• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Revel M105 Copy DIY build

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
747
Likes
1,456
Location
NZ
After seeing the success of Directiva,

But knowing the price, I wanted a nice small pair of passive bookshelf speakers for use in the second living room. I didn't need massive SPLs (if you need more SPLs the Directiva is much better - however that build is DSP XO + Multi Channel amplified). The Revel M105's looked like the business. As I am all the way down in NZ we cant easily get Revel speakers here. There is one dealer that brings them in to order but I'd end up paying way more than they are worth. So without a model to clone, I set out to make a copy. Here's what I've come up with, and I'm posting the design up free for non commercial use. (I think if I were in the US I would've just ponied up for an actual pair, but no fun right?)

Grain of salt and Disclaimer: All measurements made with 4.5ms gating spliced to NF at 550hz as per @napilopez amazing guide HERE. Spin measurements are also not taken at 2.83V so sensitivity should not be taken from spin measurements, but I did take a 2.83V/1m sweep to capture sensitivity. I use a combination of focusrite solo and EMM6 mics and a UMIK1 for captures and confirmation. Software used is the awesome REW and VituixCAD. Just like music producers don't need multi million dollar studios but can now make amazing music in their bedrooms, one can now make a rather good speaker in the garage with that software.

I also am by no means an established speaker builder - so please wait for peer review or ask here if you desire to build this speaker (as you will see later, I doubt many will want to copy this build, and its more of a prototype at this stage, I may make changes to the cabinet yet)


Preference Rating as built (per VituixCAD, I would imagine not quite so high through a klippel NFS)
SCORE: 6.4
SCORE w/sub: 8.2

Sensitivity: 81dB (as this is a full 6db BSC design, the woofer is 87dB half space)
Frequency response: +/-1.5dB 48Hz-12kHz
Frequency response: +/-3.0dB 42Hz-18kHz (I believe the acoustic lens/phase plug that revel has in their M105 waveguide allows the boost in the top octave to get flat to 20khz)


Even though the sensitivity is low, theoretically the max SPLs if crossed to a sub would be the same as the M105, just with a bit more power required. I was too scared to do a 102db sweep and 96db was as high as I went. I may do one later but start from 100hz on up to protect the woofer. I know the SB26 tweeter can do 103db.

You'll have to excuse the raw MDF finish and use your imagination here - I will apply a veneer/stain/polyurethane once the MDF has climatized to the house.


IMG_0365.JPG


Cabinet Dimensions

External: 200W x 320H x 350D
Internal: 164W x 284H x 282D - gross around 13L. Less drivers, port and brace gets you around 12L.

Woofer 25mm from frame edge to bottom, 20mm between woofer and waveguide and 25mm between waveguide and top. All vertically centered.

Front baffle is from 30mm MDF - Why 30mm? 1. Allows me to run it across a router table with a 1"/25mm roundover bit and provide the roundover. 2. Strength. 3. The 25mm radius allows easier application of veneer as smaller radii are harder unless you go normal sharp edge - no roundover. See summary for more on this.
All other pieces 3/4"/18mm MDF. One full brace with holes running side to side, front to back between the woofer and tweeter.
I use screws for extra strength and to ensure lining up of panels when gluing and stop slipping. This is optional.
Also remember to chamfer the back of the woofer hole to let it breathe.

Rear Port is 50mm/2inch at a length of 210mm or 8 and a bit inches, placed roughly behind the tweeter.

Drivers

Woofer: SB15NBAC30-8 (yes, 8 ohm, more later in impedance) available from madisound HERE
Tweeter: SB26ADC-C000-4 available from madisound HERE
Waveguide: augerpro (on DIYA forum) available from somasonus HERE - please, if you use these, consider a donation to the author - his contribution to the art is worthy. Select the 5" sb26 and you can get printed from either xometry.com or i.materialise.com, or your local friend with a 3D printer! I have pictured the 6.5 version I also have but this build uses the 5 inch waveguide. More information can be found at diyaudio, search waveguides.

sb15nbac30-8.jpg


20201130_165632.jpg

20201130_165654.jpg

20210401_132816.jpg



20210401_132829.jpg


Note: 6.5 inch waveguide shown above, not used for this build.

Simulation and Crossover


M105 Copy - simulated Six-pack.png


M105 Copy - simulated Six-pack ver directivity.png


The simulated preference rating w/o sub is 6.95! (wishful thinking)

M105 CLONE XO-schema-1.png


Note the simulated padding resistor for the tweeter is 3.3 ohms however it is 4.5 ohms as built - I often run into this with vituixCAD and I need to add 0.5 to 1 ohm to my final builds in the high frequencies. Probably a problem with my measurement rig. Either way, you can adjust this value to your taste. Too hot up top? Make it 5 or 5.5 ohm. Like more sizzle? Try the 3.3 ohm. If you're building these I would hope you at least have a UMIK to measure the final result. Driver phase matching isn't affected too much.

The crossover uses third order electrical circuits to achieve 4th-order Linkwitz-Riley slopes on both drivers and one massive advantage is that the waveguide moves the tweeter back into the cabinet allowing for a nice phase matching between the drivers, i.e. time alignment, without a stepped or sloped baffle or asymmetric crossover slopes. The series LCR network before the woofer is required to bring down the baffle boost you get around 1khz with such a small baffle.

The inductors are all air core and the 100uf cap is electrolytic, and I don't care what brand or type you use so long as they measure up. That's been proven here at ASR. Here is my haphazard crossover. It don't need to be pretty, but you cannot butt inductors up next to each other. Google speaker inductor placement.

IMG_0359.jpg



Here we go - As built Measurements

Spin (note: not at 2.83V)


M105 Copy as built measured Power+DI.png


On axis + LW

M105 Copy as built measured SPL.png


Directivity

M105 Copy as built measured Directivity (hor) line +-90.png


M105 Copy as built measured Directivity (hor) line normalized.png


M105 Copy as built measured Directivity (hor).png


M105 Copy as built measured Directivity (ver).png


Typical 2-way vertical XO holes above and below 20 degrees - usual caveat to stay on tweeter axis applies

2.83V Sweep and Inverted tweeter check

2.83v m105 copy.jpg


Impedance

Note I used the 8ohm woofer whereas I believe revel use a 4 ohm version (back when they used SB drivers) as @amirm measurements dip down below 4 ohms. The reason I went for the 8 ohm woofer is it modelled a bit deeper (more bass) at the cost of efficiency. The maximum SPLs between the two woofers are theoretically similar due to X-MAX so crossing to a sub and using high power class D the 8 ohm woofer seemed like the better choice to me. Also another reason why this isn't a clone but a copy.

M105 Copy as built measured Impedance.png


Distortion @ 86dB SPL

dist relative 86db.jpg


dist percentage 86db.jpg


Distortion @ 96dB SPL

dist relative 96db.jpg



dist percentage 96db.jpg



As you can see, 96db is really starting make the little woofer sweat. In addition to this - I believe there is quite a bit of port chuffing noise (both at 86 and 96dB), that can likely be reduced by using a flared PE precision port or similar. At this stage I only used a bit of PVC pipe with no flared ends. Don't use non flared port ends kids. Air likes to be caressed into a hole not forced.

However crossing to a sub at 80-100hz would sort a lot of this distortion out and make the mid-bass much clearer and greatly reduce IMD in the woofer. The same could be said for the M105 I would've thought.

Estimated in room response

in-room estimated.jpg


Drivers Near Field Response:


Now here is where some more problems show up - but remember this is an M105 copy, and I believe the M105 has the same problems as you will see - even though Amir conveniently(j.k.) did not measure the port output.

port woofer nf.jpg


Heres my build with our old mate port resonance. I don't think it is audible, and because the port is to the back, even less so. I think this might drop away slightly with a flared port. But almost every commercial speaker has it to some degree, including I suspect the M105 see my comments on Amir's measurements here in green;

Revel M105 bookshelf speaker Tweeter Woofer Audio Measurements.png


Revel M105 bookshelf speaker Spinorama CTA-2034 Frequency Response Measurements.png



Revel M105 bookshelf speaker CSD Waterfall Audio Measurements.png



Food for thought anyway.

BOM 1 speaker as at writing - or why you might just buy a M105 instead

Woofer - $80.50
Tweeter - $54.40
Waveguide - $ - Varies - free (your own 3d printer?) to 100USD ea. Shop around. Looks like $45 EUR/ $50USD SLS print at i.materialise and $95 USD SLS print at xometry.com. Prices also fluctuate a bit so keep an eye out.
Crossover parts, terminal cup and port (all parts express) - $88.02. I used the precision port with the big flares in this price. Match inductor DCR's
MDF - ?? varies
Finishing ?? varies

Total - ~$272 ballpark per speaker depending on where you get waveguide plus MDF as I'm not sure the cost in your area, and however wild you want to go with a finish. As you can see this is creeping up for a DIY small bookshelf speaker but nowhere near directiva levels, but this is not directiva at all. It has good directivity, just won't go as loud or sound as clear at high volume. Considering a M105 is now going for $825 USD ea, this could be worthwhile. Certainly was for me down here in NZ, even with all the freight costs.


Summary, thoughts, and listening impressions

This speaker sounds as you would imagine, probably just like an M105 or pretty close, pretty mfn good for a lil bookshelf. I remember finalizing the crossover and my ears having an ahh moment when everything clicked. The bass from this woofer is great, I was surprised at how deep this thing goes, and the tonality of the whole speaker just sounds correct. Obviously up to a certain volume level. But its certainly loud enough for the room it is in and sounds perfect. Unfortunately I don't have a M105 to compare it too which I would love to do. As you might suspect, yes I just sat there for days going through the music collection with a big smile on my face. Once the volume gets up quite high you can hear a slight reduction in sound quality as the woofers excursion is muddying up the midrange.

What might I change?

These are a full 6db baffle loss design and are rear ported. They get a massive front wall in room gain, so bringing them away from the walls is almost a requirement. I would like to try a taller, less deep cabinet with a front mounted port. It probably wouldn't affect this crossover much to have the cabinet slightly taller. The problem though is front porting will create a peak with the port resonance as opposed to a dip so some more port work may be required. Or perhaps keep the port on the back and do a lesser baffle step correction say 3db allowing closer wall placement and letting the room gain do more lifting below 100hz.

Also, if you weren't able to do the round-overs I suppose one could use a 1" or even 3/4" front baffle and the crossover should still work, I suspect the problem would be the on axis diffraction might show up a bit more, just like it does on the M105 with its flat front baffle. See here M105 spin showing a 3db bump at 5kHz, where my build is not so severe on axis:

Screen Shot 2020-07-17 at 8.00.51 AM.png


So I think the roundovers are optional. Maybe test before and after, see if you can hear the difference.

But overall I'm very happy with the speaker and plan to finish the cabinet and just keep enjoying it.

Open to comments, compliments and complaints - I know how much the audio community loves picking apart designs. Thanks for reading.
 
Last edited:

alex-z

Addicted to Fun and Learning
Joined
Feb 19, 2021
Messages
537
Likes
922
Location
Canada
The EMM6 and UMIK-1 both lack completely accurate free-field calibration by default, which may explain your simulated vs built discrepancy at high frequencies. I am working on fixing this problem in my own setup. I initially noticed adding absorption near the mic body cleaned up the impulse response considerably, but the frequency response got further away from Klippel NFS results, rather than closer as it would have if the mic was truly free-field.

Cross Spectrum will sell you free-field calibrated versions or you can send in your own mic.


As for your prominent 700Hz port resonance, you could consider adding a 1/4 wavelength resonance trap to the port and fill the trap with damping material. This has been done before in DIY designs, the theory is similar to selecting driver placement in transmission line designs.

 
Last edited:
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
747
Likes
1,456
Location
NZ
Cheers Alex-z, I looked at those cross spectrum mics back when I got my emm6 and they were out of stock. High time I grabbed one.

As for port resonance, I’m all over it with my purifi woofer build see here:

https://audiosciencereview.com/foru...ce-platform-speaker-project.20407/post-898122

As for this build, it’s more of a compromise, and tbh I can’t hear it, so I’m not gonna fill the cabinet up with piping and make the overall cabinet larger to compensate. I think the precision port with flares should help a bit and put my port chuffing issue pretty much to bed.
 

fluid

Senior Member
Joined
Apr 19, 2021
Messages
479
Likes
832
Frequency response: +/-3.0dB 42Hz-18kHz (I believe the acoustic lens/phase plug that revel has in their M105 waveguide allows the boost in the top octave to get flat to 20khz)
Nice build :)

The phase shield isn't going to boost the on axis. It looks to me that you have not compensated the mass and inductive rolloff that occurs in constant directivity waveguides. A 6dB filter between 10 and 20K (driver dependent) is usually added to compensate, there would be less need for a padding resistor as it cuts the gain from the waveguide loading.

https://www.pispeakers.com/Pi_Speakers_Info.pdf

Good diagram and explanation from the above

Pi Speakers.png
 

Madjalapeno

Senior Member
Forum Donor
Joined
Aug 13, 2021
Messages
436
Likes
1,032
Location
NH, USA
Waveguide: augerpro (on DIYA forum) available from somasonus HERE - please, if you use these, consider a donation to the author - his contribution to the art is worthy. Select the 5" sb26 and you can get printed from either xometry.com or i.materialise.com, or your local friend with a 3D printer! I have pictured the 6.5 version I also have but this build uses the 5 inch waveguide. More information can be found at diyaudio, search waveguides.
I think @McFly has done this, but for anyone else trying to make their own waveguides HP FusionJet parts will give you a strong part with a good surface finish made from either Nylon or Glass reinforced nylon. It's a little more than traditional FDM parts, but more accurate.

 

Jukebox

Active Member
Forum Donor
Joined
Apr 11, 2020
Messages
133
Likes
181
That's textbook DIY, congrats! And you still have 50 free days of listening until the Super Rugby Pacific matches begin.
Being located in NZ I'm guessing you love this sport :)
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
747
Likes
1,456
Location
NZ
Nice build :)

The phase shield isn't going to boost the on axis. It looks to me that you have not compensated the mass and inductive rolloff that occurs in constant directivity waveguides. A 6dB filter between 10 and 20K (driver dependent) is usually added to compensate, there would be less need for a padding resistor as it cuts the gain from the waveguide loading.

https://www.pispeakers.com/Pi_Speakers_Info.pdf

Good diagram and explanation from the above
Thanks fluid that’s a nice paper. I see, that makes more sense I need to configure the tweeter xo and apply a more broad Q filter with hp slope rather than the more traditional pad and slope. I can’t really imagine how that would be done although the paper mentions a parallel resistor, so I will play around with the XO in VCAD and see if I can regain that top octave. Would be nice to have that “air” even though I know I can’t hear it, I’m only good to 16.5kHz. Good thing the XO board is easy to remove, unlike some of my previous builds heh
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
747
Likes
1,456
Location
NZ
I think @McFly has done this, but for anyone else trying to make their own waveguides HP FusionJet parts will give you a strong part with a good surface finish made from either Nylon or Glass reinforced nylon. It's a little more than traditional FDM parts, but more accurate.

For some reason I cant see any option on hubs.com to 3D print - I can only see CNC machining and sheet cutting. Perhaps it is geofenced to US customers only? I'm open to more cost effective waveguide prints (aside from bulk pricing) as I might build some more of these for friends + family
 

Madjalapeno

Senior Member
Forum Donor
Joined
Aug 13, 2021
Messages
436
Likes
1,032
Location
NH, USA
For some reason I cant see any option on hubs.com to 3D print - I can only see CNC machining and sheet cutting. Perhaps it is geofenced to US customers only? I'm open to more cost effective waveguide prints (aside from bulk pricing) as I might build some more of these for friends + family
I use protolabs.com mainly. They're very competitive, and it's easier to send things their way than it is to quote everywhere, and end up with them anyway (at least for business use). I can quote something with them if you want.

Also use xometry (as you did too).

When I was in Europe we used Materialise.

These guys are in Australia - https://3dprintbureau.com.au/
 

radix

Senior Member
Joined
Aug 1, 2021
Messages
484
Likes
376
Great job. Did you hand-wind those inductors, or do you have a machine? They look very neat.
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
747
Likes
1,456
Location
NZ
Fluids post has helped me see the error in my ways with the waveguide filter. Something like this should help me regain that top octave.

M105 Copy rev 2.2 Six-pack.png

M105 Copy rev 2.2 XO-schema-1.png


The phase matching isn't as good but the power response and listening window are smoother. I will give this a go after a parts run. Simulated Preference rating went up from 6.95 to 7.1
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
747
Likes
1,456
Location
NZ
Great job. Did you hand-wind those inductors, or do you have a machine? They look very neat.
Thanks, sort of - the inductors I had here were actually too big so I wound them down, testing each time with a DATs. The 0.27 is down from .35 and the 0.15 is down from 0.30. The remaining inductors are just cheerful dayton standard or "perfect layer". I'm no Danny Richie, I couldn't give a f about expensive crossover parts or tube connectors. The drivers have the biggest impact on clarity, in my experience and opinion due to IMD, and not so much HD.
 

fluid

Senior Member
Joined
Apr 19, 2021
Messages
479
Likes
832
Fluids post has helped me see the error in my ways with the waveguide filter. Something like this should help me regain that top octave.
Did you try different crossover points in Vituix? Due to the directivity mismatch I wonder if a higher or lower crossover might work to relieve that bump dip without adding more parts or messing the offaxis up which is pretty nice already.
 

Wolf

Senior Member
Joined
Nov 21, 2020
Messages
339
Likes
373
Location
Indiana
I'm glad you lifted the top-octave, as I was going to suggest just that. Looks much better!

Brandon sure did his homework on the waveguides, that's for sure!

I think you might be tuned a touch high on the port Fb. The lower peak being higher means you tuned above Fs, which is not always preferable. If you tuned to Fs or just below, I think your HD in that range may drop a bit.

I also saw no mention of damping in the box. I hope you at least lined it, as that is the minimum for vented boxes. Just adding a handful of stuffing after that may droop the small peak before rolloff in the bass and/or droop the tuning a smidge.

I think you already have something very nice, well-documented, and I'm really only splitting 'possible' hairs.
Well done!
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
747
Likes
1,456
Location
NZ
I'm glad you lifted the top-octave, as I was going to suggest just that. Looks much better!

Brandon sure did his homework on the waveguides, that's for sure!

I think you might be tuned a touch high on the port Fb. The lower peak being higher means you tuned above Fs, which is not always preferable. If you tuned to Fs or just below, I think your HD in that range may drop a bit.

I also saw no mention of damping in the box. I hope you at least lined it, as that is the minimum for vented boxes. Just adding a handful of stuffing after that may droop the small peak before rolloff in the bass and/or droop the tuning a smidge.

I think you already have something very nice, well-documented, and I'm really only splitting 'possible' hairs.
Well done!
Hah yes the cabinet is fully lined - I did skimp on the cabinet details a bit. I did indeed tune slightly higher as I was after a bit more punch, but like I mentioned if I rebuild the cabinets I will tune for Fs and try and let the room gain do the rest (I will have the option to change the tune when my precision ports arrive). The port length gives a slight bump to the low end as pictured, this gives the speaker a meatier low end than its size suggests. I built your zingers a couple years back, they are still in heavy use at a friends house!

SB Acoustics SB15NBAC30-8 SPL.png





And I wish I could edit my first post as the new spin results after tweaking the cross over have given it a new rating too;

Preference Rating as built (per VituixCAD)
SCORE: 6.7
SCORE w/sub: 8.4

Sensitivity: 81dB (as this is a full 6db BSC design, the woofer is 87dB half space)
Frequency response: +/-1.5dB 48Hz-20kHz
Frequency response: +/-3.0dB 42Hz-21kHz (I believe the acoustic lens/phase plug that revel has in their M105 waveguide allows the boost in the top octave to get flat to 20khz) The new crossover has fixed the top octave.

Latest Spin Measurements:

M105 copy rev 3 Six-pack.png


rev 3 impedance.jpg


Estimated in-room Response:

rev 3 in-room estimate.jpg


New XO
:

M105 Copy crossover as built XO-schema-1.png


Its clear from the new measurements now there is more energy in the top octave, giving the slight flick up to the estimated in room response top end.

I think I'm calling it done! Time to work on a finish!

smile-nod.gif
 
Last edited:
Top Bottom