These are notes I distributed at an AES/Audio Club lecture, ca. 1987.   They found their way onto various audio hobbyist and DIY sites and were popular for a few years.



by Ken Kantor, NHT (~1987)

I. The Amateur (or "Blind Faith") Approach

1. Are you sure you really want to build a loudspeaker?
Consider your conscious motivations: because all commercially available speakers are overpriced, or, because you have a radical theory that you are certain will advance the state-of-the-art, or, because you think it will be educational and fun. Each the result of essentially flawed reasoning.
2. Define the desired performance specifications.
Carefully study the published specs on IRS’s and WAMM’s, then basically figure out how close to perfect you want your new speaker to be.
3. Determine number and size of the drivers required.
Meditate on the desired specs, ruminate on your personal design philosophy (minimalist vs. go-for-broke), deliberate about subtle crossover points. The number and size of the drivers you will need will come to you in a flash of intuitive inspiration.
4. Buy the best drivers you can afford from a mail-order catalog.
Always believe the manufacturer’s data, even if it seems a bitover-optimistic. Tests show that expensive European drivers with impressive cast frames sound the best.
5. Choose enclosure type and figure out dimensions.
Sealed, vented, transmission-line, isobaric, passive radiator, infinite baffle, bandpass and dipole are viable choices; you alone must decide which method is clearly superior, but vastly underrated by the "mainstream" industry.
6. Pick crossover points.
If the catalog says "midrange unit: 200 Hz to 6,000 Hz", you are home free. Try to dig out that old Audio Amateur article that proved you can eliminate all interference lobing up to 25 KHz with the proper combination of quasi-first-order networks and phase-lag compensation. Ignore input impedance.
7. Build it.
Be sure to use the densest material available to you, and brace generously. Weight equals quality in the world of speakers. This is a great opportunity to buy some neat new power tools.
8. Listen to it.
Don’t panic. Tower’s theorem proves conclusively that there exists somewhere a recording that will make any speaker sound wonderful. This is a great opportunity to buy some neat new CD’s.
9. Reverse the phase of one or more drivers.
Depending on whether you prefer an excessive peak or an excessive dip at 3,000 Hz, and whether you like a more "defined" or a more "spacious" stereo image.
10. Objectively verify your design.
Compare it favorably to your friend’s $2,500 audiophile speakers in every area but the "lower mids".
II. The Professional (or "Mid-fi") Approach
1. Consider introducing a new loudspeaker.
Because sales of existing products are slow, or because you see a hole in the market "a mile wide", or, because your dealers are bugging you for a "full-line", or, because you have a theory that youthink will fly in the underground press. All the results of capitalistic greed.
2. Define your budget.
Engineering wants a higher budget, but sales wants a lower retail price. Sales always wins until Accounting does a profit analysis 6 months later and demands immediate price increase.
3. Define target performance specifications.
Analyze the competition. Subtract a few Hz from their low-end cutoff. Multiply their power rating by 1.5. Most importantly, add 10% to their dealer margin. Also, be absolutely positive that each and every spec is better than your less expensive products and worse than your more expensive products to avoid endless dealer and customer inquiries.
4. Determine number and size of the drivers required.
Fuss with the budget, ponder the competition, second-guess the market; then pick the number and size of the drivers you will use. Analytical approaches like Thiele-Small modeling and computerized directivity analysis may be helpful, but the answer is almost always 2 or 3.
5. Submit your driver specs to your supplier or Production Dept.
Complain because your 6.5" woofer costs you more than their 8" woofer costs them. Complain because, despite the price, the factory refuses to guaranty Qts of Fs to better than 20%. Complain because delivery is 16 weeks. Find a new supplier, fire the chief engineer. It doesn’t help.
6. Choose your enclosure type, finalize dimensions.
If your company’s very first speaker product was a sealed design, dealers and press will expect a sealed system, otherwise they will expect a vented system. And that’s that. Using driver response and parameters measured from pre-production samples, reconfirm enclosure dimensions and finalize crossover points.
7. Design crossover circuit.
Try to achieve target response with one capacitor and, if absolutely necessary, an inductor per driver. A few resistors are also allowed. Whatever happens, call the input impedance "8 ohm nominal".
8. Build and evaluate prototype.
Weeks of listening tests. Hours and hours of measurements to try and comprehend the listening tests. Call on key dealers, reps and friends for input. sort through mountains of contradictory advice and conflicting opinions. Attempt to fix any obvious problems by adjusting the crossover, since 5,000 drivers are already on order.
9. Produce and sell the new loudspeaker.
If the engineers are within 20% of their original budget, they are happy. If the salespeople are within 50% of their original sales forecast, they are amazed.
10. Send the product out for review.
Stereo Review says, "Of all the loudspeakers we have ever tested, this is certainly one. Compares to speakers costing more."
Stereophile says, "Not bad for a cheap American speaker." "No, I’m sorry to disagree completely; it didn’t seem at all American in flavor."
Consumer Reports says, "82%."
AUDIO just prints Heyser spirals.
III. The NHT (or "Ideal") Approach.
1. Begin a new loudspeaker development project.
Initial product definitions arise from quarterly product meetings.  Considerations include market conditions, dealer and customer requests, technical developments and individual ideas. Schedule and cost targets are established. size and appearance approaches are suggested.
2. Preliminary engineering review.
Engineering Dept. studies project details. Development schedule and budget are outlined. A preliminary Bill of Materials showing all parts required to build project, along with estimated costs, is generated. Labor times are estimated. Rough performance specifications are drafted.
3. Official product development begins.
Engineering data is reviewed by Marketing, Production, and Finance Departments. If approved, a product development form is signed and development funds are allocated. Marketing presents sales plan and forecasts.
4. Detailed engineering review.
Computer simulation techniques are used to find required enclosure dimensions, driver parameters and crossover network topology.  Existing drivers are evaluated for applicability to product; any new technical approaches are tested in the lab. All necessary prototype enclosure, crossover and driver assembly drawings are produced on CAD system.
5. Purchasing begins.
Prints are delivered to our cabinet factory for sample construction and cost analysis. Prints are sent to various driver and crossover vendors, along with engineering suggestions about construction methods. Retail price is fixed. Typically, dealer cost is estimated at 50% of full retail price, and internal product cost (including labor, etc.) at 50% of dealer cost.
6. First prototypes constructed.
In 4 to 8 weeks samples are received and first prototypes are assembled. detailed listening tests and measurements of all types are conducted for at least 6 more weeks. Frequently 2nd or 3rd driver samples are required. Drivers are finalized first and production orders placed. This leaves about 1 month to tweak the crossover before its production order must be placed. Gradually, prototypes are refined to finished level, and final Bill of Materials is
7. Production engineering and pilot run.
Production Department plans assembly, test and packaging procedures.  When parts are available, 50 pieces are assembled as a pilot run. These are checked for performance verification. Final documentation is generated, including assembly drawings, parts list, reference curves, owner’s manual. Several pieces are supplied to packaging vendor for design and test shipping of carton.
8. Pre-selling phase.
Product is publicly announced. Sales forecasts are refined. Pilotrun units are used for sales, photo and press samples.
9. Mass production and distribution.
Production parts arrive and are assembled typically 6 to 8 months after completion of Phase 3, depending on product complexity and the number of driver sample iterations required. Monthly production per product ranges approximately from 200 pieces to 2,000 pieces.
IV. NHT Design Philosophy.
1. Good performance in a loudspeaker means the ability to recreate in a home environment, as faithfully as possible the musical and ambient information contained in a broad variety of properly made recordings.
2. Loudspeaker design should seek to balance a variety of important performance factors, rather than optimizing a few.
3. Psychoacoustics is the key to meaningful future loudspeaker improvement, through the reconciliation of objective and subjective product assessment.
4. All aspects of performance and theory need to be constantly re-evaluated; each new product should be designed individually and without an unconsidered reliance on existing products and technol
5. Product performance should be justifiable based on accepted technical standards; marketing driven engineering trends are to be avoided.
1. Providing a first-arrival that is spectrally flat from 200 Hz to 5000 Hz is the single most important job of the speaker designer. First arrival must be measured in a listening room with a high resolution time-domain system, exponentially-windowed FFT at 2 meters.
2. Obtaining a balanced steady-state spectrum at the intended on-axis listening position is also very crucial. This is determined using a swept 1/3 octave pink noise signal and a voltmeter. The goal is ruler flat response from 150 Hz to 20 KHz. Response below 150 Hz is boosted by 1 dB or so to obtain a better subjective balance in the case of an LF cutoff above 40 Hz. Deep bass is measured using near-field swept sinewaves in the listening room.
3. Short-term reflections arising from within the cabinet, within the driver or from external cabinet features have a significant effect on perceived tonal accuracy and imaging precision, and should be minimized to the greatest extent possible. This is best accomplished through careful cabinet and driver design, and through unconventional cabinet shapes.
4. Longer-term reflections from room boundaries, while unavoidable,can be reduced somewhat by the careful selection of cabinet shape and system radiation pattern. Angling the primary radiation axis away from room boundaries and towards the listener, and restricting the midrange polar pattern to the extent compatible with item 2 (above), gives significant sonic benefits.
5. Reducing IACC results in better stereo imaging and a subjectively preferable ambience presentation. This can be achieved with no penalty by aiming in the speaker radiation to arrive at the listener’s head at an azimuth angle of close to 26 degrees. This has an extra benefit of stabilizing the stereo image against head movement,
through time-intensity trading.
6. The burden of spectral accuracy is on the drivers. With properly designed units, first and second order crossovers can usually achieve excellent results. Specialized higher-order network topologies are generally not a cost-effective contributor to speaker sound, and can even degrade woofer damping.
7. A sealed enclosure is clearly preferable in systems with woofers up to 8". Vented systems can obtain a lower F3 for a given size and efficiency, but are vulnerable to infrasonics and are subject to detuning with parameter drift. for these reasons avoid vented systems with an Fb above 35 Hz.
8. Comparatively benign performance factors include phase response, time alignment and THD. No clear and consistent advantages can be attributed to any particular diaphragm construction material. Expensive crossover components that do not obviously alter the net work’s voltage magnitude response do not improve (or degrade) the sound in any way.

Post a comment or leave a trackback: Trackback URL.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: