After the Darbari, my exploration of high quality speakers acquired some rules to follow, when budgets permit, and some things to avoid. The rules to follow include:
- A 3-way is better than a 2-way, all else being equal.
- For a 3-way, the woofer-to-rest crossover is best done active, even if the midrange-to-tweeter crossover can be passive
- Baffle step compensation is not arithmetic, it’s a matter of taste. The knee frequency is confusing when enclosures are not rectangular, and the shelf depth depends a lot on the room where the speakers will finally be placed.
- Flat frequency response, on axis, other than BSC, is The Right Approach.
- You need a lot of power, a lot of cone surface area and excursion, for good bass. We keep underestimating the power needs here.
The things to avoid were:
- The physical size and build of the Darbari may not be everyone’s cup of tea. A top notch speaker may be rejected by a genuine music lover just because he can’t settle down with that baffle width.
- Some people simply will never be able to feel comfortable with an active speaker with a separate DSP appliance and multiple external amp channels. So, either incorporate all the electronics inside the speaker enclosure, or stick to passive.
- The superiority of a 3-way over a 2-way is more significant than that of an active crossover over a passive.
So the yama and niyama gave rise to the next generation of 3-way speaker designs, which split into two branches. One branch went for higher end drivers and active crossovers, and their pinnacle as of this writing are the Durga and the Shankara, closely followed by the Charukeshi Mark II. The other branch went the passive 3-way way, and tried addressing the question: how much improvement does one get if one sticks to the medium-grade drivers and components used in the Asawari IV and V, but just shifts the game to 3-way? The leader in this second line of exploration is the Adana.
The Adana incorporates all the things to avoid which emerged from the Darbari. It aims for a slimmer enclosure, works with just one amp channel per speaker, but keeps to the focus of this journey: do a 3-way. The Adana is followed very closely by the Adana Mark II, and the Charukeshi Mark III. (The Charukeshi Mark III does not avoid the first pitfall of the Darbari — it does not avoid the wide baffle and visually heavy appearance. It uses a 12″ woofer.)
Design notes
The Adana uses the same family of drivers as the Asawari Mark V, because we wanted to see how far the pure transition to 3-way would help if all else was kept unchanged. So, we have two Dayton RS180 8 Ohm metal cone drivers in parallel for woofer duty, a Dayton RS125 4 Ohm metal cone driver for a midrange, and an SB Acoustics SB26CDC tweeter for the upper octaves. It’s an all-metal symphony.
The woofers are in a common chamber with a bass reflex port. The RS line of woofers from Dayton are so similar to each other than it’s easy to mistake the RS180, 150 and 125 for each other.
The enclosure
Unibox gave me this when I specified a bass reflex box for two RS180-8 in parallel:
A 2-inch diameter port, 12cm long, would be the bass extension artefact. In-room bass extension would reach below 35Hz, covering pretty much all real music (home theatre LFE is not real music).
I designed the enclosure to be simple and non-controversial — a tall, slim box with a front baffle width as low as I could manage.
The construction is with my usual set of materials:
- 25mm marine grade high density fibreboard (HDF) for all sidewalls, top and bottom
- 25mm Baltic birch for front and rear baffles
- 18-20mm commercial ply for internal bracing, partitions, etc
- Real-wood veneer covering of top and sides. These veneer sheets in India are 4mm thick, like thin sheets of plywood.
The drawings show 25mm radius roundover of the vertical edges of the front baffle. In reality, I went with 37mm (1.5″) radius roundover, and it turned out awesome.
From the front, the enclosure looks slim and elegant.
Inside, there is an unused chamber at the bottom which is not used for anything.
The midrange and tweeter are offset to one side on the front baffle, partly to make the edge diffractions asymmetric, partly because they look cute that way.
We start by cutting the parts and putting them together. The bracing, sidewalls, and top and bottom get fitted first. All fitting is done using Araldite and wood screws. It’s always both for my builds. After the first 24 hours, the wood screws play no role in the structural strength and rigidity, but till the Araldite sets, they are vital to hold the pieces in place while the assemblies are moved around.
The video below shows a sidewall being fitted to the bracing structure. Araldite and wood screws are being used. Each screw is first dipped in Araldite and then inserted into the pre-drilled hole and tightened. This ensures that the screw never weakens or splits the plywood from which the bracing is made. After 24 hours, the ply and wood screw becomes one solid unit and will last a hundred years.
We then fitted the absorbent lining. Past projects have taught us that it’s a pain to fit it after the enclosure has been completed and sealed up. At this stage, the front and back plates are still open.
The veneer is also stuck on the top and sides at this stage.
After this, we fit the front and rear baffles. The rear baffle is removable and is simple — it will always be fitted with a dozen screws and can be removed for crossover maintenance. (Not that crossovers need maintenance.) But the front baffle is a bigger job. This is made of two sheets. The inner sheet is fitted using Araldite and wood screws. The outer sheet is then stuck on purely with clamps and Araldite, no screws at all. Screws which reach the outer surface of the enclosure can never be effectively hidden by the final touch-up and polishing, so the outer surface never has screws.
The videos below show how the front baffles are being fitted. First, fit the inner sheet of the baffle.
Note how each screw’s threads are first dipped in Araldite and then inserted into the pre-drilled hole and tightened.
After the inner baffle sheet, the outer baffle sheet is clamped down with just Araldite and G clamps
The next step is the roundover of the vertical edges of the front baffle with a 1.5″ radius roundover bit, followed by five days of hard labour for the PU polish. Epoxy, sealant, then more sealant, then more sealant, and finally, the top coat, transparent matte finish PU.
Measurements
The tweeter
I did a gated SPL measurement of the tweeter, periodic-noise, 1 metre.
This was disturbing, because I was getting a dip at 3.5K and a peak at 5K which were not there in the manufacturer’s datasheets:
Anyway, I chose to work with my own measured data.
The midrange
Took a gated farfield of the midrange, first on the midrange axis, then on the tweeter axis, 1 metre
Then I did a nearfield an inch from the driver cone:
Then merged the two in VituixCAD to get a single composite SPL plot, which I later used for crossover design:
The woofers
Did a gated farfield measurement of the woofers, on the tweeter axis, at a distance of 1 metre from the front baffle. This measurement is not expected to be of any use, because the reading does not have any valid data below about 500Hz, and the only part which is supposed to be useful for crossover design is below this.
Then I did a nearfield measurement, off one of the woofers:
And then I merged the nearfield and farfield:
Other measurements
Tweeter harmonic distortion, 30cm from the baffle:
Midrange harmonic distortion, first at 30 cm distance, for the higher frequencies, then nearfield, for the lower frequencies:
Woofer harmonic distortion, nearfield, to show low frequency behaviour.
The crossover
I started plugging the various SPL and impedance data files into VituixCAD, and after several months of part-time fiddling and experimentation, I arrived at this crossover:
The circuit shows one woofer, but that woofer represents two physical woofers in parallel.
The on-axis SPL of this crossover turns out to be:
The crossover frequencies have fallen almost exactly at 300Hz and 3KHz.
The impedance of the speaker is shown here. As you can see, it’s a lowish impedance load, so it’ll need a relatively powerful amp to drive the bass frequencies well.
I assembled the crossover on the two boards which have been provisioned for this on the rear panel of the speaker.
I had never assembled this complex a crossover before, and I had also not used such high value inductors and capacitors before. There is a 5mH inductor and two 100uF capacitors on each board. Not surprising, considering that this is the first crossover which has a Fc as low as 300Hz. Two-was crossovers are far simpler.
After assembly, I did my usual trials with a set of 30+ short test tracks which I use to test any speaker. The videos below are taken with one speaker being driven by my Roksan amp, which has adequate power to drive most speakers. The source is FLAC files delivered through a very clean Topping USB DAC.
I now started the more detailed listening tests.
The sound
It’s mandatory for every DIY audio builder to talk about the sound of his work, and for speakers, there is indeed a lot to be said about it.
- The whole stack is metal diaphragm drivers. You get what you expect here. A certain cleanness which some will consider superb, others will consider over-etched. I have encountered it in the Darbari 12 years ago and loved it, and here, so far, I am enjoying it very much.
- There is a remarkable evenness and lack of colouration in the mids and highs. Nothing seems out of place. You don’t realise that other speakers are coloured till you encounter a less coloured speaker, I guess. My Bihagda is more coloured than this, something which I would never have been able to say unless I’d compared these speakers side by side.
- The bass is clean (I like metal cone bass) but it’s also suprisingly powerful. The initial two minutes from “Hotel California” from the “Hell Freezes Over” album, for instance, sounds hard, clean, and punchy.
- I was pleased as punch that the BSC effect I had given to the SPL response in the crossover seemed to yield a very good, pleasant response. Strong bass but not excessive, in my preliminary tests in a large open, empty office.
- I still have to listen to a lot more Dylan, Sinatra, Joan Baez, Kishori, Chhannulal, to see whether the voices sound convincing and genuine, without the clean detail of the metal-cone driver stack causing problems.
- I am still curious about the tweeter response dip and peak which my measurements threw up. I built my crossover assuming that the peak exists (I notched it out), but I will measure again now to see if I really see a flat on-axis SPL in that region now.
- I’m also uncomfortable with the accuracy (or whatever you call it) of my merging of my nearfield and gated-farfield measurements of my woofers. When you have one woofer driver, you put your mic 5mm from the cone and take your swept-sine reading. When you have two woofers, in one box, what do you do? You can’t drive one woofer, because if you do, the other woofer cone becomes a passive radiator. If you drive both woofers, where do you place your mic? I don’t have answers. The Adana sounds good today, but I would like to be on stronger footing with my measurement techniques.
More later.
-x-x-x-x-x-