DM-4 Reference Upper Cabinet 2-way Speaker Build Part IV

cad_front_upperThe speaker project continues! I managed to find a few hours in the last couple of weeks to finally start the upper cabinet speaker portion of my sweet DM-4 Reference Speaker build. If you haven’t read my original introduction to these speakers, then you might want to start over there first. The inspiration for this design originates from both the Wilson Audio Watt and the Von Schweikert VR-5 hi-fi speakers. It’s sort of a marriage of the two designs, taking the best of the aesthetics of both and incorporating them into one totally awesome speaker system – and then ditching the passive crossover for a fully active DSP-based crossover with all the bells and whistles. It’s going to be my greatest achievement yet!

The dimensions of the upper cabinet are 12-1/2″ at the bottom with a 7° slope on each of sides ending up at a width of 9″ at the top. The speakers are 21″ deep at the bottom and 18-3/16″ at the top. The back baffle is perpendicular to the bottom but the front baffle is tilted back 14° to physically time align the woofer and the tweeter. The very top front edge has a 3″ chamfer cut at 14° (which mimics slightly the VR-5) which also reduces the effects of edge diffraction. There are 1/4″ chamfers around all exterior edges which finish off the cabinets (similar to the VR-5) which also aid in edge diffraction. These are the basics of the design which gives it that distinct look that I was shooting for with this whole speaker project. The sloped sides, the sloped front, the chamfered top edge all give the speaker that classic “Watt/VR-5″ look.

So just a few words on how I designed some of the shape and dimensions for this speaker because I think there are important factors with each parameter. This is not a just carbon-copy/knockoff speaker but is truly designed for the drivers I have selected. I came up with 14° sloped front baffle by drafting up side views (or cutaway views) for the 7″ Dayton woofer and 1-1/8″ Dome tweeter in Delta CAD using the .pdfs from Parts Express’s website and then played around with the angle to align the drivers until they were just right. Phase/time alignment is achieved by placing the acoustic centers of each speaker in the same vertical plane. It is believed that the acoustic center of a speaker sits roughly at or around the location of the voice coil. Since the voice coil location is not well defined in the datasheets, it can also be approximated by assuming the voice coil is centered within the top plate of the magnet assembly (not the magnet itself). I considered alignment was met by setting the baffle back such that the centers of each voice coil were aligned vertically. In the case of the Dayton RS-28F tweeter and RS-180P woofer, this turns out to be about 14° assuming the drivers are 1/2″ apart on the front baffle. This suits me just fine since most off-axis measurements are taken in 15° degree increments, so when seated directly in front of the speakers, you will be listening to both the woofer and tweeter at about 14°-15° off axis. The tweeters will sit at about 44” from the floor when resting on the lower cabinets. Based on the FR plots of both drivers, they both have a very well-behaved response at 15° off axis with only slightly more high-frequency roll-off than when listening to them head on. The exact baffle angle probably doesn’t matter too much as far as timing goes since the crossover is going to be MiniDSP-based (active), so I will have independent control over the time/alignment of the woofer and tweeter anyway. But someday I may decide to re-purpose these speakers without the active DSP so I will have to design a passive crossover for them and in that case I’ll be all set with the drivers physically already time aligned. I have to admit, my experience in this area is mostly textbook since I have not actually ever built a speaker with the drivers’ acoustic centers vertically aligned in this fashion. Like most speakers, I just allow the vertical offset to play out however it plays out on a normal, flat, vertical baffle. So this will be fun for me to see just how much it changes the way the speakers the sound.

Another cool dimensional design consideration is the 14° chamfer across the top/front baffle edge. So what purpose does this little design detail serve? It is just for looks? Did I just do it to copy the VR-5? The Wilson Watt doesn’t have it, so it is even necessary? The answer is both yes and no. Aesthetically, it completes the speaker, it’s that final detail that just makes the speakers great and sets them apart from the rest. It gives it that hi-fi, high-end look to it by simply cutting that corner at the top edge and smoothing its transition from the front to the top so that it, well, just looks cool. Sonically it should also reduce the effects of baffle edge diffraction around this surface. Baffle edge diffraction occurs anywhere there is a discontinuity on a surface where sound waves are present. I believe the Wilson Watt combats this problem with a type of acoustic foam attached directly to the front baffle around the tweeter and woofer. This way the sound waves traveling across the front of the baffle are attenuated before they even get to the edge, therefore potentially reducing the audible effects of edge diffraction by the listener. While in principle this should work, it also looks ugly. It’s probably one of the main reasons most speakers do not have some sort of acoustic foam on the front baffle surrounding the drivers, it can take a great-looking speaker and make it look terrible. Plus, I think it complicates the design in an area that does not need to be made complex. The construct of the tweeter, the voice coil, dome materials, doping methods, glues, the quality of construction, crossover type, frequency, etc. all play a much larger roll in whether or not a speaker sounds good. Once you start focusing on just baffle edge diffraction, even with a $50 tweeter, you’re really looking only to fine-tune what should already be a great-sounding driver. It’s taming that last little ripple in the response that your microphone is still picking up in the measurement that you just can’t let go. But that’s not to say that some speakers aren’t designed with edge diffraction as their sole feature with the entire focus being around proper driver/cabinet dispersion. Any speaker enclosure with huge rounded edges or extra ordinarily wide front baffles are taking into consideration these edge effects. Take almost the entire line-up of Thiel speakers for example or Sonus Faber. These guys believe whole heartily in the ill effects of edge diffraction and have designed in oversized front baffles and large-edge roundovers.

This effect can be modeled and there are tools out there for doing so. One of them is Diffraction and Boundary Simulator by our favorite Excel guru Jeff Bagby. Not only does the edge cause unwanted disruptions in sound but the location of the tweeter/woofer on the front baffle plays heavily into how bad these disruptions really are. Offsetting the tweeter from the center changes the distance from the radiating source such that the edge effects occur at different times with respect to each other. This can help reduce the overall effect of edge diffraction making the speaker’s FR response smoother. Unfortunately, the best-simulated design for diffraction tends to also be the worst-looking. So a designer must balance form with function in this sense. The complexity of the shape of the DM-4 cabinet doesn’t lend itself to an easy edge diffraction model and Jeff’s tool doesn’t exactly allow the sloped sides or a single 14°chamfer across the top to be modeled. But after researching it a bit more I came across another boundary simulation tool from the guys over at FRD Consortium called Baffle Diffraction Simulator. This tool will let me create the exact shape of the enclosure and then model its response. I’ve been playing around with it as well and it’s been pretty interesting the results. You can see a couple different screenshots below. I can say that the 1/4″ chamfer on all the edges, as small as it is, does provide as much as ±0.5 dB improvement in frequency flatness above 2 kHz. And the tapered sides and sloped front baffle provide a measurable difference as well when compared to a flat rectangular box with no chamfered edges. So moving on…

Another critical element to reducing edge diffraction effects is to flush-mount the drivers. Measured response plots can reveal the effects of edge diffraction for a tweeter mounted on the surface of a baffle (which might be only 5mm thick) compared to being flush mounted. Beside flush mounted drivers look soooooo much better. Finally a win-win for diffraction and aesthetics! I always like to flush mount my tweeters and depending on the design, will also do the woofers. Though the benefit of flush-mounting the woofers which have a lower crossover point is minimized due to the longer wavelengths. It still looks better to me!

So I’ve written plenty for this one post. The only thing I didn’t go into was the bracing technique I used. I’ll save that for another post, but the bracing definitely has some intentional design behind it and isn’t random. Recently I’ve adopted this design technique that starts with the bracing and enclosing a box around it, instead of building a box and adding bracing to it. It makes building the speakers really fun and is actually quite easy. The main consideration for doing the bracing this way is to brace each panel sufficiently such that the fundamental panel resonance is pushed out of range of the woofer (>2 kHz) as much as is practical. Also it mimics the lower cabinet bracing design. This box will be insulated using the same technique as the lower cabinets (with the carpet pad, cotton fill and polyfil) and should be extremely damped and extremely dead overall. The front and rear baffles are 1.75″ thick and the sides top and bottom are 1″ thick. Tapping/rapping on the cabinets with my knuckles provides a nice, dead, thump in response. So they sound very solid, very dead. Anyway, here’s the pictures, you may draw your own conclusions of the bracing design. Is it enough to do the trick?

Alright, enough talking, here’s all the pictures. Check them out below. And check out the video montage as well. Enjoy!

Posted in DIY Audio | Leave a comment

Damping Methods, First Listen and Measuring the DM-4 Speakers Part III

dampingdm4smallThe time has finally come! The moment when a speaker design has reached the stage where you finally get to install the drivers and have a listen to your masterpiece. I have to admit, I struggled through this middle phase of building these speakers. The part where I spent hours and hours tediously cutting strips of carpet pad, spray-gluing them into the cabinets, only to repeat the process with another layer of carpet pad and glue, then two more layers of a cotton/poly (80/20) blend fabric (and glue) and finally another two layers of 100% polyester batting and once again, more spray glue. This whole job just stunk. My garage was HOT (middle of summer here in AZ) the glue was sticky, it got all over everything, including me and halfway through the process I realized the glue wasn’t holding and all the pieces were coming off! Complete nightmare. I remember the days when all I used to do was cut a big piece of fiberglass insulation and staple-gun it the box and be done, 1/2 hour job, tops. (Okay, I did that just recently, but still).

So I don’t know what I was thinking when I decided to do this whole multi-layer-different-material-glue job catastrophe. It didn’t help that each panel in the speaker is so compartmentalized, what with all the bracing every 4″ to 8″. Every piece had to be measured and cut to fit a specific location and then repeated over and over again. It was so stinking time consuming and honestly I have no idea if it’s any better or any worse than just putting a big wad of fiberglass in there. For the sake of argument, I am just going to say that it is better. So much better than I am super excited to get to do it ALL OVER again when I build the upper cabinets for the midrange driver and tweeter. Yeah. The fun has just begun.

So enough bemoaning, I chose this hobby, it’s my own fault. The basic idea behind the damping method was to start off with a relatively dense material up against the MDF and progressively use lesser and lesser dense materials working my way outward (or inward?). The idea being that in order to absorb the most bandwidth, I would need multiple different types, or densities of materials. There really is no one-size-fits-all when it comes to good broadband sound absorption. I started off with 2 layers of 6-lb carpet pad which makes up about a 3/4″ layer of sound absorption that I suspect does a good job in the low-to-mid frequency range. Next up I have the 80/20 cotton/poly fabric winch is also about 3/4″ thick. This was sort of an impulse buy but I liked it because it felt denser than the 100% polyester but lighter than the carpet pad. A perfect “in-between” weight for those “middle” frequencies. I used two layers on all the sides except for the back panels which has 4 layers. And last in the stack-up is the famous and favorite, 100% polyester batting. This material probably works well over a fairly broad frequency range and its effectiveness is more or less just dependent on how much you use. I lined what was left of the inside of the boxes with with another 2 layers (which looks more like 4 layers) of the polyester fill making up another 3/4″. So in total the walls are lined with this custom-fabricated, multi-layer, sound-absorbing compound that is 2.25″ thick and should have good overall acoustic properties. When you consider the wood in the mix, that’s another 1″ for a total of 3.25″ of sound-deadening, sound-absorbing action. The only thing you will hear coming from these speakers will be from the actual drivers themselves. That’s the hope anyway. It’s a concoction all right, but it’s my concoction and I like it. One day I will recreate each layer in a sounds booth or something and measure them independently just to see what it really does.

So along the way I did very unscientific sound checks with each new layer which consisted of me sticking my head in the box and singing different tones and simply listening for how much the boxes echoed or resonated. The boxes without any damping had a very apparent resonance in the mid-to-low vocal range. The lowest note I could sing, the boxes just resonated like crazy. Once completed however, the boxes felt very “dead” and had very little echo or resonance. In hind sight, I wish I had measured the cabinets throughout the process, so I’d have something scientific to back up my damping technique, but like I mentioned earlier, I hated this part enough already, add in the complexity of trying to take frequency and impedance plots along the way, I probably would have gone crazy. It seems like most people just take the trial-and-error approach to speaker damping anyway (me included) and so I fell victim to the same ploy, I guess, just out of shear laziness on my part. Sorry folks, but I will say this, this damping technique seems to have worked out perfectly in this 4th order cabinet to satisfy the requirement of “walls lined” and matching the corresponding Absorption (Qa) factor. To the best of my measurement ability, the measured frequency response and impedance plots line up very closely with the modeled performance. My plan is to carry this method into the upper mid/tweeter cabinet but instead of merely lining the walls, the entire cabinet will be filled with the 100% polyester batting in addition to the layers of carpet padding and cotton/poly blend fabric as previously defined. But since that enclosure will be sealed, and will have a high-pass active filter at around 300 Hz, I am really only looking for maximum sound absorption over a very broad frequency range and not too worried about over-damping or loss of bass response. So I am going to fill that cabinet up.

Okay, so now it’s time to drop in the drivers and the port and take some measurements! For now the only measurements I am going to show is nearfield FR and impedance plots. I have my entire setup in my garage and so room modes were apparent in all my measurements and I wasn’t in the mood to tear it all apart and re-build it outside in the backyard. That is coming, just not yet. My main thing for now was just to do a quick check on the damping and tune the boxes so everything is as close to modeled as I can get before I stain and paint the boxes and call them done. And of course just to give them a quick listen!

My measurement setup consists of an old laptop running REW 5.16, an ECM8000 microphone and stand, a Behringer 1202EQ Mixer (for phantom power) and a Denon AVR-1801 receiver (for the amp). This was the first time I had done actual impedance plots using REW and it worked out flawlessly. I love REW, it impresses me every time I use it! Thanks again to the guys who developed it for people like me. =) Impedance plots can serve several purposes: they show you box tuning, box damping, can reveal if there are leaks and of course, they show you the overall impedance vs. frequency of your speaker system. It’s more useful in that sense when combining multiple drivers and complex crossover networks, but even a simple 2-driver set-up it’s nice to see a solid 4 ohm impedance for two 8 ohm drivers in parallel.

So what I chose to do was deliberately cut the port long and take measurements with progressively shorter and short port lengths just to show how the port length affects FR and impedance. I didn’t really do any exhaustive listening tests since the speakers were being driven full-range and I was doing the measurements in my garage, which is not the final destination for these puppies. So as I mentioned earlier, the alignment I was shooting for in this design was a BE4 (Bessel) alignment according to Vance Dickason which should result in improved transient response (for a 4th order design) and has the lowest tuning of all the various alignment options. Which in a 78 liter enclosure is 32 Hz given the T/S parameters of the RS225P-8 drivers.

So what the plots show is that for every inch I cut off the port, the enclosure Fb increased by ~1 Hz. Enclosure Fb can be measured using the Impedance measurement tool in REW in addition to a 100 ohm resistor. Fb is located at the lowest point between each of the two largest impedance peaks. A more accurate way to measure Fb is to measure each peak, label them Fl and Fh and then seal or plug the port and measure the peak again can call if Fc. Then you can calculate Fb with the following equation: Fb = sqrt(Fl^2+Fh^2-Fc^2). So I made 4 measurements with the port length starting at 10″ and ending up at 7″. The tuning corresponded by starting at 29 Hz and ended up 32 Hz in these tests. This can be seen in the impedance plots below. I am happy with the 32 Hz tune for now. It represents the lowest tune that still fits within a classical alignment type and doesn’t give up too much in way of lower midbass extension. But it does not represent a flat response either, which would have needed to be tuned way up at 38 Hz with a modeled f3 of 43 Hz. That just wasn’t low enough for my liking especially considering how big these cabinets are. I’m a little disappointed that I won’t get extension into the 20’s. But that’s okay, because I already have a matching subwoofer in the plans to compliment this whole setup that will get me into the teens. So there’s no sense wasting that low frequency stuff on these little 8 inch speakers. Besides, I did get a change just to play some extremely bass heavy music (just to break in the drivers a little bit) and even with a 32 Hz tune, they can drop some extremely deep bass. At one point my entire garage was bumpin’. Of course, not the intent of these speakers, but they really sounded quite awesome.

So I ran a whole bunch of frequency response plots with the speaker in the corner and in the middle of the garage, with the garage door open, near-field, mid-field, far-field, near the top cone, near the bottom cone, near the port, etc. There’s a huge room mode at about 25.5 Hz, which showed up in all the plots. That calculates out to be exactly 1/2 wavelength at 22 feet – the distance wall-to-wall in my garage. So I left it as good enough for now and will wait to do the measurements outside once I finish the upper cabinets. The frequency plots looks very good though. They match up nicely with the predicted responses as modeled in Unibox. See the Excel screenshot above. Even the impedance plots look very good as well. The impedance peaks are a little higher in my enclosures which suggests I do not have quite as much damping as the model, but that only means I can easily go up from here and add more as needed. Besides, if I had to error on one side or the other, I’d rather be a little less damped and pick up some extra bass then to be overdamped and bass shy.

So that’s about it for now. The plan from here forward will be to start the upper cabinets for the mid and tweeter. I have finally finalized the design and am ready to start cutting wood! I’ve got the materials all ready to go, I just need a nice long weekend to get to it. Hmmm, wonder when that will be? I’m thinking probably Christmas. That way it’s cooler outside and my garage won’t be a sweat shop. Enjoy the photos until then! And of course the cheesy Google video montage.

Posted in DIY Audio | 1 Comment

DM-4 Reference Loudspeakers Part II – Bottom Cabinets Nearing Completion

So it’s been a couple of weeks since my last post and I’ve made significant progress on the ‘ole speaker project. I figured it was time to do a quick write-up and post some more pictures. Last I left off I had completed 5 of six sides and left all of the internal bracing exposed. The plan for damping is still in work as well as figuring out some means of measuring the enclosures in stages with progressive amounts of damping with different materials in different quantities until I achieve what I would call “reasonable expectations” with regard to resonances. Or in other words, I want to get the structural and acoustic damping “just right” and only my ears and some measurement equipment is going to be able help me get there. Which means the speaker enclosures needed to be done so I can get to point of dropping in the speakers and taking some measurements. Besides, I still haven’t come up with the perfect combination of glue/foam/carpet pad/cotton/egg crate/wool/polyfil to dampen these puppies just right.

So I left off with the second back panel, main front panel and one side incomplete. So I glued and screwed on the second side to each speaker. I then installed the back panel which is glued to the oval-patterned frame/brace which makes the back panel 1.5″ thick but not throughout the entire panel. There are oval cutouts in the first panel while the second is solid. This stiffens the back panel without adding too much additional mass and buys me over 1 liter of extra internal volume.

The front panel went on next and was measured with a makeshift compass (nail, envelope from Ace Hardware and a pencil) and cut with a jigsaw to mount each 8″ Dayton driver and the 7″ port. The driver opening is oversized by 1/16″ of an inch to allow some wiggle room for mounting each driver. The backside of each opening was routered with a 1/2″ round-over bit to provide “breathing room” for the back/rear of the driver into the cabinet. Note that the internal panel, or the first main front panel, the opening was oversized by 1″ which leaves even more breathing room for the rear of the driver. The front panel is now 1.5″ thick with a pair of 2″x3/4″ cross braces between the drivers and between the lower driver and port. The cross braces also connect to the sides as well as the first window brace internal to the cabinet. The idea here being the more braces that are structurally tied to one another, the better damped the braces will behave, and therefore the entire enclosure. Doing a simple knuckle tap test reveals just that, this sucker it well braced and solid! But it still sings like a canary since nothing is actually damped, yet.

At this point I am basically done with the 3/4″ MDF (except for the 2nd bottom piece which will go on last) and the boxes look awesome but they also look terrible! Screws and holes and glue are everywhere! There’s no way I’m going to try and putty and polish this thing up to paint it. It just wouldn’t be worth my time. So instead I did my old trick of using sheets of 1/4″ MDF panels to “finish” the speaker cabinet. The 1/4″ front panel provides the exact depth required to flush mount each driver. Then a piece on the top and back provide a super-nice and even finish which I will ultimately be able to sand and paint to a beautiful semi-shine. All those crummy MDF edges and screws get covered up in one shot. The 1/4″ MDF plus the 3/4″ MDF means the top is now 1″ thick and the front and back panels are 1.75″ thick. The enclosure is getting stiffer and beefier by the minute and not to mention, a lot heavier!

So while I was considering installing all these 1/4″ MDF panels, I decided I didn’t want a uniform finish on all six sides of this speaker. I didn’t want just an all-black speaker and I didn’t want to to have to try and paint the entire speaker and make it look perfect, because I knew I would not succeed. So the idea crossed my mind to use a real-wood veneer for just the sides of the speakers and then do the other 3 faces in the 1/4″ MDF. This is a design scheme that I’ve grown rather fond over the years and a technique I’ve done on several speakers I’ve built in the past. The real wood stained combined with a nice semi-gloss black should look awesome together. Plus I can stain wood only slightly better than I can paint it, or at least staining seems to be more forgiving for me. So if I can get the stained sides to look great, then all I really need to get perfect is the front and the top. I think I can handle that.

I picked up two 4’x4’sheets of 0.20″ (1/4″) Birch plywood from Lowe’s. I rummaged through the only 8 pieces they had and picked two that had some nice, unique natural wood grain patterns that I thought would look cool when finished. I cut each piece to fit the sides of each enclosure and oversized each piece by 1/8″. Each piece was glued to the cabinet wall using Liquid Nails. I applied pressure via my own body weight to press the Birch into the glue as much as possible. Once it was well seated, I popped in a some 1″ (18) gauge brads into each corner to prevent the panel from drifting, flipped it over and did the other side. Then I did the other speaker and stacked them on top of each other. I added two really old speakers I use in the garage for music along with 240 pounds worth of QuickSet concrete bags (one of which got wet from the rain and is literally a solid block of concrete). This was just enough to keep some even pressure on the entire surface of the newly glued pieces of Birch plywood. I let that dry for 24 hours before tearing my mass tower down.

This is the point where I added the 1/4″ MDF to the front, top and back. The front baffle was cut using a Jasper Circle Jig from Parts Express. This tool is awesome! Since this is a visible cut, I didn’t want to use my jigsaw, I needed that cut to be a perfect circle. With 1/16″ increments, I made some practice cuts to pre-fit how much gap I wanted between the edge of the frame of the drivers and the edge of the baffle. One size was too big, so I dropped it down, then it was too small, so I picked a spot right in the middle. Perfect fit! The overall size ended up being 0.031″ larger than than driver (or ~1/32″) which leaves about a 0.015 inch gap all the way around the driver. This gives me some wiggle room to fit the driver as well as some room to grow from paint/polyurethane. You can see from the pics though, the fit is perfect, if anything I went too small and drivers won’t fit after I paint them! I am going to have to be careful and watch that inner diameter. I may even need to mask it off after a certain number of coats to keep it from being undersized when I’m done.

Anyway, I used liquid nails again on all the 1/4″ pieces mainly because the viscosity of Liquid Nails is so much thicker than wood glue that I have a much better chance of controlling the squeeze out. The reason this is so important at this stage is because both of the enclosure sides have that beautiful Birch ply finish which does not take kindly to a smattering of runny wood glue. Since it’s only Birch ply, the actual Birch is extremely thin and doesn’t give me much option to sand down/away any place that the wood glue would have touched. With the Liquid Nails I can ensure that the Birch is untouched while the other panels are being glued in place. Even then I still covered the Birch with 2″ blue painters tape, just in case. The old speakers and 240 pounds of concrete sat on top of the speakers for another 24 hours while all the 1/4″ panels dried in place. Again, the 1″ brads held the panels in place to keep them from walking while it was drying.

So now that all the panels are installed and dry (and they aren’t coming off, ever), since I had oversized all the pieces, I needed to use my router to trim each piece flush to each side. This was accomplished using a 1/4″ flush bit and a router. I took the router to each side and zipped along its edge at just the right depth so as to only trim the MDF without touching the Birch. Once complete, this left a perfectly flush edge on each side of the speaker from the MDF to the birch and the MDF to the MDF (front-to-top-to-back). Oh yeah, I had done this already to the Birch plywood before installing the 1/4″ MDF since it was also oversized when installed. That was made near-flush to the 3/4″ MDF leaving some room to account for the thickness of the Liquid Nails (about 0.015″).

I am almost done at this point but there’s one more finishing cosmetic touch to add, and that’s the 1/4″ 45° chamfer on all the 1/4″ edges. This look, I have to admit, I borrowed from the VR-5’s but it really does give the speakers just a little bit of extra class, something notable, something different, that overall fits the look and style of this speaker perfectly. I will still have to seal this edge, since it’s just MDF, but having that angle to it on each side meet at a point at each of the four corners just looks…so cool. Totally subjective sure, but for the cost of a 45° chamfer router bit and the 20 minutes it took to router it, in my mind, it was well worth it. From a baffle edge diffraction perspective, a roundover would have had less diffraction than the 45° chamfer but it’s still better than a 90° edge. And in practice to get any significant benefit or reduction in edge diffraction, the roundover needs to be quite large, which in my opinion, ruins/changes the aesthetic of the speaker. Arguably the upper frequency limit of these cabinets will neither benefit nor suffer from edge diffraction effects but I intend on carrying this design into the upper enclosures which will run all the way to 20 kHz. So it’s at least important to keep it in mind. The overall shape of the upper cabinet should help with diffraction effects also, but I’m getting off topic, and ahead of myself.

So with that, that’s about it, I’ve reached by 1000+ word limit so it’s time to just put up the rest of the pics and let them tell the rest of the story. Enjoy!

Posted in DIY Audio | Leave a comment

The DM-4 Active Reference Series Speaker Project Has Begun!

20160403_102731I have officially broken ground on my latest and greatest speaker project – The DM-4 Active Reference Series Loudspeaker. This is one of those projects that has been in the back of my mind for what seems like forever that is finally coming to fruition. I wrote a lot of the history and background over on my website Audio Innovation (which is dedicated to all of my speaker projects) so I won’t go into it again here. But I find it much easier to do write-ups and post pictures in WordPress than trying to do it on my old-school web site (which I still use Frontpage to create!) So to document some of the build process, I am going to post pics and details here out of simplicity (or laziness) on my part.

So this will be Part I of the DM-4 Active Reference Series Loudspeaker build write-up and pictures blog. As I get further and further along I will continue to update this site with pictures and descriptions of how the build is coming along. I expect this build to be a work in progress and to extend over the course of the next few months. The goal would be to complete this project by Christmas time this year (2016). I’ve waited this long to start this project, I might as well take my time actually putting it together.

So yes, if you haven’t noticed, the inspiration for this design stems from two fairly popular high-end commercial speakers – the Von Schweikert VR-5 and The Watt/Puppy by Wilson Audio. So read my introduction over at Audio Innovation if you’re curious as to why I would even want to build a pair of speakers similar to these insanely expensive commercially available speakers. Call it nostalgia, call it crazy, call it whatever you like, but the inspiration for this design originates from these two popular speaker systems. While an exact copy is not my intent and I hate to even use the term copy-cat, with that said, there is plenty of DIY thrown into the mix in terms of enclosure volume, physical size, tuning, shape, fill material, driver selection, crossover, finish type, wood selection, etc. So enough of the intro, let’s get onto the build pics!

I’ll try to keep the narrative to a minimum and let the pictures do most of the talking. I’ve CAD’d up the design using DeltaCAD (which is a great tool for drafting speakers, or any other project for that matter) as well as I just sketched up some drawings on paper of what the speakers will look like (my kids think the only thing I know how to draw are speakers – they may be right). I’ve only cut wood so far on the lower cabinet. I’m still tweaking some of the design/dimensions on the upper cabinet as well as drafting up the cut sheet. So far now I have just got the lower cabinet started.

cad-drawing-1Each cabinet measures 12″W x 30″H x 21″D and is made from 3/4″ MDF with extensive window and cross bracing. Front and rear baffles are double-stacked 3/4″ MDF with an additional 1/4″ MDF baffle board for flush mounting the drivers. Total expected volume will be 75L-80L for both drivers. Tuning frequency will be between 32 and 39 Hz using a Precision 4″ flared port from Parts Express. Volume and tuning was initially modeled using Unibox 4.08. This design represents nearly a perfectly aligned “Standard Tuning” enclosure with minimal fill and minimal leaks. I picked up my old copy of Vance Dickason’s Loudspeaker Design Cookbook and calculated some different 4th order alignments given the formulas in his book. Unibox doesn’t assign an alignment descriptor such as the ones that Dickason described. I guess the assumption is that since you can plot the desired response you don’t need to design to any of the traditional filter types since the possibilities are near infinite. I’m not even sure any of the modern design tools really mention filter type/alignments anymore. So I calculated three other classic alignments, the SBB4, QB3 and a BE4 (Bessel) and put those values into Unibox so I could plot them and see how they compare. Note that Parts Express also recommends a volume based on BassBox 6 Pro which appears to align somewhat with a SBB4 or a QB3 alignment but oddly enough they don’t define the tuning frequency of the box. Stating only f3 does us little good, without fb one cannot know what the correct tuning is suppose to be, which the the most important part of a 4th order vented design! Anyway, I may shoot for the BE4 alignment, for two drivers the enclosure should be 77 L with a tuning frequency of 32 Hz. It’s got a gentler slope than the Standard Alignment or the SBB4/QB3 alignments which should result in improve transient response, according to Dickason. The Step Response function in Unibox confirms this theory to some degree though the difference in timing between alignments is marginal.

DSC_4908The drivers that will compliment these cabinets are a pair of Parts Express 8″ Reference Paper Series woofers which were recently released (2015). I haven’t seen too many builds with these speakers but when they first came out, I knew they would make a perfect driver for my VR-5/Puppy lower cabinet. While more VR-5 than Puppy (due to the phase plug) they model very well in the enclosure size I was shooting for. They have a matched set of 7″ drivers for the midrange plus that great Dayton Audio 1-1/8″ dome tweeter. All around, I thought this would make a great speaker system to flaunt the Dayton Audio Reference name from top to bottom.

Just some quick notes about the cabinets so far – the design intent for the enclosure is to increase/maximize stiffness of wall/panels while reducing mass, i.e., maximizing internal volume – light but stiff panels to increase the resonance frequency of each panel with the objective being zero audible panel vibration within the passband (and beyond) of the two drivers. There is a lot to read about the subject, as I have learned, and as it would seem there’s more than one way to brace a box from the overkill, to the simplistic, to the complex to the, how the heck did they do that? After reading all about calculating panel resonances of MDF, I had originally set out to design the perfect balance of bracing, stiffness, volume and mass using straight up science (as in, math) but later bored with the all the numbers and figures and decided to just, well..wing it, as I think a lot DIY’ers do. But all is not lost, even though the project is already underway, it’s still fun to figure out just what I’ve got and perhaps I can tweak the bracing if I haven’t already over-engineered it. I have found some great tools online that I have been playing around with, LISA being one of them ( which is a FREE Finite Element Analysis modeling tool for up to 1300 nodes. This program is so cool! I’ll go into the details another time, but so far I am able to model and show animations of the panel modes and calculate the frequency of several modes. It’s pretty slick.

2unbracedpanelmode1Anyway, so what I have come up with is both simple to cut, easy to build and should provide no single panel surface area of more than about 20-25 before being constrained by either a 2″x3/4″ brace, a full 3/4″ cross brace or a panel edge. So let’s break this down a bit, if you take the largest surface area on this speaker box, the massive sides, you end up with a panel that is 630 square inches. Without any bracing, worst-case scenario, this panel is held, supported, constrained, fixed or “bound” on a total of 4 sides. This panel will have a fundamental resonance at a single frequency, let’s call it fs (which is a function of Young’s Modulus, Poission’s Ratio and the density of MDF). This will be the main mode, or the 1st mode (and in most cases will follow a classic drum mode), with the center of the panel flexing in and out at fs with an amplitude of z. There will be other modes, higher order modes, 2nd, 3rd and 4th and so-on, but with each higher-order mode, the frequency increases and the z amplitude decreases for the same input energy. So emphasis can be placed on killing the first couple of modes with all other higher-order modes improving respectively as bracing is added and stiffness is increased. It should be noted that the z axis motion, or amplitude of the resonance corresponds to damping. Increased damping results in the decrease of the amplitude of the resonance and does not move or change the frequency of resonance. Both concepts have real implications in speaker enclosure design and both principles need to be considered when shooting for a “solid” speaker enclosure. My plan was to shoot for optimizing (increasing fs and stiffening) panel resonance modes first via bracing using MDF and then attacking damping second via some other materials. Lastly is to consider internal cabinet fill material (like polyfil or insulation) which serves a much different purpose than the aforementioned techniques (which is mainly to attenuate the massive internal reflections that result from a basically square box made of wood which really likes to reflect sound). But that requires a blog for another day!

In fact, this entire post deserves another post for another day. So let’s just get to the pics and I’ll stop talking for now. Enjoy!

The following week I accomplished a little bit more. They are starting to really take shape now! Also the upper cabinet design is getting a few bracing tweaks.

Here’s a video montage of the above picture gallery and some short video clips just for fun.

Posted in DIY Audio, General, Home DIY, Music | Tagged , , , , , , , , , , , , , | Leave a comment

Weekend Project: Two-way Bookshelf Speakers with Spunk

Mini J Two-Ways (42)UPDATE:After 6 years with these speakers I’ve made some tweaks to the crossover to improve the response. See below for details. Now back to the original post…

We listen to a lot of music in my family, there are a vast array of speakers and stereos and mp3 players in this house. Whether on the go, in the car, or relaxing at home, there’s always a song playing somewhere. As my oldest daughter’s birthday was approaching, I decided I wanted to build her a small pair of bookshelf speakers for her room. Something that she could enjoy for many years to come. Something that was made by Dad just for her.

And so I present the “Mini J” two-way bookshelf speaker system. This speaker starts with a Dayton Audio Designer Series 5″ woofer and a 1″ Vifa soft dome tweeter. The DS series woofers from PE represent an excellent value, have a great look, and surprising bass response for a driver this small. Decent Xmax and a low fs make this little driver perform very well in a small cabinet size. The great upper frequency extension make crossing over to almost any tweeter a walk in the park. And for the tweeter I picked the Vifa DX25TG59-04 which has great power handling, low fs, and a smooth, flat response out to 20 kHz. It’s got a wide-roll surround around the dome and just has a great overall look to it. Besides, I’m a big fan of Vifa drivers and have been using them for many years.

miniJcadAfter picking the drivers I started with the cabinet design. I drew up some initial plans and started tweaking the cabinet volume and shape until I came up with something I really liked. I used Unibox to model the speaker response and ended up going with the Standard Design model which yields an f3 of 59 Hz in a 5.4L cabinet tuned to 56 Hz with no hump or dip in the response. This requires an enclosure size of 11.25x7x9 (HxWxD) using a mixed panel thickness of 1/4″, 1/2″ and 3/4″ MDF. Parts Express had also recommended a similar enclosure volume but they also recommended specific dimensions that meet what’s called the Golden Ratio. So I gave it a shot and I intentionally made the height and width conform to this Golden Ratio which is 1.618:1. There’s some great reading about the Golden Ratio over on Wikipedia if you’re interesting in killing a few more minutes. Aesthetically the most pleasing rectangular shape to look at and quite possibly exhibits superior sonic properties than other ratios. Hey, but I won’y get into that here. They look great to me and they sound fantastic too, but now I’m getting ahead of myself.

With the cabinet design complete, I moved onto the crossover design. I love the Designer Series drivers from PE because they provide FR and ZMA data for easy import into crossover designing tools such as Passive Crossover Designer. While the Vifa tweeter did not have raw data, I still made good use of SPL Trace to create data from the datasheet by tracing the FR and ZMA plots into data I can actually use. I don’t know why ALL driver manufacturers don’t provide this type of data. A picture of a plot in a .pdf file is hardly enough for doing any kind of real crossover design. You could argue that the manufacturer raw data isn’t exactly ideal either, but it’s a start.

summedresponsemodeledI’ll try and be brief on the crossover design and my methods of doing crossover design because quite frankly it’s just that, my method, and I’m still tweaking and proving my method with each new speaker design. I’m not sure I’m there yet, but this is actually a big part of the fun of speaker building. I always shoot for the simplest crossover with the fewest elements to achieve the flattest FR and a decent impedance. I typically add a Zobel network to the woofer to flatten the impedance above fs which helps with the high-frequency roll-off. The woofer is a 12dB/octave set at about 2,700 Hz with a cap value that is about double the textbook design value. This provides a sharper roll-off without peaking just before cutoff. The tweeter also ended up being a 12 dB/octave but wired in phase with the woofer. The inductor in this case is also slightly tweaked to be lower than the textbook value which also increases the slope slightly and according to the simulation blends/sums well with the woofer on-axis. I also added a 4 ohm series resistor and a 20 ohm shunt resistor (aka L-pad) to pad the tweeter and match the overall level of the woofer. It also brings my impedance of the system up to around 8 ohms which is where I wanted it for an easy load to even the cheapest amps.

crossoverimageNow that I had some crossover values assigned I started putting parts in my cart at Parts Express. Another fun part about building speakers, getting to buy everything. Since the crossover design at this point was just a model, I do like to buy multiple different values of capacitors and resistors so I can tweak things in a listening environment. I didn’t buy a bunch of different value inductors because they are quite expensive and I just hoped that my simulation was close enough to allow me to only tweak the caps once I got closer to being done. Caps are cheap, especially electrolytics, so I bought every value from 3.3uF all the way up to 15 uF which would allow me several tuning options on both the woofer and the tweeter and the zobel network. I also bought several values of resistors so I could adjust the tweeter level as needed. One day I would like to have every standard value inductor/cap/resistor just so I can have the ultimate freedom to tweak but that project will have to wait for another day.

I was just about ready to cut wood at this point, so I drew up a cut sheet and got started. The cutsheet in this case is a little unique. The box design consists of (3) panels of 3/4″ MDF that make up the front, brace and back each piece being 6″x10.25″. The sides and top and made up of 1/2″ MDF which are cut at 8.75″x10.25″ (sides) and 8.75″x7″ (top/bottom). These cuts were a breeze on my table saw and were designed this way so as to allow the table saw to be set to each dimension only once and every cut made so that every cut that is dependent on a flush fit when assembled is exactly the same size, even if the saw isn’t cutting each piece at exactly the width it should. It doesn’t matter because all the pieces that fit together that require that dimension just end up being the same. The only cut that matters is the width of the top and bottom pieces which need to be cut to whatever the width of the front/brace/back ended up being +1.0″. With a table saw, every cut comes out near perfect anyway but even if they didn’t, this design allows for a little slop in each cut while still providing a perfectly flush fit.

Anyway, I cut the wood, built the boxes, sanded the boxes, painted the boxes, built the crossovers, tweaked the crossovers, measured the responses, installed the drivers, added some polyfill the port and the terminal cup and alas I was finally done. And just in time for my daughter’s birthday the next day. I was able to whip these out in only 4 days mainly just working a few hours each night after the kids had gone to bed. So they are a super easy project but were a lot of fun to design and build and they sound absolutely fantastic. I like the fact that I could tailor the sound a little knowing what type of music will be played on these speakers. While they don’t have a lot of low-end presence, they make up for it with a smooth, solid-sounding midrange and treble. I’m not all about that bass (no treble) and can appreciate a speaker’s ability to produce vocals without coloration. The boxes are rock solid too, very little resonances despite the 1/2″ MDF. But enough fluff, here’s the meat and potatoes. Click here for a parts list from Parts Express. Check out some of the build pics below as well as pics of the final product. Hope you enjoy!

Finally, scroll past the pictures for in-room measurements and near-field plots of various crossover options and L-pad values I had considered. In the end, I chose what “sounded best” to me regardless of what the plots actually looked like.

So with most speaker projects just getting that first listen is an awesome feeling. These speakers definitely sound great. Good balance between the woofer and the tweeter and they have a really good soundstage, so music doesn’t sound like it’s just coming straight out of these two drivers. You close your eyes and listen and you can’t really tell where the speakers are located. The music just fills the room in front and behind. And I love that. No harshness out of the tweeter either. You can put your ear right up to it and it sounds clean and smooth. So here’s some FR plots. I don’t try and read too much into every little bump and valley, I’m mainly looking for an overall balanced sound. The slight dip near the crossover freq is mainly due to the close proximity of the microphone in between the drivers which actually shows up in the PCD model as you move the listening position closer to the speaker. I suppose if I had a little more time and a few extra cap/inductor values on-hand I may have tried achieving a slightly flatter response, as I am sure it is achievable, but for now I’m going to call it good enough and overall I am happy with the results.

The last thing I did was take my measurement gear outside and ran some FR plots in basically a near-anechoic environment. I took a bunch of measurements at several different distances with both the tweeter wired in-phase and wired out of phase (meaning + to + and – to – on both woofer and tweeter) therefore actually making the speakers acoustically out of phase due to the 12 dB/crossover yet quite possibly back in phase due to the delta in physical relationship between the acoustic centers of the woofer and the tweeter. Confused yet? These plots at least tell me that everything is summing properly on-axis which yields the flattest FR. Notice that these speakers do not have baffle step compensation as is evidenced in the plots below. I decided not to incorporate it since these speakers will almost always be backed against a wall in a very small space, I didn’t want the bass and lower-midrange region to be too aggressive. You can see from all the measurements, in-room included, that there is a definite rise in amplitude between 500-700 Hz. John Murphy of True Audio came up with a quick formula for approximating the -3 dB point given the baffle diameter (or width) which is f3 = 380/W where W is the width of the baffle in feet. These speakers are 7″ wide, or 0.588 feet which amounts to a -3 dB frequency equal to 655 Hz. That matches well with the measured data shown here. Even without BSC, the speakers sound fantastic in-room, but if I were to implement it, I wouldn’t do more than 3 dB and I would shoot for a corner of about 655 Hz.

2021 Update: So I’ve recently got back into measuring and tweaking some of my speakers and this was one pair that I always wanted to improve. Looking back at the response plots it’s pretty obvious that the crossover design fell just short of providing an nice, even, flat response. It looks like baffle losses and maybe a bad summing at the crossover point result in this kind of bumpy, uneven response. I think at the time I was just pushing to finish these before my daughter’s birthday the next day I called it good enough. This weekend I spent a little more time and with a handful of additional crossover parts on-hand, I made some minor adjustments that really make this speaker shine now. If you built this speaker based on the original design then I present an updated crossover which provides about 3 dB of baffle step compensation, moves the crossover point of the woofer and tweeter down a bit and lowers the overall output of the tweeter to balance it with the new baffle step network. The result is a much flatter response, more bass, less midrange presence and better matching with the tweeter at the crossover point.

The new 2021 crossover can be seen here. The basic adjustments are as follows: first off I increased the series 0.8 mH inductor to 1.3 mH and the shunt cap from 8 uF to 17.7 uF. This pushes the start of the crossover, or the start of the baffle compensation circuit, to about 300 Hz, where the slope is rolling at about 6 dB/octave. By the time the response hits 2 kHz we are down 4 dB from where we were before. The larger cap helps to set a new 12 dB corner of about 1600 Hz at which point the crossover transitions to a 12 dB octave slope so we are 30 dB down by around 6.5 kHz. There’s some leftover cone break-up modes at around 4.5 kHz which you could clean up by shunting a 1 uF cap across the 1.3 mH inductor, but I believe it was not necessary as the peaking is pretty subtle. The original Zobel network was left unchanged. Overall this brings the response from about 400 Hz to 2 kHz down a bit making that region in the response much flatter than it was before.

The tweeter now needed to be attenuated a bit to match the new lower response of the woofer plus it needed to be pushed lower in frequency to fill the gap that the woofer left due to the lower crossover point. This was accomplished by increasing the series cap from 3.3 uF to 7.8 uF and increasing the 4 ohm series resistor to 8 ohms. This provides a pretty decent crossover point to the woofer and provides the flattest on-axis response shown here. I’ve given these speakers another listen and believe they sound much better than before, more neutral, more bass. If you’ve got them near a wall, the treble might sound a little flat, you can opt for a 6 ohm series resistor in the L-pad instead of the 8 and brighten it back up another dB or so. Anyway, that’s the update, check out the before and after response plots below and let me know what you think!

Posted in DIY Audio | Tagged , , , , , , , , , , , , , , , | 12 Comments

The Ultimate DIY Hall Tree

DSC_5566I don’t know exactly when it happened or how it came about exactly, but one day my wife said to me, “I want a hall tree”. I replied, “Um, you want a tree in the hall?” She said, “No, crazy, I want one of those things, like an entryway bench, but with coat hooks so you can hang your coat and the kids can hand their backpacks.” Ah yes, I did know what she was talking about and a quick Google image search confirmed my suspicions of exactly what this so-called hall tree looks like. “Can you build me one of these?” Sure I thought, no problem, looks easy enough, how big do you want it? We walked over to the proposed location of the hall tree and she started at one end of the hall and said, “How about we start it here” and then she walked and walked and walked until she reached the other end of the hall and said “to here”.

“What? that’s like 10 feet long. You want a 10-foot long hall tree?”
“Yes, with 6 cubbies, one for everybody in the family.”
“Including the baby? But he’s only two! What does he need his own spot in the hall tree for?”
“Well, one day he will grow up and he will need a spot for his stuff too.”

Good grief. Okay, let’s draw it up and see what we can do – a 10-foot long hall tree with a spot for everybody in the family, including the two-year old baby. So basically what I came up with was a single 5-foot wide hall tree with 3 spots or cubbies and I would just build two of them. Just from the standpoint of trying to move it and the weight alone, this made sense. I had a few constraints that sort of defined this size, which in my opinion can really be scaled to any size that works in your home if need be. But this thing was actually going to go in our hallway to the garage. We rarely leave or come in the front door, so the hall tree would be going in our hallway which is only about 5 feet wide. So this hall tree had to be shallow, I mean at most, I wanted it only a foot deep, thus not to intrude into the hallway and be a nuisance. Next I capped the height at 72″ or 6 feet. My plan was to build the this hall tree entirely out of 3/4″ common pine which come in 6′ pieces at various widths ranging from 1-1/2″ to 12″. Also I have a light, a sconce, that is about 6′ from the floor on the wall and while I could have moved it up to allow a taller hall tree, it wouldn’t have been in the same location as the other sconces throughout the house so I decided just to leave it. I also could have removed it and added an overhead light but honestly didn’t see a need for the hall tree to be much taller than this. So with the basic exterior dimensions settled on, I started to design out the rest of the unit.

For purposes of this write-up I will assume we are building just one hall-tree, or half of what I built here.

It starts with a base structure made from 3 pieces of 1x12x6′ common pine which make 3 cubbies that are 17″ wide x 15-1/4″ high by 12″ deep. (Note that 1×12 common pine is actually 3/4″x11-1/2″).

Base Cut Pieces:
Top (1) = 1x12x54″
Bottom (1) = 1x12x52-1/2″
Exterior Sides (2) = 1x12x20-1/4″
Interior Cubby Sides (2) = 1x12x15-1/4″
Bottom Face (1) = 1x4x52-1/2″
Base Board Molding = 1/2″x4-1/4″x54″+12″

I will attempt to make some drawings of how this goes together since I did not take near enough pictures to show how it goes together. But it’s not too hard to figure it out from the finished pics. Butt joints are made where every piece is glued and screwed together. The top part of the hall tree is made up of the following cut pieces:

Upper Cabinet Cut Pieces:
Sides (4) = 1x10x49″
Top and Bottom (1) = 1x10x54″
Shelves (3) = 1x10x17″
Back (hook part) (3) = 1x6x17″
Trim Top (1) = 1x11x55-1/2″

Same thing with the bottom, all the pieces are butt joints which are glued and screwed together. With this part complete, it doesn’t look like much actually. It isn’t until the base molding and side facing and cove molding is added that it starts to actually look kinda nice. Those are all made up of 1×3 pieces of pine which are just cut to fit the sides. I did leave a 1/2″ overhang on the back to cover the 1/2″ pieces of red pine which make up the back. The 11/16″ cove molding sits nicely on the inside corners of each piece of 1×3″ pine to again give it a finished look. The back is what I really love. I stumbled upon these 5″ wide tongue and groove style red pine boards in the lumber section and thought they would look great as the backing to my hall tree. They give it that great french country look having a real groove every 5″ for that bead-board look without being fake. And since they are real wood, I decided not to paint them and instead finished them with a couple coats of a satin polyacrylic just to bring out the natural colors in the wood and make it look nicer. It contrasts with the white everywhere else and to me just finishes off the whole thing in style.

Anyway, here’s some pictures of just the wood, before being painted, and then done. I will try and get some more drawings of how it goes together if anyone is interested. As well as a complete parts list. Though honestly I don’t imagine anyone trying to duplicate this design exactly, for us this was a pretty custom solution that met a specific need for us. But if nothing else, I hope it might inspire others who are looking for something similar to at least consider what can be built with just a few pieces of lumber from your local hardware store.

Posted in Home DIY | Tagged , , , , , | Leave a comment