Hi JMacll, Thank you very much for sharing the detailed Bitches Brew build thread, this is an excellent reference for those of us looking to build these open baffle speakers in the future! Would you please share what amplifiers you are driving your BB speakers, are they same amplifiers for both the high frequency drivers and the the passive mid range cross over with the two SB 15OB350 woofers? I am also interested to see a nice photo of the mid range crossover with the two SB 15OB350 woofers following Perry's original schematic, like you posted for the high frequency boards, if you would be willing to do this. Your build looks excellent, I look forward to building the Bitches Brew myself!!
For those who use tube amps, the zobel circuits for both bass and treble sections flatten the impedance and will reduce interaction with the output stage and increase consistency between various amps. The difference will be audible on tube amps.
Here’s the other board, sorry for the poor light but it’s mounted inside the wing. There’s less going on with this one. The inductor in parallel with the midwoofer is mounted separately up high on the inside of the wing. What you’re seeing here is the zobel circuit which is the large red cap and resister in parallel with the signal circuit and then the 330uF cap which is in parallel with the woofers…the wire from the 330uF cap goes up and rejoins the ground out from the second woofer which then goes to the mid-woofer (with the inductee in parallel). Physically it made more sense to me to do more of the wiring off board for the lows. The 330uF cap is a potential upgrade down the road. Using the Wago connectors makes that easy…I’ve never done any breadboarding or circuits like this before and I found using the Wagos made it very easy to connect and visualize things. Not sure if they are audiophile approved or not but they work lol! FWIW this is the first speaker I’ve ever built. Sort of a jump into the deep end.
In terms of amps, right now I’m running the highs off my Aleph J I built with help from this forum and the lows of a McIntosh MC202 that I’ve owned for 20+ years. The miniDSP allows you to adjust gain independently for each channel. Because the bass and highs have different sensitivities you would probably have to adjust gain even if you used four identical channels of amplification. Right now I just tuned it by ear. I will try to do some measurements and dial it in more but it’s not hard to adjust the gains to a balanced presentation. The DSP also allows you to have up to 4 profiles so you could for example have a night time setting where you turn the bass gain down. I’m considering trying a class D amp, the Aleph has just a little bit of noise in it and these speakers are super sensitive. I’d like to hear the sort of pitch black SNR that a class D would give here.
Attachments
Hi JMacll, Thanks for the photo and your added comments on your BB build, this is very helpful!! DiyAudio is a tremendous community for project sharing and support! FYI. My first speaker build was a derivative of the Liionidas design described by Lii Song Electronics using their Fast-15 full range driver and W-15 woofer. I use a miniDSP Flex active crossover as the preamplifier. I am very happy with the performance, but really like what Perry has done so will build the BB for my next project. I do use two VTV stereo Hypex NCxxx amplifiers with the miniDSP. I can tell you they are very quiet even with the high sensitivity full range driver. My only comment on this class D amplifier is that the high frequencies can be a bit harsh and fatiguing. If you are willing to post more photos, I am interested to see more of the wiring on the back side of your project, so I may follow the full wiring path from your BB's binding posts to the four drivers. Like you, I am am a total novice and comparing a real photograph the the schematics that Perry has provided is huge help. Also, thanks for sharing your use of Wago connectors, these look awesome and I will plan to use these with a hot glue gun to fabricate crossovers as you have done. Thanks again!! D🙂
@Dachshund - I’d be happy to take some photos and I can also draw out an “as built” of how I did it.
@perrymarshall - I played around with the minidsp EQ and REW software, with mixed results yesterday! Did you do the original EQ by ear or using the REW approach? It’s funny because the “predicted” output of the REW generated filters looks really flat but when I load them it was….anything but. More experimentation needed.
@perrymarshall - I played around with the minidsp EQ and REW software, with mixed results yesterday! Did you do the original EQ by ear or using the REW approach? It’s funny because the “predicted” output of the REW generated filters looks really flat but when I load them it was….anything but. More experimentation needed.
I did original EQ and crossover with measurements using Fuzzmeasure mostly at close distances.
At 1 meter they should measure quite well “out of the box.”
Then as a 2nd step I took measurements at listener position and combined them with listening to arrive at an overall room correction curve. This step added some notch filters to correct room resonances.
The biggest “by ear“ part is adjusting the overall tilt of the spectrum across the band. What sounds best to me is a flat curve that is tilted somewhat downwards so the bass is 7-10dB hotter than the treble at listener position.
At 1 meter they should measure quite well “out of the box.”
Then as a 2nd step I took measurements at listener position and combined them with listening to arrive at an overall room correction curve. This step added some notch filters to correct room resonances.
The biggest “by ear“ part is adjusting the overall tilt of the spectrum across the band. What sounds best to me is a flat curve that is tilted somewhat downwards so the bass is 7-10dB hotter than the treble at listener position.
Last edited:
Hi JMacll, Awesome, thank you! I'd love to have photos and your drawing for guidance. Having your and Perry's examples and instruction on this thread will support all of us just getting started on this journey to replicate Perry's designs.
I use the controls on the output channels to achieve flat response with the microphone at nearfield distance (0.5M - 1.5M). Then I use the Parametric EQ on the input channels to correct measured response at listener position. I typically use high Q notch filters to get rid of room modes in the bass and low Q shelf filters to get the overall balance between bass, mid and treble right. @JMacII
This is response in my room at listener position, average of 12 positions near my couch. Bottom octaves are 6-8dB stronger than top octaves.
I take the nearfield measurements pretty seriously, but the listener position measurements I take with a grain of salt. I just try to fix the most obvious 2-3 room modes and get tonal balance right.
This is response in my room at listener position, average of 12 positions near my couch. Bottom octaves are 6-8dB stronger than top octaves.
I take the nearfield measurements pretty seriously, but the listener position measurements I take with a grain of salt. I just try to fix the most obvious 2-3 room modes and get tonal balance right.
Last edited:
Thanks for this! With the Dirac version I don’t have the input PEQ options. I’ll post my frq response later today. Im getting there and it’s sounding really good. I’m wondering what you would think of using the cross over for the low/mid in the two way set up? Like a 12db or 24db/octave at 1500hz instead of a high shelf filter in the PEQ. That would free one of my PEQ filters.
That will screw up the MF/HF phase relationship. You can try it but I think it will create a cascade of other issues that you’ll still have to fix with more parametric filters.
The goal of the design is to get as close as possible to a 12dB acoustic slope on all drivers near the crossover points.
I’m guessing there’s another parametric filter somewhere that you can skip and which DIRAC can then fix for you instead. Maybe the low-end rolloff for example. I believe DIRAC asks you to shape your target low frequency curve in the setup.
The goal of the design is to get as close as possible to a 12dB acoustic slope on all drivers near the crossover points.
I’m guessing there’s another parametric filter somewhere that you can skip and which DIRAC can then fix for you instead. Maybe the low-end rolloff for example. I believe DIRAC asks you to shape your target low frequency curve in the setup.
Some people would be tempted to mess with the 3 kHz notch and replace it with a second order low pass, but I don’t think that’s a good path.
If you want to free up a PEQ band, you have several good choices. You could free up EQ3 which is the 15Hz infrasonic filter (you could swap in a 12dB /oct high pass instead) or EQ5 (in my room it corrects a 75 Hz room mode) or EQ9 (corrects a 40 Hz room mode).
I'm guessing you're going to have a couple of room modes of your own, which may be different frequencies than mine, but you'll still have them. So the 15Hz band is your best bet and like I said, you can add a 12dB/octave high pass at 15Hz in the Xover menu or you can handle it with the DIRAC.
Basically you can push the Bitches Brews has hard as you want down to 30Hz and then below that frequency you start hitting diminishing returns. People who listen at insane SPLs should probably roll them off below 30.
Normal sane people can push them all the way to 20 Hz without any trouble. The curve you see above is a very reasonable setting that will work for most folks. These guidelines help you set the "curtains" of the DIRAC lower frequency range.
The way I approach this minimizes ringing and phase shift, which I personally believe is a pivotal reason why the BB's sound so open and transparent. This crossover doesn't look anything like what most speaker designers would normally do.
If you want to free up a PEQ band, you have several good choices. You could free up EQ3 which is the 15Hz infrasonic filter (you could swap in a 12dB /oct high pass instead) or EQ5 (in my room it corrects a 75 Hz room mode) or EQ9 (corrects a 40 Hz room mode).
I'm guessing you're going to have a couple of room modes of your own, which may be different frequencies than mine, but you'll still have them. So the 15Hz band is your best bet and like I said, you can add a 12dB/octave high pass at 15Hz in the Xover menu or you can handle it with the DIRAC.
Basically you can push the Bitches Brews has hard as you want down to 30Hz and then below that frequency you start hitting diminishing returns. People who listen at insane SPLs should probably roll them off below 30.
Normal sane people can push them all the way to 20 Hz without any trouble. The curve you see above is a very reasonable setting that will work for most folks. These guidelines help you set the "curtains" of the DIRAC lower frequency range.
The way I approach this minimizes ringing and phase shift, which I personally believe is a pivotal reason why the BB's sound so open and transparent. This crossover doesn't look anything like what most speaker designers would normally do.
Last edited:
Thanks for this Perry. I'll post my frequency responses I've been getting. I have been trying to use the REW software Minidsp recommends to create PEQ files. The EQ at .5M/1M looks good but at the listening position the responses are more ragged (though I have not done an average across multiple micorphone positions). I think I might try doing manual filters based on the bands and what I've saved as the "baseline" files you provided. In my room the baseline files have a bass bump but a more relaxed upper mid range than what the REW software is giving me.
Listening position WILL be ragged especially below 1Khz at any specific spot. You have to average 10-15 locations to get a somewhat smoother curve, and you still need to take it with a grain of salt. The nearfield measurements give you the most accurate picture.
These were all taken at .5M with the microphone 1M off the ground. I turned down the volume for the bottom traces so there is some seperation. The red is my "working" PEQ that I did a couple days ago. The lower collection is a slightly different set of PEQ files. The yellow trace is LF only, the black trace is HF only, the blue is full range. These two mixes are both 0 on the LF and +2 on the HF gain.
The red trace sounds more forward in the mids and with less bass than what I'm calling the "standard" PEQ which is what I downloaded. That "base" configuration is shown as the single red trace in the second image, you can see it's got the healthy bass bump. That's actually with the gain -1 on the LF circuits and +2 on the HF as well.
The red trace sounds more forward in the mids and with less bass than what I'm calling the "standard" PEQ which is what I downloaded. That "base" configuration is shown as the single red trace in the second image, you can see it's got the healthy bass bump. That's actually with the gain -1 on the LF circuits and +2 on the HF as well.
The black trace is the 15cxn88 tweeter, correct? Is that with or without the DSP in place?
Also, I suggest you measure on the axis of the 15CXN88 which is probably higher than 1M off the floor.
Also, I suggest you measure on the axis of the 15CXN88 which is probably higher than 1M off the floor.
Last edited:
the black trace is the tweeter with DSP, and yes that's not on axis, This magenta line is on axis with the same DSP files...pretty good!
Good, but something is off. There is a shelf filter that should be rolling the tweeter off below about 1500.
Can you post a DSP tweeter PEQ screen shot?
Can you post a DSP tweeter PEQ screen shot?
This is roughly what the nearfield on axis curves should look like for the 15CXN88 with DSP xover in place. The red lines show the tweeter acoustic slope is about 12dB/octave and the woofer is about 9dB/octave.
The DSP has a 6dB slope at 6KHz like this:
The black curve is for a subwoofer tap that's not used.
This is the PEQ curve for the tweeter section:
The DSP has a 6dB slope at 6KHz like this:
The black curve is for a subwoofer tap that's not used.
This is the PEQ curve for the tweeter section:
Last edited: