Active 4-way with reaction cancelling woofers

In recent threads:

https://www.diyaudio.com/community/...crossover-frequency-in-the-near-field.401809/
https://www.diyaudio.com/community/...er-w-pd-21-drivers.402264/page-5#post-7433034


there is some discussion of the benefits of side-firing woofers. So I thought I would describe a recently completed build of mine which employs this method of woofer mounting. I have borrowed ideas from past commercial loudspeakers such as the AR9, NHT 3.3, various vintage Allison Acoustics models.

These are the features of my build that are unusual from a typical commercial enclosure:

1. Separate isolated sub-woofer enclosure, with reaction cancelling drivers.

2. 2” edge round-overs for diffraction effects control.

3. Low sub-woofer position for enhanced room boundary bass re-inforcement.

4. Tri-laminate layer (mdf/mdf/hardboard) 1.25” panels for increased rigidity.

The drivers chosen are:
sub-woofer 8" SB23MFCL45-8 (x2 in parallel), mid-woofer 5” SB15CRC30-8, upper-mid 4” SB12MNRX2-4, tweeter SB26STAC-C000-4.
I have a great respect for SB Acoustics drivers. They have great build quality, good performance and are affordable. My previous builds have used ScanSpeak drivers, but ScanSpeak prices have increased beyond my comfort zone for DIY, and I find the SB range comparable in performance. The SB drivers are affordable and offer a wide range of cone materials in the same driver format so that they can be compared to decide on a preference.
The 4” SB12MNRX2-4 upper-mid is the only dedicated mid-range driver in the standard SB range, with shortened voice coil and linear travel compared to their 4” mid-woofers and is suitable for a 4-way, where it can be high-passed crossed over well above the bass frequencies.
The 5” SB15CRC30-8 mid-woofer was chosen after comparing listening tests with the Norex and aluminium cone versions. The cone material is a Rohacell/carbon fibre laminate.

The upper enclosure is a compact sealed 3-way, decoupled from a sealed sub-woofer enclosure by 50mm diameter Sorbothane hemispheres. The chosen dimensions keep the tweeter-floor height at 1m, which is right for my listening position.
 

Attachments

  • IMG_0453.jpeg
    IMG_0453.jpeg
    829.8 KB · Views: 222
Last edited:
  • Like
  • Thank You
Reactions: 4 users
I paid the usual attention to internal bracing, interior panel damping, fibreglass enclosure stuffing, driver flush rebates. The overall wall tri-laminate thickness is 1.25”. In the past I have used constrained layer panels, but I have found that stiff, braced panels together with reaction-cancelling, reduces the woofer enclosure excitation to a greater degree, without the significant weight penalty and complex panel jointing.
I found that two 1/2” mdf panels glued together with PVA glue, results in a stiffer panel than a single sheet of 1” mdf. Adding 1/4” hardwood stiffenss the panel further. I tested this empirically with tap tests on identical size test panels. The resulting test impacts audibly went from “bonk” (mdf) to “pock” (mdf/mdf/hardwood tri-laminate). There was an audible difference in the decay. I did no measurements, just used my ears.
All panels and interior braces are connected by 6mm rebated glue joints. The images show the partial dry assembly fitting of panels and braces.

My driver choices were guided by extensive listening comparisons with different versions of each driver from the SB range. There was not a lot of measured differences but certain preferences arose after installing and listening to the drivers. In some cases, up to 3 sets of drivers were compared before finalising.
The side mounted sub-woofer frames are connected through the mounting holes by 4mm tensioned threaded steel rods. This may not be as efficient reaction cancelling as rigidly connecting the magnets, but is relatively easy to implement. It could be argued that in any case, reaction cancelling at the driver/enclosure interface is the goal.
The crossover is active dsp 192/24 with LR24 and LR48 crossover topologies. The drivers are phase aligned at crossover by delay. Crossover frequencies are 80 hz, 320 hz, 2560 hz.
 

Attachments

  • IMG_0286.JPG
    IMG_0286.JPG
    358 KB · Views: 135
  • IMG_0288.JPG
    IMG_0288.JPG
    469.4 KB · Views: 119
  • IMG_0294.JPG
    IMG_0294.JPG
    423.8 KB · Views: 100
  • IMG_0295.JPG
    IMG_0295.JPG
    305.4 KB · Views: 100
  • IMG_0298.JPG
    IMG_0298.JPG
    297.4 KB · Views: 108
  • IMG_0307.JPG
    IMG_0307.JPG
    395.9 KB · Views: 106
  • IMG_0312.JPG
    IMG_0312.JPG
    431 KB · Views: 130
  • IMG_0310.JPG
    IMG_0310.JPG
    465.9 KB · Views: 141
  • Like
Reactions: 1 users
I set up with the best crossover phase matching I can achieve, essentially flat on axis. In one important way, these loudspeakers sound different from my many previous builds. They seem to drive the room differently. There is a distinct reduction in low frequency acoustic output to the sides of the enclosure. This is due to the bipole effect of the subwoofer mounting.
This means that sub frequencies (< 80 Hz) are attenuated towards the sidewall. A subjective effect of this appears to be less soundstage width than I am used to. The outside-the-speaker imaging seems to occur less frequently than I experience with front baffle mounted drivers. Why there should be such a difference with the subs only operating over such a limited frequency range, I don’t know. However Stereophile
did discuss a similar effect for the NHT 3.3. Maybe it is a property of the upper enclosure driver arrangement, which is similar but the woofer is single in the NHT 3.3.

With very little EQ, the frequency response at the listening position is as shown. The peak at 50 Hz is probably mains hum since the RTA mic gain is very high. The roll-off at 20 kHz is untypical when measured at 1m on axis. The system easily reaches to 20 Hz with no EQ, which is surprising to me with only 2x8" sub-woofers. I think the floor boundary is the reason. The dsp crossover allows very good phase matching at the listening position between the sub-woofers and the 5" mid-woofer. This would be much harder to do with a passive crossover because of the large spacing between the sub-woofers and the upper enclosure. Also a passive 80 hz crossover would require some humongous, expensive coils. I have used active x-overs for over a decade and no way will I go back to passive.

IMG_0496.jpeg
 
Very cool. What are the overall dimensions of the cabinets?
Top box 285x405x450 (WxDxH), lower box 285x405x720.
The dimensions were finalised to allow the tweeter height at 1m. The sub-woofer enclosure internal volume is ~75 L.
If you had your time again, knowing what you know now, what would you do differently? ;)
Not much differently.
I have had great consistent results with the fabric dome tweeter. I am quite picky about tweeters and I can't say the ScanSpeak Revelator tweeters I have used previously, sound better.
I do have a curiosity around the Satori 2.5" dome midrange
https://sbacoustics.com/wp-content/uploads/2022/03/2.5in-SATORI-MD60N-6.pdf
used in place of the 4" midrange.
https://sbacoustics.com/wp-content/uploads/2021/03/4in-SB12MNRX2-25-4.pdf

Also, would the 6" mid-bass
https://sbacoustics.com/wp-content/uploads/2020/02/6in-SB17CRC35-8.pdf
improve on the 5"? The original selection was driven by the intention to keep everything as compact as I could manage. As I get more feeble, the big monkey coffins lose much of their appeal, regardless of how good they sound. My experience is that the larger cone drivers have a bit more dynamic presentation. I don't think this is always due to resonance or cone break-up as I hear it even at low spl. I would like to give it an audition because the Rohacell/carbon cone drivers have a smooth sound to my ears. Reminiscent of those Vifa/Peerless polycone midrange drivers from last century.
 
I paid the usual attention to internal bracing, interior panel damping, fibreglass enclosure stuffing, driver flush rebates. The overall wall tri-laminate thickness is 1.25”. In the past I have used constrained layer panels, but I have found that stiff, braced panels together with reaction-cancelling, reduces the woofer enclosure excitation to a greater degree, without the significant weight penalty and complex panel jointing.
I found that two 1/2” mdf panels glued together with PVA glue, results in a stiffer panel than a single sheet of 1” mdf. Adding 1/4” hardwood stiffenss the panel further. I tested this empirically with tap tests on identical size test panels. The resulting test impacts audibly went from “bonk” (mdf) to “pock” (mdf/mdf/hardwood tri-laminate). There was an audible difference in the decay. I did no measurements, just used my ears.
All panels and interior braces are connected by 6mm rebated glue joints. The images show the partial dry assembly fitting of panels and braces.

My driver choices were guided by extensive listening comparisons with different versions of each driver from the SB range. There was not a lot of measured differences but certain preferences arose after installing and listening to the drivers. In some cases, up to 3 sets of drivers were compared before finalising.
The side mounted sub-woofer frames are connected through the mounting holes by 4mm tensioned threaded steel rods. This may not be as efficient reaction cancelling as rigidly connecting the magnets, but is relatively easy to implement. It could be argued that in any case, reaction cancelling at the driver/enclosure interface is the goal.
The crossover is active dsp 192/24 with LR24 and LR48 crossover topologies. The drivers are phase aligned at crossover by delay. Crossover frequencies are 80 hz, 320 hz, 2560 hz.
could I ask where you got your corner roundover sections ? I assume you bought in if not hoe did you make them ? km doing something similar and have got myself a 1.5inch roundover bit but mitreing hardwood strips sounds like a lot less work !!
 
@jccart
If you look in the top left of the first photo in post #3, you will see offcuts of 2" quad round hardwood. It is a standard timber moulding in Oz, but usually only sold at a proper timber yard. I have a router table and would prefer to make my own but I never found a 1/2" rounder bit of that size. I never use pine because it dents easily at the slightest nudge.
 
@jccart
If you look in the top left of the first photo in post #3, you will see offcuts of 2" quad round hardwood. It is a standard timber moulding in Oz, but usually only sold at a proper timber yard. I have a router table and would prefer to make my own but I never found a 1/2" rounder bit of that size. I never use pine because it dents easily at the slightest nudge.
thanks I thought I had already replied to this never mind . thanks what do they use them for over there I've never seen anything like this in the UK. I managed to find some mdf ones but hardwood would be nice and durable like you say
 
hi bon, nice speakers - but i am curious, do you have other response curves other than that in post 4 ?
The response curve in post #4 is in phase pink noise through both channels at the seated listening position, mic straight ahead mid-way between the speakers.
Mic position is tweaked to give the least response to out-of-phase pink noise by equalising the distance to each tweeter. The responses contains room acoustic interference but gives a good idea of the subjective spectral balance. Shown at 1m on tweeter axis for a single speaker, the response is flatter. My ancient RTA is 1/3 octave at 48kHz/24bit, so the discrimination is below what is available nowadays. It gives me sufficient info to tune the system to my preference.
Broadband level changes of 0.5 dB are easy to hear comparatively, but won't be apparent on the RTA screen. My ears are always the final tool.
 

Attachments

  • IMG_0507.jpeg
    IMG_0507.jpeg
    490.7 KB · Views: 59