why the TN28
It all has to do with using a 1st order crossover. I initially used the Audax TW010E1 because all it needed was a 2uF capacitor. With that, I was able to cross the TW010E1 at 4kHz to the CHP-70.
Though it worked well, I want to see whether I can cross lower, somewhere in the region of 2.5kHz. This will not only improve the vertical dispersion but enhance the midrange quality. I went through my collection of tweeters and finally selected the HiVi TN28. This is the only tweeter that is designed to be mounted on top of an enclosure. Because of that, it avoids the issues of reflections and diffraction arising from baffle mounted tweeters.
Fig 1 – CHP-70 with TN28
The Red plot in Fig 1 is the HiVi TN28 with just a single capacitor and an L-pad. The Blue plot is the CHP-70 with a BSC and an LCR. I tuned the two filters to cross at about 2.5kHz.
Fig 2 – Overlay with Summed Response
The Black plot in Fig 2 is the frequency response of the CHP-70 with the TN28. The summing is broad, spanning 1kHz~6kHz. Their droopy passband look very much like a Bessel alignment. Note that there are no cancellations in the crossover region. This indicates proper summing.
Fig 3 – Frequency Response of Time Aligned CHP-70 with TN28
For clarity, I removed the individual driver plots, leaving only the Time Aligned CHP-70 with the TN28. There is a bit of upper midrange presence (1.2kHz~2.5kHz) which I hope will not make the speaker shouty.
Fig 4 – Null response
To see how well the two drivers are crossing, I flipped the TN28 wires around. It resulted in a deep notch. Considering I designed this crossover through trial and error, to have the tip of the notch at exactly 2.5kHz is quite an achievement.
Fig 5 – Step Response of Time Aligned CHP-70 with TN28
Now, it’s time to check on the time alignment. Fig 5 display a beautiful transient, virtually a straight line. There is still a tiny fraction of delay in the tweeter as is shown in the inverted tip. With more work, I can eliminate this downward tip but it serves no practical purpose. As far as I’m concerned, this is as good as being time-aligned.
Fig 6 – Toneburst Energy Storage of Time Aligned CHP-70 with TN28
Fig 6 shows some excess energy at 5kHz. That probably is caused by the cone breakup of the CHP-70. The Blue slices from 1kHz stretching to almost 3kHz look like some minor ringing.
Fig 7 – Spectrogram of Time Aligned CHP-70 with TN28
The spectrogram in Fig 7 gives me a better picture of the excess energy when referenced to time. It is clear that all the ringing is below 3kHz. None of them are serious enough to affect the sound as they die off after 5 msec. The excess energy at 5kHz are inaudible. They last for only 1 msec.
is the TN28 better than the TW010E1
Personally, I prefer the TN28. The vocals are mellower when I cross the tweeter at 2.5kHz. But it’s more than the quality of the midrange. The sound stage has improved. The singer is now more forward. This is the way I like my music to sound like.
An unexpected advantage with using the TN28 is the sensitivity. With this new crossover, the satellite’s sensitivity now matches the Toucan-SF. There’s no necessity to pad down the CHP-70 and the TN28. That’s absolutely fine by me. I abhor padding a midrange anyway.
For the time alignment measurement, my microphone axis was leveled to the bottom of the TN28 tweeter. Please note that the step response will change if the microphone is higher or lower.
Adding in the Bass
Fig 8 – Toucan-SF Bandpass subwoofer integration
The CHP-70 and TN28 cannot be used alone. It needs to be supported by a bass woofer. Fig 8 shows the Toucan-SF (Brown plot) crossing over to the CHP-70 at 250Hz. Crossover is by an external 24dB/oct unit. Note that measurements below 500Hz includes my room reflections.
Fig 9 – Full Bandwidth Response
Fig 9 is the final response with the Toucan-SF subwoofer. Taking 80dB as the reference, the response is within +/- 5dB. From 500Hz upwards, it’s +/-3dB. There is some upper midrange emphasis in the bump from 1.2kHz~2.5kHz. I personally am in favor of this as it adds some presence in the vocals. I find that when a response is too flat, it sounds dull.
March 15, 2020Projects