My Thoughts and Know-how

6 Nov 2013

The 2-way Boston Acoustics Classic Series 26 speakers (6" woofer, 1" tweeter) offered mostly faithful reproduction of sound at a low price ($125 each). Measuring their frequency response yields relatively flat pass-band except for an audible notch at 3kHz, resulting from a poor cross-over and destructive interference between the woofer and sound reflected from the back wall of the speaker.

CS26 Frequency response*. Ignore the sub-100Hz part due to a limitation of my equipment.

Speaker/room EQ can easily correct for the dip, so the real weakness of the speaker is lack of deep bass, which here we will correct (I listen to them without a subwoofer).

The speakers' bass reflex port can be improved in 2 ways.:

• The port was originally tuned at 55 Hz. I wanted to bring the resonance lower for deeper bass - basically consciously trade off SPL at 50-60 Hz for reachability of 30-40 Hz. 
• The port is right behind the woofer, passing the mid-range straight trough. These waves then reflect back from objects behind the speaker, creating a noticeable interference pattern, as seen in the below figure. That's quite common with all straight port designs - play a 1 kHz tone through your speakers and move your head around - the notches and peaks will be very obvious.

The bass reflex passes mid-band signals straight-through, which interfere with frontal sound.

The port length is roughly inversely proportional to the square of the resonance frequency. I wanted to bring the resonance to around 45 Hz, or a 50% increase in port length. That would make the port longer than the depth of the speaker. Fortunately, the way I solved this, also mostly eliminated the mid-band pass through problem - I made a turn inside the port, midway through.

Serpentine bass port has lower resonance and blocks unwanted signals. I used sound blocking rubber on the outer side of the turn.

I constructed the new port from 2" ABS (aka gutter) pipes and ABS cement (which actually disolves the plastic and makes the final pieces appear molded).

15 Sept 2013

RX-V2500 is a powerful receiver. The respectable 130W/ch at 0.04% THD is well complemented with the user tweakable 7-band/ch programmable filters that allow for changing their frequency, gain and Q-factor (or YPAO, in Yamaha lingo).

Although the receiver delivers outstanding audio performance, it operates rather inefficiently and has a mild temperature problem. The Yamaha engineers went for a linear power supply and a traditional class A/B power amplifiers making the unit dissipate much more energy as heat than what is delivered to the speakers (at "normal" listening levels). The operating temperature is further exacerbated due to obstructed air convection. The heat channel cools the unit effectively only when the fan spins which never happens in practice, due to an overly aggressive acoustic design point.

The unit is equipped with a fan blowing air from the transformer into a channel made from the aluminum heat sinks of the power amplifiers and LDOs.

I was determined to make the receiver run cooler. Altering the efficiency, particularly at idle, would increase harmonic distortion, so I was left only with the option of modifying the fan rpm vs. temperature curve. I opened the receiver up and studied the fan control circuitry for a tweak. For a surprise, the design, though simple conceptually, was implemented in literally 10x the number of components I would have expected. To make matters worse, the simple fan control was distributed to several boards.

With details abstracted, here is the fan control diagram: