ScanSpeak Discovery 18 DIY (Troels Gravesen)

Discussion in 'Speakers' started by Jeb, May 16, 2017.

  1. Jeb

    Jeb Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    391
    Trophy Points:
    63
    Location:
    Bristol, UK
    I decided to build these - http://www.troelsgravesen.dk/DiscoveryW18.htm.

    Reasons for the choice were:

    1) The mini 4" ScanSpeak driver I tried before gave me a taste of the clear resolving sound of the discovery drivers
    2) Price was about right - around £600 all in. Still the most I have ever spent on speakers, DIY or otherwise!
    3) Building the Overnight Sensations has temporarily turned me away from single drivers towards 2-ways.

    The crossover components come in a kit from Jantzen. Even including shipping from Denmark I couldn't have sourced the parts any cheaper locally. Also included are ports, damping materials, wire and binding posts: €280 for the version with Jantzen Superior Z-caps. Drivers from Falcon Acoustics - £80 for the R2604 tweeters and £110 for the 7" Discovery Woofers.

    All the information is on Troels Gravesen's website, including the theory behind addressing time-alignment issues with the stepped baffle. My build is documented below in case anyone else fancies having a go - I highly recommend it! Listening impressions and measurements are towards the end. Rest assured, these sound terrific.


    PLANS:

    I asked Troels if it would be okay to use 18mm ply throughout. No problem, he replied. Here are some plans I drew up in SketchUp to help me. If you do use these, please only look on them as a guide and refer to Troels' plans as the ultimate reference. These use mitre joints - you don't have to!


    [​IMG]


    CONSTRUCTION:

    1) Gluing up mitre joints can be a great way to raise your blood pressure into unhealthy levels. I cut slots for biscuits in the mitres to make alignment easy. A new toy! - a Lamello biscuit jointer. This thing is awesome. Rock solid and 100% accurate. You can just trust it.

    [​IMG]


    2) I also cut rebates for the back panel and slots for the horizontal braces. I use a router/guide rail system and cut both sides at the same time to help ensure alignment. The Dewalt system is not the worst POS, in fact it's fine in many respects. It lacks a bit of quality.

    [​IMG]


    3) I made the brace holes with a hand drill and hole-saws. I'm drinking the Festool Kool Aid (Festool-Aid) these days.

    [​IMG]


    4) Here are the pieces for 1 speaker frame ready to go.

    [​IMG]


    5) Glue-up can be done with strap-clamps. Several other regular bar clamps were needed to pull it into place and to put a squeeze on the braces.

    [​IMG]


    6) Front Baffles: The pivot point for the curved outer baffle sits just inside the recess for the woofer (more or less where the bottom mounting screw is located). After cutting that, I moved on to the rebates, driver & port cut-outs and chamfers on the reverse.

    [​IMG]

    [​IMG]

    [​IMG]


    7) DAMPING

    An optional step is to use 4mm bitumen pads on the inside walls to help with cabinet resonances. The cheapest I could find was 'Bitflex' on eBay. It's only 2mm, which meant 2 layers were required - a bit of a pain! As recommended by Mr Gravesen - use elastic vinyl flooring glue.

    [​IMG]


    ...then a layer of wool felt. Strategic pieces of acoustilux are also used, as per the plans.

    [​IMG]


    8) One thing that I've found a bit annoying with previous builds is how the baffles will tend to slip around during glue-up. To eliminate this headache I used dowels and biscuits for alignment. It's really nice to be able to just concentrate on a good clean seal, and not worry about slippage while clamping.

    [​IMG]


    I've switched to Titebond Extend for glue-ups. The longer open time is really useful and much less stressful than Titebond I which becomes unmovable in about 2 minutes! The baffle overhangs the sides by a mm or so, so it can be trimmed flush. At this point I flush trim the back too. It's really a nice way to get a clean finish on the exposed laminations.

    [​IMG]

    [​IMG]


    9) Cutting the chamfer (25 degrees) on the front can be a bit daunting. I use the other speaker to help support the router and clamp a bit of wood to the exit point to avoid tear-out.

    [​IMG]


    10) FINISHING:

    Since all edges were trimmed with the router, very little sanding was required. A cabinet scraper can be great for carefully removing any glue residue. I then applied 2 coats of hard wax oil and then wax polish. The wax polish is a labour of love - lots of buffing! In one of the pics below you can see how large these speakers are next to the Overnight Sensations.

    [​IMG]

    [​IMG]

    [​IMG]


    Next...crossovers!
     
    Last edited: Jul 21, 2017
  2. Jeb

    Jeb Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    391
    Trophy Points:
    63
    Location:
    Bristol, UK
    11) CROSSOVERS:

    The wiring diagrams that come with the kit & those on the website are clear and great for someone inexperienced like me. The crossovers are mounted vertically on the back panel so I used an idea from Troels' FAQ section to support the heavier coils with a dowel through the board.

    [​IMG]


    For the other components I used a silicone rubber adhesive/sealant recommended for electrical components. It worked pretty well, even for holding those large capacitors - it takes a while to set!

    [​IMG]

    [​IMG]




    The binding posts can be attached and connected by reaching through the woofer hole and then through the lower brace. It's a bit fiddly - the tool of choice here was a mini 1/4" ratchet with a deep drive socket. The ports are a press fit so the hole needs to be tight - if using a Jasper circle jig, that means undersizing the hole and sanding it until it accepts the port with a bit of force. Test fit before you mount the baffles. A bead of sealant was also used just for good measure.

    and here they are finished....

    [​IMG]

    [​IMG]

    [​IMG]




    MEASUREMENTS:


    [​IMG]

    I measured these outside at 1m and also in a listening room set-up.


    Outside, 1m, 0 degrees

    [​IMG]


    Outside, 1m, 0 degrees, Left and Right Comparison


    [​IMG]

    Outside, 1m, Off-axis 15/30/45 degrees approx

    [​IMG]


    Outside 1m, Comparison of Discovery 18 (blue) with Overnight Sensations (Red):

    [​IMG]


    Listening Position : approx 2.5m

    [​IMG]


    COMMENTS:

    Outside measurements:

    From 500Hz all the way up these are definitely the flattest, smoothest measuring speakers so far. They also show the greatest extension and most gradual roll-off in the low end. The hump at around 200Hz is present with all the speakers i've measured in that outside space.

    Off-axis, the high end roll off is the most drastic of any speaker i've measured, including all the MarkAudio wide banders - especially at 45 degrees. Something to consider with this ring radiator tweeter. Even before measuring, I had developed a preference for positioning the tweeters to the insides and toeing in the speakers towards the listener. The domed D2604 is another option for these with minimal changes to the crossover.

    The comparison with the OSMTs seems a bit unfair since the mic is raised higher for the Discovery 18s, and they were measured on different days where the conditions are somewhat changeable. Nonetheless, the difference its there to see.

    Inside: wow look at the room effect - some notable 10db or more increases in the lower end. This particular room is carpeted and a bit squishy. Despite that, the theme of an overall flatter response seems evident in amongst all the interactions. At least to my untrained eye.



    LISTENING:

    Beginning to take measurements has shown how significant the role of the room is. Another thing to bear in mind is that I am coming from 3-4" drivers so the step-up feels especially pronounced.

    I let these play for 100 hours or so, dipping in now and again just to relax and enjoy the music and let the post-build glow wear off a bit. Later, when committing some notes, the first thing to note is how balanced, resolving and exceptionally clear throughout the range these are compared to anything else I've owned.

    Secondly, what struck me was how much more dynamic and commanding of a larger room these are compared to the OSMTs. Probably not a big surprise.

    Thirdly, the leap in low-end performance going to a 7" woofer. Sheer presence, extension, impact and clarity are all notable. The bass line in Alt-J's 'Tessellate' feels pretty monstrous, whilst simultaneously resolving more of the finer textures of the percussion too. 'Runaway' by The National seems more dramatic because you feel those thumps in your body when you're supposed to. The drop in 'Hey Now' by London Grammar gives the room a very good rattle. Film soundtracks and orchestral pieces have greater depth and instruments feel more separated and distinct.

    I need my speakers to sound good with a wide range of modern music - and for the most part, these deliver. Much of my music sounds really excellent - the best I have heard at home with this system. But I find these speakers to be quite a bit more revealing and unforgiving than the OSMTs. More stuff than before now sounds average and I'm becoming more aware of what might be a good recording and what isn't.

    Also, while we've taken a big step forward in resolution and clarity, now I notice a slight absence of humanity/body in the upper reaches of vocals. I need to think more about my system (Naim 50W integrated, Rega DAC via USB), which I doubt does these speakers justice. A rethink is long overdue. I need to see what happens once we get some tubes and a better source involved. Overall, I'm pleased.
     
    • Like Like x 32
    • Epic Epic x 1
    • List
    Last edited: Jul 29, 2017
  3. fraggler

    fraggler A Happy & Busy Life

    Pyrate
    Joined:
    Oct 1, 2015
    Likes Received:
    5,085
    Trophy Points:
    113
    Location:
    Chicago, IL
    I love you and hate you for sharing your amazing cabinet building skills with us.
     
  4. msommers

    msommers High on Epipens

    Pyrate
    Joined:
    Nov 6, 2016
    Likes Received:
    2,750
    Trophy Points:
    113
    Location:
    Calgary, Alberta
    Home Page:
    This is absolutely fantastic work!

    It also showcases what a clusterfuck speaker building would be for me LOL
     
  5. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Beautiful!
     
  6. murphythecat

    murphythecat GRU-powered uniformed trumpkin

    Pyrate Banned
    Joined:
    Apr 4, 2016
    Likes Received:
    1,201
    Trophy Points:
    93
    awesome job. some of you DIY guys really are good with wood
    you should have seen my 4 diy projects lol
     
  7. AdvanTech

    AdvanTech Friend

    Pyrate BWC
    Joined:
    May 13, 2016
    Likes Received:
    1,667
    Trophy Points:
    93
    Location:
    NYC
    This thread is awesome. @Jeb I know it's early but do you think these might be able to keep those SCM40's off your mind?
     
  8. philipmorgan

    philipmorgan Member of the month

    Pyrate BWC
    Joined:
    Sep 28, 2015
    Likes Received:
    3,790
    Trophy Points:
    93
    Location:
    In the wind, so to speak
    Home Page:
    Amazing work!
     
  9. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Haha, when I first saw the title of this thread I expected a huge three-way with Scan Speak mids and highs with an 18" woofer that I hadn't seen on the troelsgravesen site. Turns out it was 18cm.
    Once again awesome work! The cabinets (and pictures) are really nice. Those are some expensive crossover parts.

    The measurements look close enough, so how's the subjective tonality compared to the OSMTs? Are they more unforgiving because of a different tonality or is it a transient response thing? How's the treble texture on those ring radiators?
     
  10. Jeb

    Jeb Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    391
    Trophy Points:
    63
    Location:
    Bristol, UK
    Probably for the time being. It's hard to compare the two without listening side by side, other than to say they both sounded clean and clear. The midrange through that ATC soft dome driver was really fantastic and set a benchmark in my mind because it sounded transparent whilst also still being substantial. However, on a personal level things have changed because - for better or worse - DIY, learning new skills, and exploring different designs has become one of the main factors of enjoyment of music for me. I have no doubt that i'll probably end up spending as much on DIY that I could have just gone out and bought the ATCs in the first place - but where's the fun in that ? ;)


    Lol, yes it was tempting to just go straight to the monster 3-ways! I wanted to get a bit of a feel for what kind of sound Troels Gravesen aims for (if that's possible through just 1 project) before committing. As you say, the crossovers tend to be quite an investment. But the trend is definitely towards bigger drivers...

    In general terms I think the OSMTs sound warmer, mainly because there's that bit of mid-bass prominence to my ears. Also, when focusing on instruments, they sound a bit more full-bodied and fleshy via the OSMTs. Since the Discovery 18's have greater scale, dynamics, separation and depth the experience is bigger and more dramatic even though tonally they might be a bit leaner in the mids. Not to the point of being thin, I'd say. I could see some people thinking that they don't want to go leaner in the midrange than the OSMTs - and that's a fair concern I think. Troels uses a tube amp with these (possibly also a turntable) and i'm using a solid state amp that is well known to be a bit thin, so I can't say what's what right now. I feel like these would benefit from a bit of inner warmth.

    I did my best to volume match the speakers and then switch them over between each track. The Discovery's ability to better resolve small and larger shifts in volume is very noticeable to me, whereas the OSMTs sound more compressed in comparison. This, combined with the sense that transients through the Discovery 18s are faster, sharper and cleaner (maybe lacking a bit of that more natural, perhaps slightly extended decay that I liked via the MarkAudio wide full rangers) makes them a bit less forgiving... I think. The bass is tighter and has much greater impact, but feels as though it is probably in it's rightful place, like a weapon to be unleashed, rather than being a continual fuzzy warmish presence.

    Treble sounds mostly natural and smooth enough. In some of the modern recordings I listened to on Tidal there's a bit of strain and roughness in the very upper reaches of vocals. The in room measurements seem to show that the room was causing a bit of an uneven high end response. More than I got with the OSMTs, which were measured in a different room when I did that write-up.

    One track that impressed was 'Take Five' by Dave Brubeck. I know nothing about Jazz drumming save for what I gleaned from that cool film 'Whiplash'. Anyway, if the speakers can't portray the micro/macro dynamics, impact and finer textures then a lot of the emotion and drama of that drum solo seems to get lost and it can turn into a lifeless sideshow. It just sounded really fresh and new to me with these speakers.

    Please pull me up on anything I have said that might be contradictory.
     
    Last edited: May 17, 2017
  11. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Fabulous build! Troels has said a few times that if you had a choice between expensive drivers with cheap crossover components vs cheap drivers with expensive crossovers than you should always go for the cheap drivers. You simply gain much more from a good crossover network.

    I feel like you're going to get sucked in even deeper and build a Jenzen transmission line or one of his 4-way builds next.
     
  12. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    Just keep in mind that Troels is closely affiliated with a crossover component vendor. In my opinion he sometimes does go overboard with component quality.
     
  13. Jeb

    Jeb Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    391
    Trophy Points:
    63
    Location:
    Bristol, UK
    The crossover kit with the fancier capacitors was another €70 so it definitely gave me pause for thought. I have no idea if I would be able to discern a difference or if my system/source is even at a level to resolve such differences yet if they exist.

    On the one hand a big motivator for my DIY journey is getting value for money and learning what is sensible in a hobby that can get pretty crazy. On the other hand, when I'm putting a lot of work into it I want it to be the best it can possibly be. So, in the end I thought what the hell. It's just 2 capacitors. Some of those larger designs use 10 or more of those big caps.
     
  14. DigitalMaven

    DigitalMaven Acquaintance

    Joined:
    Nov 13, 2016
    Likes Received:
    25
    Trophy Points:
    18
    Location:
    tucson
    Since Troels likes to crossover higher (2500-3000 hz) and used mixed orders I've done my own theoretical mock-up of this speaker with my own third order crossover on both the low pass and high pass. You can use expensive caps but my design does not require that but it does require low dcr numbers in the woofer part as that effects the phase quite a bit. Also, I would like to add that this is a theoretical sim crossover using the manufacturers baffle measurements. I feel fairly confident that this sim is very very close to an ideal crossover using these speaker components. This is a half baffle-step compensation as you would lose way to much trying to do more with this woofer driver in a simple two way with much greater impedance resistance problems with larger inductors on the woofer. I chose a crossover point around 1450 as both of these drivers are very good and can handle a lower crossover point and this helps with the baffle step compensation a bit. If you want to experiment with the tweeter part to adjust the phase a bit that's what I would tweak only... the small inductor can be changed between .27 mh -.30 mh. If you would like to change the main woofer inductor from 2.2 mh to 2.5 mh it won't change the phase much but will increase the impedance at the crossover point by 3 ohms. You can change the resistors for the series part between .8-1 ohm and maybe the parallel between 5-6 ohm. Believe it or not these resistors do play a much larger role than people can believe by changing the electrical circuit (impedance) as well as the tilt of the crossover slope (up or down) and also the mild fine tuning of the final phase adjustments. I would like to add that I've taken careful measurements for speaker placement with the distance between the woofer and tweeter (vertical) at 160 mm with an offset of the tweeter and woofer of 25 mm. I do not believe that Troels speaker is exactly time aligned (close) so I added a -10 mm to the woofer for a 20-22 mm baffle thickness (total = -32 mm if flush with each other). My only caveat is that I'm not sure how the baffle diffraction between the woofer and tweeter happen as they are on different planes and also very close to each other ( the better the roundover or angled bevel between the woofer and tweeter helps reduce diffraction). This is maybe why a lot of people might shy away from this design but maybe the improved time alignment adds more than detracts so...

    Woofer crossover is comprised of 2.2 mh (dcr .15 aprox Jantzen powder core or superQ), 25 uf, and .60 mh (dcr ,23 Jantzen 15 gauge aircore ideal)
    Tweeter crossover is comprised of 16 uf, .30 mh, 39 or 40 uf use any capacitor with 400 volt or higher rating
    L-Pad after crossover and between tweater driver is .8-1 ohm series and 5-6 ohm parallel (my crossover uses .8 and 5 ohm in pictures) Please use a high quality resistors like mills or mundorf mox 10 watt (feel free to double them up with two 1.6 ohm and two 10 ohm in parallel)

    3rd order Butterworth-like Crossover with perfect phase transfer
    [​IMG]

    Impedance Friendly
    [​IMG]
    Raw Drivers vs. SPL crossover
    [​IMG]

    Transfer Function
    [​IMG]

    For those looking to do just the Flat Baffle vs. Stepped Baffle I've modified the crossover in the tweeter part. Keep the woofer Exactly the same but change The tweeter inductor from .3 MH to .37 mh (dcr .28) and the resistors to 1 ohm series and 8 ohm parallel. The phase is not as good but it;s very close.
    [​IMG]

    If you don't mind a slight bump at the crossover point on the flat baffle speaker I changed the parameters from .37 to .50 MH on the tweeter. You add a little db at the crossover point but the phase is better and the impedance is much better. Also I changed the l-pad to 1.5 ohm series and 15 ohm parallel. Here are the graphs. If I we're to build this crossover I would build this one for the flat baffle even though there's a bump.

    [​IMG]

    Impedance
    [​IMG]

    Inverse Polarity
    [​IMG]

    The big difference between the Flat baffle and the stepped baffle is the increase by 2 db in the crossover region.
     
    Last edited: May 19, 2017
  15. Jeb

    Jeb Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    391
    Trophy Points:
    63
    Location:
    Bristol, UK
    What's the plan here? Have you heard these particular drivers or any from the Discovery range? Are you going to build these cabinets with your revised crossovers and baffle dimensions and tell us what they actually sound like?

    Until you do that I don't see any reason for somebody to go to the trouble of building the cabs then deviate from a tried and tested design. A design that was born from not only having vast experience with ScanSpeak drivers but also from building and listening to 4 different iterations of baffle/crossover combinations with real music.

    I'm all for moving things forward. I'm sure your design is well thought out and you seem to have solid theoretical reasons for your choices. It might sound better. That's why I'm looking forward to your build thread and listening impressions.
     
  16. DigitalMaven

    DigitalMaven Acquaintance

    Joined:
    Nov 13, 2016
    Likes Received:
    25
    Trophy Points:
    18
    Location:
    tucson
    I've made plenty of mistakes on speaker design choices that can only be learned by trial and error. I've played around a lot with vifa's forerunner to this driver before all the designers went their separate ways (SB Acoustics, Scanspeak, Peerless-vifa) with the glass fiber cone weave in the XG18 woofer 8 years ago (the discovery drivers are like a hybrid of these drivers). My current two way consists of a seas h1212 and sb acoustics nrx17uc 6.5 inch driver. It took awhile to learn how to sim but it optimizes multiple things that you can see simultaneously and choose between the trade-offs because everything in speaker design is a trade-off. There are a lot of things troels does well conceptually but after looking at his crossover choices I do question whether he does know what he's doing by choosing to always try to crossover above 2500 hz no matter what in his high pass crossovers even if the woofers are not designed for it. You do not have to build this crossover but others might be interested in an alternative that performs better in the base frequencies as well as the midrange area. I like his website because he does have good choices in cabinet design and speaker components and some of his testing is pretty good. Most of all I like the manufacturers he uses to test out and play with. When I get some free time I'll try to reconstruct his existing crossover by plugging in his values. One thing that's missing from his design pages are some distortion figures and impedance graphs.

    Edit...... I just found his final crossover...which looks like it might work fairly well...it look to me like a 4th order on the woofer and a third order on the tweeter. The 4th order is probably necessary due to the high crossover point. Given that the tweeter has an FS of 500hz I wonder why he chose to crossover so high but I do believe that this is where he tries to do his crossover designs in most of his projects. Human hearing is most sensitive in the 500-3000 hz range.
     
    Last edited: May 19, 2017
  17. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Do you use the W18 on standmounts or on a desktop?

    When he's doing test of individual drivers he'll have distortion but won't usually post for a finished speaker. However he typically does post impedance for his speakers, including the Discover 18. Which you can see at the bottom of the following link:

    http://www.troelsgravesen.dk/DiscoveryW18_crossovers.htm
     
  18. Jeb

    Jeb Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    391
    Trophy Points:
    63
    Location:
    Bristol, UK
    They're on those Ikea bar stools you can see in the picture. They're a bit high. I need to chop the legs down a bit or sort out some proper stands.

    I'm not so sure about putting these on the desk - not only are they pretty big but I also tried various listening distances and felt that a little further away was ideal. Eventually settled on about 2.5 metres or so, which is where I took the in-room measurements.
     
  19. DigitalMaven

    DigitalMaven Acquaintance

    Joined:
    Nov 13, 2016
    Likes Received:
    25
    Trophy Points:
    18
    Location:
    tucson
    I especially like this at the bottom...

    [​IMG]

    "Final system impedance of LR4 crossover. Minimum 2.5 ohm @ 3 kHz. Don't worry, no transistor amp will mind."

    I think many a cheap amp out there might weeze on this...:)
     

Share This Page