Dayton PK165-8 with Audax TW010E1
The Audax TW010E1 is not the best choice for the Dayton PK165-8. The reason I’m doing this is simply out of convenience. Since I already have the Dayton PK165-8 mounted in a bass reflex, it’s the quickest way for me to listen to the Audax TW010E1. The problem with this arrangement is I will have to sacrifice some sensitivity on the PK165-8 to match the Audax. Worse than that, crossing the PK165-8 at 3.2kHz is less than optimal. The cone breakup will surely interfere with the crossover. Nonetheless, here goes.
Fig 1 – Blue plot = PK165-8. Red plot = TW010E1
I used a 1st order crossover (Fig 1) just to have a feel of the two drivers. As I had expected, the cone breakup of the PK165-8 is intruding into the crossover region. This will certainly impact the summing of the two drivers. By right, the PK165-8 should not be crossed above 2.5kHz if quality sound is the objective.
Fig 2 – Black plot = Summed Response
The Black plot in Fig 2 is the summed response of the PK165-8 with the Audax TW010E1. Due to the high crossover frequency of 3.2kHz, the PK165-8 cone breakup has actually lifted the treble.
Fig 3 – Dayton PK165-8 and Audax TW010E1 Frequency Response
For clarity, I removed the Low Pass and High Pass responses, leaving only the summed response (Fig 3) of the PK165-8 with the TW010E1. In the treble, the bandwidth from 4kHz~8kHz is a bit excessive. As I mentioned earlier, this is not from the Audax tweeter. It is caused by the cone breakup of the PK165-8. Generally, with a crossover like this where the midwoofer cone breakup intrudes into the crossover, one can expect some brittleness.
1st order Sound Quality
I spent a few days listening to this 1st order network and came to the conclusion that it’s not too bad. Obviously it’s not the right way to do things but the speaker doesn’t sound atrocious. I only picked up a hint of brittleness in the treble during auditioning. Is the speaker good enough to use?
Most certainly, especially so when the budget is tight. This speaker will outperform most ceiling speakers in a line distributed system. Other applications include houses of worship, food outlets like restaurants, diners etc and even in the home as surround speakers for movies.
Improving the Sound
My next step is to try to get a cleaner summing by using a 2nd order filter for the PK165-8. I’m leaving the Audax TW010E1 as a 1st order because I need it to extend down to 3.2kHz.
Fig 4 – Blue plot = PK165-8 with 2nd order Low Pass filter. Red plot = TW010E1 with 2uF.
Fig 4 is of the PK165-8 with a 2nd order low pass at 3.2kHz (Blue plot). The Black plot is the new summed response. The improvement in the crossover region is significant. There is only a slight peak at 4kHz. This comes from the first peak as the cone is rolling off. Subsequent peaks are sufficiently suppressed resulting in no artificial treble boost.
Fig 5 – Frequency Response of PK165-8 with TW010E1 with Asymmetric Crossover
The plot in Fig 5 is the frequency response with a 2nd order Low pass for the PK165-8. I removed the low pass and high pass plots of the PK165-8 and the TW010E1 for a clearer view. In this plot, I decided to spliced in a Nearfield response of the PK165-8. This is a more accurate reflection of the PK165-8 as the earlier ones recorded room reflections.
Fig 6 – Null response with TW010E1 wired in absolute phase
Fig 6 is with the TW010E1 wired in absolute phase. Considering the cone breakup, the null response is actually quite acceptable. Though it’s not a clean null, a deep notch is still recorded at about 4.5kHz. If the 4kHz peak is not present, I expect to see a deeper null somewhere between 3.5kHz~4kHz.
Fig 7 – Step Response of PK165-8 and TW010E1 with Asymmetric crossover
The Step Response in Fig 7 clearly shows the acoustic centers of the two drivers are not aligned. The first transient is the TW010E1 tweeter, followed by the next peak which is the PK165-8. Looking closer, it appears the PK165-8 is late by about 250 microsec. That works out to 3.375″ displacement. The easiest way to align the acoustic centers is to move the PK165-8 forward by 3.375″.
Fig 8 – Cumulative Spectral Decay of PK165-8 and TW010E1
The PK165-8 with the TW010E1 waterfall plot in Fig 8 doesn’t look too bad. There’s very little artifacts from 6kHz onwards. The bulk of it seems to be around 4kHz. This is quite close to the crossover frequency of 3.2kHz. My guess is that they are caused by the PK165-8 cone breakup.
Fig 9 – Toneburst Energy Storage
The Toneburst Energy Storage plot in Fig 9 shows excess energy at about 4kHz. This is in line with the waterfall plot in Fig 8. Fortunately, it doesn’t last very long. Only up to 10 cycles.
Fig 10 – Spectrogram
The spectrogram in Fig 10 shows the excess energy in the time domain. At 4kHz, the excess energy is dissipated by 2 msec. Basically, the smearing in the tweeter is insignificant.
Fig 11 – Harmonic Distortion
Fig 11 shows the 2nd (Red) and 3rd (Violet) harmonics. Generally, they are about -45dB below the fundamental. What is of some concern is the 3rd harmonics dominate. Whether it will affect the sound, I can only tell when I play some music.
Asymmetric Crossover Sound Quality
There is a notable improvement in the sound quality in this version. That slight brittleness in the treble disappeared. More than that, the midrange is clearer. Vocals are now more distinct.
This asymmetric crossover version can be used for HiFi. It’s not the best but usable. A much better performance can be achieved by crossing the PK165-8 at 2.0kHz~2.5kHz. The Peerless H26TG45-06 horn loaded tweeter will match the PK165-8 perfectly.
As for the Audax TW010E1 tweeter, I cannot find any fault with it. It may be cheap but it’s really fantastic value for money. I will mate this tweeter in future with the Fountek FE85. I believe they will sound great together.
Crossover Values & Box Dimensions
If you’re interested to build this project, kindly send me a mail for the crossover and box dimensions. Free for DIY only. Not for commercial use.
Note: Unless otherwise stated, all measurements were made with the mic at 36 ins, tweeter axis. Impulse Window=5ms. No smoothing applied.