Hi all,
Some 10 years after my first DIY speaker project, DIY Response 2.5 Clones, I finally was in the position to go for something a little more ambitious. I had my eye on Troels Gravesen's Ekta mkII for some years and, with the help of lockdown, expanded on it to become a translaminated active design. This project has taught me a huge amount over the last year or two. Still plenty to go I'm sure!
Crossover: miniDSP 4x10HD
Tweeters: Scan Speak D2608/9130 (8 Ohm)
Midrange: Scan Speak 12MU/8731 (8 Ohm)
Woofers: Scan Speak 18WU/4741 (4 Ohm)
Amplifiers: (very old) Audiolab 8000p on the tweeters and some over-specced LM3886 amplifier modules I also designed in lockdown on the mids and woofers.
Source: PC connected to miniDSP via XMOS USB to Coax bridge
The cabinets were designed in Fusion 360, using the same internal dimensions and driver spacing as the Ekta mkII. I have very limited/no access to much in the way of a workshop and so the plan was to get the parts CNC machined so it was a simple case of gluing and clamping, and gluing and clamping.... and gluing and clamping.
The assembly process did indeed turn out relatively straightforward. Ran into a couple of issues with tolerances height wise because the thickness of the MDF was about 10% greater than I had accounted for. Nothing that a few minutes with a router couldn't fix to trim down one of the layers.
Here we can some some snaps of the build process. Walls are lined internally with 5mm mass loaded vinyl, 8mm felt and polyester dampening cloth. Almost 40kg each fully assembled.
Next step was implementing the active crossover.
Raw driver measurements. Tweeter and mid are gated measurements. Bass Total is the sum of nearfield (as close as I could get) measurements of the woofer and mouth of the rear port.
Drivers EQ'd with Auto EQ.
The (theoretical) frequency response with LR4 crossovers at 350 Hz and 3200 Hz. Note the wiggles around 1k arose from using Psy smoothing because the midrange gated measurement didn't extend quite far enough down to its XO point. EQ for the final system was based on the gated response.
The response as measured in room is... as expected (left speaker measured @ approx. 1m, Var smoothing applied):
I believe the nulls at 30 Hz and 60 Hz to be a destructive room modes, 100 Hz is SBIR and... not sure what is going on from 150 Hz - 220 Hz. I've got 4 inch thick 2 foot by 4 foot acoustics panels behind each speaker, but that doesn't seem to be doing anything for the (potential) SBIR. You'll note that I added a high shelf filter in the PEQ of the miniDSP to bring the treble down 3 dB by 20 kHz.
I've had a few attempts at following the miniDSP app note regarding time aligning the drivers. I seem to get slightly different figures on each attempt. The tweeter is currently running 0.13 ms being the mid, which itself is 0.2 ms behind the woofer. Room to improve?
....Good?
One challenge with the translam design was moving house. They were build in a damp flat and a short while after moving into the dry, newish build flat, they duly started to rip themselves apart, I am guessing as the wood contracted as it actually dried out. There didn't seem to be any problems at the top of the cabinet where the mid 'enclosure' is. The nature of the design makes that section solid as a rock. Needed to apply go faster stripes (filler) to plug the gaps in main body of the cabinet.
Speakers in room (I've since hidden the chair that was directly in front of the right speaker...). Panels behind each speaker and 2 more directly behind the sofa at the listening position. The speakers remain unfinished as I have spent the last few months tweaking things to try and get them to sound decent. The expense of the drivers and lengths gone to with the enclosure, I was expecting to be 'wowed' but never fully got that feeling. To be fair, my latest attempt playing around with the crossover (as detailed in this post) is certainly better than the settings I had before and presently the speakers sound pretty nice.
So, long post, but just wanted to see if there were any opinions on where things can be improved. A couple things I would have probably done differently - because I didn't use a stepped front panel (can use DSP to align the drivers), I could have moved the tweeter and the mid much closer. I find that as sounds get close to the left or right speaker, they quickly get anchored to the mid driver. Not sure if this is due to the driver placement, room/layout or possibly poorly implemented active crossover! I did the LEDR test and it really seems to struggle with the height test - the sound doesn't really leave the tweeter much at all.
Another thing to experiment with would be sealing the port - it seems that most/some active speakers utilise sealed cabinets.
The miniDSP is fed from my computer by an XMOS USB bridge which gives the option for convolving various DRC and phase correction filters. I've had a bit of an experiment with these things but am yet to make any significant/substantial improvements. I did learn the hard way what pre-ringing sounds like, though, with one of my 'optimisations'. Am interested to look into these things further, though.
One day would like to move to class D amplification, the Purifi modules look very interesting. The physical buzzing of the transformers in my DIY amp is quite impressive and the relatively poorly implemented internal wiring does create a bit of unwanted hiss if you go close to the speakers.
Anyway, there we go.
Oli
Some 10 years after my first DIY speaker project, DIY Response 2.5 Clones, I finally was in the position to go for something a little more ambitious. I had my eye on Troels Gravesen's Ekta mkII for some years and, with the help of lockdown, expanded on it to become a translaminated active design. This project has taught me a huge amount over the last year or two. Still plenty to go I'm sure!
Crossover: miniDSP 4x10HD
Tweeters: Scan Speak D2608/9130 (8 Ohm)
Midrange: Scan Speak 12MU/8731 (8 Ohm)
Woofers: Scan Speak 18WU/4741 (4 Ohm)
Amplifiers: (very old) Audiolab 8000p on the tweeters and some over-specced LM3886 amplifier modules I also designed in lockdown on the mids and woofers.
Source: PC connected to miniDSP via XMOS USB to Coax bridge
The cabinets were designed in Fusion 360, using the same internal dimensions and driver spacing as the Ekta mkII. I have very limited/no access to much in the way of a workshop and so the plan was to get the parts CNC machined so it was a simple case of gluing and clamping, and gluing and clamping.... and gluing and clamping.
The assembly process did indeed turn out relatively straightforward. Ran into a couple of issues with tolerances height wise because the thickness of the MDF was about 10% greater than I had accounted for. Nothing that a few minutes with a router couldn't fix to trim down one of the layers.
Here we can some some snaps of the build process. Walls are lined internally with 5mm mass loaded vinyl, 8mm felt and polyester dampening cloth. Almost 40kg each fully assembled.
Next step was implementing the active crossover.
Raw driver measurements. Tweeter and mid are gated measurements. Bass Total is the sum of nearfield (as close as I could get) measurements of the woofer and mouth of the rear port.
Drivers EQ'd with Auto EQ.
The (theoretical) frequency response with LR4 crossovers at 350 Hz and 3200 Hz. Note the wiggles around 1k arose from using Psy smoothing because the midrange gated measurement didn't extend quite far enough down to its XO point. EQ for the final system was based on the gated response.
The response as measured in room is... as expected (left speaker measured @ approx. 1m, Var smoothing applied):
I've had a few attempts at following the miniDSP app note regarding time aligning the drivers. I seem to get slightly different figures on each attempt. The tweeter is currently running 0.13 ms being the mid, which itself is 0.2 ms behind the woofer. Room to improve?
One challenge with the translam design was moving house. They were build in a damp flat and a short while after moving into the dry, newish build flat, they duly started to rip themselves apart, I am guessing as the wood contracted as it actually dried out. There didn't seem to be any problems at the top of the cabinet where the mid 'enclosure' is. The nature of the design makes that section solid as a rock. Needed to apply go faster stripes (filler) to plug the gaps in main body of the cabinet.
Speakers in room (I've since hidden the chair that was directly in front of the right speaker...). Panels behind each speaker and 2 more directly behind the sofa at the listening position. The speakers remain unfinished as I have spent the last few months tweaking things to try and get them to sound decent. The expense of the drivers and lengths gone to with the enclosure, I was expecting to be 'wowed' but never fully got that feeling. To be fair, my latest attempt playing around with the crossover (as detailed in this post) is certainly better than the settings I had before and presently the speakers sound pretty nice.
So, long post, but just wanted to see if there were any opinions on where things can be improved. A couple things I would have probably done differently - because I didn't use a stepped front panel (can use DSP to align the drivers), I could have moved the tweeter and the mid much closer. I find that as sounds get close to the left or right speaker, they quickly get anchored to the mid driver. Not sure if this is due to the driver placement, room/layout or possibly poorly implemented active crossover! I did the LEDR test and it really seems to struggle with the height test - the sound doesn't really leave the tweeter much at all.
Another thing to experiment with would be sealing the port - it seems that most/some active speakers utilise sealed cabinets.
The miniDSP is fed from my computer by an XMOS USB bridge which gives the option for convolving various DRC and phase correction filters. I've had a bit of an experiment with these things but am yet to make any significant/substantial improvements. I did learn the hard way what pre-ringing sounds like, though, with one of my 'optimisations'. Am interested to look into these things further, though.
One day would like to move to class D amplification, the Purifi modules look very interesting. The physical buzzing of the transformers in my DIY amp is quite impressive and the relatively poorly implemented internal wiring does create a bit of unwanted hiss if you go close to the speakers.
Anyway, there we go.
Oli
Last edited: