21W 6600 WG

2way  loudspeaker project with 8inch midwoofer was something I have been considering for a years. The trouble was to find suitable midwoofer. There were few new choices appearing recently, 21WE and Purifi 8inch, unfortunately budget did not allow. Some of the optins, like PA rivers from FaitalPro, 18Sound, B&C would likely be fine, but I did not like the appearance.

21W Classic caught my full attention after seeing Hificompass measurements. I studied also Troels project with 7100, which convinced me that 21W might be suitable for midwoofer duty. I was clear I will need to take care of breakup 3200Hz, my initial target was to use LR4 crossover with Fc ~1500Hz, and to try to suppress breakup at least 30dB below FR of the loudspeaker. The other thing that worried me a bit was distortion in the  midrange 500-1000Hz area. HD3 and HD5 show the mirroring of the breakup. All worries aside, I had to try, listen, and make conclusion for myself.

Prototype waveguides that I used in Ellips-A projects seemed suitable, and were already ready for Scan Speak drivers. This time I needed robust tweeter able to coper with low crossover frequency, and of course perfectly sounding. Quite recently I finished ClassIllu project, 2way with 6600, and 6600 sounded in that implementation incredibly well, being perfect match for 18W Classic. I did not hesitate to chose it again, though its measured performance in WG was still a question mark. I measured it in WG few years ago, and I recalled it looked quite good. 

I had a suitable cabinets already available, I just had to modify the rebates for the waveguides.

As listening tests so far were very satisfactory I will make plans for final version, with optimized baffle for improved tweeter off axis performance, and mild baffle tilt.

Drivers measurement

Both drivers were measured in the cabinet. SPL calibrated. FR responses 0-90degrees, responses shown no gating + smoothing 1/12 and gating 5ms + no smoothing. Distortion measurement done at 2.83V, mic distance 30cm.

6600 tweeter measurements

Nice, consistent off axis behaviour, considering 20mm chamfering as baffle edge treatment is far from optimal (30mm roundover would be much better). There is mild diffraction at 2200Hz, baffle edges roundovers would definitely improve that.

Situation above 10kHz is not perfect, better optimized WG would likely improve that. Generally tweeters with thick half roll surround are more difficult to design waveguide for as surround contributes to radiation as well. There is also 18kHz bump that is presented in the measurement of tweeter without waveguide, and also in the datasheet. Listening test will show if 18kHz bump will have to be addressed/lowered in the crossover and how many dB. From the left, horizontal FRs 0-90deg gating 5ms, horizontal FRs 0-90deg smoothing 1/12, distorsion 30cm 2.83V.

21W/8555-10 measurements

21W FR measurement confirmed that expected system sensitivity ~85dB/2.83V/1m was right.

Breakup is at 3200Hz, during the measurement it was clearly audible, though significantly less compared to rigid cone drivers.

Generaly I was pleased to see such a nice measured performance, especially off axis responses and distortion measurement below 100Hz.

Small dip at 700-800Hz might be inherent cone-edge resonance of the driver, or interference coming from the port, so I will look at this more closely soon.

Crossover modelling

Just a few facts..... Fc 1500Hz and 4th order acoustical as initially planned, excellent phase tracking at Fc and on the both sides of Fc, sensitivity ~85dB, expected power response dip at Fc is quite shallow, off axis response clearly show what WG is about and its benefits. 

On the left sixpack with responses with 1/12 smoothing, on the right no smoothing + gating 5ms.

Below is the initial version for to start sound evaluation with....

And this is current version........

So far I am very please with the overall sound. As always it took a lot of hours of crossover tuning and listening sessions, and also drivers need to burn in, but it all really paid off. I also tried version without series resistor in tweeter section, and this definitely proved to be the right step, as I experienced several time before. Of course tweeter level adjustment must be done the other way, in this case series RC in parallel to tweeter. And because there are always two sides of the coin, every technical solution has its pros and cons, strengths and benefits or compromises and obstacles and price to pay, it is the same here, so here we need to carefully tune RC values and system impedance above 20kHz often lowers to <2ohm. Any decent amplifier will not complaint but is is something to be aware off and designer mustn't let impedance to dip too low.  

Horizontal and vertical polar maps are below. Vertical off axis nulls are symetrically at 30 and -30 degrees. At higher angles around 60 and -60 degrees drivers add to each other, they do not acoustically subtract, and this is the reason for even power response at crossover region. Typically, when also vertical response are included to power response and DI calculation, there is power response dip at Fc. In this case Fc 1600Hz equals to wavelength 215mm, CtC of the drivers is 220mm.

Horizontal polar map and responses normalized to 0deg response are below. Mild diffraction signature 2-3kHz is clearly visible.

Loudspeaker measurements

Mic distance 30cm, pointing between the middle WG and 21W. SPL calibrated. 2.8V graph on the left, 6V graph on the right. Measured with Earthworks M30.

Listening impressions

Measurements does not tell whole story, at least not the measurements above, the standard set of off axis FRs, distorsion, etc. Loudspeakers are made to listen to, to enjoy, to feel, so I actually like and will try to, at least with my simple, limited english, describe my impressions of what I hear, and compare it to previous experiences. 

Initial flat FR tuning showed to be slightly bright sounding, so I had to take care of the FR above 2kHz, and especially 7-10kHz area. After mild attenuation of that area by ~1dB, the sound lost mild hiss signature which attracted too much attention to the tweeter itself, and begin to sound open, detailed, with the body yet light and airy. Again it reminded me why I like the sound of Scan Speak silk domes. Rigid dome tweeters sound different usually, and people could complaint about silk being less accurate, but Scan Speak silk dome are among the best from silks, and sounding very enjoyable and involving. 

21W models well in closed enclosure, so I tried this alignment first. Though very good, fast and precise, it sounded also quite too short and dry, missing the full weight of the instruments. I played a lot with the amount of damping material but after all I decided to add port, tuned to 21Hz. This, and few adjustments of the amount of damping material brought the effect I was looking for. Midrange from 8inch cone is interesting experience. I expected, actually I was afraid, that midrange quality would not be the strength of this loudspeaker. I played CD by CD, trying to find the flaw, but actually the opposite happened. With each new CD that I lestened I was more and more sure about midrange qualities. There is certain sense of fullness, compared to 6inch midranges or smaller. For example 12MU sounds more focused, precise, yet 21W is still overall very enjoyable and I actually do not miss anything. At the beginning of the project I had concerns of -30dB breakup suppression would be enough. So far all listening tests proved there is no issue.

And the bass, well, it is Scan Speak speciality, it goes deep, has the structure, even complex drum recordings sound clean and separated, and there is the sense of energy, kick, richness and effortlessness. Something that actually not often to hear from other drivers. 

Photos