You are here

New Audio Test Gear 2008

The 2008 initiative to improve SPCR's audio testing process involved the construction of an anechoic chamber and upgrades of test equipment to take advantage of the lower noise floor in the chamber. The chamber development is detailed in An Anechoic Chamber for SPCR; this article covers the new test equipment details.

The essence of the challenge was to acquire audio test equipment capable of making full use of the super-low noise floor of the new anechoic chamber. The accuracy of our existing gear was limited to perhaps 18~20 dBA, below which the results could not be trusted. The new anechoic chamber could bring the noise floor down to 10 dBA, so I was seeking audio test gear capable of measuring down that low accurately.

The choice from the start was...

A. Get a new microphone for better recordings and a new SLM to measure lower levels. Discrete tools mean no interdependence; if one fails, the other keeps working. A portable modern digital SLM would be very convenient, as they are small yet extremely powerful. The main problem: Cost. The cheapest microphone suitable for our purpose would be $2,000 and it would take $10,000 to obtain a SLM capable of <15 dBA readings.

B. The alternative was to combine the two functions: A new microphone with a computer-based SLM. A PC-based audio spectrum analyzer can make use of the incredible power of today's computers, and the data input microphone could be the same one used for recording. The cost would naturally be much lower than option A. This was the obvious choice.

EXISTING TEST EQUIPMENT

A Brüel & Kjær model 2203 analog sound level meter made in the 1970s had been the core of SPCR SPL measurements for years. Its absolute accuracy is probably questionable, but it has the capability to read lower levels than most modern digital SLMs priced under $10,000. The meter's range is just 30 dB, but switchable from -10 dB to 140 dB. The main limitation is that the internal noise of the 1" microphone and the meter electronics is about 16 dBA. Any lower than that, the readings are highly questionable. It's likely that we're hitting the limits of this SLM when checking ambient noise levels in the house after midnight. An upgrade seemed unavoidable.



Good down to 16~18 dBA when new.

A Sennheiser ME 66 had been our recording microphone for several years. It is rated for an equivalent noise level of 10 dBA or CCIR-weighted 21 dB, which is very quiet. It also has a very high output level, which is useful in keeping the noise down. The 10 dBA self-noise claim may be exaggerated, however. Research into microphone noise and discussions with technical reps from a variety of microphone manufacturers suggested that such a low self-noise is not achievable with a microphone relying on +48V phantom power.



Highly directional Sennheiser ME 66 mic is designed foremost for voice intelligibility.

The ME 66 has a highly directional ("super-cardioid") pickup pattern that intentionally isolates the subject from the background noise. It is intended for use in the film and broadcasting industries, where it is popular among independent and documentary filmmakers. Its frequency response is not perfectly flat, as the visual below shows. The 7~12 kHz peak is audible; it helps with voice intelligibility. The dropoff below 400 Hz is also audible; it sounds thinner compared to another omnidirectional, more linear mic we have on hand.



The bass rolloff and high frequency peaks are both audible, so the ME 66 is history.

I came to the conclusion after several weeks of research that the Sennheiser ME 66 is not adequately quiet or linear enough to stay with us to the next step.




M-Audio FireWire 410 atop M-Audio Tampa
digital mic preamp: They remain in the new test system.

An M-Audio Tampa professional microphone/instrument preamp with integrated 24-bit / 96-kHz A/D converter is the primary hardware interface between the microphone and the computer software used for recordings and signal analysis. Now discontinued, the specs remain excellent: Noise is rated at 110 dBA. Although its digital signal is fed through an M-Audio FireWire 410, the Tampa is responsible for both signal gain as well as analog-to-digital conversion; The FW410 doesn't actually do anything to the digital signal. The FireWire 410's main function here is for playback and monitoring via headphones. Both M-Audio gear are good enough to continue using for the time being.

At this point, there was a choice:

A. Get a new microphone for better recordings and a new SLM to measure lower levels. Discrete tools mean no interdependence; if one fails, the other keeps working. A portable modern digital SLM would be very convenient, as they are small yet extremely powerful. The main problem: Cost. The cheapest microphone suitable for our purpose would be $2,000 and it would take $10,000 to obtain a SLM capable of 15 dBA readings.

B. The alternative was to combine the two functions: A new microphone with a computer-based SLM. A PC-based audio spectrum analyzer can make use of the incredible power of today's computers, and the data input microphone could be the same one used for recording. The cost would naturally be much lower than option A. This was the obvious choice.

CONTENDERS

Months were spent poring over articles, spec sheets, product brochures and other sources of information on microphones, mic preamps, PC-based audio spectrum analysis software, sound cards, and SLMs. I consulted independent acoustics experts, company representatives and other individuals who had hands-on experience with the gear. To go over all the twists and turns in the road that led to the final equipment choices would be far too tedious for you to read... and for me to write. Suffice it to say that these were some of the brands still left on the list near the end of the search:

Brüel & Kjær - Still a top brand, still very pricey, still one of the very few SLM makers that reach below 20 dBA, still beyond the budget. Their top models easily exceed $10,000 depending on firmware.

LinearX Systems - Developer of the LEAP loudspeaker modeling software (an early DOS version of which I've happily used), their pcRTA is a precision Type-1 true RMS 1/3 octave real time analyzer for PC. The use of an ISA full length PC slot card was the deal breaker; I just could not justify reliance on ISA, which was abandoned by the mainstream PC industry many generations of hardware in the past.

DPA - Mic division spun off from Brüel & Kjær in 1992, now a major quality microphone manufacturer with a huge product range. Their Type 4041-S Large Diaphragm Microphone with Solid State, 130V preamp uses a 1" diaphragm much like that used on the old Brüel & Kjær 2203 SLM, and it is rated for just 7 dBA self noise. It would have been the perfect choice. The ~$3,500 price tag was just too much of a reach, however.

THE FINAL CHOICES

The final choices were made on the basis of suitability, price and availability, with roughly that order of priorities. It's very possible that there are better options at similar or even lower prices, but there was time pressure to get it done.

ACO Pacific PS9200KIT Type 1 Measurement Microphone System

Most of the acoustics experts and engineers who were consulted agreed that a 1" diaphragm mic was probably the only way to make recordings and measurements at the super low target levels. Several people pointed towards the ACO Pacific measurement mic system. It is a kit of mic components that comes housed in a custom carrying case. A 1/2" capsule is the standard option, but the quieter ACO Pacific 7022 1" titanium diaphragm mic capsule was chosen instead. With the preamp, it has a self-noise of under 10 dBA. Low noise is probably the single most important parameter in a microphone when trying to measure and record extremely low sound levels.

The kit came with an individual frequency response plot for the specific mic capsule provided. Interestingly, this 1" mic capsule is 100% interchangeable with the 1" capsule on the old Brüel & Kjær 2203 SLM. The cost was around $2,000. All the available information led me to believe this was the best, yet cost-effective choice for SPCR needs. The complete list of components:

  • PS9200 - 2 Channel 200V Power Supply
  • 4012 - 1/2 Inch Preamplifier w/CA4012-5 Preamp Cable
  • PS9 - 110VAC Power Adaptor
  • WS1 - 3 inch Windscreen
  • 7022 - 1" titanium diaphragm mic capsule
  • 1/2" to 1" adapter for 7022



ACO Pacific PS9200KIT



1" diaphragm capsule with adapter on 1/2" preamp.



Frequency response on axis is within about ±0.5 dB from below 10 Hz to ~18 kHz.

The significance of the second trace is not known.

(Click on image to enlarge.)

Note that the microphone runs on a 200V power supply. The power is fed through a 5-conductor cable/connector, not the usual 3-conductor XLR used with 48V phantom power. This power scheme is apparently one of the keys to the ACO Pacific mic's low noise.

SpectraPLUS - Audio Spectrum Real Time Analyzer

SpectraPLUS is a highly capable spectrum analyzer with a huge array of functions which have been developed continuously for more than a decade. At the same time, it's reasonably simple to learn and use, as the 30-day trial allowed me to discover. There are many audio RTA software packages for use with a PC. Some require dedicated hardware; SpectraPLUS can be used with any PC with a good quality sound card. The M-Audio FireWire 410 and M-Audio Tampa
digital mic preamp used for the last few years are well suited for SpectraPLUS. The complete $1295 package of the base analyzer with all options was chosen to ensure we'd never outgrow its capabilities.



SpectraPlus screenshots.

A few of the features most important for SPCR:

  • 24-bit sampling precision with highest sampling frequency available on sound cards (192 kHz)
  • Calibration, compensation and scaling allows accuracy of SPL and frequency measurements with any good microphone
  • Wide array of displays for signal analysis: Time Series, Spectrum, Phase, 3-D Surface Plot, Spectrogram

    There's far more to this acoustic analysis software package than can be adequately described here. Please check the specifications and features at the SpectraPlus web site. Note that you can download the program and use it free for 30 days.

Landtek ND9 Microphone Level Calibrator

For a SLM or spectrum analyzer to report decibel values accurately, it must be calibrated with the microphone used to a known reference. The commonly used calibration device is a small cylinder which fits tightly over the microphone's diaphragm and creates a cavity of fixed volume. The calibrator emits a tone of 1,000 Hz, at 94 or 114 dB while the system's sensitivity (in SpectraPLUS) is adjusted to give the correct 94 or 114 dB measurement. Standard practice is to check calibration periodically and reset if necessary. IEC942 is the main standard for these devices, which usually combine electronics and a tiny speaker at a fixed distance in a cavity over the mic. The precision of the microphone calibrator in both frequency and loudness is obviously important for SPL accuracy.

Brüel & Kjær offers something called a pistonphone, which in theory should be the most accurate microphone calibrator as it uses a physical piston to produce a 250 Hz tone. The physical dimensions of the piston cavity are the determining factor in the resulting pressure and it is supposed to be less affected by environmental conditions than the electrical circuitry in a mic calibrator. Its price is beyond our reach; even a refurbished current model (Type 4228) was listed recently for $2,400 by a test equipment vendor.

The ND9 Microphone Level Calibrator is a low cost alternative with excellent specifications:

  • SPL Accuracy ± 0.3 dB (20°C, 760 MM Hg)
  • Frequency of 1000 Hz ± 0.1%



It looks like a simple cylinder.



Insert 1" mic on this end; rubber grommet makes tight seal.



Turn 1,000 Hz tone on at 94 dB or 114 dB at this end.

The ND9 came with no individual test results. The SPL accuracy cannot be checked without some other reference, but SpectraPLUS confirms that the frequency is 1000 Hz, give or take 10 Hz, as claimed. The inexpensive $115 unit was purchased from an eBay store called Easy Life Products, which claims to be "a professional Internet seller based in Hong Kong" specializing "in electronic instruments, laboratory equipment and other gadgets." It's clear they are not the manufacturer, which appears to be Landtek of China.

With the help of Professor Murray Hodgson at the Acoustics and Noise Research Group of the University of British Columbia, the ND9 was compared to several other mic calibrators and a high precision digital SLM. In the mix was a Brüel & Kjær pistonphone that works at 250 Hz, 124 dB. Amazingly, all the calibrators in the UBC sound lab gave results with 0.1 dB of each other, and the ND9 calibrator fell within that range. It's certainly high enough accuracy for our purposes.

A NEW AUDIO PC

Last June, the Shuttle Zen which had been our audio PC broke for the second time in three years. Again, it appeared to be associated with the VRM; a bad capacitor problem is suspected. But rather than replace the board this time, a new PC was assembled for our audio test duties. A newer motherboard with more uptodate features and better capacitors might not be a bad thing.



The new audio PC is also a HDD test bed. Note easy access to SATA and power cable atop.

The system was built with the following components:

  • Antec NSK-3480 micro-ATX case - chosen for good airflow potential, and separate thermal zone for the PSU.
  • SilverStone ST30NF 300W - Still our favorite fanless PSU, it runs cool in the Antec NSK-3480 case.
  • Abit NF-M2 nView micro-ATX AM2 motherboard - A discontinued but quite nice product based on the nVidia 6150/NF430 Chipset, it has many features, including passive cooling, Firewire, optical S/PDIF In/Out, 4 RAM sockets, and 4 SATA ports.
  • AMD Athlon 64 X2 4850e - This 45W TDP dual-core CPU running at 2.5 GHz has plenty of processing power for our needs while ensuring low power consumption, especially at idle, which is the operating state 90% of the time.
  • Sycthe Ninja (gen. 2) heatsink with a Scythe SlipStream 120mm SL fan at 500rpm was more than adequate to cool the CPU in this system.
  • A pair of 1GB Corsair CM2X1024-6400 DDR2 RAM ensures plenty of memory for the system.
  • A Western Digital single-platter 320GB Caviar SE16, one of the noisier samples, in a Smart Drive Enclosure sitting atop a bit of soft foam at the bottom of the case ensured that the hard drive noise would be absolutely minimal.
  • Windows XP Pro operating software

There are no fans other than the one on the heatsink, and the only other noise source in this PC is the Smart Drive enclosed WD drive. I'm getting a bit ahead of myself here, but its measured SPL (with the side cover on) is under 15 [email protected] in the new anechoic chamber.

One of the really handy aspects of this Antec case is that it pulls apart easily without tools if the top piece is just friction-fitted into place. (See the details in our review.) The top is actually left off most of the time, as there's no noise from the fanless Silverstone PSU. The cables up top make this system very useful for hooking up hard drives temporarily for testing, formatting, wiping, etc.

OTHER TEST GEAR UPGRADES

Long time readers of SPCR know that we have a broad range of test gear, not only for audio analysis but also thermal and fan testing. The precision and suitability of some of this gear is not beyond reproach. A decision was made to go ahead and add a few more items for a wholesale improvement to all our test processes.

Kanomax 6803 Anemometer

The SPCR lab is already equipped with with two anemometers for testing fans. The first is an inexpensive vane anemometer which proved imprecise at lower airflow levels, and subject to strong variations depending on the twisting. cyclonic action of fan airflow. The second anemometer is a more serious tool, an Extech Model 407123 hot wire device described in detail on page 6 of "A New Way of Testing Fan Airflow". While the Extech is a significant advance from the first vane anemometer, its 3% accuracy is assured only to 40 ft/min velocity, with a resolution of 10 ft/min.

Part of the problem is that we're seeking precision and accuracy down to very low airflow, but anemometers are generally designed for checking weather conditions or for heating, ventilation and airconditioning (HVAC). In both applications, the airflows of concern are many times higher than that of slow spinning fans for PC cooling.

The Kanomax 6803 is another vane anemometer, but its precision and resolution is far better than either of the other ones. It is described as a tool not only for HVAC but also for "Laboratory Control, Cleanrooms, IAQ Investigation, Industrial Hygiene and Quality Control." The specified accuracy is within 1%, and the resolution is 1 ft/min. Like the ACO Pacific microphone, this $650 tool came packaged as a kit in a tough carry case. The most impressive part of the system is the Pacer 275 vane probe, formed from machined aluminum or steel. Its bearings have such low friction that just walking while holding it is enough to get the vane blades spinning, and for the meter to produce readings. The vanes also seem essentially impervious to variations in reading due to the cyclonic action of fan airflow, but this has to be confirmed with more extensive testing.



Kanomax 6803 vane anemometer kit.



The 2.75" diameter vane probe with short handle, attached via cable to the anemometer.



The low friction of the calibrated Pacer 275 vane probe is astonishing.

FAN RPM MEASUREMENTS

A laser Neiko Tools USA digital tachometer has been used to measure fan RPM during testing. It works by pulsing a tight beam of light against the rotating element. A piece of reflective tape is affixed to one of the blades, and the tachometer measures the rate at which the light beam is reflected back. This works fine as we have access to the blades, but there's been no way to confirm the measured RPM. The accuracy is claimed to be 0.05%, but this is a claim I'd take with a LOT of salt.



Inexpensive laser tachometer with reflective tape: 0.05% accuracy?!

Another RPM measuring device was needed to check the results from the laser tachometer. A strobe with calibrated pulse rate seemed like a good option as it requires no contact with the fan and not even a piece of reflective tape, which is useful in some cases where the blades are not accessible. The item chosen was another from the eBay store, Easy Life Product. This $120 AC powered handheld device is claimed to have the following key specs: Resolution : 0.1 FPM (50 - 999.9 FPM), 1 FPM (over 1000 FPM); Accuracy : ± (0.05%n + 1d). Again, the claim is difficult to accept.



Handheld strobe as backup for laser tachometer.



The flash rate is displayed.

Here's how the strobe is used to measure fan speed: The strobe is turned on and the light pointed at the spinning fan blades. Sometimes it helps to dim the ambient light. The speed knobs are adjusted until the fan blades appear stationary. This can occur when the flashing rate is at any number of multiples (or fractions) of the actual fan RPM. In other words, when 1043 RPM is displayed and the fan blades appear stationary, the actual fan speed could be 522, 1043 or 2086 RPM. Affixing a piece of reflective tape on a blade makes it more positive: The tape appears stationary only when the strobe flash rate matches the fan RPM.

So far, testing comparisons between the two fan RPM measurement devices shows good correlation; often dead on, and usually within 5%. Which is the more accurate is difficult to gauge, but the alternative measurement tools are reassuring.

* * *

FUNDING THE PROJECT

In April 2008, I asked the SPCR community for help to fund this project. Corporate sponsors were also contacted. Both readers and corporate sponsors responded generously, with cash donations from readers, and product gifts, randomly awarded to the individual donators, from the corporate sponsors. It was a successful fundraiser; some $11,000 was raised for the project in about one month. The progress of the fundraiser was documented as it unfolded in this lengthy forum thread.

There are too many individual contributors to name them all (some 300), and most want to keep their privacy. To all of you: My heartfelt thanks! Without you, this project would not have gone forward.

The corporate sponsors also deserve recognition, thanks and praise for their part in offering giveaway prizes for individual contributors. Their support is much appreciated by everyone involved with SPCR!

* * *

Other SPCR articles of related interest:

An Anechoic Chamber for SPCR

Audio Recording Methods Revised


SPCR's Test / Sound Lab: A Short Tour

SPCR's Fan Testing Methodology

* * *

Discuss this article in the SPCR forums.

Sections: 

Google

www SPCR