On DIY Speakers 101 I mentioned finding SQB4 alignment giving the inexpensive DC160-8 mid-woofer a nice response starting from 47 Hz. Since then I've been playing around with alignments and custom configurations. For each I was able to see the frequency response, the phase, the group delay and the step response. Although I am capable of making sense of all, I lack the experience to make a good compromise. Needless to say, compromise is everywhere and speakers are no exception.
Incidentally
this is how I came across Paul Carmody's free designs. For those unaware of his existence, Paul is a well-known loudspeaker designer, famous for the Overnight Sensations, Carrera, Classix II and the Hitmakers. Guys that have made speakers before and know what they're talking about are writing good impressions about his designs. Moreover, his Classix I and II models use the same Dayton speaker I found for my project, which is great. That's the experience I've been looking for.
Paul's Overnight Sensations and Hitmakers are monitor-like speakers with impressive sound stages but rather fatiguing to listen after a while. Paul did mention that he couldn't get much work done around his OSs and in my experience this means they're reproducing too much detail, very articulate voices, crisp instruments and so on. Undoubtedly they're great, but not exactly what I'm looking for. So, after evaluating all candidate designs, I chose to give Classix II a try. This vented configuration looks promising: a response starting from below 40 Hz at half power, a good sound stage and just enough voicing not to be annoying. The speakers are among Paul's favourites and that means something.
At a first look one may observe that his box has 16 litres, roughly twice the volume suggested by SBB4 or SQB4 alignments. More volume means lower bass at the cost of poorer transients. Again, compromises! Paul's remark "filling... probably fattening, too" may be suggesting that. Such a speaker tends to have an out-sticking fat bass. He concludes their description by saying that they're " so much fun!". I could go on dissecting his words but I'd rather check-out the facts.
Frequency response
Estimating the volume of bracing, crossover and speaker backs at around 0.65 liters, we get an effective volume of 15.35 l. He described the vent to have 38.1 mm in diameter and tuned to 40 Hz, meaning it is 109.8 mm long. There's no mention about the Ql for his 19 mm thick MDF construction but that's generally measured at the end of the build. Choosing a typical value of 7 we get the following response:
The Classix II model for an estimated Ql of 7. Total frequency response of the system.
This yields a slightly non-linear response characterized by a 1 dB variation, with a local peak below 0 dB at 47 Hz, a minimum at 90 Hz and rising asymptotically to 0 dB. Not ideal, but this model reshapes the DC160's SPL to a more linear response. Smart! However, looking closer at his model - the thick grey curve - we see a small hump but no minimum whatsoever.
The Classix II model for an estimated Ql of 15. Total frequency response of the system.
A Ql of 15 - more appropriate for low volume boxes - gives a similar curve with different values. The peak is now above 0 dB, correcting the speaker's response further more. However, the shape still doesn't resemble the one in his designs. Either I am wrong or he simplified the drawing to eliminate the model non-linearity. Ignoring the minor inconsistency, for the remainder of this analysis I'll assume Ql = 15 solely because it gives a F3 in the mid 30s, 36.93 Hz to be precise, as Paul said his box does.
Transients
Classix II manifests a maximum group delay of 19.5 ms at 40 Hz which, compared with the 12.2 ms given by the SQB4 alignment, seems rather high. The local minimum at 20 Hz is 9.7 ms, giving a difference of 10 ms between the two over a 20 Hz gap. The SQB4 manifests an insignificant 3 ms wiggle over the same frequency range. Anyway you look at it doesn't seem that good. The shape of the curve is the same for both but that's normal for all vented designs.
The step response isn't that impressive either. The oscillation stops to a intelligible zero after 85 ms, while the SQB4 gives a slightly more damped response, grinding to a halt after 65 ms.
The conclusion
is, as expected, that nothing comes for free and compromises need be made. Modeling the box for a deeper bass comes with poor transients but surely Paul Carmody was aware of that. It must have been the experience gathered over the years that made him choose this design. He knew how far to go correcting the speaker's SPL, how much group delay is too much and the level of humanly perceivable transients.
In his wisdom, he predicted that if his design turned out too much of a compromise, he could block the vent and still get a good speaker. A sealed box of the same volume gives a Qtc of 0.5 - perfect transients - with a 3 dB cut-off frequency of 60 Hz, 6 ms of group delay and a superior step response. Thumbs up for that!
Update: Check-out the build process and initial listening impressions.