Print this page

Moore’s Law

Rate this item
(0 votes)
November 21st Written by 

SILICON SYMPHONY

Motorola announced in January 2001 they have developed a digital audio amplification process, dubbed Symphony, which B&K Components is using in their upcoming DA-2100 amplifier. The DA-2100 will do the Symphony processing in a Digital Signal Processor (DSP) program using Motorola's 56300 DSP family. Motorola plans to integrate Symphony processing into future DSPs, much like Dolby Digital, HDCD, AC3, and other algorithms are integrated now. A vendor need only add output power transistors and an output filter to get a high power and high efficiency digital amplifier.

WHAT'S THE BIG DEAL?

The first trend to note is the proliferation of high fidelity digital audio amplification techniques and hardware fast enough to implement them. Until recently, digital amplifiers did not have enough distortion free bandwidth for use over the entire 20Hz to 20KHz spectrum and were relegated to subwoofer use only. 

The second trend is the integration of these amplifiers with DSP chips to create the first end-to-end digital audio solution for consumer audio. The audio signal stays entirely in the digital domain up to the speaker inputs. These inexpensive integrated modules and chipsets will allow companies to easily implement designs that have better sound and are cheaper to produce than traditional mixed analog/digital approaches. Let's refer to this type of solution as Advanced Digital Audio (ADA). 

In the next few years ADA is going to be a disruptive force in all segments of the audio industry, from portable MP3 players to high-end home stereos. Moore's law, the doubling of the number of transistors on a given chip every 18 months, will quickly push down prices and improve the sound quality available from consumer audio gear.

DIGITAL AMPLIFIERS

Digital amplifiers are a type of switching amplifier. Switching amplifiers rapidly switch the output devices on and off at 100KHz or higher, and then usually low-pass filter to recover the audio portion. Older Pulse Width Modulation (PWM) switching amplifiers, called Class-D, controlled their switching with analog circuits. These designs suffered from poor fidelity and high Radio Frequency Interference (RFI). 

A digital amplifier generates its switching signal using digital logic. Purists would say that digital amplifiers must accept a digital input signal and do all processing in the digital domain. By controlling the switching signal with digital logic, advanced signal processing can be employed to compensate for the switching distortion. Most digital amplifiers avoid the Class-D moniker to distance themselves from analog switching approaches. 

Switching amplifiers have been pursued with interest since they offer higher power amplifiers at a lower cost than traditional class A or A/B amplifiers. Switching amplifiers' output devices are switched entirely on or off. This means that the output transistors do not have to dissipate power that is unused at low volume levels as they do in Class A and AB amplifiers. A Class AB amplifier may be 50% efficient at maximum output power while a switching amplifier can achieve 90% efficiency. The story is better than the specs indicate since at low power levels a digital amplifier could be as much as six times more efficient than Class AB. The increased efficiency allows for amplifiers with smaller power supplies and smaller heatsinks with equivalent output power to non-switched designs. Both of these components are costly and bulky, so shrinking them reduces the size and cost of the whole amplifier. 

Until recently though, they were not suitable for high fidelity applications. In the past few years companies like TacT, Spectron, Sharp, and Bel Canto have released digital amplifiers with sound quality on par with traditional analog amplifiers. In fact, some feel that their fidelity surpasses traditional audio amplifiers. While these digital amplifiers cost upwards of $2,000, they are only the vanguard of the coming revolution. 

TABLE --  DIGITAL STEREO AMPLIFIERS TECHNOLOGY

ADVANCED DIGITAL AUDIO: STEREO ON A CHIP

The second trend highlighted by Motorola's Symphony announcement is the upcoming ADA phenomenon. In 2001 major semiconductor companies will start to release digital chipsets that can handle every audio processing task including amplification. Currently, only Texas Instruments has their solution available in volume. These chipsets will be inexpensive, especially when compared to the many analog and digital parts they replace. For instance, the Pulsus chips go for $7 to $10. 

This phenomenon is being driven by the rapid convergence of cheap powerful DSPs, advanced digital PWM control theory, accurate psychoacoustic models, pervasive digital audio, MP3, and home theater. Market forces and recent technical advancements are making ADA audio technically possible, cost effective, and in demand. 

Not convinced it's happening? Lets see what some of Motorola's competitors have been shopping for lately: 

JUL 1999, Tripath licenses DPP to STMicroelectronics 

Tripath nonexclusively licensed their digital amplification process for use in commodity markets. In return, the company obtained favorable wafer prices and wafer supply availability. 

27 JUL 1999: Cirrus Logic Acquires AudioLogic; Gains Revolutionary PWM and Low-power Audio Technologies 

AudioLogic has several patents on low power DSP and feedback techniques for digital amplifiers. It appears that AudioLogic's feedback scheme is being incorporated into Cirrus's amplifier. AudioLogic's low power DSP could be useful for portable applications but may not have much bearing on amplifier performance. 

16 MAR 2000: Texas Instruments Acquires Danish Toccata Technology 

Toccata developed the EquiBit PWM ampifier process as used by TacT in their $10,000 Millenium amplifier. TacT has a layman's description of the process and its benefits here. 

02 OCT 2000: Cirrus Logic Expands PWM Technology Portfolio Through Purchase of Patents From B&W Loudspeakers 

Cirrus will introduce 5 amplifier chips using sigma-delta techniques. Four will produce a 110-dB dynamic range, and the highest-power product will have a 120-dB range. 

02 DEC 2000: TI releases 4-chip solution for digital audio. 

TI is the only vendor currently making volume shipments of an ADA solution.  

07 FEB 2001: STMicroelectronics gets exclusive license for Apogee's All-Digital DDX Amplifier Technology 

As you can see, digital amplification techniques have become very popular purchases. Philips, MicroSemi, Linear Technologies, and National Semiconductor are not included here because they seem to be using older analog PWM control techniques and focusing on lower cost car and portable applications. STMico has Apogee's technology but has not announced any high power products yet. 

It is odd that Analog Devices, who has a huge presence in audio codecs and DSP, has not made any announcements about a digital amplification strategy. Their SHARC DSPs are popular in audio products like the Sony TA-E9000 ES, Bose Lifestyle, and Denon AVR3300 home theater boxes. Their digital to analog converters and asynchronous sample rate converts are well respected. Unless they've got something in the labs, perhaps they should snap up Tripath or Korea's Pulsus. Tripath, being a mixed signal design, might prove too hard to put onto a single chip with a SHARC DSP. 

TABLE -- MARKET CAPITALIZATION AS OF FEB 27, 2001

 

Why are large DSP and digital audio houses scrambling for digital amplification intellectual property? Maybe it's because they need it to compete for a piece of the projected $3 billion dollar market in two years time. 

"PC audio applications represented the largest opportunity in 1998 with nearly half of the $1.55 billion market. According to market research firm Forward Concepts, consumer applications will represent the largest segment by 2003, with a compound annual growth rate of 24 percent and a total available market (TAM) of nearly $1.9 billion. The firm predicts that the overall market, including PC, consumer and professional applications, will represent an opportunity of more than a $3 billion by 2003." 

If PC, consumer, professional, and high fidelity products can all use similar chips then the ADA vendors can target most of the audio market with a only a few chipsets. Can you say "economies of scale?" TI and Crystal already have fixed function cores specifically specifically for digital audio. 

It's heartening to see that the large semiconductor companies feel that high fidelity reproduction is worth pursuing. The variety of digital amplifier techniques is also good news since it will give vendors a choice and allow market forces to weed out low fidelity approaches.

WHAT DOES ADA MEAN FOR HIGH FIDELITY AUDIO?

Some audio hobbyists will argue that ADA robs them of the chance to mix and match their favorite speakers and amplifiers. Also, some will undoubtedly prefer to use tube electronics. Perhaps the industry can accommodate these segments while still bringing the unprecedented benefits of ADA to High Fidelity's mainstream. The benefits of ADA in high fidelity include cheaper front-end electronics, higher fidelity input to speakers, and much more speaker design flexibility. 

1. A penny saved.... 

No more sinking money into huge transformers, massive heat sinks, expensive crossover components, multiple chassis & power supplies, or exotic cables. In many high-end loudspeakers the inductors in the bass crossover alone can cost more than the digital audio chips we're talking about here. 

ADA will replace many components of a traditional audio system and allow systems designers to shift budget to speaker drivers. Most designers will agree that using high quality drivers is money well spent. While ADA will be priced like any commodity computer chipset, good speaker drivers will remain high price items. Advanced magnetics, high precision mechanical assembly, limited markets, and hand fabrication will conspire to keep quality driver prices relatively dear. 

Even if ADA only equaled the quality of our current rat's nest of preamps, amps, DAC's, cables, and crossovers, the increased expenditure on drivers alone would improve fidelity at any given system price point. This is only the tip of the iceberg. 

2. Garbage In, Garbage Out 

There's a school of thought that says that a speaker can only sound as good as the electronics feeding it. Replacing the speaker correction DSP, DAC, preamp, amplifer(s), and analog crossovers with a tightly integrated digital solution will provide higher signal fidelity at lower cost. With ADA there is no preamp, no analog parts variance errors, no analog parts drift, no compression in analog crossovers, less loop area for RFI problems, etc. Digital amplifiers are receiving good reviews, and they will only get better as the market focuses its resources. 

Some high enders are so concerned with signal fidelity that they use outboard power supplies like the $1000 PS Audio Power Plant 300. These devices resynthesize an AC waveform from household AC to get a more constant and noise free voltage for sensitive electronics. Some digital amplifiers, like Cirrus's, monitor power supply voltage and take it into account when calculating the output. Yet another $1000 to be spent elsewhere in an ADA system since a $15 chipset will remove the need for the component. 

3. Synergy is the speaker designer's friend. 

Digital audio pioneer Meridian has been producing highly acclaimed active loudspeakers for years. Meridian's designs exploit the advantages of integrating DSP, DAC's, and (analog) power amps into the speakers. There are no exotic drivers or revolutionary amplifiers and yet their systems are some of the highest rated in the industry. 

By performing all signal processing in the digital domain and designing each stage to work in tandem with the next, Meridian is able to extract a high level of performance from the components used. The same chips, amplifiers, and drivers used in more traditional stand-alone equipment would result in a system of lower performance at a higher cost. 

ADA will allow companies without digital hardware expertise to perform similar feats but with digital amplification and at lower costs. Sony experimented with digital input audiophile speaker systems back in 1998, as did Dunlavy Audio at the 2000 Consumer Electronics Show. Dunlavy showed a modified version of their SC-IVa with S/PDIF digital inputs, digital crossovers, and a Spectron 600 Watt PWM amplifier for every driver. 

There are many advantages in designing a multi-amplified system with active crossovers, some of which are listed here. There are extra perks to be gained from implementing some or all of an active system in the digital domain. 

 Digital amps are smaller, provide more power, produce less heat, and at a lower cost than their analog cousins. There's an inefficient driver you really like the sound of? Go ahead and use it. Want to biamplify or triamplify? Do it for the cost of a single analog amp. 

 Broader transducer choices for designers. DSPs can provide transparent EQ and crossover flexibility. Know of a driver with great time domain performance, but it's not completely flat? Flatten its frequency response with the DSP. Doing the same complicated EQ in the analog domain can be tricky. 

 expensive, and degrade sound quality. If ADA becomes pervasive, we may even see new drivers that tradeoff frequency response flatness for low non-linear distortion. 

 Crossovers impossible for analog designs are possible with digital crossovers. You need 4th order crossover slopes, but like the linear phase properties of 1st order designs? With digital FIR filters you can have both. 

 More decor friendly packaging could boost sales. ADA solutions could be designed with less boxes and cabling since there are fewer components. Speaker enclosures can be made more acceptable to consumers. Dedicated high power amplifiers can reduce bass cabinet size, and DSP delayed signals can align drivers' output without unusual baffle designs. Wireless networking such as IEEE-1394 could even remove the need for cables from the audio system to speakers with integral amplifiers. 

 With crossovers determined by coefficients in the DSP code, designers can test many more crossover shapes much faster than before. It's quicker to change coefficients than solder up a new board. Designers could A/B test crossover curves at the press of a button. 

See Meridian's site for additional ideas on digital integration. 

4. Upgradeable & Customizable speakers. 

This could actually create a new trend in the industry. Currently designers spend (we hope) a lot of time measuring and listening to the crossovers before going into production. How many times have we seen a manufacturer release a Mk II, or a factory upgrade to change crossovers in production systems? 

With DSP based crossovers, if a revision is warranted it could be published online and downloaded instantly at no cost to consumer or manufacturer. Perpetual Technologies has already started down this path with downloadable codes for their P-1A speaker correction DSP box. A company could treat the initial release of their speakers as a "beta" release. Hundreds of listeners in the field could then provide their input for the next crossover release. 

RETHINKING POWER SUPPLIES

Say we go from the 50% efficiency of class AB amplifiers to 90% for our fancy digital PWM amps. We want 500 Watts RMS per channel so we have plenty of headroom. We still need fairly expensive transformers and capacitors for a 550+ Watt linear power supply. 

Can't we use a switch mode power supply (SMPS) for its cheaper, smaller transformer and better regulation under load? Accepted in the mass market and pro audio worlds, switching supplies do not have much presence in the high fidelity segment. It's difficult to suppress switching noise, and switching RFI plays havoc with nearby circuits and wires. Implementing a low noise SMPS requires much more engineering expertise than a quiet linear supply. However, some recent integrated SMPS controller chips have made high quality designs much easier to implement. 

Power semiconductor companies like International Rectifier are producing integrated solutions that incorporate "soft switching" logic such as Zero Voltage Switching (ZVS) and Zero Current Switching (ZCS). Both techniques have better power density, lower RFI, and reduce stress on the switching transistors improving reliability and product lifetimes. Integrated controllers can also provide Power Factor Correction (PFC), reducing AC current demand by 40%. The semiconductor companies are already developing versions of these components tailored for digital audio. 

In September 2000, Cirrus logic and International Rectifier announced they are working together to optimize power supply design, and high power MOSFETS optimized for the PWM output stages of digital amplifiers. Crystal's digital amplifier chipset has a sync-lock to easily integrate with a switching power supply. 

Clearly, the time for switching power supplies in high fidelity amplifiers has come. They have many sound quality advantages over linear supplies if implemented correctly, and can be cheaper. QSC Audio has been using a low EMI resonant switching supply they designed in-house for their acclaimed PowerLight and PLX professional amplifiers for years now. High-fidelity companies will start finding it possible and profitable to move to switching supplies since the new integrated SMPS controllers will lower costs and reduce required design expertise. 

It may already be happening. TacT's high fidelity PWM amp uses a switching supply. The highly regarded model 10 & 12 analog amplifiers from Jeff Rowland use a ZVS/ZCS switching power supply. California Audio Lab's CL-2500 MCA also uses a ZVS design to squeeze 5 x 500 Watt channels into a single chassis. These are all expensive products, but there is no barrier to prevent the technology to trickle down to mid priced units. 

Some advanced SMPS controllers use digital domain PWM control as do the audio amplifiers discussed in this article. After all, a regulated power supply is really just a type of amplifier. Digital amplifier designs may eventually integrate the PWM audio signal generation directly into the SMPS control logic, lowering the component count even further. Such designs would process the incoming AC power directly into the desired output signal without any conversion to DC. With this level of integration, it wouldn't be surprising to see a 300-Watt consumer ADA box for $ 300 or less in the next 5 years. 

The increased demand for advanced SMPS designs is due to demand for higher efficiencies and lower EMI/RFI. New European RFI regulations, longer battery life, efficient motor controllers, and computer & networking equipment are just a few of the driving factors. The worldwide market for integrated SMPS controllers and power semiconductors will grow rapidly in the next few years, and will result in lower costs, better specs, and tighter integration. All consumer audio has to do is sit back and enjoy the ride.

HIGH FIDELITY: THE ROAD TO ACTIVE DIGITAL LOUDSPEAKERS

Moore's Law and rapid commoditization will drop the prices of ADA solutions to a point where they will be far cheaper to implement than the traditional DAC, amplifier, and crossover combination. The high-end industry will have an inexpensive solution to every part of the reproduction chain but the speaker drivers and cabinetry. 

So far it looks like Texas Instruments, Motorola, and Cirrus logic are in the lead for high fidelity ADA solutions. Don't count out consumer electronics powerhouses like Sony and Sharp, but their solutions are more likely to be used in-house. Things could get even more interesting if Analog Devices gets on the bandwagon this year. 

To accelerate the acceptance of ADA in the high-end market, some enterprising manufacturer could produce a branded or OEM multi-channel ADA unit for use in digital active speakers. Add some filter design software and half the industry would be knocking on your door. QSC has the right idea with their DSP-3 module and filter design software for their analog amplifiers. 

Speaker design houses could integrate ADA boxes into their designs, or they could be purchased separately by the consumer and the appropriate software downloaded from the speaker manufacturer's website. The convergence of digital audio makes it likely that this type of product will be developed by both amplifier companies and digital audio companies alike since the cost of adding one to the other will be low. 

High end amplifier companies will need to enter the ADA ring as even mid-priced digital power amplifiers will challenge their fidelity. Some high-end companies will probably become nothing more than valued brand names and distribution channels for repackaged OEM amplifier modules. However, when everyone is using the same five amplifier chipsets the problem for companies will be differentiating themselves from the pack. Since high performance will be easier to achieve, manufacturers may try to add value with unique packaging and hardware/software features targeted at niche markets. For instance, ADA boxes with integrated high quality Analog to Digital Converters like might be a selling point for those with large record collections. 

Luckily the coming transition to ADA won't force companies to abandon their current customer base. With the inclusion of an analog to digital converter ADA boxes could be used just like traditional amplifiers. Sharp is doing this with their SM-SX100. Manufacturers will be able to serve different audio market segments with the same electronics. 

Most of the digital amplifier chipset vendors are building products at three power levels. Portable, consumer, and Professional / Audiophile. This is a sensible strategy as the market exists today. However, if the cost and quality of ADA electronics causes the audiophile market to transition to active speakers, there will be reduced demand for high power amplifiers. With multi-amped active speakers, high fidelity companies may end up piggybacking on the midpower mass market segment. 

In multi-amped designs it is common to see every driver's amplifier under 100 Watts. A 30 Watt module, bridged to provide 60 Watts for a driver would satisfy many active driver requirements. With clipping distortion in digital amplifiers mitigated by transparent signal processing, designers will be able to size the system's amplifiers better, and without the risk of expensive warranty repairs for blown drivers. Mid-power mass-market modules could offer the incremental cost savings of high volume parts. 

With the DSP power and programmability of consumer products like Perpetual Tech's P-1A and Sony's TA-E9000 ES, and innovation from pro audio companies like QSC we're already halfway to high quality ADA active speakers. We can't be sure exactly how events will play out, but high fidelity enthusiasts have a lot to look forward to in the years ahead. 

Want to add new information to the pot or just argue a point? Post a message to the discussion group. To get posting rights and sign up for future news bulletins join WaveFront by following the link under the Members box on the upper left of this page. 

Otherwise, send comments to Dan Barrett at This email address is being protected from spambots. You need JavaScript enabled to view it.

 

Read 104541 times Last modified on Wednesday, 25 December 2013 16:36
Published in Industry