Build of Keramiskas DIY Speaker

[chych7]

Just wanted to share my build and some measurements of the Keramiskas DIY speaker (Meniscus Audio Link), designed by Craig Salin. This is my first ever speaker build and complex wood working project. I picked this speaker since it looked like a good beginner DIY build, and numerous people who've listened to them have commented on their great sound. I thought they came out reasonably well (the finish could have been better), and sound great too. Some build pics are in the attachments.

I had a few modifications from the original build plans; I opted for the component upgrades, used Baltic Birch instead of MDF, made my back panel removable, and used more and thicker tile on the inside (for constrained layer damping) than specified in the design, which led me to enlarge the box dimensions slightly. One speaker weighs a hefty 32 lbs. I also embedded magnets for a magnetic grille (not built yet).

My (not well controlled) measurements were done in my HT room using a Denon 3600H AVR with a UMIK-1 placed 12" from tweeter on its axis; REW software was used. Also plotted are my old Ascend Acoustics CBM-170 (reviewed on ASR) measured in a similar fashion, which is my point of comparison. I have some acoustic panels around the test setup, but there are a lot of room modes and artifacts showing up in both speaker's response.

Freq Resp 1/48 smoothing (Green = Keramiskas, Brown = Ascend CBM170)

Distortion plot for Keramiskas:

Sound wise, bass extension is much lower than I would have expected; roll off starts at 40 Hz, and I would estimate a -3 dB point of around 30-35 Hz (Ascends are listed to have -3 dB around 53-58 Hz, approximately indicated by the cursor on the plot), which is great for a 2-way bookshelf with a 5" driver. I wonder if the high amount of internal cabinet damping from the tiles is helping bass extension, which speakers don't normally have. Highs have a dip in the 6-10 kHz region (Ascends are noted to be bright sounding), and don't roll-off at 20 kHz. Distortion plot is also included, shows increasing distortion for bass freqs (Ascends are lower, <1% everywhere). Subjectively, the stereo imaging of these speakers is excellent, instruments are noticeably more pin-point than the Ascends, and they throw a wide sound stage. Highs are smoother/less harsh sounding, and bass is clearly stronger. Sensitivity is a lot lower than the Ascends (Ascends are -7 dB to match levels), so these are power hungry.

Open to ideas of how to improve these measurements with the limited tools I have, I'm relatively inexperienced with these type of measurements. Hope someone finds this useful; I thought this was a fun project and yielded a great sounding speaker.

 

[bigjacko]

Nice wood working! The speakers look very professionally built. Do you know what was the SPL for distortion test? How did you make the constrained layer damping and what was the lossy material?

 

[dougi]

Try to do a FR plot using impulse response gating as well to get a better idea of mid/treble response. You can also go the whole hog and do near field and blend them all together. MiniDSP has a good guide on how to do this. I am well surprised by the bass extension at first glance! If you can calibrate the SPL measurements and do the distortion at known SPL like 85, 90, 95 that will be a better indication. Bass distortion looks pretty good though.

 

[chych7]

Nice wood working! The speakers look very professionally built. Do you know what was the SPL for distortion test? How did you make the constrained layer damping and what was the lossy material?

Thanks! The distortion data was taken from the same data as the FR, which was about 85 dB. The constrained layer damping is just ceramic floor tile stuck on the interior walls of the speaker with a generous amount of Silicone adhesive. This results in a very dead sounding cabinet.

Try to do a FR plot using impulse response gating as well to get a better idea of mid/treble response. You can also go the whole hog and do near field and blend them all together. MiniDSP has a good guide on how to do this. I am well surprised by the bass extension at first glance! If you can calibrate the SPL measurements and do the distortion at known SPL like 85, 90, 95 that will be a better indication. Bass distortion looks pretty good though.

Here's the FR plot with 4 ms gating for both speakers, it still looks like some reflections are showing up. Do you have a link to the specific guide for the near field measurement? Not seeing it on the minidsp website.

For SPL and distortion, this was taken at 85 dB. I can't go higher than this, REW indicates clipping. Apparently the gain in my UMIK-1 is set too high (factory default), and I need to open the microphone it to change it... which is proving to be challenging... I did measure the other (right) speaker and it has lower distortion, so now I'll have to figure out what's going on with left (both have similar freq response).

Right speaker distortion, 85 dB:

 

[dougi]

Yeah, the near field guide must be somewhere else. You still then need to simulate box diffraction to also take that into account. Your 1m measurements look good and distortion fine. My commercial speakers have a fair bit of difference between them too. I can't help with the umik issue though!

 

[bigjacko]

The constrained layer damping is just ceramic floor tile stuck on the interior walls of the speaker with a generous amount of Silicone adhesive. This results in a very dead sounding cabinet.

I have researched on the lossy material, people say that silicone is not good because it does not convert the movement into heat, the movement will not be damped but continue to go. The material that damp the movement has a property called viscoelastic, the rebounding speed is slow when you press into it.

 

[chych7]

I have researched on the lossy material, people say that silicone is not good because it does not convert the movement into heat, the movement will not be damped but continue to go. The material that damp the movement has a property called viscoelastic, the rebounding speed is slow when you press into it.

Yes that is true a viscoelastic material will absorb vibrational energy (which is frequency dependent) than an elastic material. There are various types of silicone sealant/adhesive/caulks, and they have different elastic/viscoelastic behaviors, depending on their formulation. The silicone sealant I used was advertised to be flexible window/door sealant. If I stretch the stuff and let go, it goes back but slowly, not instantly (like a rubber band) - that's an indication that it is viscoelastic. I have used bathroom silicone sealants that dry to a very hard, inflexible caulk; this would be more elastic and not appropriate for the constrained layer damping use. Also, even if the adhesive material was fully elastic, the heavy tile creates mass loading; this by itself will reduce vibrations transferred to the outer cabinet.

 

[bigjacko]

@chych7 Have you tried this before? https://patents.google.com/patent/US20060231327A1/en It is designed by Gedlee. The layer is not between two walls, but two sticking out rod. I think this way there will be longer shear distance.

 

[chych7]

@chych7 Have you tried this before? https://patents.google.com/patent/US20060231327A1/en It is designed by Gedlee. The layer is not between two walls, but two sticking out rod. I think this way there will be longer shear distance.

Looks like an interesting idea, I'd be uncertain of how to tune such a bracing. I'd think it would have to have a certain amount damping for the brace to critically damp the flexural vibrations of the cabinet. Too much damping and it's the same as a regular brace, too little and it doesn't function as a brace. This probably requires some actual modeling, experimentation, and damping materials understanding to optimize.

 

[Wolf]

I thought Craig did a great job on those, as I've heard the original pair, and they do disappear really well.

Nicely done!

 

[dtaylo1066]

I have two of those woofers and am just waiting to pair them with a tweeter in a similar sized two-way monitor. This Keramiskas design had caught my eye a while back as it seems to have gotten good reviews.

Here is the other design I am aware of for the SBA 8 ohm ceramic 5-incher: http://iamp.diyaudio.pl/?sb-monitor-sb15cac30-08-sb21rdc-c000-4

The tribute design to the late Jeff Bagby by Selah Audio looks sweet but is for the 4 ohm version. I am not a crossover guru so I want to build from a proven design. I think those white cones look fantastic.

 

[chych7]

I have two of those woofers and am just waiting to pair them with a tweeter in a similar sized two-way monitor. This Keramiskas design had caught my eye a while back as it seems to have gotten good reviews.

Here is the other design I am aware of for the SBA 8 ohm ceramic 5-incher: http://iamp.diyaudio.pl/?sb-monitor-sb15cac30-08-sb21rdc-c000-4

The tribute design to the late Jeff Bagby by Selah Audio looks sweet but is for the 4 ohm version. I am not a crossover guru so I want to build from a proven design. I think those white cones look fantastic.

Looks to be a good design with a dome tweeter. The Kermaiskas have the ribbon tweeter (crossed at 3 kHz with a sharp 4th order slope), and are a little larger, being the primary differences.

Another design is the Travelers, which also has an 8" woofer and extends lower, but this may be too large for what you are looking for.

 

[dtaylo1066]

Yes, I have seen the Travelers. Not looking for a box that big, though I think they must kick ass. I have never heard the Hi-Vi mylar tweeters.