Meeting Topic: Sound System Design for Automobiles
Moderator Name: Rob DiVito
Speaker Name: Patrick Dennis, BSEE, Principal Engineer, Nissan
Other business or activities at the meeting: Blair Francey welcomed everyone to the meeting. He thanked the evening's sponsors: Nissan, HHB, and Sonotechnique; as well, Ryerson University for the facilities, and the support of all the section members. He also thanked everyone for paying their dues. He noted Doug McClement will present at the April meeting entitled Tales of Sochi. He then turned the proceedings over to Rob DiVito.
Meeting Location: Ryerson University; Rogers Communications Building; Toronto, ONTARIO
Rob briefly stated his motivation for considering car audio (like many of us using it as a testing ground for mixes and masters) as a meeting topic before introducing Patrick Dennis and providing his background. As part of a team Mr. Dennis is responsible for patents for the 'Audio Broadcast Processing Method' and one pending for the 'Vehicle Voice Interface System Calibration Method'
Mr. Dennis thanked everyone for attending. His presentation offered a broad overview and was accompanied with visuals including photos, charts and graphs.
Before beginning his actual presentation, he began by going over a brief history of car audio starting with the first (expensive) radio ($130 compared to $540 for the car itself). One of the milestones happened in 1982 when Bose developed the first premium car system. He briefly discussed car audio associations.
Since the majority of people listen to music in their car, according to surveys, the goal of car audio systems is to reproduce recorded audio as accurately as the original or artist intention.
The presentation was in two parts: the first dealing with audio system design; the second dealing with testing, philosophies, and 3D audio systems.
Topics covered under system design included imaging, challenges of design, speaker design and packaging (critical for good sound quality), and NVH (noise, vibration, harshness) considerations.
Some more detailed points:
Under the topic of imaging, the target for the phantom centre is between the steering wheel and the centre of the dashboard.
Some of the main challenges of car audio design is that while listening positions are known, the listener is not seated ideally with respect to speaker location. Off-axis response, time delays, and speaker location are used to optimize not only imaging but perceived frequency 'flatness'.
Speaker design topics included driver and woofer design; packaging topics included location, mounting, and grills.
Infinite baffles are the most cost effective for woofers as far as design considerations are concerned.
Speaker placement plays an important role in imaging. Not only left-to-right characteristics are considered for imaging, but also front-to-back.
Depending on the reproduction range, there are optimal locations woofers, midrange, and tweeters.
Woofers have the most freedom for location as their range is non-localized, can use corner loading for the most efficiency, and can take advantage of cabin gain. Cabin gain happens when half the wavelength is longer than the dimension of the interior cabin, typically increasing 12 dB/octave below this point.
Cavity resonance is a concern with the midrange as this cannot be 'tuned' out with equalization. It's mounted as high as possible to promote better imaging.
While DSP algorithms have been developed, the best practice is to mount speakers properly. Amplifiers contain parametric filters with independent left/right controls to eq appropriately for the passenger and the driver. Time delays are also used to optimize imaging. Loudness parameters are provided (Fletcher-Munson curves) to compensate for ear sensitivity at different levels.
Mr. Dennis closed the first half with a summary.
Before the break, there were a few questions. The responses:
Studies are planned in the coming months to test and validate the imaging design considerations, and accurately determine listener's wants.
User controls for audio compression are not employed as the manufactures don't want to 'confuse' the consumer.
Nissan typically do not use ferrofluids in their tweeters.
A brief discussion of noise cancellation systems ensued when noise floor (as high as 105 dB) was brought up and its impact it has on the audible frequencies.
After the break, testing was discussed. This consists of both objective level and subjective level varieties. Many objective level tests are done to convince management of design decisions!
The target response curve is not flat. The Localized Sound Power Method is used. Because of the great variations in the midrange results, a 6 mic array is used to get an average, repeatable response
Responding to an audience question, no phase coherency tests are done.
For Subjective Level Tests a good home audio system is the target response. One method used is the LiT Rating Method which scores 5 categories: tonal balance, spatial reproduction, time distortion, dynamics and distortion, and noises and distractions.
Other points noted: Driver's seat takes priority since that's guaranteed to be filled 100% of the time. Priorities for the passenger is balanced frequency response.
For audio system tuning, the goal is to maximize reproduction so that it is as close as possible to a real audio event. Controls available are parametric EQ's, delays, polarity inversions, and channel gains.
He discussed the two basic philosophies for tuning: one has independent L/R for the front channel EQ. This improves imaging for the driver's seat though ultimately actual listening tests are the final judge.
The other philosophy has common L/R front channel EQ. This makes tuning easier and requires less internal memory.
The tuning guideline goal is to bring out artist's intentions. EQ cuts are preferred, but overall level must then be monitored to maintain consistent before and after levels.
On 3D audio, Mr. Dennis stated 5.1 is dead (as the industry doesn't want to spend money creating these mixes), manufacturers are creating surround by extracting info from source and sending that to the appropriate channels.
He concluded by briefly going over various systems from different manufacturers including Harman JBL's professional quantum logic surround and Meridian's Audio System.
More responses to questions:
Pink noise works best as source material. Impulse response measurements are also used. He added they're moving away from binaural recordings because one is tied to the track one's recorded.
Interior design does get changed occasionally to accommodate audio, but it does depends on the importance of audio for the product, and on the budget. Audio is still a secondary concern, unfortunately.
Active noise cancellation systems are used typically involving polarity inversion. He noted another thing the audio system is used for is noise enhancement, for example, making a four cylinder car sound like a V8!
In-vehicle testing is limited as few test prototypes are built and many parties require to do their own tests.
Once the work is complete on the prototype, no changes are done on production models basically. The window for development is very narrow.
Different tunings are required for different interiors, leather as opposed to cloth. So the same car model with different interiors would have different tunings.
With active noise monitoring, and real time monitoring of the car interior, adaptive tuning systems have also been implemented, notably Bose' audio-pilot algorithm.
On the after market, it all comes down to styling. Also the system installed is tuned to the car. Swapping components can do more damage than good.
Finally, they do not yet have modelling software that re-creates the interior of the car. Cost is the major issue. But once implemented, it will improve results by providing more objective results during the interior design phase.
Blair thanked everyone for attending. He presented Mr. Dennis with a Toronto AES Certificate of Appreciation and coffee mug.
Written By: Karl Machat