Special Report Semiconductors

February 2007--Each February, Pro Sound News queries audio design engineers about their semiconductor usage. While the February issue of the magazine offers a Special Report based on the responses, the complete transcript of responses follows.
Author:
Publish date:
Updated on

February 2007--Each February, Pro Sound News queries audio design engineers about their semiconductor usage. While the February issue of the magazine offers a Special Report based on the responses, the complete transcript of responses follows. This year's respondents are:

Bob Bauman
Title:Co-founder, Chief Hardware Engineer
Company: Lynx Studio Technology
www.lynxstudio.com

Vince Capizzo
Title: Owner/Designer
Company: Midnight Designs
vincecapizzo@voyager.net

Tino Fibaek
Title: Chief Technology Officer
Company: Fairlight
www.fairlightau.com

Dan Lavry
Title: President
Company: Lavry Engineering
www.lavryengineering.com

Jim Odom
Title: Founder/President
Company: PreSonus Audio Electronics
www.presonus.com

David M Thibodeau
Title: VP
Company: DakingPlus
www.dakingplus.com

Lucas van der Mee
Title: Sr. Design Engineer
Company: Apogee Electronics
www.apogeedigital.com

ANALOG

Dan Lavry [Dan chose to give a sweeping answer to cover several of our analog queries]: I take issue with all questions regarding analog components. In audio, there are a lot of folks talking endlessly about which opamp is good, which is bad. The same goes for resistors, capacitors and everything else. Such generalizations most often lead to wrong conclusions. Let me explain with an example:
An opamp has many characteristics. Lets focus on noise, output drive capability and supply noise rejection. A given circuit may require a device with a lot of output drive. But if signal level is high having low noise is a lesser issue. Say the circuit power is very clean thus supply rejection is also a non-issue. Clearly, we select an opamp with good drive capability, while noise and power supply rejection are not important.
So someone listens to the unit (or measures it) and concludes that the opamp is great. Such "conclusions" often spread all over the industry.
Now lets use the same opamp in a different circuit, where the signal level is very low, thus noise is very important. Say the power supply is noisy, thus now the supply rejection matters a lot. And say the load is high impedance, so we do not need the drive capability. In fact, for this circuit, we need a low noise, good supply rejection device, and the output drive does not matter, thus the same opamp is a real poor choice.
There are thousands of opamp types out there, each offering a different set of performance levels and compromises. Each device specifies a lot of parameters, and no device can be "all things to all circuits." You improve a couple of parameters and something else "has to give."
When I began analog design, I thought that design was made of "step one"--to draw a circuit, followed by "step two"--putting parts into circuits. But I did learn long ago that the art of analog design is about the interaction of parts and circuits. A good part in one application can be terrible in another and visa versa.
What I said about opamps is true for other parts. I see people changing all their caps to, say, film caps, where a ceramic would do better. I see people going for a certain type electrolytic because it worked better in a whole other circuit in a different product. I see people changing opamps for the same reasons. Parts selection should be the domain of a person with in-depth understanding of many aspects of electronics. One cannot select analog parts without complete understanding of all the parameters. That best be left for experienced design engineers.


1. For analog design, have you discovered any new chips recently? Are you using different components than you have traditionally?

David M Thibodeau: Yes, Our experience with the THAT Corp. 1200 series balanced receivers have been quite favorable. They have proven to provide excellent specifications in an actual product vs. just a test jig. They also do not color the sound and so therefore do not change the sound of our products.

Bob Bauman: Not really, although I am sure there be will parts that spark my interest as we start some new analog projects later this year.

Tino Fibaek: We have not changed much in out analog circuits recently. We largely use BurrBrown and NJR opamps, and Mic Pres from THAT Corp.

Jim Odom: THAT Corporation continues to be the innovator in this arena. We watch them closely.

Lucas van der Mee: Yes, contrary to what I experienced a few years ago, there are actually some very interesting new developments in the opamp world. At that time silicon companies seemed to be only interested in low power, rail-to-rail type solutions as required for battery powered and portable applications. Surprisingly, now there are a few new releases that are aimed again at the high-end audiophile applications running on higher voltage rails. Examples are the extraordinary LM4562 opamp from National and the ultra low-noise AD8599 by Analog Devices.
Another very interesting development is the growing number of differential opamps; there are some real good ones, now running on higher voltage rails.
I would say my biggest change in component use is in the power management application. The use of linear regulators is getting smaller and smaller. All of my late designs now have custom designed switching power supplies. The range, quality and possibilities of DC/DC converters have made an incredible flight. It totally changes the way we think about power handling and how we apply it.

Vince Capizzo: National may have a winner with the LM4562 dual opamp. Vanishingly low THD and IMD, low noise, drives 600 ohm loads, good DC specs. Worth a listen.


2. Are you using different components than you have traditionally?

Vince Capizzo: To some extent, yes. Innovation isn't limited to digital parts. THAT Corp. has introduced some clever line driver/receiver chips. Both TI and Analog Devices have introduced low distortion amplifiers with fully-balanced inputs and outputs, intended for driving double-ended ADC inputs.


3. Are you using any of the modular analog building blocks-mic pres, VCAs and balanced drivers and receivers on a chip, for instance? If so, elaborate where and why.

Bob Bauman: We have used balanced drivers in the past, but recent designs have been more discrete. We typically build these functions using high performance amplifiers and custom precision resistor networks. This allows us to tweak noise, distortion, and CMRR optimally, based on the gain and signal level required. There some nice balanced receivers from TI that we still use.

Tino Fibaek: We use these where applicable--for example, we use BurrBrown line drivers on a number of designs. The performance of these devices is now at a level that is hard to compete with, and the lower component count reduced cost and increase reliability.

Jim Odom: We have used THAT VCA's exclusively in our analog compressors.

Lucas van der Mee: Yes, I have been looking into several of those. I have been using some VCA's in the past, for the compressor/limiter in the Mini-Me.
On the newer designs two companies stand out, one is TI with their digital controlled mic-pre PGA2500. They made that feature available to a much wider audience, by integrating it all in one chip. It is a surprisingly good sounding and well performing design. I was very impressed. I am using this chip in the Ensemble, but honesty forces me to say I did a few extra tweaks to get the performance I wanted for this design.
The other is THAT Corporation with their new balanced driver and receiver chip. THAT's "OutSmarts" driver solves the current into ground problem of an unbalanced load to a traditional cross-coupled feedback output driver.
I have developed my own solution for this problem in the past, but my application requires an extra opamp and a lot more components. However mine is still able to output more current, so I am not sure if I will switch.

Vince Capizzo: Yes. These devices often come in handy when space is critical and the highest performance isn't required. VCAs, on the other hand, are generally better in modular form rather than as discrete designs, since DC drifts due to temperature have been carefully compensated.
Another advantage, when design cycles are short, is that the modular circuits come pretty much pre-bullet-proofed.

David M Thibodeau: Yes, the THAT Corp., 1200 series balanced line receivers.


4. Are there any newly or recently released components that are improving your analog designs? If so, what are they and in what areas are you seeing improvement (ease of design, performance aspects, etc)?

Tino Fibaek: We are searching for possible implementations of remote controlled MicAmps. There are some offerings on the market, but we're not sure that they are quite ready yet...

Jim Odom: We are using a variety of switching power supply IC's as well as some new low power components.

Lucas van der Mee: Well, I think the answer to this has been pretty much given in question 1 and 2. The opamps, because they are giving me better performance for the same price and the modules mentioned in 2, because they hugely simplify the circuitry and reduce the amount of components required to do the job.
However, I tend to be a bit wary of using a single source chip solution for solving a problem. If there are no crosses, what are you going to do when the part is suddenly in short supply? It is a dilemma that becomes more and more common. On the one hand being flexible, but a big BOM and higher cost--and on the other hand a relatively cheap module/chip, but being very dependent of the supplier. Lately the balance is definitely towards the latter. The downward price spiral our industry is in gives you no choice.

Vince Capizzo: Both. The trend seems to be that the performance gap between medium- and high-end designs is shrinking. Multi-channel converters (six or eight channels in a package) simplify the design, but aren't quite good enough yet for the really top end stuff.
David M Thibodeau: I am hoping to have the opportunity to test the THAT Corp., OutSmarts balanced drivers soon. We are not planning on changing our transformer balanced output stages because that is part of our sound but these parts will be handy for things like insert sends where a transformer is not needed for its sound, just for balancing. This will help to lower the product costs without sacrificing performance.

Bob Bauman: See # 1.


5. Are there components that you have traditionally used (analog or digital) that are becoming hard to source because of the EUs RoHS initiative or other reasons? If so, how are you responding to the shortfall?

Jim Odom: Tubes are a problem still, because of the lead on the pins. We are working with our vendors to provide compliance.

Lucas van der Mee: We have been really lucky so far; we started to implement lead free parts in a very early stage and haven't experienced any problems with the parts that have been replaced with lead free crosses. Other than a headache for the people involved with the logistics the transition has been pretty smooth.

Vince Capizzo: Many op-amps are no longer available in through-hole packages, and the smaller surface mount types seem to have greater temperature gradient issues than their larger obsolete predecessors. There's no magic in getting around these issues. Since all engineering involves tradeoffs, the way forward is making new and different tradeoffs. Mostly this involves searching out new components that do comply, learning their shortcomings and then optimizing them in the configuration required. Sometimes this means developing new configurations.

David M Thibodeau: Electrolytic and Film caps have been a bit of a problem because of the higher temperatures required for lead free solders. Many of the parts are not rated for these temperatures. In the past few months I am seeing some improvement with the availability but there is a long way to go. Hand soldering surface mount film caps adds a lot to the production costs but in Pro Audio good film caps make all the difference so they have to be used.

Bob Bauman: There is at least one D/A converter that is now obsolete and not available in a RoHS compatible version. The product that utilizes this device will be re-designed in the near future. Overall we have been "lucky" that the majority of the parts we use are available in RoHS versions. I must say, the transition to RoHS production has been very labor intensive and time consuming. Some part manufacturers have supported this transition with new "green" part numbers and excellent web-based documentation. Others did not change part numbers, and rely on date code as the criteria for RoHS compatibility. The latter method proves to be very difficult to track in some cases. Besides the product mentioned above, all of our products are now available in RoHS compatible versions.

Tino Fibaek: Our recent switch from DSP to FPGA processing eliminated all projected component obsolescence issues.


6. As audio production becomes increasingly computer based, simpler, interface bus-powered I/O is becoming popular. Does this low voltage approach present challenges in the analog circuitry proceeding or following conversion? Do these challenges mean performance compromises or present restrictions in analog performance?

Lucas van der Mee: Yes, of course it does. High-end analog circuitry is not exactly low power, so having to design for efficiency is limiting your options a lot. The good news is that there are more parts nowadays that are efficient and have reasonable performance. So it is possible to make something really decent on bus power as long as the design is small, i.e. a limited amount of analog I/O.

Vince Capizzo: Yes, big time.


7. Do these challenges mean performance compromises or present restrictions in analog performance?

Vince Capizzo: Supply rail voltages have been steadily shrinking for the past 10 years, at least. There are ways to cope. Rail-to-rail op-amps and balanced operation are commonly used to gain back some dynamic range. Current-mode operation is another approach. The best solution depends on the application at hand. Thinking is required, but there's almost always a way out.

David M Thibodeau: Yes, That is why we do not use bus-powered designs. We feel it is to limiting to the analog performance.

Dan Lavry: In principle, lower voltage environmental restrictions can be solved, because one can generate a localized higher voltage. Say you need to use a varactor diode with control range of 1-25V. Such control voltage requires very low current, so a simple and inexpensive voltage multiplier can do the job.
But when the power requirement grows, voltage boosting becomes more costly, and it calls for attention regarding both conducted and radiated noise. In recent years we have seen an increased use of differential (balanced) signals in both AD and DA's. The differential scheme effectively extends the range by a factor of 4, thus a 0-5V device with say a balanced +/-3V range offers an equivalent 12V single ended range. Extending the voltage range is only one of the reasons for the use of differential balanced analog.
As a rule, the reduced voltage range does make life a bit more difficult. I use higher voltage range whenever I can, scaling the range down only when I need to.

Bob Bauman: Yes, particularly with PCI Express cards. Whereas in the past we utilized +5V, +12, and -12V power from the standard PCI bus, PCI Express provides just +3.3V and +12V. The loss of the -12V supply makes if partially difficult to support typically line level analog I/O. We are investigating a few options for generating the required voltage rails while keeping noise at a minimum.

Tino Fibaek: So far, we have elected to not manufacture any bus-powered products; instead, we have stuck with internal power supplies, in order to ensure professional level audio specifications.

Jim Odom: PreSonus is very keen on the design of preamplifiers; power supply design is a key component of the sound. In our bus-powered products, we create what we call a 'power plant' from the single voltage supplied by the bus. The "power plant" provides all of the traditional voltages: +/-16.5Volts, +48V, etc. needed for proper analog design.



DIGITAL

8. Digital to/from analog converter parts seem to be developing in two directions at once, upwards to higher performance, and more mass market where cost and power consumption are more key. What criterion determines your selection of converter parts?

Vince Capizzo: Generally the selection is made by juggling the product specifications, converter feature-set, availability, and price. This narrows the field, and final determination is made by listening.

Bob Bauman: Since we don't try to compete in the mass-market arena, we focus on parts with the highest performance primarily. For high channel-count products, power consumption is used as a secondary criterion.

Tino Fibaek: It all depends on the target application. As mentioned earlier, we are not restricted on the power budget. Also, our designs tend to be more system orientated, with elaborate sync and control possibilities. In this environment, the extra cost of the converter does not drastically affect the overall system price.

Jim Odom: Dynamic range, clocking, and streaming interface are the key factors.

Lucas van der Mee: We use two criteria dependent on the application. Our high-end gear pretty much requests the best we can get. Price and or power consumption is less of an issue, but of course there is a limit to this. Otherwise we would be building our own converters.
For our budget oriented products we do have to look at price though. That is inevitable. I would say that the quality of cheaper converters has increased a lot more than the high-end. In other words, the bottom is coming up, just like anywhere else in technology.


9. Have you adopted any new conversion parts this year? If so, which and why?

Bob Bauman: None yet, but the folks at TI have a few nice parts hitting the market that are very attractive.

Tino Fibaek: We have recently designed in a couple of 2/6 codecs from Cirrus Logic. The specifications, cost, and feature set fit very well with the specific application that we were developing.

Jim Odom: No comment.

Lucas van der Mee: We started using a Crystal codec CS4272. It is the younger brother of the much-loved CS5361/81 series we have been using for quite some time now. It shares the same characteristics, but is lower in price and performance. A very nice chip though. It was the perfect solution for making a budget converter unit and still having the same sound and feel.

Vince Capizzo: Cirrus Logic's CS5381 is an excellent ADC with impeccable performance and low latency. TI's PCM1792A DAC is also formidable, and can be used in both PCM and DSD applications.


10. Where at one time, the golden ears eschewed sample rate conversion circuitry except where unavoidable, even many high end users are accepting the performance of modern parts. Is this a trend you accept? Are you using SRC parts in your designs?

Tino Fibaek: We're currently evaluating an FPGA-based SRC solution. This approach has the advantage that we can balance how much processing power we're allocating to the conversion, vs the quality and sound-we're not stuck with what a general-purpose (must fit all) solution provides.

Jim Odom: N/A

Lucas van der Mee: We are only implementing SRC as a user option, specifically with the use of digital conversion in mind. The application of using it for solving a jitter problem with AD and DA conversion we have stayed away from, for a number of reasons. One is and although this argument is declining in value, SRC is still not a lossless process. Two, it will increase the latency of the conversion path dramatically. Our converters are being used consistently in tracking sessions, so every gain on latency reduction is a must. Three, we have a few very solid PLL designs, that really make the "jitter" problem a non-issue. So there's your answer.

Vince Capizzo: Yes. They've gotten much better.

11. Are you using SRC parts in your designs?

Vince Capizzo: They still aren't perfect, but in some applications SRCs solve the problem so elegantly, and simply that even with their minuscule shortcomings they give us the best sounding product.

Bob Bauman: I am still a purist--if you can avoid using SRC, do so. However, the performance of recent SRC devices is improving to the point where the noise added by the SRC becomes insignificant relative to the noise in the rest of the signal chain. You must still be cautious of cascading multiple sample rate conversions in a signal transmission.
An SRC can be used at the receiving end of an AES/EBU signal as an effective jitter attenuator. In this case the sample rate is not changed, but an incoming signal is processed to a produce a new signal referenced to a local low-jitter crystal oscillator. Again, the noise added by the SRC may be lower than that generated from residual jitter of a typical PLL in the digital receiver and other downstream PLL's.


12. Do you incorporate Sample Rate Conversion in your digital I/0? If so, what parts are you using and why?

Jim Odom: No.

Lucas van der Mee: Indeed, for digital I/O, we have been using it. We used the AD1896 from Analog Devices because that was the first SRC chip with acceptable performance. Later we switched to TI's SRC4192 when that came out, because of its better performance.

Vince Capizzo: Yes, where appropriate. For example, they are very useful for products used in a predominantly video environment where the studio operates exclusively at 48KHz, but audio sources of all types may need to be incorporated.


If so, what parts are you using and why?

Vince Capizzo: CS8421, AD1896. See above.

Bob Bauman: Yes we do offer SRC in the digital I/O of some our products. The user has the choice of enabling this function or not. We have used the TI SRC4192 because of its excellent performance at any sample rate conversion ratio.

Tino Fibaek: We have been using the 8420 from Cirrus Logic for a while now. It sounds good, and combines a number of features that fit well in the specific architecture.


13. Formats such as USB, Ethernet, AES50 and FireWire are being increasingly used for audio purposes. Are you incorporating new protocols into your designs? If so, what devices are you using for interface?

David M Thibodeau: FTDI is the greatest thing since sliced bread for USB designs.

Lucas van der Mee: USB and FireWire we both have quite some experience with. USB 1.1 is way too limited to further expand. USB 2.0...just nothing happens there. FireWire developments we are deeply involved in. As many readers know, we do have a very intimate working relationship with Apple in order to develop and complete this format for audio standards.
The other formats, like Ethernet based ones, we are closely monitoring. We'll certainly adopt one, once they become more standard or once we see an application for our products.

Vince Capizzo: Not currently.

Bob Bauman: We have just announced a FireWire expansion card for our Aurora A/D D/A converters. This product is primarily aimed at satisfying the needs of customers requiring a connection to a laptop computer for mobile recording. FireWire is convenient and works fairly well with proper hardware and well-written host drivers. I would not consider it the ultimate choice for high-performance, multi-channel audio though. The protocol has tons of overhead, there are latency issues to deal with, and users are typically at the mercy of FireWire chipsets and drivers used in off-the-shelf computers.
AES50 shows promise in that it was designed from the ground up as a high-performance digital audio interface. Its latency is very low and, unlike FireWire, has dedicated clock signals-a key feature that helps reduce system jitter. I suspect that AES50 will supplant MADI for multi-channel digital I/O with consoles and other applications.
Another newcomer in the external interface realm is PCI Express over a cable. The spec for this was just made official. Now you can basically take a PCI Express add-in card design, put it in a box, and connect it externally to a desktop or mobile computer. Design time is minimized since drivers written for an add-in card work transparently for an external device.

Tino Fibaek: Generally, unless there is a compelling reason not to, we tend to move as much as possible into FPGA technology. This allows better integration, more reliable product, and field-upgradeability.

Jim Odom: We currently use a variety of components for FireWire, Ethernet and USB products. We use FireWire parts from a number of vendors and we will continue to do so, though we don't want to point our competitors in the directions we are headed (we already feel like PreSonus should have won the award 'most copied product' for the past four years).
I will say that the JetPLL clock in the DICE series of parts from TCAT (which we licensed for our Digimax products) is incredible. You should hear the detail in the converters--simply amazing. It is proof to me that the most overlooked spec in any digital audio product is the jitter. Most companies don't have the tools to measure jitter, as it is a very small distortion of clock timing (picoseconds). We hired a nuclear physicist (not joking) to teach us about measuring extremely small variations in time and how that affects placement in the stereo image.


14. Is the current crop of interface devices (as in Q10) adequate to the task? Any tools you would like to see?

Bob Bauman: No comment.

Tino Fibaek: There are more and more companies developing IP-cores (for FPGAs), instead of or as well as traditional chip level interfaces. We think this is a very exciting trend, as it allows for faster and more flexible development of new designs.

Jim Odom: Yes, there are ample ways to achieve the physical and logical connections to stream audio at this time. Cheaper, faster, smaller solutions are always in fashion.

Lucas van der Mee: We actually decided to release our own proprietary standard (Symphony) because none of the formats mentioned above has the bandwidth and price/quality required for a large channel count session with very low latency in and out of the computer. For lower channel count though, FireWire is very sufficient and a cheaper alternative.


15. If you are involved in DSP design, why have you chosen the components you use? Do you see advantages in one family of processors over another? In the tool sets available for DSP programming?

Jim Odom: We develop using SHARC processors. I believe there are huge advantages in floating point architectures, the least of which being unlimited headroom for processing audio...

Tino Fibaek: Two years ago, we made the decision to move away from traditional DSP designs, and instead amalgamate all our DSP processing requirements into a single FPGA chip. For the large-scale audio engines that we use in our products, we felt that FPGAs were a better, more flexible route into the future.

Bob Bauman: Using FPGA's for DSP.


16. Have you experimented or employed FPGA processors for DSP tasks? If so, why?

Bob Bauman: Yes. It's amazing the amount of signal processing capability available in today's cheap FPGA's. The upfront development and implementation of an algorithm in an FPGA is more time consuming, but for raw speed they are tough to beat. FPGA resources, such as multipliers, can be run in parallel and organized to optimally support a particular algorithm. As opposed to general purpose DSP, I/O peripherals, microcontrollers, and custom logic can be easily added.

Lucas van der Mee: Yes, we have. Reason is that FPGA's are becoming bigger, better, more powerful and more advanced. So why add extra chips if you can do it all in one. Since our DSP functionality is pretty light an FPGA would suffice in many cases.

Vince Capizzo: Yes.


If so, why?

Vince Capizzo: If the DSP requirement is not too complex, implementation in an existing FPGA is feasible, and where it's sensible we do it this way. It's faster, saves board space, power, etc.

Tino Fibaek: Yes, we have now replaced our previous DSP design (which included 64 SHARC processors) with a single FPGA device. Again, the benefits that we saw were better reliability, more cost effective, future proofing, and re-deployment into different applications. We think of our new design as "virtual hardware"...

Jim Odom: Yes, our FireStudio is based on XILINX FPGA's. We like the deterministic nature of an FPGA for data formatting and clocking. However, advanced DSP algorithms are delegated to the SHARC.


17. Are there advances in DSP technologies that you are particularly excited about?

Jim Odom: Moore's law working here as well.

Vince Capizzo: As a rule I derive excitement from things other than DSP technologies.

Bob Bauman: Nothing specific, it's all good.

Tino Fibaek: Yes, The emergence of FPGA-based processing cores.


GENERAL

What advice would you give a consumer who is trying to intelligently assess a purchasing decision-as a designer, do you have any guidance to share?

Tino Fibaek: The CPU manufacturers (AMD and Intel) have both recently made announcements that endorse co-processors for the acceleration of process intensive applications such as media creation. In both cases FPGA has been identified as the technology platform that will deliver new performance benchmarks. Consumers will need acceleration especially when they begin to work with more diverse and enriched media formats. Our advice would be to look at accelerated platforms and avoid being convinced that host only supported applications deliver guaranteed performance.

Jim Odom: Many products claim equal specifications but are indeed quite different. I would recommend reading reviews and talking to more experienced consumers. Microphone preamplifiers for example, can have similar specifications, but sound completely different. Some specifications are not published, but have a huge effect on the quality of sound, such as jitter in a clock. Ultimately, stick to brands that have a proven commitment to quality and performance.

Bob Bauman: Research and listening is required of course. Try to get real-world specs of different products that allow an "apple-to-apple" comparison. Avoid products that quote just the spec of the A/D chip as opposed to that of the entire analog signal path.
Don't rely on rumors or unfounded opinions. Case in point-for years it was thought that a PCI add-in card was incapable of delivering professional quality audio just because it was enclosed in a "noisy PC." With proper supply filtering, ground planes, component selection, and circuit topology, dynamic range and S/N ratios in excess of 116 dB can be achieved. This performance is exhibited by our LynxTWO/L22 product line.

Lucas van der Mee: First of all, to the consumer it really doesn't matter what parts are inside the box. Two separate products from different manufacturers can be build around the same chip and still be radically different. It is all about the implementation. I have seen gear with very expensive parts, totally underperforming and I have seen products with really cheap parts doing an amazing job. So as a buyer, don't waste your time trying to figure out how it has been built. What is important is how well the company supports your product. Do they have a good tech support? Are they loyal to their customer? Do they have a website with a lot of useful information? Have they been around a while so you don't have the danger of buying a product that becomes obsolete before you know it? Be wary of the forums, there is a lot of good information, but a lot of it is also driven by commercial interest, or simply herd mentality; afraid to be different. The best test is always to take it home and test it on your own [where there's] no pressure of sales people and all the time in the world. It is surprising how quick you can find out this way if something works for you or not. And last but not least, trust your ears.

Vince Capizzo: There's an art to reading data sheets. It's important to notice what's not being spec'd. If possible before purchasing, direct your applications questions to the support people at the company. This can be very informative. In the end, listen carefully. No amount of features will compensate for a product that doesn't sound right.


Any parting thoughts on ICs and their application? Trends you'd like to comment on?

Vince Capizzo: National Semiconductor is starting to take notice of the professional audio community. This is a Good Thing. There are some innovative products coming out of That Corp. and Wolfson.

Lucas van der Mee: I mentioned it before, the universal trend is towards cheaper one chip solutions. Which means that products will have a limited life cycle and will become more disposable. Think of your DVD player. Nobody will go to a repair shop to have it fixed. You simply buy a new one. This is of course a waste of resources; then again the actual products are getting smaller and smaller too and will use less material. So in the end it may end up being the same. But what is does imply is sadly, that in the future, vintage pieces of gear will become a vintage phenomenon.

Bob Bauman: Hats off to the pro audio guys at TI, Crystal Semi (Cirrus Logic), and AKM for their continuing commitment to push performance levels of converters, amplifiers, and I/O devices. In the midst of an increasing market for MP3 players and iPods, I am sure these guys have a tough sell to their corporate higher-ups to invest in the development of products for this relatively small, high-end market with fickle customers.

Tino Fibaek: We believe that the world of digital hardware design is rapidly changing; it's a big and positive change, yet its benefits will not be obvious to the end user for maybe another year or so.