Need/Use Case: I’ve always wanted a decent bluetooth speaker that I could carry when traveling/hiking/camping/etc. but could never find one that really provided a quality listening experience. Most of the larger manufacturers with bluetooth devices seem to be dubious with their ratings and make speakers that appeal to the mass market with little regard to any sort of controlled directivity (not that omni is necessarily a bad thing when it comes to BT speakers).
In addition to this, I run a sound system for small-medium sized events and have been looking for a no-frills fill box that I can throw in a corner and feed a signal to without having to worry about it, so the appeal of a battery powered “set it and go” box is high for me.
Background: I’ve had a fascination with multiple entry horns over the last year or so having prototyped a few larger format MEHs at this point. If you know anything about MEHs, there are some considerations regarding the size, flare rate, and horn profile to achieve what they do. Going small with a MEH means you do lose out on some of these benefits if you plan to do a full-range build.
However, the primary guiding reason of this speaker’s arrangement is not necessarily to benefit from hornloading the 6.5” woofers, but to offer a highly compact arrangement that wouldn’t be possible with a flat baffle. As an additional benefit this places the acoustic centers of the drivers very closely and provides a good directivity match between the woofers and compression driver, despite the steep crossover. What it doesn’t do is give a substantial sensitivity boost/impedance match to the woofers that you would get from a larger conical MEH.
Build Requirements: I set out to design a speaker that would attempt to hit these metrics:
The Design
Acknowledging that I wasn't constrained by a straight wall conical horn, I opted to utilize Ath to design a waveguide that would provide a more uniform wavefront than would be possible with a pyramid shaped horn. For this reason, I’m hesitant to call it a horn and lean more towards a waveguide (I’m loosely defining this based on my understanding of Geddes’ classification).
For components, I opted towards using the new B&C DH350 compression driver, two 6NDL38 6.5” woofers, and two Scanspeak SB16PFCR-00 passive radiators (ports were out of the question due to the small enclosure size). This blend of components lends itself nicely to a crossover point somewhere in the 1.0-1.2kHz region, so that is what I targeted in Ath/Hornresp.
To achieve that, there was a good deal of fine tuning/tweaking the models in Ath/Hornresp. The basic idea was to really nail the directivity/impedance curve in Ath, then (loosely) translate that profile into Hornresp so that I could leverage the MEH wizard to integrate the woofers and find the point at which they could meet the CD.
Knowing that my crossover would be somewhere in the 1-1.2kHz area, I then sized out and placed the woofer taps in Fusion along the corners of the horn where the centroid of the taps is about ¼ wavelength @xo from the CD exit. I then modeled a flange for mounting the woofers leaving a flat area open so that I could come in later on and add cone fillers if needed.
My 3D printer bed isn’t large, so the model is split into 3 pieces - a left and right half with a throat adapter. I used 2 part clear epoxy to join the halves together. The woofer mounting nuts are heat-set M4 inserts.
Because of the proximity of the drivers, the 2 M4 fasteners holding the compression driver in may be hard to reach. I used a stubby hex drive to get to them when installed into the enclosure.
The enclosure is a simple hexagonal shaped box made from 12MM birch plywood with about a 15 liter volume after accounting for the displacement of the horn/drivers.
Unfortunately I misjudged the amount of clearance I'd need for the passive radiators, so they do butt up against the rear of the woofers when installed. If I do another box, I’m going to add a couple extra mm clearance there. The good news is that there is a hole in the basket of the passive radiators that lines up perfectly with the 6NDL38’s pole vent, so I don’t anticipate any overheating issues there.
During the initial testing I taped the front of the waveguide to the enclosure. This went about as expected, but I was at least able to get the data I needed. I use an outdoor ground plane measurement setup according to this guide from Scott Hinson. Measuring with a Dayton audio UMM-6 with a 20 cycle FDW applied at 1 meter (2 meters for off-axis). 30°C at 72% humidity.
For this first prototype, I wanted to validate the Abec/Hornresp model and check physical fitment - wasn’t going for a final rendition. The thought was to build the initial prototype and gradually add cone filler until I achieved the desired woofer response. For the most part, this was a success, but for whatever reason there was a secondary peak in the woofer response that I hadn’t anticipated. I’m going to consider my failure to anticipate this a happy accident, because it meant that the woofers conveniently worked out perfectly to meet the compression driver on the first try. If you have any insight/theories into why that secondary peak might exist, I’d love to hear.
Barring that, the unfiltered responses of both drivers overlaid looks like this.
For the woofers, I opted for a 1st order 80Hz LPF to flatten out the response, which got me 80% of the way to where I needed to be. After applying some corrective EQ, I was able to get a solid response from 60Hz-16kHz. I’m applying a 4th order 45Hz HPF to protect from overexcursion. After generating the spinorama I realized I probably could have done more on the EQ front to correct the 500-2k region, but I’m happy enough with it and don’t really wanna do 36 more sweeps just to make a squiggly line less squiggly.
Distortion as measured with stepped sine tones:
Group Delay:
And finally polars (normalized on axis):
Given these results, I decided to go ahead and finish out this prototype. There’s some details in the physical construction of the box I might make in a future version, but overall I was quite happy with the performance. I chose a Dayton Audio KABD-4100 amplifier paired with one of their 6 cell 18650 BMS boards, mostly due to availability and balance of features. One of the outputs was left unplugged, so it's not the most efficient setup but I found it works quite well for the price. The sigmastudio file is provided in the zip file.
Regarding the enclosure, whatever you build make certain it's well sealed. You may want to modify the flange on the waveguide for a cleaner mount. These little drivers move a lot of air and I spent the better part of an afternoon working out caulking/gaskets to fully take care of air leaks.
(I've since added a Neutrik combo XLR jack to the rear)
Specs
Compromises: These are the areas that I’ve identified as weaknesses in this build or areas for improvement.
In addition to this, I run a sound system for small-medium sized events and have been looking for a no-frills fill box that I can throw in a corner and feed a signal to without having to worry about it, so the appeal of a battery powered “set it and go” box is high for me.
Background: I’ve had a fascination with multiple entry horns over the last year or so having prototyped a few larger format MEHs at this point. If you know anything about MEHs, there are some considerations regarding the size, flare rate, and horn profile to achieve what they do. Going small with a MEH means you do lose out on some of these benefits if you plan to do a full-range build.
However, the primary guiding reason of this speaker’s arrangement is not necessarily to benefit from hornloading the 6.5” woofers, but to offer a highly compact arrangement that wouldn’t be possible with a flat baffle. As an additional benefit this places the acoustic centers of the drivers very closely and provides a good directivity match between the woofers and compression driver, despite the steep crossover. What it doesn’t do is give a substantial sensitivity boost/impedance match to the woofers that you would get from a larger conical MEH.
Build Requirements: I set out to design a speaker that would attempt to hit these metrics:
- Full range with at least 100dB output down to 50Hz
- Wide 60x90 pattern with well controlled directivity in both horizontal and vertical axes
- Highly portable (<15 LBS and <25 liters)
- At least 8 hours battery life
The Design
Acknowledging that I wasn't constrained by a straight wall conical horn, I opted to utilize Ath to design a waveguide that would provide a more uniform wavefront than would be possible with a pyramid shaped horn. For this reason, I’m hesitant to call it a horn and lean more towards a waveguide (I’m loosely defining this based on my understanding of Geddes’ classification).
For components, I opted towards using the new B&C DH350 compression driver, two 6NDL38 6.5” woofers, and two Scanspeak SB16PFCR-00 passive radiators (ports were out of the question due to the small enclosure size). This blend of components lends itself nicely to a crossover point somewhere in the 1.0-1.2kHz region, so that is what I targeted in Ath/Hornresp.
To achieve that, there was a good deal of fine tuning/tweaking the models in Ath/Hornresp. The basic idea was to really nail the directivity/impedance curve in Ath, then (loosely) translate that profile into Hornresp so that I could leverage the MEH wizard to integrate the woofers and find the point at which they could meet the CD.
Knowing that my crossover would be somewhere in the 1-1.2kHz area, I then sized out and placed the woofer taps in Fusion along the corners of the horn where the centroid of the taps is about ¼ wavelength @xo from the CD exit. I then modeled a flange for mounting the woofers leaving a flat area open so that I could come in later on and add cone fillers if needed.
My 3D printer bed isn’t large, so the model is split into 3 pieces - a left and right half with a throat adapter. I used 2 part clear epoxy to join the halves together. The woofer mounting nuts are heat-set M4 inserts.
Because of the proximity of the drivers, the 2 M4 fasteners holding the compression driver in may be hard to reach. I used a stubby hex drive to get to them when installed into the enclosure.
The enclosure is a simple hexagonal shaped box made from 12MM birch plywood with about a 15 liter volume after accounting for the displacement of the horn/drivers.
Unfortunately I misjudged the amount of clearance I'd need for the passive radiators, so they do butt up against the rear of the woofers when installed. If I do another box, I’m going to add a couple extra mm clearance there. The good news is that there is a hole in the basket of the passive radiators that lines up perfectly with the 6NDL38’s pole vent, so I don’t anticipate any overheating issues there.
During the initial testing I taped the front of the waveguide to the enclosure. This went about as expected, but I was at least able to get the data I needed. I use an outdoor ground plane measurement setup according to this guide from Scott Hinson. Measuring with a Dayton audio UMM-6 with a 20 cycle FDW applied at 1 meter (2 meters for off-axis). 30°C at 72% humidity.
For this first prototype, I wanted to validate the Abec/Hornresp model and check physical fitment - wasn’t going for a final rendition. The thought was to build the initial prototype and gradually add cone filler until I achieved the desired woofer response. For the most part, this was a success, but for whatever reason there was a secondary peak in the woofer response that I hadn’t anticipated. I’m going to consider my failure to anticipate this a happy accident, because it meant that the woofers conveniently worked out perfectly to meet the compression driver on the first try. If you have any insight/theories into why that secondary peak might exist, I’d love to hear.
Barring that, the unfiltered responses of both drivers overlaid looks like this.
For the woofers, I opted for a 1st order 80Hz LPF to flatten out the response, which got me 80% of the way to where I needed to be. After applying some corrective EQ, I was able to get a solid response from 60Hz-16kHz. I’m applying a 4th order 45Hz HPF to protect from overexcursion. After generating the spinorama I realized I probably could have done more on the EQ front to correct the 500-2k region, but I’m happy enough with it and don’t really wanna do 36 more sweeps just to make a squiggly line less squiggly.
Distortion as measured with stepped sine tones:
Group Delay:
And finally polars (normalized on axis):
Given these results, I decided to go ahead and finish out this prototype. There’s some details in the physical construction of the box I might make in a future version, but overall I was quite happy with the performance. I chose a Dayton Audio KABD-4100 amplifier paired with one of their 6 cell 18650 BMS boards, mostly due to availability and balance of features. One of the outputs was left unplugged, so it's not the most efficient setup but I found it works quite well for the price. The sigmastudio file is provided in the zip file.
Regarding the enclosure, whatever you build make certain it's well sealed. You may want to modify the flange on the waveguide for a cleaner mount. These little drivers move a lot of air and I spent the better part of an afternoon working out caulking/gaskets to fully take care of air leaks.
Specs
- Frequency Response: 60-16kHz (+-2dB)
- Coverage: 75Hx60V
- Maximum SPL: 106dB (about where the 6NDL38s and PRs reach xmax - can reach higher if you HPF at 80Hz.)
- Battery Life: Subjectively well over 8 hours, but I’ll be doing an M-noise stress test later at some point to get a real validated runtime number.
- Weight: 14.9lbs (6.75kg)
- Outer Dimensions: 18.5” wide x 10.2” deep x 7.9” tall (47x26x20cm)
- Internal Volume: about 15 liters
- Components:
- 2x B&C 6NDL38 Woofers
- 1x B&C DH350 Compression Driver
- 1x Dayton Audio KABD-4100
- 1x Dayton Audio LBB-6S
- 6x Samsung 35E 18650 cells
- 2x Scanspeak SB16PFCR-00
Compromises: These are the areas that I’ve identified as weaknesses in this build or areas for improvement.
- Coverage is a bit narrower than I would have liked it to be. I think I need to refine my Ath skills a bit more since this didn’t necessarily match the polars I generated in ABEC.
- The steep crossover slopes may not be ideal for some listeners and introduce a small amount of group delay (about 1.3ms at 1200Hz.) This could probably be corrected with FIR.
- The DH350 breaks up at around 16kHz, so don’t expect full extension up to 20kHz
- Distortion is higher than I’d like it to be on the low end - no doubt due to nonlinearities with the passive radiators. They’re cheap - buy better ones if you can. Also - you should be able to achieve much better bass performance by increasing the box size by about 25%, so if space isn’t an issue that should be considered.
- Subjectively, the DH350 lacks a sort of upper midrange clarity that I’ve heard with other (larger) compression drivers. I’d wager this is due to some diaphragm resonances, but haven’t measured it. I’d recommend buying one to prototype with other horns/waveguides to assess this yourself before determining if this build is for you.
How are these sounding now that you've lived with them for a while? Looks like a fun and exciting build.
I'm considering getting a big 3D printer that could do something this size in one shot (if it made sense to do so).
I'm considering getting a big 3D printer that could do something this size in one shot (if it made sense to do so).
I've had it out on a few camping trips and festivals over the summer and so far it's been quite nice. 2 complaints I can think of though:
1. The Dayton battery boards don't seem to be that great. One shorted out and stopped working so I got a replacement one in, but it has a problem of slowly draining cells past the point of no return if not plugged in over a few days.
2. The vertical dispersion is quite narrow, so you do need to get the speaker elevated to ear level. If you only have flat ground to place it on, you have to tilt it back to get decent coverage.
As far as the actual sound quality/listening experience goes though, no complaints there. It sounds much better than other battery Bluetooth speakers I've listened to and I even had a chance to A/B compare it against a soundboks gen 4. Obviously the soundboks was capable of going louder, but at equal volumes, the MEH was much less fatiguing and easier to listen to.
For the battery pack, I'm kind of stumped as it seems a custom 18650 pack with BMS/charger is the likely best approach and that's currently a bit out of my wheelhouse.
I did end up making a slightly (2x) larger version with 4 woofers and the larger DH450 comp driver that features a 90x90 pattern, so thats been considerably more flexible in terms of placement. It's paired with a 2x12 sub and ran off a DJI power station, so much less headache on the battery front, while compromising on portability.
1. The Dayton battery boards don't seem to be that great. One shorted out and stopped working so I got a replacement one in, but it has a problem of slowly draining cells past the point of no return if not plugged in over a few days.
2. The vertical dispersion is quite narrow, so you do need to get the speaker elevated to ear level. If you only have flat ground to place it on, you have to tilt it back to get decent coverage.
As far as the actual sound quality/listening experience goes though, no complaints there. It sounds much better than other battery Bluetooth speakers I've listened to and I even had a chance to A/B compare it against a soundboks gen 4. Obviously the soundboks was capable of going louder, but at equal volumes, the MEH was much less fatiguing and easier to listen to.
For the battery pack, I'm kind of stumped as it seems a custom 18650 pack with BMS/charger is the likely best approach and that's currently a bit out of my wheelhouse.
I did end up making a slightly (2x) larger version with 4 woofers and the larger DH450 comp driver that features a 90x90 pattern, so thats been considerably more flexible in terms of placement. It's paired with a 2x12 sub and ran off a DJI power station, so much less headache on the battery front, while compromising on portability.
Thanks for the update! Glad you've gotten some use out of it.
Is the limited vertical dispersion inherent to that design?
The white one in the image looks really neat! Did you post on that somewhere else? Looks like it could also make a great middle speaker for a HT setup?
Parasitic battery drains can be such a joy killer.
Is the limited vertical dispersion inherent to that design?
The white one in the image looks really neat! Did you post on that somewhere else? Looks like it could also make a great middle speaker for a HT setup?
Parasitic battery drains can be such a joy killer.
Yeah the narrow vertical is due to the waveguide profile. I think I maybe could have gotten a little wider out of it, but not sure if its worth the effort to revise at this point.
The white one is very much a work in progress/early prototype. I posted a short writeup on my instagram, but hoping it can serve as my festival rig next year as the whole stack fits nicely in the trunk of my sedan.
It should in theory be a very flexible little box once complete, so anything from a DJ booth monitor to mini-PA, HT, etc.
The white one is very much a work in progress/early prototype. I posted a short writeup on my instagram, but hoping it can serve as my festival rig next year as the whole stack fits nicely in the trunk of my sedan.
It should in theory be a very flexible little box once complete, so anything from a DJ booth monitor to mini-PA, HT, etc.
Last edited:
Your projects are simply awesome,...thanks for posting.
(and for sharing the full JMOD MEH build writeup/plans, which a friend pointed me to)
What drivers are behind the ports in the square horn above?
I'm envious of your printer skills, takes MEHs to another level.
Really like your oval reflex port channels in the JMOD project.
Were you are to get a good sense, or measure, how much lower extension adding the reflex ports gained, over just using the 12NDL88's with only its primary ports? thx
The design above uses 4x 6NDL38s. When used in full-range mode, they're quite limited by excursion, but when paired with a sub you can get considerably more output out of them thanks to the high power handling.
Regarding the JMOD ports, I've found that the ports don't have a large impact to the bass extension. What they do help with is equalizing the temperature inside & outside of the cabinet, which I think is helpful for this design being a PA cabinet.
You can see how the measured response compares to a hornresp model if the box were sealed. There's about a 3dB advantage below 80hz with the ports.
I looked through my measurements folder and couldn't find a sweep that went lower than 40hz
Regarding the JMOD ports, I've found that the ports don't have a large impact to the bass extension. What they do help with is equalizing the temperature inside & outside of the cabinet, which I think is helpful for this design being a PA cabinet.
You can see how the measured response compares to a hornresp model if the box were sealed. There's about a 3dB advantage below 80hz with the ports.
I looked through my measurements folder and couldn't find a sweep that went lower than 40hz
Definitely keep us posted about your progress!
From your instagram post, the 90x90 constant directivity goal would be interesting to achieve. How far off are you at the moment?
I've got some ways to go, but I think its achievable. You can see the simulated ABEC polar vs actual measured below.
Barring the issues with graph scaling and using 2 different tools to generate the same type of graph, the main issue is that the pattern is still a bit too narrow >4khz, and maybe a bit wider than it should be in the 2-4khz range.
The ABEC sim was done without the woofer taps modeled, so my current theory is to move the taps to the peaks of the "knuckles", and reduce the number from 8>4 to see what effect that has. I'm basing that off of some very crude AKABAK fields analysis that showed MF/HF energy building up in the valleys where the taps are currently located. So maybe getting them out of those high impedance areas will reduce any detrimental effects to the comp driver wavefront?
Doing this increases the path length of the bandpass port, which could introduce pipe resonances. It also increases the volume under the taps, so I'm adding a cone filler to make up for that.
Barring the issues with graph scaling and using 2 different tools to generate the same type of graph, the main issue is that the pattern is still a bit too narrow >4khz, and maybe a bit wider than it should be in the 2-4khz range.
The ABEC sim was done without the woofer taps modeled, so my current theory is to move the taps to the peaks of the "knuckles", and reduce the number from 8>4 to see what effect that has. I'm basing that off of some very crude AKABAK fields analysis that showed MF/HF energy building up in the valleys where the taps are currently located. So maybe getting them out of those high impedance areas will reduce any detrimental effects to the comp driver wavefront?
Doing this increases the path length of the bandpass port, which could introduce pipe resonances. It also increases the volume under the taps, so I'm adding a cone filler to make up for that.
Thanks 🙂
When used with the sub, where do you high pass the 6NDL38's? I'm thinking 150Hz might be good to 125 dB ???
My bet is when you move their ports out onto the knuckles, towards cone center, low output picks up a little.
On every driver I've tried in a MEH, from 4" to 12", when the port moves towards cone center, bottom end has been helped.
Interesting re using the JMOD relex ports more for cooling than for extension (& thx for searching for meas)
The one reflex MEH I did, used two 8"s which only got to about 80Hz handily.
Tried adding reflex ports to a pair of 18"s mounted in a big MEH, but the ports were a disaster.....actually decreased low end extension.
I've come to believe for reflex ports to work on a MEH, box volume is a necessity (just like with regular designs).
Very cool you can model so extensively. And check out ports, reflex and primary.
You're a one stop shop...modeling, printing, woodwork, passive/DSP, and measurements !
Right now I've got them high passed at 100Hz. I could probably go higher and squeeze out a bit more SPL with a different sub design, but with this arrangement I'm able to get about 118dB, which is plenty for my needs.
Next iteration in progress. Hoping to have the rest of it printed and some measurements made in the next couple of weeks.
