At the risk of coming off as overdemanding & scattered, I ask for help with another project: create a passive crossover to go between a GRS version of the NEO3 planar tweeter & the old Peerless/Tymphany NE123-8. https://loudspeakerdatabase.com/Peerless/NE123W-08
These are used in a 4-way active speaker that was a clone of the Linkwitz LX521 that I built last year for my son. I just converted it to a monopole closed box bass speaker. My son changed his mind about what he wants to do with the speakers: before, sound great, no need for real loud. Now, real loud, especially for deep bass. Force-cancellation OB Peerless SLS10 woofers did not cut it.
It took a lot of swearing and work but I managed to shoehorn a Dayton RSS265HF-8 to replace the original woofers. Now they are more like Juhazi's Ainograients -- monopole closed bass, OB above. No question this produces more deep bass at higher volume than the original config.
The new bass section has internal volume of ~1.3 cubic ft, stuffed with around 10-12 oz of teased long wool hair, a Q=0.68, Fs=39Hz, and within a foot of the wall behind, with +8 dB @ 35Hz, is down -3 dB at 30Hz. I doubt it can stay uncompressed to 100 dB, but that's probably good enough. So far, I have not heard the driver or 2x150W amp overload with music.
Before:
Now:
But as you might know from my recent forum posts, my son has expressed an interest in even more power/depth in the bass, which is mostly what is driving my investigation of subs and the SB Audience Nero 15" 800.
One of the challenges is the need for a 5th active band -- ie more than an 8-ch miniDSP OpenDRC-DA8 can handle -- and all the crossover complexity & expense this entails. In my son's home, such complexity is not really ideal.
SO... my thinking is to go passive between tweeter & high mid to reduce the active bands in the main speakers to 6 so the subwoofer, when it's introduced to the system, can run off the DA8 directly. The amplifier I built for him uses 3 of Eric A's 2x150W modules and one 2x300W module (from DIYaudio's vendor bazaar). The latter can be bridged to 600W, to minimum 6 ohm load, which suits a Nero 15SW800 well.
One day soon I have to learn to design passive crossovers with VituixCad, but that day hasn't come yet. Which is why I beg for your help. I'll provide whatever measurements are needed. I know it doesn't have to be perfect because PEQ in the miniDSP can be brought to bear, but it'd be nice to get close.
My target is acoustic (effective?) LR4 @ 3kHz. Open to LR2 & B3 as well. Tweeter needs 3-4 dB padding to compensate for its higher sensitivity.
I have a week-long window to pull this off, so there is time pressure. At least, I need to do all the raw driver measurements within a week.
These are used in a 4-way active speaker that was a clone of the Linkwitz LX521 that I built last year for my son. I just converted it to a monopole closed box bass speaker. My son changed his mind about what he wants to do with the speakers: before, sound great, no need for real loud. Now, real loud, especially for deep bass. Force-cancellation OB Peerless SLS10 woofers did not cut it.
It took a lot of swearing and work but I managed to shoehorn a Dayton RSS265HF-8 to replace the original woofers. Now they are more like Juhazi's Ainograients -- monopole closed bass, OB above. No question this produces more deep bass at higher volume than the original config.
The new bass section has internal volume of ~1.3 cubic ft, stuffed with around 10-12 oz of teased long wool hair, a Q=0.68, Fs=39Hz, and within a foot of the wall behind, with +8 dB @ 35Hz, is down -3 dB at 30Hz. I doubt it can stay uncompressed to 100 dB, but that's probably good enough. So far, I have not heard the driver or 2x150W amp overload with music.
Before:
Now:
But as you might know from my recent forum posts, my son has expressed an interest in even more power/depth in the bass, which is mostly what is driving my investigation of subs and the SB Audience Nero 15" 800.
One of the challenges is the need for a 5th active band -- ie more than an 8-ch miniDSP OpenDRC-DA8 can handle -- and all the crossover complexity & expense this entails. In my son's home, such complexity is not really ideal.
SO... my thinking is to go passive between tweeter & high mid to reduce the active bands in the main speakers to 6 so the subwoofer, when it's introduced to the system, can run off the DA8 directly. The amplifier I built for him uses 3 of Eric A's 2x150W modules and one 2x300W module (from DIYaudio's vendor bazaar). The latter can be bridged to 600W, to minimum 6 ohm load, which suits a Nero 15SW800 well.
One day soon I have to learn to design passive crossovers with VituixCad, but that day hasn't come yet. Which is why I beg for your help. I'll provide whatever measurements are needed. I know it doesn't have to be perfect because PEQ in the miniDSP can be brought to bear, but it'd be nice to get close.
My target is acoustic (effective?) LR4 @ 3kHz. Open to LR2 & B3 as well. Tweeter needs 3-4 dB padding to compensate for its higher sensitivity.
I have a week-long window to pull this off, so there is time pressure. At least, I need to do all the raw driver measurements within a week.
Attachments
Last edited:
What order crossover is most desirable to you? Along similar lines, is the baffle configuration fixed or are you open to aligning acoustic origins of the tweeter and mid?
The current setting in the minidsp is LR4 @ 4kHz. But I've tried LR2, and that works ok, too. No changes in driver positions possible -- just too hard to make it clean.
Can you please describe your measurement gear? Which measurement software do you use?
Best regards,
Michael
Best regards,
Michael
REW is what I am comfortable with. I use both UMIK1 USB mic & Dayton EMM6 with TASCAM mic preamp/usb audio interface. This TASCAM is a touch noisier than I'd like but it works fine.
Very good, I also use REW for loudspeaker measurement purposes. I also would recommend to use your Dayton-mic with the TASCAM Audio interface because of the possibility to measure with an electrical timing reference, very useful especially for developing passive crossovers by using simulating programs like VituixCAD, if not mandatory.
Can you please descibe your usual measuring process?
Best regards
Michael
edit:
Worth reading: https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf
...., although I have sometimes made compromises in this respect....🫣
Can you please descibe your usual measuring process?
Best regards
Michael
edit:
Worth reading: https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf
...., although I have sometimes made compromises in this respect....🫣
Last edited:
Usual measurement after drivers mounted in final baffle/box:
Start with very close measurements to establish core driver response -- 1" -- from the middle of my fairly well damped 22x19' studio.
Then I go to a foot, try a few sweeps -- 0, 15, 30, 45, 60 degrees.
Try again at a meter. These are usually the worst with room effects.
I tend to use 12-24", avg 0-15-30 measurements to first PEQ the driver responses, then work out filter F & slopes, usually working with natural driver rolloff.
The 2nd half of the crossover development process is sweeps at different SPL & distortion measurements interspersed with extensive listening.
This has all been with DSP for the last decade.
This 2nd LX521 clone sounded very good to about 95~97 dB peaks. With the new sealed woofer, it's better in the bass, to about 100 dB.
I am open to new & different ways of doing things, especially for anyone willing to use the data to develop the passive filters.
Start with very close measurements to establish core driver response -- 1" -- from the middle of my fairly well damped 22x19' studio.
Then I go to a foot, try a few sweeps -- 0, 15, 30, 45, 60 degrees.
Try again at a meter. These are usually the worst with room effects.
I tend to use 12-24", avg 0-15-30 measurements to first PEQ the driver responses, then work out filter F & slopes, usually working with natural driver rolloff.
The 2nd half of the crossover development process is sweeps at different SPL & distortion measurements interspersed with extensive listening.
This has all been with DSP for the last decade.
This 2nd LX521 clone sounded very good to about 95~97 dB peaks. With the new sealed woofer, it's better in the bass, to about 100 dB.
I am open to new & different ways of doing things, especially for anyone willing to use the data to develop the passive filters.
I actually only make near-field measurements (i.e. as close as possible) to determine frequency responses for the bass range, whereby a baffle step correction is applied to these measurements.In your case, I think such near-field measurements are unnecessary.
Measurements from a distance of 30 cm (about one foot) are well suited for distortion measurements. Distortion measurements over frequency are one criterion for finding a favorable crossover frequency (directivity is an other)
I would make the actual measurements to import into VituixCAD from at least one meter away to adequately account for the influence of the front baffle. I make such measurements from 0 to 90 degrees in 10-degree increments.
The room influence can be suppressed by setting a time gate, whereby the gate does not have to be particularly long here, as the crossover frequenzy will be comparatively high.
Best regards
Michael
Measurements from a distance of 30 cm (about one foot) are well suited for distortion measurements. Distortion measurements over frequency are one criterion for finding a favorable crossover frequency (directivity is an other)
I would make the actual measurements to import into VituixCAD from at least one meter away to adequately account for the influence of the front baffle. I make such measurements from 0 to 90 degrees in 10-degree increments.
The room influence can be suppressed by setting a time gate, whereby the gate does not have to be particularly long here, as the crossover frequenzy will be comparatively high.
Best regards
Michael
So much data can be generated so quickly it's often mind-boggling.
I started by calibrating the level of the Dayton EMM6 mic & Tascam UH-7000 preamp/interface at 94 dB with my trusty mic calibrator -- accurate to well within a dB. Then I started doing some sweeps of the speaker as a singular system, then individual drivers, with all the crossover filters & PEQ.... then realized this is not useful.
Started a new config in miniDSP, imported only the mid & tweeter settings, then removed all filters except a LR4 high pass at 500Hz to protect the drivers. The sweep was set for 1 to 18kHz, run on axis, then 10, 20, 30, 40, 50 & 60 deg. Loopback was used as timing reference. All shown in the 2 images below.
I presume the data files are needed?
What more would be useful?
For example, I could show the PEQ'd sweeps of both drivers, too, at the same range of angles.
No time windowing applied, btw.
I started by calibrating the level of the Dayton EMM6 mic & Tascam UH-7000 preamp/interface at 94 dB with my trusty mic calibrator -- accurate to well within a dB. Then I started doing some sweeps of the speaker as a singular system, then individual drivers, with all the crossover filters & PEQ.... then realized this is not useful.
Started a new config in miniDSP, imported only the mid & tweeter settings, then removed all filters except a LR4 high pass at 500Hz to protect the drivers. The sweep was set for 1 to 18kHz, run on axis, then 10, 20, 30, 40, 50 & 60 deg. Loopback was used as timing reference. All shown in the 2 images below.
I presume the data files are needed?
What more would be useful?
For example, I could show the PEQ'd sweeps of both drivers, too, at the same range of angles.
No time windowing applied, btw.
Not sure what causes the 5-6 kHz peak in the planar tweeter. It's evident in both of the speakers. And plainly audible.
I forgot to mentions -- all measurements are at 1m, with mic at the height which corresponds to the the point between tweeter & mid.
I forgot to mentions -- all measurements are at 1m, with mic at the height which corresponds to the the point between tweeter & mid.
Last edited:
How does it look like when using time windowing? May be that a gate length of 2 ms is long enough for your purposes. To have more control you can let REW show you the impulse response (in "%", not in "dBFS"). An example (but with significant more distance to reflecting surfaces):
Note: the settings in "IR windows" must be identical for all measurements. You can do your preferences using one measurement and apply them to the others using the button "Apply windows to all" (I recommend not to use "Apply to all, keep ref time",
Using the new preset-feature in "IR windows" may be another option, I've not test it yet because it's a new feature for me (I've installed the newest version of REW a few minutes ago).
Best regards
Michael
Note: the settings in "IR windows" must be identical for all measurements. You can do your preferences using one measurement and apply them to the others using the button "Apply windows to all" (I recommend not to use "Apply to all, keep ref time",
Using the new preset-feature in "IR windows" may be another option, I've not test it yet because it's a new feature for me (I've installed the newest version of REW a few minutes ago).
Best regards
Michael
It is also there in the midrange, but less peaky. It could eventually be the dipole peak, when the rear and front wave add up constructively.
(I see similar baffle width for both drivers).
Uups, I've forgotten that this is about a MHT-unit built as an open baffle. Unfortunatly I've not experience in developing such constructions. 🙁
Best regards
Michael
Best regards
Michael
Yeah, of course. How quickly I forget -- both null & dip in all narrow dipoles... but not always audible.
So you're bailing? Pls don't! I think the passive filters only have to get reasonably close based on averaged frontal 0~30 deg sweeps. The entire response will be PEQ'd afterwards with up to 10 bands anyway, then further tweaked with the input PEQ for "room curve" in the final destination, anyway.
A 4kHz crossover point is mostly below the effect of the 5.5-6kHz bump in tweeter response (which is pretty narrow), and also below the more gradual rise @6+ kHz in the mid. Achieving an acoustic 4kHz LR4 would be my target -- never mind the bumps. I can eliminate most of it with PEQ in the OpenDRC-DA8.
The measured response of the whole (revised) speakers shows ~2dB rise from 5kHz to 6kHz, then a quick dip between 6-6.5kHz, but I really don't hear the effect with a variety of music. I'm pretty sure I can achieve that when both tweeter and mid are on the same amp channel.
The measured response of the whole (revised) speakers shows ~2dB rise from 5kHz to 6kHz, then a quick dip between 6-6.5kHz, but I really don't hear the effect with a variety of music. I'm pretty sure I can achieve that when both tweeter and mid are on the same amp channel.
A reassessment of driver measurements, particularly FR & directivity makes me realize 3kHz would be a better passive crossover with 3~4 dB reduction of tweeter level.
The .mdat-file doesn't include impedance measurements of the drivers. They are necessary for developing passive crossovers.
Best regards
Michael
Best regards
Michael
Hmmm... I'd never shared REW measurements before; I thought all the data that's available on the various tabs would be included in that zip folder. It's the only thing that gets saved, iirc, & every tab of the REW screen gets filled with relevant data when the folder is opened. I'll take a close look when I get back into the studio in the morn.
If this helps, the tweeter impedance is a flat 3R5. Straight line. The mid is slowly rising from 8R @1kHz to 10R @ around 10khz; iirc, at 3kHz, it's just below 9R.
If this helps, the tweeter impedance is a flat 3R5. Straight line. The mid is slowly rising from 8R @1kHz to 10R @ around 10khz; iirc, at 3kHz, it's just below 9R.