A walk-through from 2005 to 2025
In 2005 my wife insisted to replace my then GAYA loudspeakers with a design that was more fitting in our interior. Thus a trip to hifi shops, which was a sobering experience for my wife and me, be it was what I expected. So we agreed that I would create a design together with my wife. This resulted in the enclosure shape we have now. And when in 2022-23 I decided to upgrade the system, obviously the enclosures were a given.
The enclosures are made from phenolic-paper laminated plates, the oval outside from several 1/16” layers glued together with epoxy, and in the middle a CLD layer specific for stiff polyester boat hulls. The outside veneered with cherrywood. The cavity dividing elements from ½” thick plates, and the main baffle from 2 thick layers with again CLD layer in between. The midrange and tweeter part, the top ~30cm of the inside is massive filled with layers of MDF with only a cavity (3 Liters) for the midrange and the tweeter. The old drivers were Accuton drivers. The woofers were the ScanSpeak 8” woofers (weight added), and in the bottom space the filter was housed.
The filter was originally designed in XSIM, then there always was a discrepancy between the designed response and the measured response. Around 2018 or so VituixCad provided the option to freeform design a passive filter, and then the discrepancy was gone, the design and the measured response matched. So I altered the filter to get a better response. However one aspect kept bothering me, that was the need to swap polarity of woofer to midrange to get the right response. For me this felt as fundamentally wrong. So I began to focus on design such that no pol-swap was needed, and ultimately came across electronic filtering.
In 2022 i purchased a license for Acourate and after the required practicing I for first time realized a more or less polarity phase and response correct system, what a revelation.
Next was to improve the off-axis response of mid and tweeter by reshaping the baffle. In 2023 after some pondering I decided to go for fully active, to replace the ScanSpeak and Accuton drivers with PURIFI drivers, and for the electronics an Okto Dac8Pro and 3 PURIFI Eval1 2-channel amplifiers. We were already using Roon together with a Tidal subscription, and all CD’s we had have been ripped. So obviously for the active filtering Acourate was the chosen platform.
In autumn of 2023 I also ran into an artist’s-block. I was a successful fine art printing specialist, and what I already did before the Covid-period, I wanted to hold art-expo sessions in our house, and to augment that with appropriate music. So the plan was not only it to Purifi the main loudspeakers, but also to have nice speakers in the atelier on the first floor. The artist’s-block made me decide to stop the printing business, and a period of recovery started which took up to early 2025. It also resulted in a rescheduling of all activities.
To get acquainted with the PURIFI drivers test-boxes were made, and once it was clear that the choice for PURIFI was right, I designed in 3D the shape of the baffle, again printed a part of that shape to correlate the simulation with the reality. So in May 2024 i began to renovate the enclosures to accommodate for the Purifi drivers: PTT8.0X08-NAB02 , PTT4.0M08-NAC4, and as upgrade path towards the Purifi tweeter, an SB26ADC with 4” waveguide based on a design of Augerpro, with dimensions of Purifi tweeter received from Lars.
The renovation took a few months
To accommodate the PURIFI mid and tweeter, the baffle was renovated with phenolic-cotton weave based plates with a CLD layer in between. To improve the baffle diffraction within the physical limits of the enclosure, the outer shape has been optimized using 3D CAD and BEM/LEM using AKABAK.
The PTT4 required a much smaller cavity, and the tweeter a separate cavity, so that space was reduced using phenolic based plates and epoxy. Also for the midrange to support its magnet extra elements glued in place. The PTT8 woofers were shape and cavity volume wise almost a direct replacement of the ScanSpeak woofers. Only new screw holes needed to be drilled and Qts around 0.6 instead of 0.75
The end result baffle shape is realized with an additional 3D printed layer and shaping of the phenolic plates to get the desired smooth rounding shape. The woofers and midrange are slightly recessed relative to baffle front and have a 3D shaped and printed flexible rim for smooth shape connection to baffle.
The speaker fronts were already almost 20 years old, the cloth used to be black but are grey now. So these are replaced by a flat frame covered with cloth, and mounted with a distance of 1cm from the baffle.
In the bottom of the enclosure is an open space, once holding the passive filter, and now only passive filter elements to optimize the driver response.
For the loudspeaker cable connectors I decided to switch to Speakon after some sobering experiences with so-called HiFi connectors (WBT chassis , Banana plugs, Spades). Speakon connectors now in use both for the amplifiers and the enclosures.
The choice of crossover frequencies, based on how human hearing works, a combination of these factors, researched by ENT specialists in late sixties - early seventies:
- Critical bands
- Doppler effects
- Lateral directivity sensitivity
Based on information received mid-seventies and based on what a couple of ENT specialists and a ECG specialist researched and concluded upon as a range starting at ~433 (up to 480 or so) and 3464Hz (up to 3840 Hz or so). Instead of traumas they focused on how the human hearing works in combination with 2-channel stereo. So different publications at that time about the auditory system, basilar membrane cochlea etc , and also research by scientists from Philips NatLab while during WW2 hiding from german occupiers they spent their time on 2-channel stereo in combination with the way our human hearing works, such as: The hearing directivity vs frequency research, and doppler effects related to used bandwidth of drivers, cone diameter, our head and torso, shape of head (f.i. effect of nose on directivity) . Unfortunately practically all my notes, copies etc from that period are thrown away, except for a few sheets that luckily were misplaced
It is interesting to see that critical bands (Bark etc) and cochlear implants studies still show a lot of overlap.
With drivers mounted, the design of the active filters started, it also required to get proficient in using the Acourate functionality, so given the mental recovery I mentioned above it took some time, but found it worthwhile as it helped the recovery, and now I can create a modification very easily. The active filters are linear in amplitude and time(phase).
For me a house-curve or target-curve is the outcome of the loudspeaker-room interaction, and not an objective/target per se. But for the response measurements at MLP the house-curve is a key element to take into account, be it the listening distance has an influence on the drop off notably.
I based mine on the predicted in-room response as calculated in VituixCAD based on the on and off-axis responses of woofers midrange and tweeter, which correlated well with measurements of the whole system thus confirming the choice for house-curve. All this with listening distance of 3000-3500mm.
Yet the prediction and the measurement are just indications, is the drop-off 6 or 8dB and does it start at 20 or 150Hz etc, so relatively small variations of the house-curve will have an audible impact. Also the Purifi tweeter radiation behaviour stays wide up to 20kHz or so, thus almost flat instead of a drop-off at high frequencies. All in all making room-curve shape an aspect of (fine) tuning or tinkering perhaps.. Luckily it is an easy mod in the filter creation, and the system allows for A-B tests, so interesting choices ahead. In my working life I was responsible for projects, and was wary for the 99% complete syndrome. With the active filtering and DRC it is always 99% complete, and that is fine
So this total system is playing, and we are quite happy already, the improvement over the previous set up was really significant, especially is the dynamics, clarity, the sound stage, the silence in between notes, the whole perception of music and the peak into the soul of musician. And the nice addition, at any place in the living room or even with doors to garden open, on the terrace the music sounds remain coherent and enjoyable.
Schematic diagram of system:
Picture of room with enclosures and installation:
Clearly an asymmetric setup! Some dimensions: width main room 5.2m, ceiling 2.45m, main listening position (my ears) some 3.5m from enclosures, and trying to get it to ~3m.
And now having received the PURIFI tweeters, I decided to one last project to finish the purifi-cation of the GAYA2.
