at the jitter spectrum of the Anedio D1 DAC through the SPDIF interface. Shown here is the spectrum of the D1 DAC output, measured using the APx525 ADC. Notice that practically all the discrete tones in the APx525 loopback measurement is now gone, suggesting that it was the APx525 internal DAC that produced most of the discrete tones. The FFT spectrum of the D1 DAC plus the APx525 ADC looks remarkably clean. The only discrete tones that are symmetrically positioned around the signal are at +/-1 KHz, -150 dB, which translates to about 0.7 ps rms jitter.
At this point, measuring jitter down to the picosecond level can only be tentative because we can see only part of the picture. The FFT-based jitter measurements cannot distinguish between white noise and random jitter-induced noise. Yet these measurements, limited as they are, show that the multi-stage jitter reduction implemented in the Anedio D1 DAC performs their job effectively and consistently for both USB and SPDIF interfaces.
Designing a Low-Noise System
Often an audio component may measure well on a test bench, but when connected to real systems found in typical homes, it may generate much more noise than that reflected in the bench-top measurement. At Anedio, we are acutely aware of this discrepancy, and we strive to design our equipment to perform well in real audio systems.
Ground Loop
In a typical home audio system, there are usually three or more components, e.g., a CD player, Blu-ray player,TV, computer, music server, preamp, and power amplifiers. When these component are connected together, the total noise of the system is not merely a sum of individual noise sources. It may increase far more drastically because of synergistic interactions among components' grounds.
The main culprit is the ground loop, formed when two or more components are connected together. The presence of ground loops is evident in a noise spectrum that is in multiples of power-line frequency. Shown here is a poorly-designed system comprised of a computer sound card, a preamp, and a power amplifier. In the FFT measurement, you can see multiples of 60Hz, e.g., 120, 180, 240, 300, etc. What is alarming is that the ground-loop noise reach well into the critical midrange and even upper midrange, where the human ear is highly sensitive. Since these frequencies are not harmonically related to the input signal (1 KHz), they are more detrimental to our perception of music than the harmonic distortion at 2KHz, 3KHz, etc.
What then causes ground loops to be formed? After all, the grounds are shorted together with wires, so how can they be the cause of noise?
Perhaps, the most erroneous assumption in audio circuit design is to consider all the grounds to be equal or simply at 0 volt. But the grounds are never at exactly the same voltage. Ground is merely a local reference point, connected to other reference points called "ground" through a low-impedance path. The voltage differences may be ever so slight, but are sufficient to cause a drastic rise in the noise. Ground loops are a reality to be faced whenever two or more components are connected together.
The reasons for the differences in ground voltages are too numerous to cover adequately here, but among the many are stray magnetic and electric fields, leakage currents from the AC mains and other sources, and ground conductor resistance. These slight differences in ground voltage may not cause any harm as long as the components are left unconnected, but as soon as they are connected together through interconnects, loops are created, and unwanted currents circulate through the loops.
The simplified diagram shows the ground layout of a three-component system (e.g., a CD player, preamp, and power amp). For the sake of illustration, only the grounds are shown without the signal paths. Inside each component are chassis ground and circuit ground -- each at a slightly different voltage. In reality, there can be multiple circuit grounds within a single component: digital ground, USB ground, analog ground, etc. But for simplicity, we've lumped all these into the "circuit ground." But even with such simplification, there are already six grounds and six different ground voltages.
Once the three components are connected together through interconnects, ground loops are formed via the shield (ground) conductor of the interconnects, and unwanted currents circulate, as indicated by the pink lines. If we were to add another component to the preamp, we would add three additional loops, and the whole thing becomes rather complicated quickly.
Minimizing the Effect of Ground Loop
How then do we minimize the adverse effects of ground loops? In our design process, we believe it's important to keep the whole audio system in view and try to optimize the whole, not just individual components in isolation. We pay attention to where ground currents flow and optimize the circuit topology and PCB layout, we try to break the ground loop wherever possible, we galvanically isolate all digital inputs, and we carefully shield magnetic fields. There is not much innovation here, but simply old principles implemented with meticulous care.
The result of such optimization is shown here in the FFT spectrum of a complete system -- a computer playing a 1KHz tone through its USB port, driving the USB input of the D1 DAC, which in turn driving the A1 Power Amplifier. The noise due to the power line frequency and its multiples is below -118 dB, on the order of one part per million. Note that this is the actual noise levels you will get in a complete system when playing music through the Anedio D1 DAC and A1 Amp. It does not matter which digital source you use since all the digital inputs are galvanically isolated.
Steps You Can Take
To ensure the lowest possible ground loop noise in your audio system, we recommend the following:
Use One Power Strip for All Components: We highly recommended that the power cords of all components be connected to a single point — usually a single surge protector with multiple outlets. This keeps all components referenced to a single point, as far as it is practically possible, thus minimizing the differences in ground potentials among the components.
Isolate Cable TV Ground: If you have a cable TV connected to your audio system, ensure that the cable ground is galvanically isolated from that of the audio ground. The cable TV ground, tied to the earth outside your home, is at a significantly different potential and can cause audible hum if it is not isolated properly. If you suspect this is the case, insert a cable TV ground isolator inside your home, just before your cable set-top box or TV RF input.
Bypass the Preamp If Possible: If you listen to music only from digital sources, we highly recommend bypassing the preamp completely. The built-in volume control in the Anedio D1 DAC allows power amplifiers to be connected directly to the D1 DAC. Moreover, all digital inputs of the D1 DAC are galvanically isolated using high-performance digital transformers, cutting off potential ground loops right at the source. In most systems, you will find that bypassing the preamp is one of the most significant steps toward a higher level of transparency. With the ground loop noise eliminated, you can hear more easily reverberations of the hall against blacker background, and you can experience another layer of sonic space opening up.
Why IC Power Amplifiers?
There was a time when the use of ICs (Integrated Circuits) in high-end audio was almost universally regarded as undesirable. The first ICs designed in the 1960's and 70's had poor slew rates and a host of other problems, making them inferior to discrete designs. However, times have changed. Over 40 years of collective design experiences have transformed the scene. We now understand more about various mechanisms of distortion and have the advantage of more advanced fabrication technologies and faster transistors. We at Anedio believe that IC power amplifiers built using National Semiconductor's LM3886/LM4780 offer truly exceptional performance, on a par with or even superior to discrete designs.
Its sonic potential was first noted by 47 Laboratories, which introduced an IC-based amplifier called GainCard. It was a paradigmatic shift in the philosophy of high-end amplifier design, and it started a small revolution in the DIY world, with countless hobbyists confirming the quality of its sonic performance.
The beauty of the IC amplifier is its simplicity. It is remarkably easy to build, and even novice hobbyists can build a good sounding amplifier out of it. At the same time, it requires just as much attention to details as any other amplifier to obtain the best possible performance. With careful design and tuning, its sonic performance can truly blossom.
"Of all our human resources, the most precious is the desire to improve" -- Anonymous
We started our own investigation by building several versions of IC power amplifiers based on National Semiconductor's Overture series. Through our investigations, we discovered that its potential had not yet been exhausted and that it could be pushed even more to a higher level by optimizing the circuitry around it and the PCB layout. The result of this optimization has demonstrated that the performance of an IC power amplifier can equal or exceed that of expensive discrete designs. The only limitation is its moderate power capability, which still is adequate for most home listening environment.
Key Advantages of IC Power Amplifiers
What are the key advantages of IC power amplifiers over conventional discrete designs? Three factors stand out:
Small Current Loops: With power amplifier ICs, a fully-functional amplifier can be built using merely a handful of capacitors and resistors. This austerity lends itself to a compact layout, with tiny current loop areas compared to discrete designs. Why is the loop area important? The smaller the current loop, the less noise picked up by the circuit. The inside of the amplifier enclosure is actually a noisy place, with lots of high-order harmonics from the power supply (for Class AB amplifiers) and multiples of 60Hz from the transformer floating around. Given such a noisy environment, the small loop area is a great antidote to unwanted noise. The compactness of the IC power amplifier results in a very clean noise floor, well below the realms of audibility.
Small Parasitic Capacitances and Inductances: The integrated-circuit technology reduces drastically the capacitances and inductances associated with the layout. Within the IC, wiring distances are orders of magnitude shorter compared to discrete designs. Also, outside the IC, the layout can be compact, as already mentioned. All of this, along with the sound circuit design of LM3886/4870, contributes to superior speed without resorting to expensive technologies. Even with a 2uF capacitor load, just about the worst capacitive load it will drive, the amplifier remains stable and well-behaved. Its excellent transient response is well suited to driving all types of loudspeakers.
Close and Fast Thermal Tracking: Another key advantage of IC amplifiers is the close thermal tracking of all the transistors. Since all the transistors are on the same chip, any excessive heating in the output transistors can be detected fast and reliably, and actions can be taken quickly to protect the amplifier and the loudspeaker (See National Semiconductors Application Note AN-898 on SPiKe protection circuit). Moreover, the bias devices for the output stage can track the output transistors closely in real time over a wide range of temperatures. The result is consistent performance whether the amplifier is cold or hot and whether it plays softly or loudly. The amplifiers settles into its optimum operating condition within a few minutes of turn-on, and its performance remains solid over a large range of temperatures and power levels.
These three factors, we believe, are the most significant ones that contribute to the excellent performance of LM3886/4870 amplifiers.
Are there then any limitations to IC power amplifiers? The power handling capacity of a single IC is moderate. But this is not an inherent limitation. The power output can be increased significantly through the use of bridge and parallel techniques (See National Semiconductor's Application Note AN-1192).
Understandably, some are concerned about thermal distortion, which arises from the heat generated in the output transistors flowing into the input transistors. The first generation IC op amps indeed suffered from this effect, but again, a significant progress has been made. By carefully controlling thermal gradients and by using the common-centroid layout for the input devices, the thermal distortion can be made negligible. If it is present, it would manifest itself as an increase in harmonic distortion at low frequencies. In Anedio's implementation, no increase is observed, and the total harmonic distortion at 20Hz remains below the threshold of measurement, about 0.0002%.
In conclusion, IC power amplifiers, such as the LM3886/4870, can be pushed to state-of-the-art performance by carefully optimizing the circuitry around it and the PCB layout. And such a level of performance can be achieved at a fraction of the cost of typical high-end amplifiers, making them accessible to all audiophiles. Most important of all, it offers exceptionally musical sound that engages the listener, powerful transients that communicate the emotional impact, and extraordinary definition that draws the listener closer to live performance.
Assumptions in Audio Design
High-end audio is both stimulating and perplexing because of the diverse assumptions about the goals of sound reproduction and the process by which one achieves those goals. We at Anedio start with a set of assumptions about how to optimize an audio system for the most faithful reproduction of music. Other designers start from different sets of assumptions. These fundamental assumptions shape the design process and how the whole system is conceived.
Our fundamental assumption is that an optimal solution requires a holistic approach -- an artful integration of analysis and listening in view of the whole system. It is both art and science, in which we try to keep the following triad in mind: (a) whole system, (b) measurement & analysis, (c) extended listening.
(a) Whole System
Perhaps, the most important factor is to consider the system as a whole, and not merely as an assembly of individual components. Often a designer would invest enormous amount of resources in tackling a single issue, only to run up against diminishing returns. Without a holistic view of the sound reproduction system, it is easy to forget the real bottle neck that limits the performance of the whole system. In the language of optimization, each designer tends to focus on achieving a local optimum for a single product, while neglecting a global optimum for the audiophile.
For the consumer, to gain a holistic perspective is even more difficult since the marketing of high-end audio is highly fragmentary, and each manufacturer claims that investing in their equipment will make the most difference. Unfortunately, the interpretation of such advertisements is left up to the consumer.
We need to learn to set our course by the stars, not by the lights of every passing ship. -- Omar N. Bradley
In order to achieve a global optimum, what is required is an integrative approach, considering all the elements in the audio reproduction chain -- the source, preamps, power amplifiers, cables, crossovers, loudspeakers, and the room. Integration is the most difficult, yet the most rewarding task. A well-integrated audio system can reproduce recorded music with exceptional fidelity at a surprisingly low cost.
With the holistic view of the system, the designer can invest more resources on those parts that have greater impact on the experience of listening to music. In our view, the two most significant factors in shaping the overall listening experience are the loudspeaker and its interaction with the room. Especially critical is the loudspeaker's radiation pattern over a wide frequency range and how it illumines the room. Of course, the source and the amplification chain are important, as we strives to perfect them, but they must be seen as part of the whole system, serving the greater purpose of reproducing recorded music faithfully.
(b) Measurement & Analysis
Another aspect that we keep in mind in the design process is that our listening experience must be informed by objective reality. Often, a perceived improvement might actually be a euphonic aberration due to a slight deviation in the frequency response or an addition of the second order harmonic. It may sound euphonic, and we may even prefer that sound, but it has little to do with accurate sound reproduction. Objective analyses supported by meaningful measurements would temper the tendency toward imaginary interpretations of the differences we perceive in listening.
We are not satisfied with the assertion that an audio component measures well but sounds terrible. The discrepancy usually lies in the limitation of the measurement, the interpretation of it, and the artificial condition under which it was administered. The standard measurements, such as THD+N (Total Harmonic Distortion + Noise), by themselves, are inadequate. A single number simply cannot capture the reality of psychoacoustic phenomena or the complexity of underlying distortion mechanisms.
Measurements that can meaningfully represent slices of reality need to be multi-dimensional, not merely a single point. For example, FFT (Fast Fourier Transform) spectrum, a commonly used frequency-domain analysis, is a 2-dimensional slice, cut along the amplitude and frequency plane. This 2-dimensional information is more meaningful than a single number since it reveals the shape of the noise floor and non-harmonic noise as well as the detailed behavior of harmonic distortion. It also opens up room for nuanced interpretation, as higher-order harmonics and non-harmonic noise are more objectionable to auditory perception than 2nd and 3rd order harmonics.
Measurements must also reflect the typical condition in which the audio component will operate. A component may perform well on a test bench, but when it is connected together with the rest of the system, it may falter because of its undue sensitivity to ground-loop noise. In order for measurements to be meaningful, the condition must not be artificial, but reflect a realistic setup found in typical homes. When properly administered in a realistic condition, measurements do represent slices of what we actually hear.
Every creator painfully experiences the chasm between his inner vision and its ultimate expression. -- Isaac Bashevis Singer
Having said this, we also believe that sound reproduction and psychoacoustics are incredibly complex processes with a multitude of variables that interact in ways that are not always understood. Any measurement or analysis is bound to be reductionistic, merely a slice of the reality. The process of reduction is useful in that it allows the designer to focus on a manageable aspect and to understand the underlying physical principles. As long as one understands its limitations, it can serve as an essential tool for guiding the design.
(c) Extended Listening
The third essential aspect that we keep in mind in the design process is extended listening. The human ear is an incredible instrument with an extraordinary dynamic range, capable of hearing minute details while absorbing thunderous passages. It also has the amazing ability to perceive space and to locate the sources of sound in 3-dimensional space. We emphasize extended listening because something that is not immediately apparent might become more recognizable as we listen to a wide variety of music over an extended period.
The ultimate reference point for the design of an audiophile system is the live, unamplified, and unprocessed sound. As Linkwitz insightfully notes, the art of listening is different from the art of wine tasting. Even though both disciplines have subjective elements, listening has an objective reference point, namely, unamplified and unprocessed natural sound. It is to this reference point that we must return over and over.
We need to develop the senses for live, natural sound, and let our memory of it serve as the objective basis for listening tests. This means regularly attending live concerts to train one's ears. It does not need to be a first-class orchestra or jazz band. Even in student recitals and in less than glorious ambiences, much can be learned about the characteristics of live, unamplified sound. It can even be in daily environment -- the sound of birds chirping, cars passing by, school choirs singing, and kids playing. Once we train our ears to live, unamplified sound, we are in a better position to evaluate the reproduced sound of audio systems.
The ultimate goal of designing audio equipment is the enjoyment of music. When all is said and done, the bottom line is the emotional impact of the music reproduced through the system. We cannot measure the beauty of music with test equipment. At its best, high-end audio is about those moments when the equipment fades in the background and all that remains is the echoes of awe and wonder.
Bi-wiring and Bi-Amplification
One of the overlooked issues in designing audio systems is the crossover hidden inside the loudspeaker box. The crossover has the job of splitting the signal into the low and high frequency components (e.g. in a 2-way system). It is deceptively simple, consisting of a handful of passive components. Yet, it has the exceedingly difficult job of carrying high current and high voltage signals -- about 1,000 times higher current and 20 times higher voltage compared to the line-level stage.
The signal path through the crossover is fraught with problems that can generate more distortion than the rest of the amplification chain, preventing the audio system from performing at its full potential. There already exist several approaches that can improve the fidelity of this interface. In this article, we describe the progression from the traditional approach to active bi-amplification, with progressively improving fidelity.
Traditional Approach with Passive Crossover
The traditional approach with passive crossovers suffers from serious limitations, some of which are:
- The crossover presents a complex impedance between the amplifier and the speaker drivers, making it difficult for the amplifier to control the drivers directly.
- Especially outside its passband, the crossover impedance rises significantly, undermining the amplifier's ability to control out-of-band resonances in the drivers.
- The inductors used in the crossover usually have ferromagnetic cores, which can generate high levels of distortion, typically 0.1-0.2% even at low levels. Often, this is the largest source of nonlinearity in the entire electrical chain leading up to the loudspeaker drivers.
In spite of its inherent limitations, the passive crossover is predominantly being used, mostly because it requires only one amplifier to drive a loudspeaker. However, with the cost of high-quality amplifiers falling, it makes more sense to consider bi-amplification, using two or more amplifiers per channel.
Bi-wiring
The first step toward improving this interface is bi-wiring, referring to the use of two cables to drive a loudspeaker, one for the low frequency section and another for the high frequency section. Most high-end loudspeakers offer bi-wiring terminals (two pairs of terminals per loudspeaker). The low and high frequency crossovers are separated, and the signal current flows through two separate cables. Even though both paths see the same voltage, the currents are now split into low and high frequency paths. One cable carries only the low frequency currents, and the other only the high frequency currents. With this separation, the interaction between the low and high frequency currents are reduced, resulting in lower intermodulation distortion.
In our experience, the audible improvements from bi-wiring are subtle. Some experts, however, believe that the effects are not so subtle. According to Vandersteen, "The improvements are large enough that a bi-wire set of moderately priced cable will usually sound better than a single run of more expensive cable." Jon Risch, through the use of multitone testing, also concluded that bi-wiring significantly reduced intermodulation distortion between high-frequency currents and low-frequency currents ("A New Class of In-Band Multitone Test Signals," AES Convention, August 1998).
Passive Bi-amplification
We can go a step further by using two amplifiers to drive the low and high frequency paths separately. This configuration is usually called "passive" bi-amplification because it still retains the passive crossover. Notice that the two paths are now completely isolated from each other, and any potential interaction between the two paths is eliminated. The only residual interaction would be through the electromagnetic coupling between the two cables. Moreover, the impedance of each path rises outside its passband, significantly reducing the output current requirement of each power amplifier.
Passive bi-amplification makes subtle but discernable improvements beyond that of bi-wiring. The sound stage appears more sharply focused and detailed and the dynamic contrast increases, enabling music to be more engaging. Even in loud, complex passages, music seems to breathe more freely.
Active Bi-amplification
The greatest gain in the quality of sound comes from eliminating the passive crossover completely and replacing it with an active crossover before the amplifiers. This raises the definition, clarity, and dynamic contrast to another level.
The advantages of active bi-amplification are too numerous to list here (see Loudspeakers by Philip Newell & Keith Holland chapter 5 for more comprehensive discussions). In addition to solving all the problems mentioned above, it reduces the power requirement of amplifiers drastically. Since the signal is divided into low and high frequency components by the active crossover, each amplifier needs to handle only a limited range of frequencies. Moreover, active bi-amplification offers a much more direct control of loudspeaker drivers across the entire audio spectrum and provides adequate damping over all audio frequencies. It also eliminates the nonlinearities associated with passive crossovers, reducing intermodulation distortion and improving clarity and definition.
Considering all the advantages of active bi-amplification, we can no longer be enthusiastic about a single high-power, high-current amplifier driving a loudspeaker. No matter how much current it is capable of pouring into the loudspeaker, it still can do nothing about the bottleneck.
With the advances in digital crossovers, we believe that active bi-amplification will become increasingly more accepted in high-end audio. The digital crossover offers exceptional flexibility in customizing not only the amplitude response but also the phase response and delay. As with digital technology, the performance-to-price ratio of digital crossovers will continue to improve, and it is only a matter of time before they will become more accessible.
Unfortunately, at the moment, most loudspeakers are not designed with the active crossover in mind, and it involves removing the factory-installed passive crossover. Many audiophiles may not want to go this extent.
For most people, the best compromise between performance and practicality might be passive bi-amplification, using a pair of amplifiers to drive the existing loudspeaker. A pair of amplifiers with moderate current capacity, when used in passive bi-amplification, is sonically superior to a single amplifier with massive current capacity.
Benefit of Anedio’s Passive Bi-amp Implementation
Anedio's implementation of passive bi-amplification offers an important benefit: immunity to ground noise when two amplifier channels are tied to a single input.
A pitfall in bi-amplification is the potential ground loop introduced when the two inputs of the stereo channels are tied together. Depending on how the amplifier grounding system is designed, the additional ground noise can be quite high. That is why, in some systems, passive bi-amping actually degrades the sonic quality.
In the A1 Amp, the bi-amplification feature was conceived at the very beginning of the design stage, not as an afterthought and is optimized for the lowest ground noise. Even when the two amplifier channels are configured for bi-amplification, the noise level remains extraordinarily low. Moreover, the bi-amplification mode is activated with a built-in switch, and there is no need for an external Y cable.
Until high-performance digital crossovers become easily accessible, passive bi-amplification can serve an intermediate step that is sonically superior to a single amplifier with massive current capacity. The A1 Amp provides a robust way that enables the listener hear another layer of sonic information from existing loudspeakers.