My wife is going to be scratching her head (left side new project vs previous build on right)… but, but I did change up the stain color (slightly) ha.
I’ve been a fan of coaxials (especially this Celestion with it’s common motor / magnet, and the Beyma’s with their extended horn design), so I’m hoping the MarkAudio is the star of the show just being supplemented on the extremes.
Front vs rear vented measurements- front with slight rise ~ 300 Hz downward (as per co’s measurements) should work well relative to BSC (this Xover will be active, btwn MarkAudio and W). Vented reduced back pressure vs a sealed enclosure.
And potential opportunity of setting / positioning Super Tw 10 to 15 degrees inward (relative to MarkAudio) to fill inward (listening area) response. I’ll be playing with active crossover levels and orders- then trying to match with passive components…
My modeled (Xovers) vs real world don’t often rhyme.
So, I focused first on the beyma tw and it’s looking like a 3rd order may be the way to go. I was hoping, as with Nelson Pass’s JBL L300 Xover mod to go with a single 1uf cap, but it’s not rolling it off smoothly. Targeting ~8k +/- btwn MarkAudio (looking at the off center measurements) and the Beyma super tweeter. I’ve got a lot of playing around with the values yet to do, but tossed together a quick 2nd order LP for the MarkAudio (I’ll need to mod to bump it a bit before its roll) just to be able to play a few tunes for the eve. Been sweeping all day…
What I do find the modeling good for is once I’ve got an initial design is to be able to adjust one value/ component at a time and make a note of it’s affect (and to toss in add’l resistors here and there) to take back to the real world passive I’m playing with - reviewing the resulting measurements.
My best matchup (real vs sim) I've gotten sofar is by taking better control of the mic and speaker positioning and using the summed response alignment method for acoustic offset win WinPCD. Though this method is considered outdated by now. It still works (relative to the mic position).
I got to the point that any component value changes either didn’t improve or was worse, so I placed some capacitor orders (had the inductors in inventory).
The MarkAudio will have a 2nd order LP with Jantzen Z-Silver caps (it will be active btwn it and the woofer on the low end, thinking 4th order LR).
The Beyma compression S tweeter will have a 3rd order HP, with C1 a Jantzen Z-Silver and a ClarityCap CSA in C2. And, if the ClarityCap is the weak link I’m good with it- have used them in my previous builds. Final Xover ~6600.
Just playing now with the new build on the left side w/o woofer (but with an active 2nd order HP @105), along with my previous build on the right. I’ll post a comparison when the new build gets completed.
@DrewsBrews said:
My best matchup (real vs sim) I've gotten sofar is by taking better control of the mic and speaker positioning and using the summed response alignment method for acoustic offset win WinPCD.
I revisited the on axis measurements of the MarkAudio (and the Beyma) and changed the cap orders moving back towards a 8kHz Xover, seeing that the MarkAudio response is fairly stable (off axis) through this point, and moved the HP of the Beyma (Super Tweeter) back up accordingly.
Measurements (microphone) aimed btwn Mid and Tw at 30 inches >< these drivers’ response at listening area. Near(er) field measurements not as important as at listening area. Recently read an interesting article on hearing vs measurement at near field and comb filtering. How the mind / hearing (naturally) deals with comb filtering issues. Focusing crossover on individual driver responses at listening area and interaction at the far field (less emphasis on near field constructive / deconstructive interaction).
. . . Recently read an interesting article on hearing vs measurement at near field and comb filtering. How the mind / hearing (naturally) deals with comb filtering issues. Focusing crossover on individual driver responses at listening area and interaction at the far field (less emphasis on near field constructive / deconstructive interaction).
Interesting. Would you have a link to this article? I think our mind is playing tricks on us. We probably have "built-in" compensation filters that are flattening out the FR curve. Just like our eyes/brain have the ability to correct the color balance of a scene in a room. But when we take a picture of the same scene, the camera's auto white balance software often messes up the colors.
One of many, but it adds credence to those whom proffer critical listening over measurements. I used to be in the measurement camp (cables, amps... sorta 'show me the money') but moving to a balance of both, in that our mind/hearing perception cannot necessarily always be replicated by a simple measuring microphone.
I think our mind is playing tricks on us. We probably have "built-in" compensation filters that are flattening out the FR curve. Just like our eyes/brain have the ability to correct the color balance of a scene in a room. But when we take a picture of the same scene, the camera's auto white balance software often messes up the colors.
... and how we see a sequence of static picture frames in a movie, yet 'see' a linear continuum of movement / action...
If your head is staying in one spot there is really not much to notice for comb filtering. Only when your ears move through space on the axis in question is it all that noticeable. IMO we are programed to "make the best of the situation" and being too distracted by missing frequencies to notice the sound of the predator creeping up behind us was probably not beneficial to survival. So yes we can still easily discern a guitar sound even when the tone is way off, either completely missing frequencies or others boosted. Something that a computer may have a hard time at. However there is a difference between recognition and sounding "good" or "correct", etc.
I agree both measuring and listening are equally necessary. It is hard to get in the ballpark with listening alone unless you have a ton of experience. That is where measuring comes in.
3rd order HP with 2nd order LP ~ 8200 (no eq, expected pre mid/tw passive) btwn super Tw and MarkAudio mid, wired tweeter inverted. Will be bringing L (on LP section) up to 0.28+ from .25 as shown in measurements to flatten Xover range a bit, expect slight drop in freq Xover point as well (guestimate 7800 to 8000?).
Shown vs Pass Mod L300 Xover (inverted mid) w 1st order HP on the JBL S Tw.
My set-up with the waveguide on the beyma super tweeter exhibited a bit of a shelf in the 6-8g area (when set with a 1st order HP filter), hence the 3rd order HP
Waiting for a couple of back ordered Jantzen 1.0 caps. In the meantime a few alligator clips and components on the floor (and not running the woofer). High WAF ( jk ), but she puts up with me.
Unwound a couple of inductors, per measurements, to 0.27 for the mid’s LP. This is the fun part of the builds, imo next to the concept and design.
After more listening and measuring, going with a 3.9 uF (vs 2.2) in the 2nd order LP for the MarkAudio. The resulting slope matches up better with that of the 3rd order HP on the slot tweeter and, as such, levels out the crossover range response.
I’m currently listening / comparing the MarkAudio/beyma slot tweeter (with the waveguide, and the Lpad set at the middle) combo (on the left side) vs my previous build with its Celestion coaxial (as the right speaker). The MarkAudio is a 4 in driver vs the Celestion at 5.25 (both as measured across ctr of surround). What I’m hearing through the mid and up (I believe) is that while the Celestion seems to be a bit fuller the MarkAudio/beyma sounds a bit cleaner. Maybe size difference is showing, and/or the super tweeter is stepping up? Additionally, the MarkAudio/Beyma side-to-side is superior to the Celestion, and I’m giving credit to Nate / silverD for his waveguide printing. Vertically, similar higher frequency rolloff.
My 2cents, and I’ve been biased towards coaxials, is that the coaxial (once a good crossover is established) is more stable (not necessarily more extended as more rolloff) off center than a separated mid / tw, but the latter can play cleaner. I’ve gone back with a capacitor reorder, so completion and final listening with a woofer will be delayed.
The ‘final parts’ were just delivered- but in the meantime I’ve been thinking about alternative Xovers concepts. I worked on a series Xover, and while I got the mid rolloff as I wanted, the net S tweeter frequency/ rolloff was problematic. As such I’ve concluded the series route would just put the S tweeter at risk.
I’ve included an Lpad (as per many JBL’s) for the S tweeter (again borrowing from the JBL L300 / Pass modified Xover) but, because I’m not able to go with a 1st order (slow rolloff/slope) adjustments beyond minimal +/- is not a smooth transition. Although set at a tight range +/- slight upward tilt for my old ears is on target.
I’m going to try a viable inductor set-up with the Lpad as used in full range drivers. Looking to be able to adjust the S tweeter tilt above the Xover point (vs just up and down per adjusted overall level). Playing around with modeling looks like it may be the way to go - then there is putting the model into practice- fingers crossed.
In the end, going forward with passive 2nd order LP (0.27mH inductor - 3.9uF ClarityCap) / 3rd order HP ( 4.0ohm 20w resistor - 1.0uF Jantzen Silver Z - 0.1mH inductor - 2.2uF Jantzen Silver Z / Lpad).
This was the simplest passive / best match I found for this driver combo (as measured), and I tried several. This gets the Xover ~6800 between MarkAudio and the beyma slot S Tweeter.
Will be active (bi-amped, my go-to) between Woofer and Mid/Super Tw unit (4th LR) with a bit of active eq across the board (more so attenuating peaks). The Lpad more than likely set with a tad bit of high frequency rise (I’m an old fart…).
The speakers got buttoned up, and I’ve initially set an active 4order LR @265Hz (nets ~280Hz) btwn woofer and MarkAudio/beyma, but this may get moved upwards a little (as well as level relative to woofer)? Established the Lpad max at +3dB with the 4ohm resistor at front of beyma’s filter, and expect +/- 1.5 range is use. The measurement was at 30 inches, with a bit of smoothing eq, none added at the low end currently.
That looks really nice! Since you are active between the woofer and the passive mid-tweet combo, and you can do some digital delay (time alignment), I'd at least try a LR2 xo slope around 400-500 Hz.
@PWRRYD said:
That looks really nice! Since you are active between the woofer and the passive mid-tweet combo, and you can do some digital delay (time alignment), I'd at least try a LR2 xo slope around 400-500 Hz.
Thanks.
I’ll look at a LR2 in that range. I’d like (to think anyway) that the MarkAudio can hang at 380hz, as the smaller ATC 3in dome crosses in that range. The more the broad-ranger can carry (across the voice range and beyond the 1-4Khz range) the better.
So, your thinking an L2 might provide a smoother transition?
And, thx - I hadn’t thought of playing with delay.
I've personally always like 2nd order slopes between the woofer and the mid in 3 ways. That's hard to pull off passively because the woofer accoustic center is almost always way too far behind the mid's. But with DSP you can totally fix that with some delay to the M-T to time align it to the woofer. Just remember that LR-2 needs the woofer in reverse polarity (electrically) to the mid.
Sorry for an add’l pick, this is the back side. Top (of box) venting of the MarkAudio back wave, and the 4pole speakONs (satisfying click connection lock).
@PWRRYD said:
Since you are active between the woofer and the passive mid-tweet combo, and you can do some digital delay (time alignment), I'd at least try a LR2 xo slope around 400-500 Hz.
I’m now targeting 380Hz LR 2nd (as measured vs electrical, vs 265 LR4th, per your ~direction) and have been playing with DSP and REW looking at measuring delay. I got the attached by adding a 1ms delay (per miniDSP instructions) but not sure if my measurement is correct. To get in the ballpark- via comparing the depths of the W relative to the Markaudio Mid (est motor center depths) what would be the appropriate delay for a 28mm difference?
I have the active LR2nd LP/HP electric @380Hz, nets measured ~355Hz, with a 0.09ms delay (~1.1in equivalent set-back diff, for the mid-tw combo relative to the Woofer). Green and pink pre-filters to get the levels equal.
Below measured 30in (@ mid-point of Markaudio mid and Beyma tw).
I’ll live with this awhile as set. And this is a sealed woofer, w/o eq extension at low end.
Comments
My wife is going to be scratching her head (left side new project vs previous build on right)… but, but I did change up the stain color (slightly) ha.
I’ve been a fan of coaxials (especially this Celestion with it’s common motor / magnet, and the Beyma’s with their extended horn design), so I’m hoping the MarkAudio is the star of the show just being supplemented on the extremes.
Same but different
Front vs rear vented measurements- front with slight rise ~ 300 Hz downward (as per co’s measurements) should work well relative to BSC (this Xover will be active, btwn MarkAudio and W). Vented reduced back pressure vs a sealed enclosure.
And potential opportunity of setting / positioning Super Tw 10 to 15 degrees inward (relative to MarkAudio) to fill inward (listening area) response. I’ll be playing with active crossover levels and orders- then trying to match with passive components…
My modeled (Xovers) vs real world don’t often rhyme.
So, I focused first on the beyma tw and it’s looking like a 3rd order may be the way to go. I was hoping, as with Nelson Pass’s JBL L300 Xover mod to go with a single 1uf cap, but it’s not rolling it off smoothly. Targeting ~8k +/- btwn MarkAudio (looking at the off center measurements) and the Beyma super tweeter. I’ve got a lot of playing around with the values yet to do, but tossed together a quick 2nd order LP for the MarkAudio (I’ll need to mod to bump it a bit before its roll) just to be able to play a few tunes for the eve. Been sweeping all day…
What I do find the modeling good for is once I’ve got an initial design is to be able to adjust one value/ component at a time and make a note of it’s affect (and to toss in add’l resistors here and there) to take back to the real world passive I’m playing with - reviewing the resulting measurements.
Ok so it's not just me lol.
My best matchup (real vs sim) I've gotten sofar is by taking better control of the mic and speaker positioning and using the summed response alignment method for acoustic offset win WinPCD. Though this method is considered outdated by now. It still works (relative to the mic position).
I got to the point that any component value changes either didn’t improve or was worse, so I placed some capacitor orders (had the inductors in inventory).
The MarkAudio will have a 2nd order LP with Jantzen Z-Silver caps (it will be active btwn it and the woofer on the low end, thinking 4th order LR).
The Beyma compression S tweeter will have a 3rd order HP, with C1 a Jantzen Z-Silver and a ClarityCap CSA in C2. And, if the ClarityCap is the weak link I’m good with it- have used them in my previous builds. Final Xover ~6600.
Just playing now with the new build on the left side w/o woofer (but with an active 2nd order HP @105), along with my previous build on the right. I’ll post a comparison when the new build gets completed.
Passive crossovers are tough, I struggle, but at the same time I am eager to learn.
I revisited the on axis measurements of the MarkAudio (and the Beyma) and changed the cap orders moving back towards a 8kHz Xover, seeing that the MarkAudio response is fairly stable (off axis) through this point, and moved the HP of the Beyma (Super Tweeter) back up accordingly.
Measurements (microphone) aimed btwn Mid and Tw at 30 inches >< these drivers’ response at listening area. Near(er) field measurements not as important as at listening area. Recently read an interesting article on hearing vs measurement at near field and comb filtering. How the mind / hearing (naturally) deals with comb filtering issues. Focusing crossover on individual driver responses at listening area and interaction at the far field (less emphasis on near field constructive / deconstructive interaction).
. . . Recently read an interesting article on hearing vs measurement at near field and comb filtering. How the mind / hearing (naturally) deals with comb filtering issues. Focusing crossover on individual driver responses at listening area and interaction at the far field (less emphasis on near field constructive / deconstructive interaction).
Interesting. Would you have a link to this article? I think our mind is playing tricks on us. We probably have "built-in" compensation filters that are flattening out the FR curve. Just like our eyes/brain have the ability to correct the color balance of a scene in a room. But when we take a picture of the same scene, the camera's auto white balance software often messes up the colors.
https://www.audioholics.com/loudspeaker-design/comb-filtering
One of many, but it adds credence to those whom proffer critical listening over measurements. I used to be in the measurement camp (cables, amps... sorta 'show me the money') but moving to a balance of both, in that our mind/hearing perception cannot necessarily always be replicated by a simple measuring microphone.
... and how we see a sequence of static picture frames in a movie, yet 'see' a linear continuum of movement / action...
If your head is staying in one spot there is really not much to notice for comb filtering. Only when your ears move through space on the axis in question is it all that noticeable. IMO we are programed to "make the best of the situation" and being too distracted by missing frequencies to notice the sound of the predator creeping up behind us was probably not beneficial to survival. So yes we can still easily discern a guitar sound even when the tone is way off, either completely missing frequencies or others boosted. Something that a computer may have a hard time at. However there is a difference between recognition and sounding "good" or "correct", etc.
I agree both measuring and listening are equally necessary. It is hard to get in the ballpark with listening alone unless you have a ton of experience. That is where measuring comes in.
3rd order HP with 2nd order LP ~ 8200 (no eq, expected pre mid/tw passive) btwn super Tw and MarkAudio mid, wired tweeter inverted. Will be bringing L (on LP section) up to 0.28+ from .25 as shown in measurements to flatten Xover range a bit, expect slight drop in freq Xover point as well (guestimate 7800 to 8000?).
Shown vs Pass Mod L300 Xover (inverted mid) w 1st order HP on the JBL S Tw.
My set-up with the waveguide on the beyma super tweeter exhibited a bit of a shelf in the 6-8g area (when set with a 1st order HP filter), hence the 3rd order HP
Waiting for a couple of back ordered Jantzen 1.0 caps. In the meantime a few alligator clips and components on the floor (and not running the woofer). High WAF ( jk ), but she puts up with me.
Unwound a couple of inductors, per measurements, to 0.27 for the mid’s LP. This is the fun part of the builds, imo next to the concept and design.
I enjoy the playing with (voicing) stage as well.
It's the best part - never ending possibilities and experiments . . .
After more listening and measuring, going with a 3.9 uF (vs 2.2) in the 2nd order LP for the MarkAudio. The resulting slope matches up better with that of the 3rd order HP on the slot tweeter and, as such, levels out the crossover range response.
I’m currently listening / comparing the MarkAudio/beyma slot tweeter (with the waveguide, and the Lpad set at the middle) combo (on the left side) vs my previous build with its Celestion coaxial (as the right speaker). The MarkAudio is a 4 in driver vs the Celestion at 5.25 (both as measured across ctr of surround). What I’m hearing through the mid and up (I believe) is that while the Celestion seems to be a bit fuller the MarkAudio/beyma sounds a bit cleaner. Maybe size difference is showing, and/or the super tweeter is stepping up? Additionally, the MarkAudio/Beyma side-to-side is superior to the Celestion, and I’m giving credit to Nate / silverD for his waveguide printing. Vertically, similar higher frequency rolloff.
My 2cents, and I’ve been biased towards coaxials, is that the coaxial (once a good crossover is established) is more stable (not necessarily more extended as more rolloff) off center than a separated mid / tw, but the latter can play cleaner. I’ve gone back with a capacitor reorder, so completion and final listening with a woofer will be delayed.
The ‘final parts’ were just delivered- but in the meantime I’ve been thinking about alternative Xovers concepts. I worked on a series Xover, and while I got the mid rolloff as I wanted, the net S tweeter frequency/ rolloff was problematic. As such I’ve concluded the series route would just put the S tweeter at risk.
I’ve included an Lpad (as per many JBL’s) for the S tweeter (again borrowing from the JBL L300 / Pass modified Xover) but, because I’m not able to go with a 1st order (slow rolloff/slope) adjustments beyond minimal +/- is not a smooth transition. Although set at a tight range +/- slight upward tilt for my old ears is on target.
I’m going to try a viable inductor set-up with the Lpad as used in full range drivers. Looking to be able to adjust the S tweeter tilt above the Xover point (vs just up and down per adjusted overall level). Playing around with modeling looks like it may be the way to go - then there is putting the model into practice- fingers crossed.
In the end, going forward with passive 2nd order LP (0.27mH inductor - 3.9uF ClarityCap) / 3rd order HP ( 4.0ohm 20w resistor - 1.0uF Jantzen Silver Z - 0.1mH inductor - 2.2uF Jantzen Silver Z / Lpad).
This was the simplest passive / best match I found for this driver combo (as measured), and I tried several. This gets the Xover ~6800 between MarkAudio and the beyma slot S Tweeter.
Will be active (bi-amped, my go-to) between Woofer and Mid/Super Tw unit (4th LR) with a bit of active eq across the board (more so attenuating peaks). The Lpad more than likely set with a tad bit of high frequency rise (I’m an old fart…).
Nice quality components
The speakers got buttoned up, and I’ve initially set an active 4order LR @265Hz (nets ~280Hz) btwn woofer and MarkAudio/beyma, but this may get moved upwards a little (as well as level relative to woofer)? Established the Lpad max at +3dB with the 4ohm resistor at front of beyma’s filter, and expect +/- 1.5 range is use. The measurement was at 30 inches, with a bit of smoothing eq, none added at the low end currently.
That looks really nice! Since you are active between the woofer and the passive mid-tweet combo, and you can do some digital delay (time alignment), I'd at least try a LR2 xo slope around 400-500 Hz.
Thanks.
I’ll look at a LR2 in that range. I’d like (to think anyway) that the MarkAudio can hang at 380hz, as the smaller ATC 3in dome crosses in that range. The more the broad-ranger can carry (across the voice range and beyond the 1-4Khz range) the better.
So, your thinking an L2 might provide a smoother transition?
And, thx - I hadn’t thought of playing with delay.
I've personally always like 2nd order slopes between the woofer and the mid in 3 ways. That's hard to pull off passively because the woofer accoustic center is almost always way too far behind the mid's. But with DSP you can totally fix that with some delay to the M-T to time align it to the woofer. Just remember that LR-2 needs the woofer in reverse polarity (electrically) to the mid.
Sorry for an add’l pick, this is the back side. Top (of box) venting of the MarkAudio back wave, and the 4pole speakONs (satisfying click connection lock).
Just remember that LR-2 needs the woofer in reverse polarity (electrically) to the mid.
Thx (again), simple as a click / to reverse.
I’m now targeting 380Hz LR 2nd (as measured vs electrical, vs 265 LR4th, per your ~direction) and have been playing with DSP and REW looking at measuring delay. I got the attached by adding a 1ms delay (per miniDSP instructions) but not sure if my measurement is correct. To get in the ballpark- via comparing the depths of the W relative to the Markaudio Mid (est motor center depths) what would be the appropriate delay for a 28mm difference?
FW its worth, I’m getting a range of 0.078 and 0.15ms. ???
I have the active LR2nd LP/HP electric @380Hz, nets measured ~355Hz, with a 0.09ms delay (~1.1in equivalent set-back diff, for the mid-tw combo relative to the Woofer). Green and pink pre-filters to get the levels equal.
Below measured 30in (@ mid-point of Markaudio mid and Beyma tw).
I’ll live with this awhile as set. And this is a sealed woofer, w/o eq extension at low end.
Thanks for everyone’s input here!