As a teenager completing my summer job with most of the income available for my DIY audio hobby, I had the desire to build a 2 way high end monitor/bookshelf speaker. Having limited confidence in my own abilities, I looked for a proven high end design. A speaker design published in Speaker Builder magazine caught my eye. The measurements were impressive as were the transducers. As the article appeared in a print magazine, I assumed it was considered to be a very good design.
I thought these transducers were very expensive at a cost of over $600 for a set to build a pair of speakers. To ensure all parts of the speaker were of appropriate quality, I spent another $100 on high end crossover parts and meticulously duplicated the crossover in the magazine article. Everything was double checked and upon completion, I was quite certain that I had built a very high end pair of hi-fi monitors.
The next step was to listen to my reference CDs through the speakers. Initial opinions were positive. So much detail and transparency! It’s as though they’re performing right in front of me! Definitely better than any DIY speaker I previously built.
The initial listening level was quite low. I turned it up but the sound began to hurt my ears. It was fatiguing. There was very little bass. As I listened to additional CDs, it sounded as though some frequencies were missing. I checked the crossover wiring for a third time. Auditioning of these speakers carried on for months and the speakers were only used at low listening levels. They unfortunately ended up gathered dust in the corner as I went on to build another speaker system.
Let’s fast forward to when I started working with Paul Barton at PSB. We were using the NRC anechoic chamber every few months and I had asked Paul if it was ok to bring up a set of speakers to measure during one of our trips to Ottawa. At the end of one of our days at the NRC, I measured both left and right high end DIY speakers in the anechoic chamber. The measurements were at least consistent among the left and right speakers.
Shock was my initial reaction followed by relief because my ears were not lying to me. Paul laughed and no comment was required to confirm that the performance of these speakers was pathetic especially compared to the $250/pair PSB speakers we were working on at the time!
The transducers may have independently been excellent performers. The author of the article obviously designed a much better speaker than I had built. Somehow my build did not match his and I was not interested in spending more time to investigate so I tore apart the speakers and sold the transducers.
After many years in the industry and looking back on this experience, my lessons learned are as follows.
Transducer manufacturers sometimes change the design and materials used in their products. This may be due to a material sourcing issue, cost, consistency, quality, etc. and they may have no choice. The change may affect the frequency response. Even though that transducer change may be positive, the change may negatively impact the accuracy of the speaker that was designed with the initial driver.
The manufacturing tolerances guaranteed by even the best transducer manufacturers from batch to batch can still be too large to achieve appropriate sound consistency. Sensitivity and frequency response tolerances can be evaluated separately on their production line and result is a larger combined frequency response deviation from the reference. On the other end of this continuum, no acoustic or electrical quality assurance is in place during manufacturing of the transducers used in some of today’s popular low cost speaker kits.
A computer model of a speaker design will usually not exactly (>1+db deviation) match a physical measurement. One should not expect their DIY speaker build to perform exactly as a computer model indicates.
Passive component manufacturers do not all use the same equipment to measure primary component values. For example, if a design requires a 10uF electrolytic capacitor as listed on the component, a seemingly equivalent part may measure differently because the measurement frequency is different, the original part is on the low or high side of the +\-10% tolerance, the designer or manufacturer measurement equipment is out of calibration or a combination of all these factors. Even if this primary value is correct, there are still other specifications that may not be matched. In this case, a sufficient maximum voltage rating and matching of the equivalent series resistance are important.
These potential deviations can all be migrated. Measuring each component, transducer and crossover before it is packed into a kit (as we do) will ensure your speaker closely matches the sound of our reference speaker. Evaluating actual kit measurements as opposed to computer model results is important.
Why did my high end DIY speaker build ultimately fail? While I’m not able to confirm this, I suspect the materials in the transducers changed which caused a change in the frequency response. There were likely component primary and secondary value discrepancies as I did not have the equipment to measure the values at that time. Wish I could have purchased a complete professionally engineered and fully tested kit instead!