Canary-II (Silver Flute W17RC38-08 with Vifa D27TG35-06)

Silver Flute W17RC38-08 (Shielded vs Unshielded)

When I first worked on the Canary in 2013, I was sorely disappointed by the shielded Silver Flute W17RC38-08. The midwoofer sounded lethargic to the point of being dull. I was so completely put off by it that I wrote it off.

5 years later, in 2018, I bought the 8″ W20RC38-08 Silver Flute only to discover that it sounded pretty good. Nothing similar to the shielded W17. It was then that I decided to buy the unshielded Silver Flute W17 from Madisound to compare it to my shielded one. To my surprise, the unshielded W17 sounds quite decent. So 7 yrs later, the Canary-II is born. 

Unshielded W17RC38-08

Fig 1 – Silver Flute RAW Response  •  Baffle Width=10″

The plot in Fig 1 is the RAW response of the W17 in a box with a baffle width of 10 ins. No smoothing is applied. Measurements below 500Hz are in nearfield.

I was quite horrified when I saw the frequency response. Most notable are the spikes at 5kHz upwards. They are caused by the cone breaking up as it reaches the end of it’s bandwidth. However, for a paper/wool cone to exhibit such sharp peaks is something I’ve not seen before.

Another area that caught my attention is the series of waves from 600Hz to about 3kHz. They appear to be cone ringing. It is unusual for a cone to ring so early but it sure looks like it. 

Fig 2 – Toneburst Energy Storage

Just to be sure, I did a Toneburst Energy Storage sweep (Fig 2). In an ideal woofer, there won’t be any light blue slices but with the W17, the stored energy is quite obvious. The cone breakup is clearly visible at 4kHz to 10kHz. I’m not too concerned about that as they will be outside of my low pass filter eventually. What can be more problematic are those from 1kHz to 2.5kHz. At 1kHz, the stored energy last for 4 cycles (periods). Converted to time, it lasts for 4 msec. It is still not too bad. 

Fig 3 – Silver Flute W17RC38-08 with Low Pass Filter

Getting the best out of the W17 is a challenge. If given a choice, I would cross the W17 at 1.5kHz but that would mean using a compression driver and a horn. Personally, I have no objections with that but seeing how some home users are adverse to horns, I opted for a dome tweeter instead. This will mean I will have to cross higher. After much deliberation, I decided on 2.5kHz (Fig 3 – Blue plot). It is not the best place to cross because it’s right at the point of inflection but anything lower and I risk straining the tweeter.

Fig 4 – Silver Flute W17RC38-08 LPF  •  Vifa D27TG35-06

Fig 4 shows the responses of the W17 and the Vifa D27 tweeter with their respective networks. They appear to be crossing at my targeted frequency of 2.5kHz. 

Fig 5 – Crossover Passband

The Black plot in Fig 5 is the summed response of the W17 and the D27. This is an absolutely beautiful crossover. There are no cancellations at the crossover passband (Black plot).

Fig 6 – Canary-II Frequency Response

The final frequency response of the Canary-II is in Fig 6. It is as flat as can be from 600Hz to 10kHz. I am frankly surprised that I managed to get this response. I was expecting a dip at 2kHz but the summation was so spot on that it leveled up to flat. 

Fig 7 – Null response

I flipped the tweeter wires around to check on the summation and it resulted in a symmetrical notch at 3kHz. It is not -40dB deep but I’ll take this anytime. 

Fig 8 – Step Response

The step response of the Canary-II is quite clean. There are no breaks in the leading edge of the W17. What is perplexing though is the plateau at the apex. I appears that the cone came to an abrupt halt.

Fig 9 – Waterfall

Fig 10 – Toneburst Energy Storage

The artifacts from 4kHz upwards in the waterfall plot in Fig 9 are the light blue slices in Fig 10. These bunch of unwanted energy are not from the Vifa D27 tweeter. They are from the W17 cone breakup peaks that are first seen in Fig 1. 

Fig 11 – Spectrogram

When referenced to a 2-dimensional frequency/time domain, the spectrogram (Fig 11) gives a better view of the artifacts. As expected, the excess energy in the treble does not present any issues. The cone ringing from 1kHz to 2kHz is clearly visible but they are not harmful. They are fully dissipated by 6 msec. 

Fig 12 – Impedance

The Canary-II is a very friendly speaker. The lowest is 6Ω at 5kHz. Any competently designed power amplifiers will have no problems driving it.

Sound of Canary-II

I must admit I’m a bit surprised at the performance of the unshielded Silver Flute W17RC38-08. On paper, it doesn’t measure that well but in the Canary-II, it sounds so much better than the shielded W17.

What I like with the Canary-II is the tonal balance. Nothing is lacking or exaggerated. The treble is not too hot. Neither is the bass too soft or the midrange too laid back. 

There’s also a distinct punch in the upper bass. It doesn’t overwhelm the midrange but loud enough to be noticeable. As for the lower bass, it goes down to 50Hz (-3dB) which is quite normal for a woofer of this diameter.

On a personal level, the Canary-II is too laid back for my liking. As a designer, my expectations are different but that doesn’t mean it’s a bad speaker. It is excellent for users that just want a pair of speakers for casual listening. In commercial use, the Canary-II is ideal for soft background music in the workplace. Other venues include restaurants, boutiques and the like.

Crossover is available on request. Free for DIY. Not for Commercial use.
Unless otherwise stated, all measurements were made with the mic at 36 ins, tweeter axis. Impulse Window=5ms. No smoothing applied.