What exactly does a speaker designer do?
Speaker designer Colin Whatmough explains exactly what it is a speaker designer does.
Many people have asked me “Where do you start when designing a speaker system?” This is a reasonable question, considering the number of possible drivers, different speaker configurations and different speaker sizes.
The actual design process starts long before we have any speaker system in mind. We measure every driver with potential that we can lay our hands on and store these measurements on computer. Whether a driver has potential depends on many factors such as cone material, size and profile. Many drivers that at first appear promising measure badly and are rejected at this stage. Those that make the grade are stored on file for future reference.
ADVERTISEMENT
When we begin to design a speaker system, we must decide what price point, size and configuration we are aiming for. If we plan to design a speaker system retailing for $1000, we are not going to select drivers costing hundreds of dollars each, nor would we look at drivers costing $5 each, not that drivers this cheap are likely to make it onto our reference file.
If we were designing a two-way bookshelf speaker of around 15 – 20 litres, we would opt for a bass/mid driver of about 175mm diameter. A larger driver would be unlikely to suit a cabinet of this size, while a smaller driver would not offer the same bass authority as a 175mm driver. Cabinet size has virtually no effect on tweeter choice. We would choose the best tweeter we can within our budget. If this speaker were to be part of a range of speakers, we would use the tweeter that matches the rest of this range.
The crossover point between the bass/mid driver and the tweeter would be between 2000Hz and 4000Hz. Below 2000Hz would infringe on the all important midrange area (where our ears are most sensitive) as well as minimizing tweeter choice (very few quality tweeters can perform well below 2000Hz). Above 4000Hz, a tweeter should outperform any bass/mid driver, so there is no point crossing over higher than this.
At this stage we can check through the potential drivers we have on file, to find some that are suitable for the system we are planning. Out of twenty or thirty drivers, we generally only find a few drivers which appear to fulfil all the requirements. The likely candidates are re-measured thoroughly to establish frequency, phase and impulse response as well as impedance measurements. All the relevant measurements are stored in our computer for use during the design stage. Having made these in-depth measurements, we re-assess these drivers and choose the most suitable for our new design.
Although the actual design is yet to commence, choosing the most suitable drivers is crucial to the finished product and can be very time consuming.
Using computer-modelling techniques we can establish the optimum cabinet size and shape for our new system. We can also accurately estimate the bass response of the system. A proto-type cabinet is built, fitted with the chosen drivers, and measured. The result should be very close to our computer-modelled response.
We measure both drivers mounted in the prototype cabinet, including their frequency response, phase response and impedance verses frequency. These results are stored on the Computer. From these measurements we can calculate a crossover network circuit. This circuit plays a number of roles. Firstly it must filter high frequencies away from the bass/midrange driver above the crossover frequency and, secondly, filter low frequencies from the tweeter below the crossover frequency. The output of the bass/mid driver and the tweeter must be equalized (this usually involves attenuating the tweeter). While this is being achieved, we must also minimize phase shift. In a speaker with no phase shift at the crossover point both tweeter and bass/mid driver would move out or in together at that frequency. In a speaker with 180 degrees phase shift, as the tweeter moves out, the bass/mid would move in. The units would cancel each other out, causing a large dip in the frequency response.
Once we have modelled the desired response on our computer, we build a prototype crossover network (this is used externally for convenience). The response of this system is now measured. We expect this to be close to our computer-modelled (optimized) response, but it still may need some fine tuning. With the aid of our computer, we will usually achieve a response as good as or better than that originally aimed for.
It is now time to build a pair of prototypes and have a listen. To the uninitiated this is always a let down, as they never sound quite right. What I am listening for is their overall virtues. I expect them to sound a little lean or a little fat and/or slightly bright or dull, but these are minor problems that can be overcome by fine tuning the crossover network or some cabinet alterations. This fine tuning may take days or weeks and possibly months for a large, more complex speaker. The improvements made at the fine tuning stage cannot be over-emphasized. It is often the difference between an average and an excellent speaker.
The new design is tested in various environments; on a number of systems using equipment of an appropriate value (it is pointless testing a $500 speaker using a $10,000 valve amplifier or alternatively testing a high-end speaker with a budget amplifier). When we are satisfied that the speaker achieves its objectives, it goes into production.
-
ADVERTISEMENT
-
ADVERTISEMENT
-
ADVERTISEMENT
-
ADVERTISEMENT