|
|
Post by marcl on Mar 8, 2024 17:37:58 GMT -5
Yeah well ... not so fast! It looked pretty good with no Dirac right? So I did the domino Dirac dance, and look what it did ... (sorry I attached the wrong one before ... this is with Dirac) View AttachmentFlat curves, but L/R are dropped down right at the place where the active crossover goes to the subs. And the center crosses to Bass Management at 100 (Fronts) and it's way high! I have no clue what's going on. This is just from a single measurement in Dirac to see if it was in the ballpark. I even reinstalled 3.9.7 and used the single measurement again. I'm stumped ... I did everything as we always have ... I need a nap! hmmmmm . . . . Is there a Level adjustment on the Active XO modules? If so, may need to play with that a bit, trying higher and then lower settings. But first, try just a single mic point run with Dirac first to see if you get a duplicate result which would indicate that you could do some quick Diracs trying various things. Yes, I spent a couple hours tweaking the LCR Hi and Lo levels in the crossover this morning until it looked like this with no EQ.  Then I did the Dirac with one measurement and it looked like this.  So I took a nap. Uninstalled Dirac 3.9.7 and installed that last production 3.8.2 (even though I had had success weeks ago with 3.9.1 and 3.9.7). Then I lowered the level of the fronts going to the subs -10db (L/R low passed at 50Hz). I did a single Dirac measurement with mic level 100%, sub -18, everybody else -25. Now the subs and the L/R measure in Dirac at the same peak level in the bass as they should. But here's the result. L/R look fine. But BM is elevated about +8db. This is the same symptom I had in May 2022. All of a sudden, after a Dirac calibration, BM is not working properly.  |
|
|
|
Post by marcl on Mar 9, 2024 10:17:14 GMT -5
Thanks for listening to my blow by blow on this Jawn. Here’s the simple state of things: • Definitely not a problem with, related to, or caused by the active crossover. That is working great!• Definitely Dirac/Emotiva related … unintended interaction between Dirac and Emotiva Bass Management • Looks identical to the May 2022 Dirac-BM related issue. Similar to December 2022 issue. Those were reported to Emotiva but no solution was evident, and I solved them with a roundabout Dirac-related process. The active crossover is only on LCR. The first plot is when I reverted to Dirac 3.8.2 and took a single measurement yesterday. L/R are perfect. C is good at the active crossover point 200Hz. But C, SL and SR are elevated +8db-ish at the Bass Management crossover to the Fronts. Second plot is from May 2022. Different crossover points, but same issue. Measurements from March 2022 do not have this issue. Emotiva Firmware had not changed before and after the problems happened. Just Dirac versions. In December 2022 a similar issue happened when going to a new Dirac version, and its symptom was that Bass Management was disabled. The “cure” in both cases was to go back to an old version of Dirac and then go forward again. Not a precise process because I was trying different things including Factory Reset, loading old Measurement files, etc. So I just have to beat through it today trying Dirac versions and remeasuring and see if the issue clears itself.   |
|
|
|
Post by webmst007 on Mar 9, 2024 17:48:56 GMT -5
Thank you for sharing this journey Marc. 😁 On another note - here is the latest 3 way release the K235 - not cheap - sublimeacoustic.com/ |
|
|
|
Post by marcl on Mar 9, 2024 18:23:54 GMT -5
Thank you for sharing this journey Marc. 😁 On another note - here is the latest 3 way release the K235 - not cheap - sublimeacoustic.com/Yes I saw that. Interesting timing. I don't need a three way, but it's good to see they have another fully integrated product. |
|
|
|
Post by webmst007 on Mar 9, 2024 18:35:18 GMT -5
Agreed 💯
|
|
|
|
Post by lhracing on Mar 9, 2024 20:15:15 GMT -5
I think this is the right place to move this topic since it has migrated from "Latest Purchase" to "Installed and down the rabbit hole". I just put together an LCR 2-Way active crossover with boards and power supply from XKitz. $240 for the electronics including shipping, plus a few bucks for the case, power switch, etc. www.xkitz.com/collections/active-crossovers-and-bi-amplifiers-1/products/linkwitz-riley-2-way-active-crossover-fully-balanced-xover-2bI installed it yesterday and it's basically working. I just have to fine tune the levels of the outputs, then rerun a Dirac calibration. View AttachmentSo what's the point of all this? It's simpler than it looks: - Cross the Fronts to the subs at 50Hz outside of Processor Bass Management, so that the Fronts remain configured as Large, and are used in the system for Bass Management (Center Sub=LFE).
- Cross the CCR Center speaker to two DWMs(Dynamic Woofer Modules) at 200Hz, bi-amping the Center channel. Center is configured Small, and crosses to Bass Management (i.e. Large Fronts) at 40-60Hz (TBD)
- The schematic also shows the Front L/R Hi outputs split to a miniDSP to drive two DWMs which fill in bass cancellations.
- Note that each DWM has two voice coils, so the L/R Hi outputs go to one voice coil in each L/R DWM; and the Lo output of the Center channel goes to the other two voice coils wired in series. DWMs are located next to the Front L/R speakers and slightly forward of the CCR Center which is above the TV.
I immediately see in the measurements that I seem to have achieved another goal which is to reduce bass distortion by high-passing the Fronts at 50Hz. Although they play down to nearly 30Hz, they don't respond well to dynamic content that low. This configuration protects them and the DWMs from very low frequency transients. The XKitz modules are well-made and the hardest part of assembly was cutting holes in the box. But it's all together and working. I'm not expecting to use the Baffle Step Compensation feature as it's intended for bi-amping drivers in a box, but I will try turning the pots just to see if there is any effect. Otherwise, adjustment is confined to tweaking the separate Hi and Lo output levels with individual 20-turn pots. And here's a measurement of LCR (blue, green, pink) with no overall PEQ or Dirac. All I did was pull the sub 40Hz down in the miniDSP, and did some All Pass filters on the two DWMs for the L/R to eliminate some dips. View AttachmentYou have my interest in your active crossovers and when you have everything setup I would like your final thoughts. I have some DYI 3-way speakers in a 2-channel system that I have actively crossed with a Behringer Super-X Pro. It works but the crossovers that you are using look to be a much more of an audiophile crossover than the Pro unit that I am using. |
|
|
|
Post by marcl on Mar 9, 2024 21:14:50 GMT -5
I think this is the right place to move this topic since it has migrated from "Latest Purchase" to "Installed and down the rabbit hole". I just put together an LCR 2-Way active crossover with boards and power supply from XKitz. $240 for the electronics including shipping, plus a few bucks for the case, power switch, etc. www.xkitz.com/collections/active-crossovers-and-bi-amplifiers-1/products/linkwitz-riley-2-way-active-crossover-fully-balanced-xover-2bI installed it yesterday and it's basically working. I just have to fine tune the levels of the outputs, then rerun a Dirac calibration. View AttachmentSo what's the point of all this? It's simpler than it looks: - Cross the Fronts to the subs at 50Hz outside of Processor Bass Management, so that the Fronts remain configured as Large, and are used in the system for Bass Management (Center Sub=LFE).
- Cross the CCR Center speaker to two DWMs(Dynamic Woofer Modules) at 200Hz, bi-amping the Center channel. Center is configured Small, and crosses to Bass Management (i.e. Large Fronts) at 40-60Hz (TBD)
- The schematic also shows the Front L/R Hi outputs split to a miniDSP to drive two DWMs which fill in bass cancellations.
- Note that each DWM has two voice coils, so the L/R Hi outputs go to one voice coil in each L/R DWM; and the Lo output of the Center channel goes to the other two voice coils wired in series. DWMs are located next to the Front L/R speakers and slightly forward of the CCR Center which is above the TV.
I immediately see in the measurements that I seem to have achieved another goal which is to reduce bass distortion by high-passing the Fronts at 50Hz. Although they play down to nearly 30Hz, they don't respond well to dynamic content that low. This configuration protects them and the DWMs from very low frequency transients. The XKitz modules are well-made and the hardest part of assembly was cutting holes in the box. But it's all together and working. I'm not expecting to use the Baffle Step Compensation feature as it's intended for bi-amping drivers in a box, but I will try turning the pots just to see if there is any effect. Otherwise, adjustment is confined to tweaking the separate Hi and Lo output levels with individual 20-turn pots. And here's a measurement of LCR (blue, green, pink) with no overall PEQ or Dirac. All I did was pull the sub 40Hz down in the miniDSP, and did some All Pass filters on the two DWMs for the L/R to eliminate some dips. View AttachmentYou have my interest in your active crossovers and when you have everything setup I would like your final thoughts. I have some DYI 3-way speakers in a 2-channel system that I have actively crossed with a Behringer Super-X Pro. It works but the crossovers that you are using look to be a much more of an audiophile crossover than the Pro unit that I am using. I'll definitely post when I get it all working. So far I think the crossover looks good. I got everything balanced but when I ran Dirac all kinds of weird stuff happened. I spent a whole day trying to sort it out and finally wired everything back without the crossover just to get back to a baseline. I really do think something weird with Dirac happened along with some weird state of the processor as eventually it froze and I had to do a Factory Reset. I'll work through it again tomorrow. I also looked at "pro" options and miniDSP and this crossover looks like a simple clean solution. |
|
|
|
Post by marcl on Mar 11, 2024 10:41:41 GMT -5
Progress over the weekend. Still not sure what happened Friday and Saturday but the fact that behavior with Dirac was erratic and eventually the processor froze and I had to do a cold/hard reboot and Factory Reset ... I'm just going to forget about all that!  Yesterday i put the system back the way it was without the active crossover. Everything was good. Then I incorporated the active crossover only for the center channel, crossing above 200Hz to the CCR center and below 200Hz to the DWM woofers. It took a lot of tweaking levels in the crossover to get the response of the Hi/Lo aligned, and when it was done I ran Dirac and it was good. Here's LCR with Dirac ... Psychoacoustic Smoothing and the center crosses to Bass management (large L/R) at 40Hz.  And here I have the center set to large and we see the overall response as well as the Lo and Hi response crossing at 200Hz, 24db/octave. This is exactly what we want!  At this point the L/R are not going through the active crossover. Testing today will be the following: 1. L/R 3.7s through XO HPF 50Hz; DWMs/Subs from 2x4 HD as they are now 2. L/R 3.7s through XO HPF 50Hz; Subs from XO LPF 50Hz; DWMs from 2x4 HD as they are now 3. L/R XO HPF 50Hz to splitter; 3.7s and DWMs direct from splitter; Subs from XO LPF 50Hz; 2x4 HD eliminated This should all work to a degree and will no doubt take some tweaking of levels in the crossover. Scenario 3 would be ideal, but it does not let me independently control levels of the DWMs and their interaction with their respective 3.7 so it may not work. I'll be happy with scenario 2.  Schematic reminder ...  |
|
|
|
Post by marcl on Mar 13, 2024 8:42:15 GMT -5
Okay, this journey into incorporating an LCR 2-way active crossover has stabilized with all components in place, Dirac calibration successful, and a much simplified configuration. Benefits: - Fronts are HPF at 50Hz which eliminates possibility of diaphragm slap with transients below 30Hz; LPF at 50Hz to same subs as LFE
- Maintains Fronts=Large with subs full response down to 16Hz, and they play Bass Management
- Significant reduction in Front channel bass distortion
- Center CCR speaker crosses over with active crossover 24db/octave at 200Hz, vs passive 12db/octave at 250Hz; Bass is biamped to two DWMs
- Eliminated miniDSP HD for simplified configuration, less latency, and eliminates crossover cancelations with small speaker Bass Management
|
|
|
|
Post by leonski on Mar 13, 2024 22:47:52 GMT -5
The ONLY thing that I don't get is HP and LP at the same frequency.
ME? I'd separate by maybe 10hz......So that as they fall, they may SUM FLAT thru the passband.
But that's just my noodle.....
|
|
|
|
Post by marcl on Mar 14, 2024 2:49:29 GMT -5
The ONLY thing that I don't get is HP and LP at the same frequency. ME? I'd separate by maybe 10hz......So that as they fall, they may SUM FLAT thru the passband. But that's just my noodle..... Well the way I did this symmetrical is what you get. You pick a crossover and they put in a resistor module that sets the frequency. But it does sum flat across the passband if the filters are designed properly, as these are. You just may not see it in a room measurement because other things are going on. But on a bench with a signal generator and scope the HPF and LPF sum at the crossover and it is flat. |
|
KeithL
Administrator 
Posts: 10,273
|
Post by KeithL on Mar 14, 2024 9:13:58 GMT -5
I suspect you don't understand how "the cutoff frequencies of filters" actually work. Filters aren't like buying beer... where, the day after your birthday, you are allowed to. ALL filters are, to some degree, "gradual" (that's what the "slope" describes"). (And, in case you weren't aware, "one octave" means "double or halve the frequency"... so, for an "80 Hz crossover", 40 Hz is "one octave below", and 160 Hz is "one octave above" it.) So, for example, let's take a look at "a typical 12 dB/octave high-pass filter at 80 Hz"... like you might find on your main speakers if you have a subwoofer... At frequencies "well above 80 Hz" its output is going to be arbitrarily close to flat... And, at frequencies "well below 80 Hz" its output is going to fall at a slope that is pretty close to 12 dB/octave. However, typically, its output is going to start dropping a bit above 80 Hz... so will really be about -3 dB at 80 Hz. In the filters most commonly used for audio the "turnover frequency" is defined as "the frequency where the response is down 3 dB". (It's sort of "the spot where that line that follows the slope would cross the zero point if it was a real straight line".) So your "typical 12 dB/octave 80 Hz high-pass filter might actually have a response that is.... 0 dB at 640 Hz (that's flat three octaves above the cutoff point) 0 dB at 320 Hz (that's still flat two octaves above the cutoff) -0.1 dB at 160 Hz (that's -0.1 dB at one octave above cutoff) - 3 dB at 80 Hz (that's -3 dB at the cutoff frequency) -15 dB at 40 Hz (that's -15 dB at one octave below the cutoff frequency) -27 dB at 20 Hz (that's -27 dB at two octaves below the cutoff) (Notice that now we're following that "12 dB/octave slope line"). So, in other words, it's about -3 dB at the "crossover point"... And it "settles into a slope of 12 dB/octave once it gets a ways past that point"... (And, again, I'm using that vague language, because different types of filters follow slightly different curves around that point in the graph) And, typically, if you have both a high-pass and a low-pass filter, both "set to the same frequency", they will both be down 3 dB at that frequency, and will try to sum to 0 dB at that frequency. Note that I said typically because this is one of the details that is determined by "what sort of filter it is"... So, for example, some filter types actually have a bump in their response right before their set frequency... And some filters "sum flat in amplitude but not in phase" or vice versa... Also bear in mind that filters have phase shift so, even if "the amplitudes add up to zero", because they are not perfectly in-phase, they will NOT sum to 0 dB anyway. (So, in essence, you can calculate things "so that each half is right" or "so that the sum of the two is right"... but not both. ) The other thing you need to keep in perspective is that small changes in value, especially small percentage changes, DO NOT tend to produce major changes in the actual response curve. For example, let's look at the numbers I gave above for a typical "80 Hz 12 dB/octave high pass filter... Now change that "cutoff frequency" from "80 Hz" to "70 Hz" instead of "80 Hz" and LOOK AT THE ACTUAL VALUES. Our new filter is going to be "-3 dB at 70 Hz" instead of "-3 dB at 80 Hz".... So, instead of being -3 dB at 80 Hz, our new filter will be more like -2.5 dB at 80 Hz. So, if you actually look at the amplitude at any given frequency, there won't actually be much difference. (If you look at the two graphs they're shifted "horizontally" but, at any given point, they aren't that far apart "vertically".) And that difference will be insignificant compared to the differences between speakers... or rooms... or moving the microphone a few inches. Also remember that "it's dB that we hear"... A 3dB difference, which is double the power, is only slightly audible... And most people agree that, even in the midrange frequencies, the minimum that would be audible would be 1-2 dB... And a 10% difference, which seems like a significant number, as a change in "output power", would probably not be audible at all... My point here is that, if you're set to 80 Hz, and you "want to try a different frequency to see how different it sounds"... You probably wouldn't even bother with 70 Hz... (And a "10 Hz gap or overlap - at 80 Hz" is probably not going to be audible either way.) Just go right to 40 Hz or 50 Hz if you want to hear "a significant difference"... The ONLY thing that I don't get is HP and LP at the same frequency. ME? I'd separate by maybe 10hz......So that as they fall, they may SUM FLAT thru the passband. But that's just my noodle..... |
|
|
|
Post by marcl on Mar 14, 2024 9:46:12 GMT -5
I suspect you don't understand how "the cutoff frequencies of filters" actually work. Filters aren't like buying beer... where, the day after your birthday, you are allowed to. ALL filters are, to some degree, "gradual" (that's what the "slope" describes"). (And, in case you weren't aware, "one octave" means "double or halve the frequency"... so, for an "80 Hz crossover", 40 Hz is "one octave below", and 160 Hz is "one octave above" it.) .................................. The ONLY thing that I don't get is HP and LP at the same frequency. ME? I'd separate by maybe 10hz......So that as they fall, they may SUM FLAT thru the passband. But that's just my noodle..... And that reminds me, does Bass Management in the processors apply both ways at the selected 12 or 24db/octave? or is one side always 12? Lots of detail on the XKitz page about how these Linkwitz-Riley crossovers are designed. One relevant quote "The high and low outputs are phase aligned to within a fraction of a degree to eliminate distortion at and around the XO frequency" ... because ... "A Linkwitz-Riley filter uses a parallel combination of low-pass and high-pass filters which results in zero gain at the crossover frequency. Because of the zero gain at the crossover frequency, a Likwitz-Riley filter behaves like an all-pass filter delivering a flat amplitude response and a smooth changing phase response. A Linkwitz-Riley filter is made by combining two Butterworth filters. The main difference between the two is that Butterworth crossovers are 3dB down at the filter cutoff frequency while the Linkwitz-Riley filters are flat." |
|
|
|
Post by foggy1956 on Mar 14, 2024 10:56:17 GMT -5
I suspect you don't understand how "the cutoff frequencies of filters" actually work. Filters aren't like buying beer... where, the day after your birthday, you are allowed to. ALL filters are, to some degree, "gradual" (that's what the "slope" describes"). (And, in case you weren't aware, "one octave" means "double or halve the frequency"... so, for an "80 Hz crossover", 40 Hz is "one octave below", and 160 Hz is "one octave above" it.) .................................. And that reminds me, does Bass Management in the processors apply both ways at the selected 12 or 24db/octave? or is one side always 12? Lots of detail on the XKitz page about how these Linkwitz-Riley crossovers are designed. One relevant quote "The high and low outputs are phase aligned to within a fraction of a degree to eliminate distortion at and around the XO frequency" ... because ... "A Linkwitz-Riley filter uses a parallel combination of low-pass and high-pass filters which results in zero gain at the crossover frequency. Because of the zero gain at the crossover frequency, a Likwitz-Riley filter behaves like an all-pass filter delivering a flat amplitude response and a smooth changing phase response. A Linkwitz-Riley filter is made by combining two Butterworth filters. The main difference between the two is that Butterworth crossovers are 3dB down at the filter cutoff frequency while the Linkwitz-Riley filters are flat." I believe the HP is fixed at 24db but the memory is fuzzy. |
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on Mar 15, 2024 15:41:58 GMT -5
Actually I believe the high-pass is fixed at 12 dB/octave. And that reminds me, does Bass Management in the processors apply both ways at the selected 12 or 24db/octave? or is one side always 12? Lots of detail on the XKitz page about how these Linkwitz-Riley crossovers are designed. One relevant quote "The high and low outputs are phase aligned to within a fraction of a degree to eliminate distortion at and around the XO frequency" ... because ... "A Linkwitz-Riley filter uses a parallel combination of low-pass and high-pass filters which results in zero gain at the crossover frequency. Because of the zero gain at the crossover frequency, a Likwitz-Riley filter behaves like an all-pass filter delivering a flat amplitude response and a smooth changing phase response. A Linkwitz-Riley filter is made by combining two Butterworth filters. The main difference between the two is that Butterworth crossovers are 3dB down at the filter cutoff frequency while the Linkwitz-Riley filters are flat." I believe the HP is fixed at 24db but the memory is fuzzy. |
|
|
|
Post by leonski on Mar 16, 2024 20:38:25 GMT -5
I might add that the 10hz I cited in my earlier post is at low frqucnies.
24db /octave from a 40hz crossover is a robust 24 down at 80hz.
At a mid-tweet frequccy of say.....4khz? Gap just be much larger.....
And it's not just 'response' but also phase that's important .
Which is why I later suggested FIR filters which have no phase shift thru passband....
|
|
ttocs
Global Moderator
I always have a wonderful time, wherever I am, whomever I'm with. (Elwood P Dowd)
Posts: 8,162
|
Post by ttocs on Apr 1, 2024 21:30:49 GMT -5
All I can say right now is, Wow! Six 12" subs in a horizontal line array along the bottom of the front wall below the tv panel. Over a year ago I was just beginning to play with new subwoofer arrangements when the fun stuff had to take a backseat to serious life altering situations affecting two people in my life. I'm finally in getting back in gear and have been playing with the subs for a couple weeks. The goal is great impulse response. Last year I laid out four 12" subs along the front wall under the tv panel, with gaps between each one about the width of each sub. This worked really well with little effort. But there was a lot more to gain from more effort. In December just before I'd undergone surgery, I bought two more subs to add to the four. I knew I wouldn't be able to move the subs for a long time, so I put two stacks of three subs each flanking the tv panel and left it like that until I was able to continue working on the project. So this is what I started with a couple weeks ago. I'd had experience with stacked subs before, 2 subs in each stack, and got really good results. I was able to get good results with 3 in each stack this time also, but two locations of subs is not the best for seat to seat variation. Enter the six sub horizontal line array. Lined up along the front wall with gaps of about half the width of each sub worked very well, but I expected better results. So I pushed all the subs together, no gaps. This proved to be immediately more successful. I'm using two miniDSP 2x4HD units, three subs each, and so far just using the Center Sub Output in Mono and then split into a Y with each RCA plug getting half the balanced output. One RCA is +HOT and Ground, the other is -HOT and Ground and the miniDSP channels are inverted for this side. Using nothing more than delay for each sub, and then Dirac taking things further, the result was extremely good right away! After an evening of tweaking while playing lots of music, things improved. While I think there’s more improvement to be had, this is sounding fantastic! Something I didn't expect - no bass bloat along the back or side walls like I had with every other subwoofer setup in my room! There is extra bass in one rear corner of the room, but not like it used to be. So the bass seems to be pretty even as I walk about the room. Then I put on Top Gun: Maverick for just a few minutes, or so I thought. It sounded so phenomenal that I ended up staying up way too late for a Sunday night and watched the entire movie! The bass is heard, felt, voluminous, but doesn’t overpower any other sounds or dialogue. It’s really, really, great! There’s more to be done, but this is way better than I would have thought possible with so little effort.  |
|
|
|
Post by leonski on Apr 2, 2024 13:24:16 GMT -5
quote from ttocs:::
'There’s more to be done, but this is way better than I would have thought possible with so little effort.'
Sorry, ttocs, the above just cracks me up. The under TV array of 6 (count 'em, 6) subs and the large center channel speaker is heavy and a lot of moving stuff around.....
Moving those 'stats must be a lot of work, too......And just the adjustments to make it 'better' could take DAYS or longer.
If you want some better photography, just let me know.......?? I think I could get some more detail without 'blowing' the highlights......
|
|
|
|
Post by 405x5 on Apr 2, 2024 13:52:10 GMT -5
“The goal is great impulse response.” ….. Beyond impulse Power from Star Trek I’m at a loss as to what exactly it is. You mean by that could you elaborate? |
|
ttocs
Global Moderator
I always have a wonderful time, wherever I am, whomever I'm with. (Elwood P Dowd)
Posts: 8,162
|
Post by ttocs on Apr 2, 2024 17:05:28 GMT -5
“The goal is great impulse response.” I’m at a loss as to what exactly it is. You mean by that could you elaborate? Impulse is one of the aspects of how a driver responds that can be measured, and can reflect on how well or poorly it's operating. If multiple speakers in a system are playing the same sound, but at different times, the impulse will be smeared, bass response will be muddy and boomy. Adjust the timing of the group of speakers so they are aligned well, the impulse will be a single spike, bass response will be crisp, articulate, and impactful. Here is an example of an impulse response that is pretty much perfect! This is from a system with 8 18" subwoofers connected to a Trinnov processor using Trinnov's WaveForming subwoofer tuning system. I've never before seen, nor even thought that an impulse from 8 subwoofers could look this good.  I'll post some of my impulse measurements in the coming days. Right now I'm enjoying the fruits. |
|
