Swift 2.5 (TMM)
The Swift 2.5 is a 22 liters mini-tower bass reflex utilizing two units of Zaph ZA14 midwoofers and a Peerless DQ25 titanium dome tweeter. This design has been a long time coming. Actually, 10 years to be precise. I first worked on the ZA14 in a 2-way back in 2010. That was the first Swift. By the time I finished with the 2-way, I had four versions. I intended to do a 2.5 way after that but it was put on the back burner as more pressing projects took precedence.
This 5″ midwoofer is not a friendly midwoofer to work with. It’s greatest strength is the midrange. Beautifully clear. The biggest problem is the nasty cone breakup. The easiest way to get round this is to use a steep crossover and cross it low. This will suppress the breakup peak but this approach has it’s own drawback. Firstly, I would need to use a tweeter that I can cross low. That will increase the cost. A tweeter like a Seas 27TDFC cost as much if not more than the ZA14. Furthermore, a steep crossover is not always the best option where sound quality is concerned.
In this 2.5 version, I will not try to save on the crossover. I want the 2.5 to sing and I believe it can be done. All I have to do is to release the reins of the ZA14 and let it do what it’s good at. I just have to find a way to tame the cone breakup without murdering the music.
Fig 1 – ZA14 RAW (Black plot) and with Low Pass filter (Blue plot)
The Black plot in Fig 1 is the RAW response of the ZA14. There is a massive +10dB peak at 8.5kHz and a smaller one at 5kHz. Measurements below 700Hz include my room reflections. Disregard the deep notch at 150Hz. That’s my mic picking up a floor bounce.
For an easier reference, I spliced in a nearfield plot, that is the Blue plot below 700Hz. It serves as a good approximation of what is happening below 700Hz. The Blue plot above 700Hz is with my low pass filter. The manner in which the ZA14 rolls off is the critical part of the Swift 2.5. Get this wrong and the Swift 2.5 goes down the drain.
Fig 2 – ZA14 with Low Pass and DQ25 with High Pass
Fig 2 shows the ZA14 with the DQ25. There are two things I don’t quite like with these plots. In the tweeter, there is a slight depression from 2kHz~5kHz. As for the ZA14, there’s that small peak at 5kHz. There’s not much I can do to correct them except by changing the drivers but that’s not an option. I’ll just have to make do and see what happens.
Fig 3 – Summed Response Overlay
The Black plot in Fig 3 is the summed response of the ZA14 and the DQ25. In this plot, we can see the effects of the 2kHz~5kHz depression in the tweeter and the 5kHz peak of the ZA14. All things considered, I’m quite happy with the summing.
Fig 4 – Swift 2.5 Frequency Response
For clarity, I removed the individual drivers, leaving the final frequency response of the Swift 2.5. Once I got the ZA14 to cross properly with the DQ25, I then proceeded to integrate the lower ZA14 for the addition bass.
Fig 5 – Swift 2.5 Null
To check on the crossover, I flipped the tweeter wires around. It resulted in a broad notch. It is not quite time aligned but close enough. But how close is close enough.
Fig 6 – Swift 2.5 Step Response
Fig 6 is the Step Response of the Swift 2.5. This plot is with all three drivers wired up, that is the DQ25 tweeter and the two ZA14 midwoofers.
This is an impressive measurement. The transient is a straight line, almost vertical. This indicates the Swift 2.5 is very close to being time and phase coherent. And it’s fast. The highest peak is at 100us.
Fig 7 – Swift 2.5 Waterfall
The waterfall plot in Fig 7 shows some artifacts in the tweeter. I believe the decay from 4kHz~7kHz are from the ZA14 cone breakup. The tweeter itself is quite clean. There are some burst at 14kHz and 17kHz which probably came from the DQ25 cone breakup but it’s very minor.
Fig 8 – Swift 2.5 Spectrogram
The Spectrogram is a two-dimensional representation of the waterfall. In this measurement, the frequency is plotted against time.
From here, we can see the artifacts in the DQ25 tweeter do not last more than 1ms. It is below 4kHz that we see some excess energy. Even then, they dissipate by 6ms. Nothing to be alarmed about.
Fig 9 – Swift 2.5 Harmonic Distortion
Frankly, I’m surprised at how well controlled the 2nd (Red) and 3rd (Violet) harmonics are in the Swift 2.5. Generally, they reside at -50dB below the fundamental. More importantly, there are no bumps in the plots. This is an exceptionally clean speaker.
Fig 10 – Swift 2.5 Impedance
The Swift 2.5 is an easy load. The lowest impedance is 5Ω at 4.5kHz. Towards the bass, the impedance is slightly above 5Ω at 200Hz. Couple that with the electrical phase, most power amplifiers will be able to drive the Swift 2.5 without complaints.
The valley at 50Hz is the bass reflex port tuning. This is exactly what was predicted in my simulation. I would have preferred to tune the Swift 2.5 to 40Hz but that would mean I’ll have to apply electronic eq to boost the bass. I want to keep electronics out of my speakers as far as possible. If need be, I can always use a subwoofer for the 40Hz~50Hz region. That will extend the bandwidth down to 40Hz.
How does the Swift 2.5 sounds like?
In one word, FAST. When I use this word to describe a speaker, these are the properties I’m referring to. The attack of the bass is excellent. There’s no bloom in the bass notes. The timing of the bass is in sync with the midrange and tweeter.
To give an idea of what I mean, I played Toxic (Britney Spears). What I’m listening to is the bass. With the Swift 2.5, the attack is instant and there’s definition. On top of that, the bass has texture. Very few speakers can do all three. Most not even one.
Another track for bass is Love of My Life (Santana-Supernatural). There’s body in the bass and with texture too.
With the bass sorted out, I moved on to the all important midrange. This is the heart of all my designs. If the definition is lacking, it is hard for me to make out the lyrics. What happens in such situations is one has to turn up the volume. Even then, it’s still difficult to hear the words.
In Ghost in the House (Midnight Matinee), Amanda McBroom voice came out perfectly. The guitar accompaniment doesn’t overwhelm her performance. The sibilance is well controlled too. It doesn’t spit at you nor is it exaggerated.
Next up is Jewel – Perfectly Clear album. Her singing is exceptionally demanding on speakers. In Loved By You (track 10), her vibrato with the Swift 2.5 is bewitching. No screaming, no shrillness. Just that vibration in her voice.
For the treble, I put on Teresa Teng. If a speaker can produce her songs properly, it is an A+. I listened to one of her most famous songs, The Moon Represents My Heart (月亮代表我的心). It was a joy. There’s no harshness in the treble throughout and the sibilant fricatives have just the right emphasis. It’s sad that she passed away prematurely from a asthma attack. She remains one of the greatest artists in Asia.
Where does the Swift 2.5 stand
Having thrown different genre of music at the Swift 2.5, I’m more than pleased with it. There are not many of my designs that reach this standard.
It has nothing to do with the ZA14. I can design another speaker just as good or better with a different midwoofer. It has to do with a design philosophy and having a clear understanding of the variables involved.
Unless otherwise stated, all measurements were made with the mic at 36 ins, tweeter axis. Impulse Window=5ms. No smoothing applied.
April 2, 2020Projects