Loudspeaker Designs and Articles for the DIY Enthusiast
A Design utilizing both
Quasi-Transient Perfect
Conventional Crossovers
One of the supposed advantages of a transient perfect design, or at least one
of the things you can measure its performance by, is the ability of the driver's
LP and HP filter to reconstruct a square wave input as a square wave output.
To the right shows the modeled response of the two topologies with a 1 kHz
square wave stimulus. The amazing thing to me is in spite of the differences,
the brain manages to interpret both of these signals as a square wave.

If you don't like looking at square waves, below is the modeled impulse
response of the two topologies. Different picture, same story...  
Above are the measured off axis plots of an earlier iteration of the
design, These QTP and NTP measurements were made consecutively
at each angle so direct comparison between the plots should be
possible. Of significance (Besides the fact that the mid doesn't perform
very well as a tweeter off axis.) is the relatively slight difference in
responses below 2 kHz. The sagging off axis performance of the QTP
design is of course due to the woofer's increased pass band in that
design, but in spite of this is only approximately 3 dB worse than the
NTP at the same angle. I had theorized the sonic differences would be
more apparent in the off axis measurements than is indicated on these
Finally to the left are HD plots of the 2 topologies as measured in the
enclosure. The peak in 2nd order products at 2 kHz is related to the
response dip of the midrange, and indicates a cone edge resonance.
These plots were done prior to the incorporation of the 'B' version of the
W4-1757S into the design. The B version changed the surround and
exhibits less of a response dip in that region, suggesting that distortion
peak will be ameliorated in the current design.   
Links to:
Design Goals and Driver Selection
Cabinet Construction and Drawings
Crossover Design 1
Crossover Design 2
Bill of Materials
Listening Impressions
The plots to the left show the measured system response at on the
design axis, which is about 10 degrees from the on axis plots used in the
modeled plots. Note that the response dip between 2 kHz and 3 kHz as
filled in substantially. These meaurements are gated, so the response
below about 300 Hz should be ignored. However, if you look at the
phase plots for both topologies, the QTP response would be obvious
even without the Verbiage. While these plots were only minimally
smoothed, I did splice two measurements at 3500 Hz. The reason for
this was the rather large distance between the woofer and midrange on
the baffle, and the relatively short, 1 meter measurement distance. The
lower frequencies were measured at a height between the woofer and
mid on the baffle, while the higher frequencies were measured between
the midrange and the tweeter.
As part of the QTP design exercise, I performed lots of 'extra' measurements, and
some were retained and displayed here.  
To the right shows the projected vertical polar response around the crossover
point of the midrange and tweeter. Due to the high transfer functions utilized, comb
filtering is only significant in a narrow range of frequencies around the crossover
frequency and was not audible in listening sessions.