SpeakerBuilding.com - The GeerS eVe II

The GeerS eVe II

By Edgar Beers
27 Feb 2001

Printed from SpeakerBuilding.com, 21 Nov 2008 03:58
URL: http:///content/diy/1104/

The GeerS eVe II

History

Eleven years ago, my first steps into the world of DIY audio involved a Visaton wideband unit in a small box. The sound of my TV being the reference, soon a tweeter, protected by a capacitor was added. Still not good enough so I decided to use a large cabinet (from a Vifa design) made from 18mm particle board and use the wideband as mid-unit. After two years of tweaking, changing etc. I winded up with two 18cm Visaton paper bass units, the wide band that I started with and a Monacor ribbon tweeter, all filtered at 12 db/oct.

This was about the time that I decided never to build again and bought (after endless auditioning of many different speakers) a pair of Acoustic Energy AE1 speakers. These speakers have been with me for nine years when I upgraded the electronics and learned to appreciate the openness, detailed musicality of the AE1. Meanwhile, developments in larger speaker systems caught up with the high-end mini-monitor market in terms of transparency and detail So after nine years I decided to compare my AE1's to larger systems in A-B comparisons Especially my "old loves" Proac, Audio Physic, Avalon, Duntech and Wilson Audio were put to the test. Conclusion: the world has moved on and a transparent imaging with a solid bass fundament was reachable. However, only the top range speakers I audited could provide what I was looking for and my budget did not provide the means.

So slowly the idea of building a system myself came back into play, but based on my previous experiences I would maybe build, but certainly not develop a speaker system. So I started to look into speaker kits and made an extensive study of available kits and compared used structures, components and philosophies with information derived from the mentioned top speaker companies. Unfortunately, most speaker kits are targeted at easily buildable cabinets, with the use of often high quality drivers, but with an affordable end result. Additional to that, I did not want to build a good speaker over the weekend, I wanted to build the best I could possibly reach. During a visit to the near by DIY speaker store I audited many designs, all failed to meet my requirement, which is not surprising, since I put the reference at world top level. Only one design was somewhat acceptable (to my ears at least) the Reference Plus design (http://www.zelfbouwluidsprekers.nl/ned/scanspeak/scan-speak.htm), a 3 way design based on two scan speak 18 cm units and the Revelator tweeter, provided sufficient musicality and detail. Only two drawbacks: the bass was far from tight and the high was too sharp to my ears. Nevertheless, this was a speaker to look into.

Then I came across Tony Gee's Andromeda design. Now there was a loudspeaker that took into account not some, but all essential things to build a great loudspeaker divided in four sections:

1. Great quality units (Focal and Scan Speak)
2. Great cabinet construction (sandwiched MDF construction with lead bitumen and internal matrix)
3. Great filter design (series filter with high quality components)
4. Long experience in speaker design

I contacted Tony who, like me, lives in the Netherlands in order to audit the Andromeda's and maybe buy them or build them. Unfortunately the speakers were sold the very same day I contacted him. Building these speakers without ever having heard them was too much, so it was back to building the Reference Plus again. However, one glance at the Reference Plus' drawings made me add some modifications nothing much, just some structural changes like a double thick baffle, a separate chamber for the tweeter and maybe a physical separation between the bass unit and the rest, creating two separate enclosures (similar to Andromeda). This I could handle, but in email discussions with Tony some new ideas came to light:

replace the Revelator with the 9700 tweeter as a cost saving measure
replace the lower 18W8545 with the 18W8545K for improved bass response
completely use the Andromeda series filter design in stead of the used parallel filter
Employ matrix bracing

Basically, we were talking about a smaller Andromeda design based on the reference plus architecture. Exciting as it is, these plans seemed way out of my league to take up and develop further. So we agreed to develop this loudspeaker together and I would actually build it. This was the start of a unique loudspeaker development solely based on Internet communication. The final result, the GeerS-eVe II might be the first loudspeaker for the new E-conomy Fig. 1. GeerS - eVe II Preview.

The Final Design

After incorporating the new units, filter architecture and enclosure design, simulations showed that this would be an excellent speaker for smaller rooms, but that usage of a larger bass unit would greatly improve the easiness of bass reproduction, leading to a more natural sound. We chose the 21W4555-01 to replace the 18W8545K. However, this also means a larger cabinet. Based on Q=0.5, with 100 filling material, I would increase the cabinet volume from 27 to 46 litres. And so, the GeerS-eVe II was born, while the GeerS I (with the 18 cm woofer) never left the drawing board Fig. 2. GeerS - eVe II CAD drawing.

Phase Coherence

It is relatively easy to build a good sounding loudspeaker. It is quite hard to build a loudspeaker that is able to catch the emotion in music. For that you need detail and optimal imaging. In my personal opinion both are a result of phase coherence. So in the GeerS eVe we did everything to maintain consistent phase behaviour. First, the system is designed as a 2.5 way. Which means that the 18W8545 unit provides the most important area of the musical spectrum. The bass unit only works up to 200 Hz, where the mid unit takes over and runs up to 2500 Hz. There the tweeter takes over.

Second, The whole Baffle is slanted 6.85 degrees. This, in combination with critical placement of the units on the baffle, leads to near-perfect phase behaviour on a listening axis about 93 cm high, from 2 meter away from the speaker. Additional to this, the angle of the top enclosure can be adjusted through the spikes, as well as its position relative to the bass enclosure. However, the design is optimised for placement where the top enclosure's baffle aligns with the bass enclosure's baffle.

Third, The use of a series filter for mid-high, improves phase behaviour over a parallel filter. The parallel filter for the low end acts as a sub-woofer.

Finally, this speaker can be bi-amped or bi-wired. Although it is common marketing practice to proclaim a loudspeaker phase coherent and therefore fit it with one pair of binding posts (the other reason might be that series filters are more commonly used; a 100% series filter cannot be bi-wired), I say keep the option open! If two cables sound better, why not provide the option (hhmm. one reason might be the additional cost for two sets of WBT binding posts).

Again, many speakers are emphasising on phase coherence these days. However, phase coherence is the result of a cocktail of measures, and not "just" slanting the baffle, or put all the drivers magnets in vertical alignment. Cabling, filtering and the phase coherence of the used equipment might make all the difference.

The Tweeter

Scan Speaks D-9700 was chosen over the Revelator for a number of reasons. First, they're basically the same tweeters. Second the speakers I have heard using the Revelator (including Sonus Fabers' Amati Homage) had a tendency to be on the bright side. However, this might be personal preference rather then a quality statement. Third, the smaller front-plate of the D-9700 allowed me to design Avalon-like cut-off edges, which I like both from a design as well as a sonic point of view Fig. 3. Top enclosure.

The mid-woofer

Scan Speaks 18W8545 was the easiest choice. It's about the only remaining element from the Reference Plus as well as the Andromeda and its quality is proven by many top speaker designs who use this unit. Application of strips of lead bitumen on the unit's ribs clearly made the whole structure deader.

The Bass Unit

We basically choose a larger 18W8545, being the 21W4555-01. This unit has a nice long throw flexible design, strong magnet and is ideal for nice natural bass reproduction. Application of strips of lead bitumen on the unit's ribs clearly made the whole structure deader. I also wrapped the magnet in felt. I honestly didn't know what the effect would be, but it was also done on the Scan Speak units employed in the Audio Data Elance, and I would not let any chance on improvement go by Fig. 4. The bass unit and its enclosure.

The Bass Enclosure: Speakers Density

When I bought my Acoustic Energy's I was told that it was easy to build a large box, but hard to actually design a good sounding loudspeaker. Now I have to disagree. It is quite hard to build a large box with similar damping characteristics as a small box. I truly believe that a loudspeaker requires a minimum density. Simply devide its weight by its internal volume. Large speakers need to be heavy in order not to let the enormous amount of energy in the enclosure interfere in the audible spectrum.

Experience teaches me that good loudspeakers have a density of around one, which means that for every litre of internal volume, you need one Kilogram of mass to control vibrations. The Andromeda has a density of 1.05 (95 kilogram for 90 litres). The eVe II has a density of 1.00 (46 kilogram for 46 litres internal volume), even the AE1 has a density of around one (11 kilogram for 11 litres). I have for instance estimated Wilson Audio's CUB's at a volume of around 30 litres, with a weight of around 33 kilograms brings its density to around 1.1. However density cannot be the only indicator. The way mass was applied makes all the difference. Building thin plywood boxes and pouring concrete at the bottom to reach a certain density simply won't work. One can easily see that maintaining a proper density for larger loudspeaker designs is quite hard. Basic mathematics shows that a small volume has a relatively large surface (read: easy to apply mass) as an increasing volume has a relatively decreasing surface (read: harder to apply mass). So, if the minimal density statement is true, than it is much, much tougher to realise this density for large cabinets than is it for smaller ones.

The Andromeda needed a sandwich structure to do it, the eVe II doesn't. For the eVe II, I used 25 mm MDF for the sides, top and bottom. The baffle was made with double thickness: 50 mm. A similar 30 mm collar around Andromeda's top enclosure was used in eVe's bass enclosure, making wall thickness varying from minimum of 25 mm to 55 mm for most of the unit's side surface. Internal bracing was provided through 18 mm MDF matrix on all three axes. Here too, a cut out lined with felt provides additional support for the bass unit. Cutting out large holes out of the back of the bass units prior to installing the collar, added little, but useful volume while incorporating the back as internal bracing Fig. 5. The bass enclosures internals.

This enclosure was designed as a closed box with a Q=0.5. However, just to have the flexibility, a slot shaped bass port was placed at the top side of the enclosure, tuned to 29.5 Hz (this slot is now notoriously known as " the mailbox") Fig. 6. The mailbox.

The baffle is constructed out of two pieces of 25mm MDF. In order to give the bass unit the ability to move freely (remember, there is also a 18 mm matrix around the magnet), the hole in the inner piece of MDF was cut under a 15 degree angle, creating a trumpet like form in the baffle.
The inside, except for the baffle was lined with 4 mm thick lead bitumen and 40 mm thick Pritex foam. BAF wadding was used to fill the remaining volume.

The Top Enclosure

The original top section of the Reference Plus design was provided with a 50 mm thick baffle, similar in set-up with the bass enclosure. A simple 18 mm piece of MDF behind the cut outs for the tweeter created its own little chamber which is felt lined. A matrix on two axes provides necessary rigidity, in combination with lead bitumen lining and pure wool filling. To prevent back-lash, 40 mm pritex was used for the enclosure's back panel Fig. 7. The top enclosure prior to closing.

In order to reduce high frequency reflections, large cut-offs around the tweeter were used. The top enclosure rests on three adjustable spikes in order to minimise energy transfer from bass- to top enclosure.

The Brain Box

This contains the cross over, which is designed by Tony Gee, it uses a second order filter for the bass, with parallel to that a second order series filter for mid-high. For a review on the design criteria for such a filter, please read Tony's article on the Andromeda. The best available components were used. However critical listening tests showed that for the value of the inductor in series with the 18W8545 should be between 0.82 and 0.68 mH. The sonically determined value (and the audible differences are enormous!) was set to 0.785 mH. This value can be achieved by unwinding a 0.82 inductance to the desired value. The complete crossover was hard wired, and WBT silver solder was used throughout. All inductors were placed perpendicular to each other to minimise magnetic influences Fig. 8. The X-over schematic Fig. 9. The Brain Box.

Wiring and Connectors

The Brain Box is fitted with four WBT terminals for bi-wiring (or bi-amping) applications. The Bass enclosure uses also WBT connectors on a thick (Trespa) plate, which is engraved with the Speakers' name. To make this speaker " fool proof", I wanted to minimise the risk of someone connecting the top enclosure (and therefore the tweeter) directly to the amplifier. Utilisation of the series filter gave me the opportunity of using a 3-way gold plated Neutrik XLR connector, which can only be connected to the Brain Box. For bass, 6mm2 OFC copper was used throughout, while the mid-high section uses Teflon insulated silver plated copper throughout. The length between the brain box and the top enclosure was braided and kept as short as possible. Needless to say, all connections were soldered with WBT silver solder Fig. 10. eVe's back.

Finishing

The eVe II's were veneered with a light cherry wood veneer on all sides (including the bottom) and finished with six layers of water based acrylate lacquer.

The Result

We spent 3 months on designing and I spent 6 months on building the eVe II. First listening tests showed the effects of all design criteria: natural bass in perfect harmony with a detailed mid and a fine present high which never becomes sharp. Is this a world-class speaker? Yes, absolutely. Could it perform even better? I thought so. There was a slight fussiness in the mid-low area which took away some of the transparency I liked so much in the AE1. Additional damping material in the bass enclosure, in combination with a light lowering of the mid capacitor by 0.035 mH solved all that Fig. 11. The tuning sessions: HiFi heaven!.

Now this loudspeaker is just not there. There's just music. All superlatives that one reads in hi-fi magazines reflect what could be said for the eVe II. But in the end, if I have to define the character of this loudspeaker, it is that there really is no character at all. It has all the fundament, depth, height and transparency in its musical picture, that one immediately focuses on the music itself and how musicians interact with each other. The effort to create a loudspeaker that excels in phase coherence and is able to catch musical emotion has been worth it! Just listen to a jazz trio, and besides hearing what they play, you can actually "see" how much fun they have playing it. Oh and where they are of coarse. Or listen to Roger Waters' album "Amused to Death". The power of Q sound (imaging sounds in the room by employing phase differences) is amazing and you really like to stand up and kick the dogs out of the backyard if you didn't know better. (If you don't know what I am talking about go buy the album, of go buy phase coherent equipment) Now I know that this album uses studio tricks and is not very purist stuff, but it takes a damn good speaker to make it work

This loudspeaker surpasses anything I could have bought by far. And to me it proofs that top quality audio is indeed not a matter of money, or the newest innovations. As a matter of fact, nothing in this loudspeaker has been done before. In that respect there is absolutely NO innovation. It's merely the result of doing EVERYTHING right, and some healthy thinking in stead of focussing on one (market sensitive) issue.

Note: this design is purely for the DIY enthusiast and may not be reproduced on commercial scale.

Visit GeerS's home at http://www.members.home.nl/edgar.beers