The bitches brews have a very controlled directivity pattern, so they radiate far less energy to the sides than most speakers. That bodes well for what you are trying to do.
I did once demonstrate the bitches brews in a large room with a lot of glass surfaces that was probably 20 x 40‘. It’s manageable in my opinion.
I suggest you put an L-pad on the rear mounted horn tweeter and then you can adjust the rear high frequency reflections to your taste.
Also you can tame the room with fluffy furniture like rugs and couches.
I did once demonstrate the bitches brews in a large room with a lot of glass surfaces that was probably 20 x 40‘. It’s manageable in my opinion.
I suggest you put an L-pad on the rear mounted horn tweeter and then you can adjust the rear high frequency reflections to your taste.
Also you can tame the room with fluffy furniture like rugs and couches.
Understood, thanks.
From browsing these threads, it looks like the officially sanctioned modifications are 1) 5º tilt instead of 10º and 2) tri-amp to omit the passive XO. Some people have modified the wings, but that will result in different low end response. The sine shape as drawn originally seems like a good idea to me.
I'm inclined to tri-amp with a miniDSP Flex HTx for balanced outputs (no FIR filters but would use DIRAC). With that setup, the only passive components would be a protection circuit for the tweeter and L-pad for rear-firing tweeter, correct?
My only other concern is noise/hiss. In a big room like this and with quiet components it seems like it should be fine, but I will have a turntable hooked up and hope that chain isn't too noisy. I'm also have affection for vintage pre-amps even if they're unnecessary.
From browsing these threads, it looks like the officially sanctioned modifications are 1) 5º tilt instead of 10º and 2) tri-amp to omit the passive XO. Some people have modified the wings, but that will result in different low end response. The sine shape as drawn originally seems like a good idea to me.
I'm inclined to tri-amp with a miniDSP Flex HTx for balanced outputs (no FIR filters but would use DIRAC). With that setup, the only passive components would be a protection circuit for the tweeter and L-pad for rear-firing tweeter, correct?
My only other concern is noise/hiss. In a big room like this and with quiet components it seems like it should be fine, but I will have a turntable hooked up and hope that chain isn't too noisy. I'm also have affection for vintage pre-amps even if they're unnecessary.
Happy thanksgiving Perry - could you review the purpose of using the WiiM streamer in your setup? How is that additive vs. simply streaming/controlling say Spotify through your phone wirelessly (bluetooth) through the MinDSP Flex 8? Just want to understand the value of the streamer? mahalo!
Bluetooth is lossy. Airplay is lossless and WiiM supports Airplay. So there is a huge difference.
Also, Spotify is lossy so I use Tidal which is lossless. A major difference again.
Put the two together (Tidal over Airplay) and having lossless 24/96 or even 24/192 is vastly superior to Bluetooth and Spotify. Especially with DSP crossovers etc.
Spotify has said for years they will start supporting lossless but still hasn’t happened.
Also, Spotify is lossy so I use Tidal which is lossless. A major difference again.
Put the two together (Tidal over Airplay) and having lossless 24/96 or even 24/192 is vastly superior to Bluetooth and Spotify. Especially with DSP crossovers etc.
Spotify has said for years they will start supporting lossless but still hasn’t happened.
Note to @JMacII and anyone else who builds the Bitches Brews with 100% active 3 way crossover, instead of the stock passive 2-way configuration:
When you build the filters in the DSP for the subs and midrange, I suggest you design them to be fairly close to the curves you see here:
https://www.diyaudio.com/community/...over-vs-parallel-for-subs-open-baffle.404680/
The signal to the two 15OB350 subs should be a 6dB/octave low pass above 100Hz.
The signal to the 15CXN88 midrange should be a 2nd order shelf filter with a Q of around 1 to 1.2 and a -3dB point of about 50. The low shelf amplitude should be -16dB; and if you add a 10dB notch filter at 30 Hz (Q=2) that will even further reduce midrange excursion.
Those settings will closely resemble what the drive signal from the passive crossover sends to those drivers. Try to replicate the curves you see in the link above, then modify as you wish after that once you're taking actual measurements.
The threads from this link explain the rationale which prioritizes low phase shift. The Bitches Brews behave more or less like all drivers have a 2nd order acoustic slope and this is a big advantage in the time domain.
The 15CXN88 using those filters requires inverted polarity relative to the subs.
When you build the filters in the DSP for the subs and midrange, I suggest you design them to be fairly close to the curves you see here:
https://www.diyaudio.com/community/...over-vs-parallel-for-subs-open-baffle.404680/
The signal to the two 15OB350 subs should be a 6dB/octave low pass above 100Hz.
The signal to the 15CXN88 midrange should be a 2nd order shelf filter with a Q of around 1 to 1.2 and a -3dB point of about 50. The low shelf amplitude should be -16dB; and if you add a 10dB notch filter at 30 Hz (Q=2) that will even further reduce midrange excursion.
Those settings will closely resemble what the drive signal from the passive crossover sends to those drivers. Try to replicate the curves you see in the link above, then modify as you wish after that once you're taking actual measurements.
The threads from this link explain the rationale which prioritizes low phase shift. The Bitches Brews behave more or less like all drivers have a 2nd order acoustic slope and this is a big advantage in the time domain.
The 15CXN88 using those filters requires inverted polarity relative to the subs.
I just posted a new thread about diffraction and Constant Directivity which includes some analysis of the Live Edge Dipole polar pattern:
Perry:
I am going to try and build your BB speakers.
First, thanks for sharing the design and answering so many questions in the various posts. I can only imagine the hours you must have invested. Second, I would like to build the speakers exactly as you recommend. I don’t know enough about crossovers or speaker design to vary from what you did. I am handy with wood working and have even built a few subs (sealed box only). I know the build may be challenging but I am hopeful I can build exactly what you spec.
I have a few questions / comments:
I am going to try and build your BB speakers.
First, thanks for sharing the design and answering so many questions in the various posts. I can only imagine the hours you must have invested. Second, I would like to build the speakers exactly as you recommend. I don’t know enough about crossovers or speaker design to vary from what you did. I am handy with wood working and have even built a few subs (sealed box only). I know the build may be challenging but I am hopeful I can build exactly what you spec.
I have a few questions / comments:
- I have the file “The Bitches Brew Open Baffle Live Edge Speakers” PDF document dated 7/4/2023. Is this the latest file showing me what drivers, crossover components, etc. to buy?
- In one of your posts, (dated 12/11/2024) you attached a zip file for a Flex Eight processor to use with the BB speakers. Is this the latest file I should use?
- I plan to use the Flex Eight MiniDSP, do you still recommend that processor?
- There is some back and forth in the posts about about using a tri-amp setup that would eliminate use of the passive crossover. Do you recommend that approach? Cost is not an issue (I already have some extra amps). However, I prefer to build something that you already know works.
Perry will have better advice, but I would offer that I have really enjoyed my Hypex FA123 DSP Amp. Hypex D-Class are top quality and the DSP processing is built in. The FA series now also supports FIR filters, so if you want to go crazy with re-phasing or zero phase filters, etc. you can. The FA123 is three channels of about 100 watts and you can bridge two of the channels. Or, there's the FA223, for 250 watts per channel. The filter programming interface is more complicated to use than the Minidsp devices, but a lot of that is just intimidation from so much functionality, and if you're just going to implement a typical crossover and filters and PEQ, it's really straightforward. I have not had great luck so far having the Hypex automatically switch from analog RCA inputs to the digital inputs, and you need to have either Toslink or SPDIF coax input, but there are many settings and thresholds for the auto detection. You can get the Hypex at Madisound.
Cheers,
Jamie
Cheers,
Jamie
Thank you, Jamie. I am open to trying any approach as long as someone like you or Perry know it works. I am such a novice here. When you say things like setting FIR filters and zero phase filters I get lost fast. I have used a mini DSP (2x4 HD) and found it manageable but never got to any advanced features. I am willing to try and experiment once I get the system up and running. I do hear really good things on Hypex, anxious to try them at some point.
Well make sure you get other people to agree, I might be wrong. But, I think you can think of using the Hypex FA123 just like you would the MiniDSP, you just don't need the amplifiers since they are built into the Hypex as an integrated device. The dsp of the Minidsp is pretty amazing for the cost, so if you already have amps and you compare costs, that's not a bad way to make the decision. I think they are very similar in quality and functionality. Personally, I'm not interested in going down the FIR rabbit hole. I don't think it's necessary, but it's a way to compare dsp engines regarding chipset and functionality, etc.
Bitches Brew Design Update
This filehttps://evo2.org/wp-content/uploads/2023/07/Bitches_Brew_Open_Baffle_Live_Edge_Speakers.pdf
has correct components to use with accurate drawings. However I've made two minor hardware adjustments:
1) I added a 6 ohm resistor to the front tweeter only, which lowers its SPL relative to the rear tweeter. To my ear, the increased fill on the rear adds some welcome ambience. The newest DSP files take this into account. You can tweak the resistor, together with the level control of the tweeter in the MiniDSP, to adjust ambience to your own liking. You can also replace the resistor with an L-Pad if you prefer.
2) When you build the cabinets, I recommend you tilt your front panel 5 degrees back instead of 10 degrees back.
DSP FILES:
Yes, I still recommend the MiniDSP Flex Eight. Very low noise and good D-A converters.
Here are the latest DSP files for Flex Eight as of 22 February 2025:
https://evo2.org/wp-content/uploads/2025/02/bitches_brew_biamp_triamp_flex_eight_22feb2025.zip // https://tinyurl.com/bitchesbrewDSPbiamp-triamp
The zip file contains 2 files:
1) BIAMP DSP settings for the design in the PDF above.
The stock design has 6dB passive LC series crossover at 110 Hz between the 15OB350 subwoofers and the 15CXN88 bass-mid.
It has an active DSP crossover at 1100Hz between 15CXN88 bass-midrange and tweeter drivers.
This DSP file is the standard version discussed and featured in AudioXpress, but with more recent DSP settings.
You will notice in the schematic that the polarity of the bass-mid is reversed relative to the subwoofers. That is intentional.
2) I've added TRIAMP DSP settings for those who wish to skip the 110 Hz passive crossover and go 100% active. That's also in the zip.
You’ll still want the LCR notch filter on the rear tweeter, and the protection capacitors on the entire tweeter section.
I use a 6dB Butterworth low pass filter on the subwoofer section. The high pass on the bass midrange is a -16dB low shelf filter with a corner frequency of 45Hz. It mimics the Bitches Brew passive crossover curve shown here. I would expect the triamp version to be a small improvement over the biamp.
As you can see the drive signal level to the bass-mid drops rapidly below 60Hz.
Note: The bass mid in the Triamp DSP section has no conventional “low pass” or “high pass” filters. It’s all parametric and shelf filters. Sometimes people are puzzled by the Live Edge Dipole & Bitches Brew crossover settings because I use shelf filters instead of low and high pass filters whenever I can get away with it. This reduces phase shift and group delay. It improves impulses. I think people who are designing an end-game ultimate reference system, and are not aiming for flat phase response, are not serious. It’s one of the easiest things to measure and I think it affects imaging and resolution.
In the TRIAMP version, the polarity of all of the front-facing drivers should be normal, + terminal on amp wired to + terminal on speaker. The rear tweeter should still be reverse polarity.
Please note, I have not physically tested the TRIAMP DSP configuration, so there’s a slight chance I made an error somewhere. If so, it will easily show up in measurements.
My earlier DSP files had 40Hz and 75Hz notch filters which tune for modes in my particular room. I have de-activated those parametric filters. You will probably need to switch those back on and then adjust the frequencies for your own room.
I am also guessing you’ll need to add or subtract some extra EQ between 50 and 200Hz to make the transition between the subs and bass-mid perfect for your own system.
Go ahead and use the FIR DSP that is baked into these files, just understand it’s specific Left and Right correction for my unique specific drivers, as well as compensating for the phase shift of the crossovers. You’ll need to tweak your EQ for your own system.
Or you can roll your own FIR files. The most important part of the FIR is that it "unwinds" the 360 degrees of phase shift generated by what is effectively a 12dB/octave acoustic slope at 110 Hz and 1100 Hz. It does this with a couple of "maxphase" filters. I discuss that in other threads, for example here and here. I also added very precise amplitude correction on the top octave. Other than the top octave I don't use FIR to fix frequency response issues, I leave that to standard IIR parametric and shelf filters.
I still insist the Bitches Brews have absolutely zero competition in the marketplace below $100,000 and very little above that point. I think you’re really going to enjoy these. Last week a DIY guy who was thinking about building them stopped by the house to hear them. He was blown away, especially by the imaging. “Yep. Exactly what I hoped for. I’m gonna build these things.”
Last edited:
Thank you. Very much appreciated. I am going to order parts and get started soon. I will post pictures and keep everyone updated.
Perry, excuse my presumption to speak for the OP or other builders, but I'm thinking that the FIR filters and phase might be confusing to some people, but not really something to worry about. Would it be accurate to say that if they use your filter files and load them into the MiniDSP, all the FIR and phase filters should just be plug and play. And, that all of the "tune to taste" EQ is going to be somewhat intuitive, or easy to understand and tweak in the MiniDSP interface?
And, for what it's worth, I think this is quite an amazing "kit" of design. FIR filters are the most advanced filter technology, I mean real bleeding edge tech, and the chipsets and easy to use UI that MiniDSP are making available to the DIY builder at a price point is astonishing. To have all of this bundled in a design with filter files for sharing to others for free, unless I'm mistaken, is just incredible.
Jamie
And, for what it's worth, I think this is quite an amazing "kit" of design. FIR filters are the most advanced filter technology, I mean real bleeding edge tech, and the chipsets and easy to use UI that MiniDSP are making available to the DIY builder at a price point is astonishing. To have all of this bundled in a design with filter files for sharing to others for free, unless I'm mistaken, is just incredible.
Jamie
Yes, you are entirely correct. And they will sound pretty good without the FIR, but as you say, the FIR already comes preloaded… so yes, a person with basic speaker building ability can build something that competes with the very most expensive rooms at AXPONA.
If you can run basic measurements and adjust the EQ on the DSP, you can dial it in to an incredible level of accuracy.
My kids are in their 20s and I took them to the AXPONA show last year and I asked them “how many rooms sounded better than what you’re used to hearing at home?”
My son said “oh…. maybe two.”
If you can run basic measurements and adjust the EQ on the DSP, you can dial it in to an incredible level of accuracy.
My kids are in their 20s and I took them to the AXPONA show last year and I asked them “how many rooms sounded better than what you’re used to hearing at home?”
My son said “oh…. maybe two.”
I am also about to start building these. Heard them, loved them. I plan to use the miniDSP Flex Balanced. I considered omitting the passive woofer XO but plan to build according to Perry’s design. I just don’t see the theoretical benefits making much of difference to me in situ in day-to-day listening. Less fiddling for me, and a single 4-conductor cable terminated with SpeakOn will be nice and clean from a design perspective.
The change I do plan to make is in the depth of the side wings. Just shrinking it by a few inches to lighten the footprint. I think I can live with the bottom octave consequence of that.
I’ve modeled them in Fusion360 and can share the files if it’s helpful to you. Since this thread title is about a different speaker, is it worth starting a new one for build progress and images?
Since we are here, and my apologies if Perry has discussed this in his build thread, but I am curious about hybrid Xover designs.
After seeing something similar in a Scott Hinson MEH design, and also from Wayne Param's recommendations, I have been using an L-Pad on compression drivers and it appears the shunt resistor helps to dampen the compression driver. I don't really know, but when I try to drive a compression driver with my Hypex, I think maybe there is some harshness since the CD is so efficient that the amp, as clean as it is, can't even really get up and going. And, Wayne shows in a couple of his Xover white papers how the L Pad configuration and shunt smooths things out. Scott did this for his midrange drivers in the MEH to dampen the coupling of the drivers.
All to say, even with a full DSP Xover design, are any of you using some few passive components to get a better match to the amp, in particular the high efficiency CD's such that you can get the amps up into some current?
I'm guessing that maybe woofers can actually benefit from being directly driven since a current capable amp can respond directly to driver demands without the problems of inductance from the driver and the high damping amp can control the driver better(?) without a passive in the way?
After seeing something similar in a Scott Hinson MEH design, and also from Wayne Param's recommendations, I have been using an L-Pad on compression drivers and it appears the shunt resistor helps to dampen the compression driver. I don't really know, but when I try to drive a compression driver with my Hypex, I think maybe there is some harshness since the CD is so efficient that the amp, as clean as it is, can't even really get up and going. And, Wayne shows in a couple of his Xover white papers how the L Pad configuration and shunt smooths things out. Scott did this for his midrange drivers in the MEH to dampen the coupling of the drivers.
All to say, even with a full DSP Xover design, are any of you using some few passive components to get a better match to the amp, in particular the high efficiency CD's such that you can get the amps up into some current?
I'm guessing that maybe woofers can actually benefit from being directly driven since a current capable amp can respond directly to driver demands without the problems of inductance from the driver and the high damping amp can control the driver better(?) without a passive in the way?
Look forward to following a dedicated thread on a BB build. I documented my live edge dipole build a couple years back (Perry’s award winner) and I think it helped a few people with their builds. And who know, it may inspire me to build another set of OBs, even though I certainly don’t feel the need to.
I could only hazard a wild guess that a highly efficient horn driven by a digital amp might expose low-level artifacts analogous to dither noise in a digital recording. Maybe that is what you are hearing, and padding them down with resistors brings the dynamic range of the amp to be more in line with the SPLs you actually use when you listen.
I could be wrong about that.
Either way, nobody in a home setting needs all 200 watts of available power for a 108dB compression horn. And many analog amps have enough noise that you can hear a hiss from the horn when no music is playing, and the padding helps that problem as well.
Keep in mind that a series resistor alone will impress the impedance curve on top of the frequency response curve, so if you put an extra resistor in parallel with the driver it will suppress some of that.
Woofers will benefit from being directly connected with no other drivers in series and no crossover components.