Hi Perry, I posted them with a picture back at post #92 on this thread. They are the GR Research NX-Treme's. I've actually built 2 pairs of those speakers and, to my ears, There has only been one system (It was a million dollar system I heard in Chicago about 20 years ago) I've heard which may have beaten them - but it was such a long time ago that it's impossible to make the comparison. Ever since I built them I have never even considered replacing them with something else, which as any audiophile can attest to, is very rare. I'm still constantly tweaking every other aspect of the system, mostly playing around with different amps, DAC's, cables, room treatments, etc. but the speakers are sublime. At this point I very rarely find a piece of gear or some other tweak which improves the sound - it's pretty dialed in.
I'm building a 2nd house and was looking to build something a little less imposing as it's sort of a small vacation home and I may even rent it on Airbnb, so aesthetic was also a consideration. But after reading your impressions of the bitches brew and also balthazarp's comparison to the avantgarde trio - it got me wondering just how good they might be. I'm pretty excited to build them and compare them to what I have now.
I'm building a 2nd house and was looking to build something a little less imposing as it's sort of a small vacation home and I may even rent it on Airbnb, so aesthetic was also a consideration. But after reading your impressions of the bitches brew and also balthazarp's comparison to the avantgarde trio - it got me wondering just how good they might be. I'm pretty excited to build them and compare them to what I have now.
So created first draw off my idea off the BB concrete speaker project...
Some thought I have: please give feedback anyone...
1. I have set the angle on 6 degree because I need it to support the tall heavy baffle by the iron armored into the cast concrete stand.
2. I also would like NOT to put sidepanel on the speaker because off a design perspective...
3. My good friend who have quite advanced diy speaker through time says its a flaw to mount speaker behind the baffle because off a "tunnel effect" - there will maybe be 3 cm "tunnel to the outside off the baffle...
Guess who the cool guy inspection is?
Some thought I have: please give feedback anyone...
1. I have set the angle on 6 degree because I need it to support the tall heavy baffle by the iron armored into the cast concrete stand.
2. I also would like NOT to put sidepanel on the speaker because off a design perspective...
3. My good friend who have quite advanced diy speaker through time says its a flaw to mount speaker behind the baffle because off a "tunnel effect" - there will maybe be 3 cm "tunnel to the outside off the baffle...
Guess who the cool guy inspection is?
1. 6 degree angle sounds great.
2. You will lose about an octave of bass without the lambda shaped wings. You can still boost the bass in the DSP and push them down to 40Hz, but I'm doubtful you'll get enough output below that. @Balthazarp can speak to that.
3. You should absolutely flush mount the drivers as I have done, I would not rear mount them. It might be OK with the subwoofers, but NOT OK for the coax.
4. Very handsome guy.
2. You will lose about an octave of bass without the lambda shaped wings. You can still boost the bass in the DSP and push them down to 40Hz, but I'm doubtful you'll get enough output below that. @Balthazarp can speak to that.
3. You should absolutely flush mount the drivers as I have done, I would not rear mount them. It might be OK with the subwoofers, but NOT OK for the coax.
4. Very handsome guy.
As a reminder, in the published Bitches Brew design, the 2 subs and midrange are driven by Amp1+DSPchannel1, and the front and rear tweeters are driven by Amp2+DSPchannel2.
This conversation caused me to try an unorthodox way to wire these:
1-Keep the front coax tweeter connected to Amp2+DSPchannel2
2-Connect the rear tweeter to Amp1+DSPchannel1 (the same circuit the 2 subs and the mid-woofer are on), with a 1.5KHz passive high pass, and then EQ the region above 2KHz so as to make it work.
3-Then I could get rid of the 6 ohm resistor on the front tweeters and also adjust the rear tweeter levels separately through the DSP EQ.
This has to sound a little confusing. But it's promising because the 15CXN88 bass-mid is rolled off at 1100Hz by a DSP shelf filter. Above that, the DSP levels flatten out at -16dB at frequencies 4KHz. The driver itself acoustically rolls off above 4K. I use notch and shelf filters instead of low pass filters to minimize phase rotations.
So I reasoned that since the rear horn is about 10dB more efficient than the bass mid around 2KHz, I could connect it to the bass-mid channel (first adding a 1500Hz 4th order passive high pass to protect the rear tweeters) and if it wasn't quite hot enough, I would just boost frequencies above 4KHz until I achieved the rear tweeter levels I wanted.
Roughly speaking, it did work. But it had issues.
The problem was that I had to raise the out-of-band level of the bass mid in the 2-5KHz range to make the rear tweeter hot enough, and then the bass mid started interfering with its coax tweeter on the front side. If I EQ'd it flat on axis, off axis was compromised and the nice Constant Directivity behavior started to degrade, as you see here. These curves are roughly 0, 30, 60 and 75 degrees off axis.:
ABOVE: Front side of the cabinet, off axis curves using my experimental (and unorthodox) circuit where the rear tweeters are connected to the bass+mid amp & DSP. The off axis response is a little irregular around 1.5-4KHz.
BELOW: off axis curves of the regular published design:
Off axis behavior in the published design is better behaved.
Side comment: You can see that the 30 degrees off axis curves for the 15CXN88 are very close to the 0 degree curves. This driver has better polar pattern than a 1" dome tweeter above 10KHz AND Constant directivity across its entire band. This is why I like this driver so much.
Last edited:
If you really want to optimise things, sell the MiniDSP kit and get Acurate or AudioLense. You might even end up with a net saving if you already have a laptop you could use.
Not necessarily a sound card. There are lots of good quality pro gear options to choose from. I bought a used RME UFX interface which has all sorts of functionality (including a mic preamp you can use for your calibrated mic) and works as a multichannel DAC. I think Topping also now make a multichannel DAC.
You can put together a low-cost DIY-oriented multi-channel DSP crossover solution - and still have real ESS or AK HiFi DAC chips, low phase-noise oscillators, easy rolling of I→V opamps (or none)....
I've used DIYINHK assembled and partly-assembled modules in the past, and they give high performance, especially for the cost.
Now, inspired my Perry's superlative designs, I have gathered some new modules for my own version of Perry's DSP and Lambda-baffle based speakers. (THANK YOU, Perry!)
Here's the lineup of Hardware & software I will use:
(You may have to mount some though through-hole PCB parts, and get some regulated supplies to feed the modules, but it's not too tough).
Start with an 8-channel USB to I2S interface:
XMOS 8-channel USB to I2S assembled PCB
For a high quality 8-channel HiFi DAC, the ES9028PRO answers well. This one connects directly to the XMOS module:
8-channel I2S ES9028PRO DAC, part-assembled
The DSP can be implemented in your Music Laptop, with Free Software - ,for example usingCamillaDSP (see the diyAudio thread) ideally on GNU/Linux, but also works on MAC or Windows.
Easiest of all: a Raspberry Pi 4 with the 64-bit version of Moode Audio - free software that streams or plays your local files, and includes CamillaDSP, tightly integrated, and easy to set up. This is a superb solution, and unbeatable value.
With CamillaDSP, the DSP functions are performed by the PC (or Rpi) processor hardware, which has few of the limitations of the miniDSP processing, like the stricty limited number of FIR taps.
Last edited:
That sounds like a very economical way of improving on the MiniDSP approach. I’m not familiar with the capabilities of the Camilla but what draws me to AudioLense is its capabiity to perform crossover duties with perfect time / phase correction and room correction based on measurements at your listening position. 96K taps makes it extremely powerful.
@Subsonic1050 it's not critical. As high above the floor as you can get it without the horn driver physically bumping into the woofers.
This is the rear frequency response and step response of the Bitches Brews. If you can position the rear horn driver so that the diaphragm is in the same vertical plane as the cone neck of the 15CXN88 (or perhaps an inch or two further from the microphone if you can manage it) then the rear signal will also have great time alignment and step response just like the front signal does. The lower it is mounted on the lambda, the further ahead the horn is timing wise from the midrange and the pulses will start to separate. I hardly think this is critical, but for the sake of perfection....
The above measurement is taken with the microphone about 5 feet off the floor, which is where the signals best align. If I were building these over again I would move the horn a little higher and that would set it back a couple more inches, and the time alignment sweet spot would be a little lower in elevation.
Most speakers don't have nearly as good step response on the front side, let alone on the back side.
I think the reason dipoles sound so natural (full dipole on front and back including tweeters, and not just tweeters on front) is:
1) It's important for room reverberant response to match direct response, and this is definitely not the case for 99% of speakers
2) The dipole null at 90 degrees off axis greatly reduces side reflections and ceiling reflections; so if you get the on axis response right, then:
3) Constant Directivity means that not only does every seat in the house get great imaging, but every seat also gets good measurements and close to flat response across most of the range, especially above 300Hz.
I don't know any other way to design a speaker that solves all these problems at once. I think this is why John Hilgers built the Bitches Brews and his wife thought he was playing his clarinet when he was actually playing a recording.
Thanks for the great responses Perry! I have built the stands for my speakers and I've mocked up some temporary baffles to get an idea of how the drivers will all fit. I don't think I'm going to be able to get the rear tweeter up that high (in line with the 15cxn88) but I guess I should position it as high as I can? I'm going to mess around with placement and see what I can do. Thanks again for your great insights!
Hi Perry, I think i'm getting close to giving the Bitches Brew a try.
Thanks a ton for its document that you have on your website.
I realized I already have a 15cxn88 ...kinda forgot it, because it's in a mono rig down at a dock.
So it would only be one for now unfortunately. But $400 would cover the rest of the drivers; and amps & processing I have in spades.
I'll just build a good stout baffle and frame out of birch ply. I'll go active FIR on all channels.
Seems like a great, relatively low cost opportunity, to wade into new territory...due to the generous guide-map you've provided
Thanks a ton for its document that you have on your website.
I realized I already have a 15cxn88 ...kinda forgot it, because it's in a mono rig down at a dock.
So it would only be one for now unfortunately. But $400 would cover the rest of the drivers; and amps & processing I have in spades.
I'll just build a good stout baffle and frame out of birch ply. I'll go active FIR on all channels.
Seems like a great, relatively low cost opportunity, to wade into new territory...due to the generous guide-map you've provided
correct. If tweeter is roughly in same vertical plane, will work just fine.
I had the same idea as Mark while continuing to build my two way MEHs this afternoon. I already have four 15OB350s and really nice Beyma 12XA30Nd coaxials. And I am sure I could find a pair of CDs and horns (I think I have some BC ME20s). The only thing that is holding me back is lack of space for listening.
Hi @perrymarshall. I haven’t been on the site for a while and just catching up on your new designs. Fantastic! Always learning new nuggets from your write ups.
I’m actually going to build a dipole now because I’ve gotten into guitar/bass and was going to build a pair of 2x12 cabinets. But then I thought I might as well make them proper dipoles since guitar speakers are typically designed for open backed cabinets anyway (or large sealed volumes). And then I thought I might as well add a horn, so I can use it as an active full range speaker with a DSP config change.
So I’ll be mining your papers for ideas.
One question I had is on the use of the angled u-baffle wings on your designs. Have you seen this analysis? I’m sure you are aware, as I believe I’ve seen your reference the resonance that results.
So my question is: what is your take on the tradeoffs of the directivity ripples at around 150hz, 300hz, etc vs the added bass extension you get from the wings? Is it that you think room modes dominate more down here, so the added octave of bass is worth the trade off? Or have you found a way to side step the issue?
That analysis makes me lean toward using just a wide front baffle. Probably wide at the bottom and tapered toward the top. I’d still have wings for support, but I can make the triangles hollow. This should get me down to bass guitar range, but I’d need to add subwoofers for anything really low and loud.
I’m actually going to build a dipole now because I’ve gotten into guitar/bass and was going to build a pair of 2x12 cabinets. But then I thought I might as well make them proper dipoles since guitar speakers are typically designed for open backed cabinets anyway (or large sealed volumes). And then I thought I might as well add a horn, so I can use it as an active full range speaker with a DSP config change.
So I’ll be mining your papers for ideas.
One question I had is on the use of the angled u-baffle wings on your designs. Have you seen this analysis? I’m sure you are aware, as I believe I’ve seen your reference the resonance that results.
So my question is: what is your take on the tradeoffs of the directivity ripples at around 150hz, 300hz, etc vs the added bass extension you get from the wings? Is it that you think room modes dominate more down here, so the added octave of bass is worth the trade off? Or have you found a way to side step the issue?
That analysis makes me lean toward using just a wide front baffle. Probably wide at the bottom and tapered toward the top. I’d still have wings for support, but I can make the triangles hollow. This should get me down to bass guitar range, but I’d need to add subwoofers for anything really low and loud.
The Bitches Brews have the biggest issue with this because the cabinet is about 2 feet deep max, corresponding to 1/4 wavelength at 140hz. Several factors reduce the effects of this:
1. Crossover is 100Hz.
2. The sine wave shape of the wings means the distance from each woofer to the rear wing edge is NOT constant. This spreads the resonances.
3. Because of #2 the resonances are pushed even higher than 140. The further we get above the 100Hz xover, the better off we are.
4. When I measure the impedance of the 15OB350s the curves show very little effect of OB wing transmission line resonance which you can see if you look closely to this post. This is the middle woofer I have measured and then put into a Vituixcad simulator:
https://www.diyaudio.com/community/...ssive-xover-no-dsp.404171/page-3#post-7479682
You can see a little 150Hz wiggle in the impedance curve. But not a lot.
I don’t have an easy way to measure the polar pattern in a large enough space but I don’t think the wings create much of an issue.
In general I’m willing to accept some polar pattern weirdness in exchange for an extra octave of bass. Keep in mind this exact tradeoff is at play with the Walnut Dipoles
https://www.diyaudio.com/community/threads/the-walnut-dipoles.404187/
Which also have wings but do not have the luxury of covering those issues up with a woofer - midrange crossover.
1. Crossover is 100Hz.
2. The sine wave shape of the wings means the distance from each woofer to the rear wing edge is NOT constant. This spreads the resonances.
3. Because of #2 the resonances are pushed even higher than 140. The further we get above the 100Hz xover, the better off we are.
4. When I measure the impedance of the 15OB350s the curves show very little effect of OB wing transmission line resonance which you can see if you look closely to this post. This is the middle woofer I have measured and then put into a Vituixcad simulator:
https://www.diyaudio.com/community/...ssive-xover-no-dsp.404171/page-3#post-7479682
You can see a little 150Hz wiggle in the impedance curve. But not a lot.
I don’t have an easy way to measure the polar pattern in a large enough space but I don’t think the wings create much of an issue.
In general I’m willing to accept some polar pattern weirdness in exchange for an extra octave of bass. Keep in mind this exact tradeoff is at play with the Walnut Dipoles
https://www.diyaudio.com/community/threads/the-walnut-dipoles.404187/
Which also have wings but do not have the luxury of covering those issues up with a woofer - midrange crossover.
Last edited: