PxPixel
Web Bonus Audio Semiconductors--2009 - ProSoundNetwork.com

Web Bonus Audio Semiconductors--2009

Each February, Pro Sound News queries audio design engineers about their semiconductor usage. A subset of replies was used for our Semiconductor Special Report in the February print issue.
Author:
Publish date:

Each February, Pro Sound News queries audio design engineers about their semiconductor usage. A subset of replies was used for our Semiconductor Special Report in the February print issue. The full transcript of the respondents replies follows:

Name: Dave Derr
Title: President
Company:  Empirical Labs

Name: John Stadius
Title: Technical Director
Company: DiGiCo UK Limited

Name: Olivier Limacher
Title: VP R&D
Company: PreSonus Audio Electronics

Name: John La Grou
Title: Founder
Company: Millennia Music & Media Systems

Name: Brent Elder
Title: VP of Engineering
Company: Universal Audio

Name: Paul Wolff
Title: President
Company:  Tonelux Designs, Ltd

Name: Craig (Hutch) Hutchison
Title: Senior Design Engineer
Company: Rupert Neve Designs

Name: Marvin Caesar
Title: President
Name: Gary Liden
Title: Chief Engineer
Company: Aphex Systems, Ltd.

Marvin Caesar/Gary Liden:
Essentially the answer to all the questions were either no or not applicable.

Analog
1. Analog components, general:
a) For analog design, have you discovered any new chips recently?

Dave Derr: Yes
John Stadius: Plenty of new chips but nothing to write home about.
Olivier Limacher: Too busy to look for any.
Brent Elder: Yes.
Paul Wolff: Talking to TI about a new low noise opamp. Seems interesting. Otherwise I use my own two discrete amps.
Craig Hutchison: Yes

b) Are you using different components than you have traditionally?
Dave Derr: Not in analog
John Stadius: Yes
Olivier Limacher: No.
John La Grou: We’ve been using fewer chips (IC’s) and relying more on discrete analog parts – FETs and bipolar transistors – for our active circuit elements. I recently designed a custom 56 channel distribution amplifier for Georgetown Masters which uses our new 120V discrete amplifiers exclusively. The 120 volt amplifiers employ a unique combination of discrete FET and bipolar transistors.
I’ve also been working on a profoundly new topology of microphone preamplifier for many years using all-discrete components. I’m very close to realizing my audio performance goals and we hope to have the design into production by 2010. A select group of our current customers may get to test beta units later in 2009.
Brent Elder: Yes.
Paul Wolff: We are converting to surface mount on just about everything as through hole parts are becoming harder to find.
Craig Hutchison: Yes

c) If so, what are they and in what areas are you seeing improvement (ease of design, performance aspects, etc).
John Stadius: Almost everything now is surface mount. Test yields and product reliability is much improved. Migrating from a multi DSP audio engine to one using just one large FPGA has the advantages of being easier to make, more reliable and less power hungry (green).
Brent Elder: Performance, ease of design, level of integration, physical size, cost.
Paul Wolff: Op-amps are getting better as the surround and higher sample rates move in. People like better sound. It’s a relief, as the chip industry always did analog chips as a joke. Now they see it as an industry.
Craig Hutchison: There are tens of thousands of new chips every year and some of them conveniently fit a purpose for us and some offer better specs or performance. For audio, maybe no integrated op-amp has surpassed the 40 year old NE5534 for price vs. performance and universality. The reason they make so many choices, is because each one is intended for a particular application and can be inappropriate elsewhere. For certain circuits and applications sometimes we need to dig a little deeper.

2. Modular Analog Components:
a) Are you using any of the modular analog building blocks—mic pres, VCAs, dynamics engines and balanced drivers and receivers on a chip, for instance?
Dave Derr: VCAs we use.
John Stadius: Yes
Olivier Limacher: We use a RMS/VCA engine on some products, but we recently moved much of the analog channel processing into a DSP.
John La Grou: No.
Modular units (TI2500, balanced IC receivers, etc.) are fine for commercial and mid-range applications, but we’ve never been in that segment of the market. We concentrate on the ultra-high-end of sonic performance. This means doing things from scratch with very strict design goals.
When we need a certain function, we try to achieve it via an all-discrete solution. For instance, we recently needed a true differential amplifier (not an opamp, but a true diff-amp) so we designed our own from discrete parts. It sounds far better than any IC-based diff-amp we tested. No comparison, really.
Brent Elder: Yes.
Paul Wolff: None. I do use the VCAs and RMS detectors from THAT Corp; they are the best. They think discretely when designing things, which makes their stuff sound better. I use the Balanced in and balanced out chips from them in my Control Room module, we moved away from the discrete amps because they had a slight color.
Craig Hutchison: VCAs and dBrms modules.

b) If so, elaborate how and why.
Dave Derr: THAT VCAs offer high performance, ease of use and small PCB footprint with the standard SIP Package
John Stadius: We have used “modular” mic pre’s and line drivers for many years now. They are very consistent and match many good discrete designs. With fewer components they also make the product more reliable.
Olivier Limacher: A senior coworker of mine used to say, “Engineering is the art of getting done for 50 cents what any Joe can do for a buck.” This exercise often requires replacing that expensive IC with a cheap analog trick without degrading the signal quality.
Brent Elder: We use analog building blocks if they are less expensive, higher performance, or smaller than the discrete equivalent.
Craig Hutchison: How - carefully, why – good stable performance.

3. Component Sourcing:
a) Are there components that you have traditionally used (analog or digital, ancillary components other than chips) that are becoming hard to source because of the EUs RoHS initiative or other reasons?
Dave Derr: NO, but we had to essentially re-specify every single part for RoHS. Cost us well over $100,000 in time and inventory, AND, we have had some parts fail because of the higher soldering temperature required.
John Stadius: We are past the problem of components being discontinued due to RoHS. Apart form normal long lead items we have no major issues now.
Olivier Limacher: The RoHS version of the Alesis 1k DSP had a different power up behavior which lead us to discontinue a product. The lead free solder is the usual suspect when production units fail, but it gets acquitted almost every time after a more thorough analysis. Other than that knock on wood…
John La Grou: Not really. But don’t get me started about ROHS. It’s a nightmare for manufacturers. The solders we’re forced to use are not nearly as robust as before. And it’s questionable whether or not ROHS really helps our planet. I’ve seen one academic study from the University of Florida showing that ROHS requirements actually cause far more net energy and ground pollution (more greenhouse gasses, etc.) to be expended worldwide.
I’m a huge fan of helping the planet (Millennia recently became a charter B-Corporation: www.b-corporation.net) , but I don’t like it when governments enact “feel-good” programs that, in reality, do little or nothing to address the real problems we face as a human family.
Brent Elder: Yes.
Paul Wolff: No, most of the parts just needed the lead removed from the leads. They did use it as an excuse to charge more, even though they didn’t have to use lead or pay to have the waste processed. Some of the caps still are problematic, as the soldering temperature is higher. We have been lead free since the directive was in place. It doesn’t look as nice, and we get raped on the solder prices, but oh well, it’s all a big screw-a-thon anyway.
Craig Hutchison: This year, mostly obsolescence of thru-hole capacitors.

b) If so, how are you responding to the shortfall?
Dave Derr: I cuss and complain and bear down to find parts that will work and NOT FAIL.
John Stadius: N/A
Brent Elder: We continue to try to find alternate sources, or to obtain waivers where possible.
Craig Hutchison: Combining or substituting.

Digital
Paul Wolff: No digital stuff yet.

4. What criteria determine your selection of A/D and D/A conversion parts?
Dave Derr: Mainly performance. Mostly the usual AKM and Cirrus parts
John Stadius: Don’t believe the data sheets. Trial and error and lots of listening and testing.
Olivier Limacher: Specs have gotten very good. Who looks at distortion specs on a DAC anymore? We check the basics: Dynamic range, channel count, flexibility of the digital interface, cost.
Brent Elder: THD+N, SNR, latency, resolution
Craig Hutchison: Our pure pleasure when listening to music. Minimal unnatural artifacts!!
Marvin Caesar/Gary Liden: performance/cost

5. New converter parts: Mostly the usual AKM and Cirrus parts
a) Have you adopted any new conversion parts this year?

Dave Derr: NO
John Stadius: Yes
Olivier Limacher: Yes. The TI PCM4104 and 4204 for our new digital mixer.
Brent Elder: Yes
Craig Hutchison: Yes

b) If so, which and why?
Olivier Limacher: Great dynamic range. This allows us to compete head to head with high quality analog mixers, while providing the extra functionality of digital. We get a good price too.
John Stadius: Unfortunately I cannot discuss as the products are not yet in production
John La Grou: It’s all about sound. A converter’s specification is rarely an indication of how it sounds. I always suggest that engineers take spec sheets with a grain of salt. Rather, they should ALWAYS compare ADCs and DACs (or any audio product for that matter) in real-world A-B listening tests before making a buying decision.
We’ve just released our plug-in 8-channel ADCs for the 8-ch micamps (HV-3D and HV-3R) and they sound fantastic- very musically pure. We tested all sorts of converter devices – cheap ones, expensive ones, all over the map. We selected the part that sounded best, irrespective of cost. And then we applied all of the very latest ADC design techniques to assure that we’re maximizing sonic performance in these designs.
Brent Elder: Still in process, but we have new product designs that require improved performance and need to hit aggressive cost targets
Craig Hutchison: We are always evaluating and comparing new ADCs and DACs that seem to be contenders on the spec PDFs, but adoption just depends on listening tests.

6. Sample Rate Conversion:
a) Are you using SRC parts in your designs?

Dave Derr: No
John Stadius: Yes
Olivier Limacher: Not using at the moment.
Brent Elder: No
Craig Hutchison: Less, because the trade-off between advantage and disadvantage has shifted in the last 3 years. Better clock recovery devices and methods are available and with the newest converters, old ASRC chip artifacts have become more audible. That balance may shift again if better SRC appears.

b) If so, what parts are you using, where and why?
John Stadius: We currently use Cirrus Logic parts on all our AES transceiver cards. The user has the option of switching them in or out as they wish.
Olivier Limacher: The DSP we use has some built in.
Brent Elder: TBD

7. Formats such as USB, Ethernet, AES50 and FireWire are being increasingly used for audio purposes.
a) Are you incorporating new protocols into your designs?

Dave Derr: Starting to
John Stadius: We have used USB and Ethernet in our products for some years now. We only use them for data and control and not for transporting audio.
Olivier Limacher: We have been using 1394 (Firewire) extensively.
Brent Elder: a) No
Craig Hutchison: Yes

b) If so, what devices are you using for interface?

Dave Derr: We like the PIC controllers
John Stadius: Through our embedded industrial PC
Olivier Limacher: TC Dice, Bridgeco.
John La Grou: We’re using Pro Tools “PRE” compatible MIDI and standard Ethernet for our remote control mic preamps. This allows us to control up to 792 micamps from one computer, and gives us virtually unlimited processing power. All instructions are carried out virtually instantaneously, no matter how many micamps are on the Ethernet network. What’s more, Ethernet allows us to control our remote preamps wirelessly, or even via the Internet if needed.
We’re also working on another product that employs USB digital input. USB is fine for up to 96kHz digital-audio data. The trick is keeping low latency, which we’re working on and getting very good results. You also don’t want to force people into loading drivers on their computers, so the USB drivers should be embedded into the hardware itself.
Craig Hutchison: USB and Ethernet interfaces

c) Is the current crop of interface devices adequate to the task?
Dave Derr: We will know soon!
John Stadius: I think so
Olivier Limacher: Yes but still over engineered and too expensive for the very low end.
Craig Hutchison: They work.

d) Are there any tasks currently difficult or awkward to design, where you would like to see dedicated devices?
Dave Derr: The world still needs ultra quiet regulators compatible with the old 7815s and 7915s (and other voltage versions). I want –100dBu signal to noise on the power supplies!
John Stadius: No
Olivier Limacher: AVB (hi quality, low latency audio over Ethernet). It’s being worked on.
Craig Hutchison: Firmware and drivers tend to be complex for modern interfaces and maybe this is just the way it has to be. The modern interface standards documents fill many, many difficult to fathom pages. However, if simpler dedicated chips were available and did what we need, we could be interested.

8. With some of the formats listed in Q7, devices are powered off of the interface bus, or for other reasons like portability, lower voltage supplies are employed than in traditional audio designs.
a) Does this low voltage approach present challenges in the analog circuitry proceeding or following conversion?

Dave Derr: Not yet.
John Stadius: As we currently do not use these formats to transport audio then no.
Olivier Limacher: Conversion and processing/data formatting is more power hungry.
Brent Elder: Yes

b) Do these challenges mean performance compromises or present restrictions in analog performance?
Dave Derr: No
John Stadius: Not applicable
Olivier Limacher: No. Cost (sometimes related to power conditioning) erodes performance more than power consumption itself.
Brent Elder: Yes

c) If applicable, what techniques are you employing inside your products to raise these smaller voltages to higher supply rails?
Dave Derr: We do use a voltage tripler in one product
Olivier Limacher: DC to DC converters.
Craig Hutchison: In the Portico range, we provide an external 12 volt switching supply, mostly for portability, hence the name. For true portability a 12 volt battery can be used. In the chassis, we use DC-DC converters to convert the 12 volts to plus and minus 18 volts and plus 48 volts. Filtering switch noise might be considered a challenge but it is worth it these days with all the international approvals that a product needs.
However, interfacing to even a modest 24 dBv signals means we see more than 34 volt peak to peak swings. Rupert has always been able to use transformers for step-up or step-down to get there. And 29 dBv signals require over 60 volts P2P. Compared to the trend of noisy +5 volt bus rails, Rupert Neve is going in absolutely the opposite direction with 90 volt rails and discrete Class-A circuits, and using substantial sized transformers for step-down and thus lower the noise floor. In our opinion, this is the more appropriate method if the topic is high-end converters and wide dynamic range.

9. DSP Processing:
a) If you are involved in DSP design, why have you chosen the components you use?
John Stadius: All our new products use an FPGA to process the audio. We do use 3 Analog Devices Tiger Sharcs in our audio engine processor card for control and special effects.
Olivier Limacher: Cost, prior experience.
Brent Elder: Performance, value (MFlops/$), aggressive development roadmap for the DSP family.

b) Do you see advantages in one family of processors over another?
Dave Derr: NA
John Stadius: In the past we always used Analog Devices Sharcs primarily for their ease in interconnectivity using their LINK PORTS. There are some great parts now from TI and Analog that have been designed specifically for audio however they are not really suitable for our particular applications.
Olivier Limacher: Not so much as a matter of personal preferences within the chosen range (fixed point/floating point) and familiarity with the peripherals, coding language, development environment.
Brent Elder: ADI’s SHARC has a value and performance edge over TI’s DSP family at this point, and is well suited for audio.

c) Do you see advantages in the design tool sets available for DSP programming for particular families of components?
Dave Derr: NA
John Stadius: Not really
Olivier Limacher: Processing latency and customization weigh heavily in the choice of software architecture and the degree to which generic design facilitators can be employed.
Brent Elder: Not sure – haven’t had much experience with TI’s tools.

10. Have you experimented or employed FPGA processors for DSP tasks (and if you are employing them, why)?
Dave Derr: Not at this company.
John Stadius: We have been developing a single FPGA audio engine design for some years now and it is now in production in our SD7 and SD8 consoles.
Olivier Limacher: Yes. They make sense for high channel counts of simple and non-parametric DSP tasks where very low level signal formatting is also necessary. If some logic is needed to decode a fast digital signal, a section of the FPGA which is doing it will also perform simple DSP tasks (mixing, switching, scaling).
Brent Elder: Yes, we use FPGA’s for lower level processing (summing, metering, gain computations) where there’s a significant amount of repetitive high bandwidth computation that needs to be done.
Craig Hutchison: Just experimenting for now.
Marvin Caesar/Gary Liden: we are still experimenting with them

11. Native Processing:
a) Are their advances in native processing that are now allowing you to perform DSP tasks formerly relegated to dedicated hardware inside a computer’s CPU?

Dave Derr: Yes
John Stadius: This is an area we have not yet looked in to.
Olivier Limacher: Absolutely. Much of the outboard processing needed some time ago has now moved inside the PC. And if it wasn’t for the bloated OS’es that come with said PCs and the tedious nature of optimizing code for every available flavor of CPUs, much more could be brought in.
Brent Elder: In theory, although we do not produce native plug-ins at this time.

b) Do you favor a particular brand of CPUs, and if so, which and why?
Dave Derr: The new Intel quad cores etc are blindingly fast and getting cheap.
John Stadius: No
John La Grou: When we first started to develop the remote micamp (HV-3R), we chose a very powerful and very complex (and very expensive!) processor called an ARM-9. Some months into the process, we realized that the ARM9 was vastly more powerful than we needed. Overkill. We could keep the cost of the product lower by better matching the CPU with our actual product requirements. So we switched to a more conservative CPU but made sure that all of our digital performance parameters remained optimized. Nothing slowed down and all required commands and features (high-resolution metering, etc.) remained in full force without compromise. Reminds me of something Einstein said, “everything should be made as simple as possible, but not simpler.”
Brent Elder: N/A

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

Dave Derr: Generally it’s the designer and not the chip. The big commercial processors are getting so fast that I’m not so convinced that DSP chips will keep up. They are not as market driven.
John Stadius: Not so much DSP but more FPGA. The major manufacturer’s have released or are about to release some absolutely enormous devices.
Olivier Limacher: Cost more than technologies. It’s now possible to equip smaller devices with DSPs, thanks to the cell phone industry.
Brent Elder: Yes, multi-core DSPs, higher performance per dollar (Moore’s law), and dedicated signal processing elements in some DSP intended for audio (e.g. FFT accelerators)
Craig Hutchison: How about that massive parallelism in some of the new video chips – in one case, 480 processor cores on a single chip. And “physics libraries” available to harness the chips for non-video tasks.

General
13. What devices are you employing for system control—do you integrate control into DSP or other processes or are you using dedicated hardware for human interface, i.e. button commands and parameter display?
Dave Derr: PIC controllers
John Stadius: Our main GUI interface is based around Windows XP and multiple touch screens. Work surface control is distributed amongst a number of PIC microcontrollers.
Olivier Limacher: I did both (DSP and MCU). I am very impressed with the new low cost chips based on ARM7 cores that so many vendors now offer. So fast, inexpensive, and well laid out. Their ubiquity is also a plus. High quality open source tools are available for most of them.
John La Grou: Again, simplicity. We’re moving our new products toward menu and software control of pure analog signal paths. We call this AELogic, digitally-controlled-analog. Look for more LCDs and jog wheels to control all analog functions. It generally means using more relays or FET switching where appropriate (no added sonic muck), and fewer panel pots and panel switches. It shortens the audio path by allowing the circuit to stay on the main board rather than making trips up and back to the front-panel.
As for display and interface control, again simplicity is the key – using just enough CPU to achieve a desired result keeps cost, power consumption and emissions down while not restricting performance.
Brent Elder: We use either separate processors or the host for these functions to avoid the extra software complexity in the DSP.
Craig Hutchison: Both old school hardware techniques and Microchip’s PIC controllers, depending on which may be the most appropriate.
Marvin Caesar/Gary Liden: we use dedicated hardware for human interface

14. 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?
Dave Derr: You have to balance between qty discount, and risking too much for inventory. Ramp up productions slowly and always try to avoid borrowing money or bringing in partners early on.
John Stadius: Try not to buy from a brochure unless it is a simple product. Get a good demonstration and asked those embarrassing questions.
Olivier Limacher: Get an eval board. If you know a rep, try to get it for free ?
John La Grou: Listen, listen, listen. We can never do enough to train our ears on subtle differences between products. Develop a consistent A-B test methodology in your own studio to assess candidate products, and work with dealers who let you borrow gear for these kinds of tests (of course, they’ll expect you to buy something now and then).
Ask the manufacturer intelligent questions about their designs. You’ll find that the designers are usually available for Q&A and in fact most of us relish those opportunities to talk with customers and prospective clients. We relish it because we take great pride in the detail that goes into our products, and we enjoy sharing these details with others who care about great sound.
Brent Elder: No
Craig Hutchison: Beware seasonal audio fashions and one-trick ponies. Develop the listening skills appropriate for equipment comparisons and with experience one learns to trust one’s hearing and methods. “Harshness” and “blurriness” are pretty good hints of something not quite right.
Marvin Caesar/Gary Liden: reliability and low power consumption

15. Any parting thoughts on ICs and their application? Trends you’d like to comment on? Devices that we haven’t asked about that you are excited about or are turning to for more of your designs?
Dave Derr: No
John Stadius: The major issue with semiconductor devices today is their longevity. You select the ideal part, spend 18 months designing the product, go into production and the device is made obsolete. The first question we ask the supplier now is not will it do the job but will it still be available in 4 or 5 years time. Re-engineering products due to obsolescence is such a waste of good R&D time.
John La Grou: IC’s, when applied with care, can sound great. But we’ve found over the years that a well-designed, properly applied discrete amplifier can simply sound better, more musical, more real than an IC. Sure, this is a generalization, and there are always exceptions, but in general we’ve found this to be a truism.
Here’s a for-instance: IC’s have a fixed amount of internal feedback which is normally quite high, whereas discrete amplifier feedback can be optimally adjusted to each individual application. Less internal feedback often translates into a more “relaxed,” more convincing audio reproduction.
Brent Elder: No
Craig Hutchison: The latest digital audio converters and clock circuits seem to be finally getting pretty good. It took about 30 years, but it looks like we could all be hearing musically satisfying digital audio in the next 5 or 10 years, ha ha.
Marvin Caesar/Gary Liden: there are parts that we are working excited about, but we would like to keep that confidential.