My blending methodology for rear A:
I first take a look at my sims to get a general idea of high and low frequency levels. The 1st pic below shows the comparison. For level matching, I used the levels around the 5kHz peak and tried to average out the levels around the measured peaks at ~1000Hz and 2300Hz. Then I look down at 200Hz to see what the SPL level is after baffle step loss in the sim. I get 77dB. So that's what I'm going to be shooting for in the blending.
Gating was 4ms so blending Fmin is about 250Hz and with a 4" cone, blending Fmax is 1077Hz.
I use the blending tools to add in the baffle diffraction to the NF measurement and I actually use the box modeling to add in the port contribution, tuned to 56Hz re your measured NF null and I used 13L (I can't remember the exact Vb. You can correct it if I'm wrong).
Using those criteria, I ended up blending between 550Hz and 1000Hz and choosing an SPL adjustment of -28dB for the NF data. Setting SPL level another dB lower might look like a better choice below about 600Hz but then the match above about 700Hz is farther off and the SPL at 200Hz is 76dB instead of 77dB, so I'm choosing a blending level that kind of averages those differences above and below about 600Hz. The 2nd pic hopefully shows what I mean.
You can double check those 3 data points against what you did but your posted results look pretty much the same. Choosing the blending data points is by nature somewhat subjective so I include that first step of comparing measured FF data to the simmed data to help give me an extra point of reference in terms of making those decisions. In this case a difference of 1dB in SPL isn't going to make or break the xo simulation, so don't sweat it.
I first take a look at my sims to get a general idea of high and low frequency levels. The 1st pic below shows the comparison. For level matching, I used the levels around the 5kHz peak and tried to average out the levels around the measured peaks at ~1000Hz and 2300Hz. Then I look down at 200Hz to see what the SPL level is after baffle step loss in the sim. I get 77dB. So that's what I'm going to be shooting for in the blending.
Gating was 4ms so blending Fmin is about 250Hz and with a 4" cone, blending Fmax is 1077Hz.
I use the blending tools to add in the baffle diffraction to the NF measurement and I actually use the box modeling to add in the port contribution, tuned to 56Hz re your measured NF null and I used 13L (I can't remember the exact Vb. You can correct it if I'm wrong).
Using those criteria, I ended up blending between 550Hz and 1000Hz and choosing an SPL adjustment of -28dB for the NF data. Setting SPL level another dB lower might look like a better choice below about 600Hz but then the match above about 700Hz is farther off and the SPL at 200Hz is 76dB instead of 77dB, so I'm choosing a blending level that kind of averages those differences above and below about 600Hz. The 2nd pic hopefully shows what I mean.
You can double check those 3 data points against what you did but your posted results look pretty much the same. Choosing the blending data points is by nature somewhat subjective so I include that first step of comparing measured FF data to the simmed data to help give me an extra point of reference in terms of making those decisions. In this case a difference of 1dB in SPL isn't going to make or break the xo simulation, so don't sweat it.
Attachments
And for the rest of the drivers, my blending data points are:
Tower Woofers: 275Hz -> 1000Hz -> -29dB
Tower Mids: 300Hz -> 1200Hz -> -27dB
CC Woofers: 300Hz -> 1175Hz -> -35.5dB
CC Mids: 700Hz -> 1400Hz -> -29dB
See how those agree with your work.
Tower Woofers: 275Hz -> 1000Hz -> -29dB
Tower Mids: 300Hz -> 1200Hz -> -27dB
CC Woofers: 300Hz -> 1175Hz -> -35.5dB
CC Mids: 700Hz -> 1400Hz -> -29dB
See how those agree with your work.
I had very similar results for the blending. I didnt blend so high, probably around 800Hz, but it didnt change the result too much when I went even higher.
Thank you for the explanation of your reasoning behind it though, that is insightful!
I am itching to move ahead tho and will mess around with XSim.
While I do want to learn as much as I can, and still strive for that. I am personally trying to finish these speakers. I need to move onto job searching more intensely haha.
Maybe I'm also impatient to hear these
So Im going to mess around with XSim to get more of a feel of it all. But I am unclear as of now of the phase and what I should be going for with that. Also, when you say that I should extract the phase is not done already, is that in XSim or REW?
Thank you for the explanation of your reasoning behind it though, that is insightful!
I am itching to move ahead tho and will mess around with XSim.
While I do want to learn as much as I can, and still strive for that. I am personally trying to finish these speakers. I need to move onto job searching more intensely haha.
Maybe I'm also impatient to hear these
So Im going to mess around with XSim to get more of a feel of it all. But I am unclear as of now of the phase and what I should be going for with that. Also, when you say that I should extract the phase is not done already, is that in XSim or REW?
Last edited:
Sounds good.
For phase extracting, you can do it right in Response Blender for the FR's at that time for the larger drivers but otherwise I just do it in XSim when I'm loading the drivers. I believe you can save files from REW with the phase info but I've never actually done it that way. Call it personal preference but if I can I like to use all the same program to extract phase for all the files and it's right there in XSim when you load your data so that's just easier for me. Plus it has the tailing process included (as does Response Blender) which helps to create a more accurate phase file. Choose frd and zma phase source as measured if the file already has the phase extracted or choose derived if you want XSim to do it for you.
For phase extracting, you can do it right in Response Blender for the FR's at that time for the larger drivers but otherwise I just do it in XSim when I'm loading the drivers. I believe you can save files from REW with the phase info but I've never actually done it that way. Call it personal preference but if I can I like to use all the same program to extract phase for all the files and it's right there in XSim when you load your data so that's just easier for me. Plus it has the tailing process included (as does Response Blender) which helps to create a more accurate phase file. Choose frd and zma phase source as measured if the file already has the phase extracted or choose derived if you want XSim to do it for you.
So before you start on the actual xo sims, you should take a moment and decide on your overall parameters, like the xo point and xo slope and any other special thingies that might need your attention. There's 3 things for sure and sometimes 4 that I'll look at if I can.
1a) On-axis FR: Driver Resonance Frequency (Fs) -> Dickason states it this way. At a minimum, keep the xo frequency 2 x Fs and better at 3 to 4 x Fs. This rule of thumb (ROT) is generally most important for tweeters. So for the RST28, Fs is 710Hz which is fairly low or what we might call a pretty robust tweeter, so your Fmin would be 1420Hz and better yet up at ~2100Hz. Except this ROT doesn't take the xo slope into account. So I find it better to express it in dB. So keep the driver Fs a minimum of ~30dB down from the fundamental, maybe more if you're going to be pushing it hard. So if the acoustic slope is 2nd order (12dB/octave), Fmin is going to need to be about 2.5 x Fs or about 1800Hz but if you go 4th order acoustic* (24dB/octave), your Fmin only needs to be a little bit above 1420Hz.
In this case with a 2-way, the max SPL is going to be limited to a single woofer's output which won't be as loud as say your tower or CC 3-ways, so in this case you won't be asking too much from it so it could be fairly low if desired.
1b) Look for the Bad Stuff: Look at the on-axis FR and identify any problems you want to keep out of the passband or will have to pay special attention to. So this is mostly about cone breakup and resonant peaks but also other frequency peaks or possible problems. Nothing to note with the tweeter but we might take notice of the 5kHz or so resonant peak to the woofer although with a paper cone instead of the metal one, this resonance isn't quite as bad but you may want to do something about it anyways since it's simple and only takes 1 extra small value part or sometimes 2 (so not expensive to implement at all).
2) Off-axis ROT: Since the total sound you hear in-room is a combination of the on- and off-axis responses, you want to consider the off-axis response too when picking your xo point. Generally you want to keep the xo point below the frequency at which the woofer's response at 45 degrees (or the lower frequency driver's response) falls 3dB below the fundamental.
Looking at your rear A 45 degree measurement compared to the on-axis FF measurement, I'd probably say you want to keep Fxo below about 3000Hz.
You might also want to look at the tweeter's off-axis response and see what that looks like too. A narrow vs a wider dispersion pattern may mean you set the tweeter level slightly higher or lower in SPL. Here with the RST28, looking at the driver spec sheet, the 45 degree off-axis response is actually still really quite loud, so a fairly narrow dispersion pattern which is a good thing for HT, so we might might want to attenuate the on-axis response just a touch more than usual, although ROT #26 is of course that your ears are always going to be the final arbiter in all things xo in the end.
3) Harmonic Distortion: Just generally, you'll want to keep the worst offender, the 3rd order harmonic distortion as low as you can in the drivers' passbands. Have a look in REW at your drivers' FF measurements and hit the distortion tab. With these drivers I don't think you're going to find much affecting your decision based on the other factors but have a look anyways.
4) Cone Resonance Behavior: And lastly, a Cumulative Spectral Decay (CSD) or waterfall plot can let you know about other cone resonances that you want to avoid or treat depending on the situation. Again REW will do this for you although I have to admit that I've been a little lazy and haven't quite figured out how to get the program to display its waterfall plots in a meaningful format like Zaph or others have presented with different measuring programs. Not to worry here anyways as I'm pretty sure you're not going to find any problems here with the paper cones and the low Fs tweeter you've selected.
All of that to say that for your 2-ways, I would choose 2nd order acoustic Linkwitz-Riley (LR2) slopes with a xo target somewhere around 2500Hz. That xo point is not set in stone, sometimes you just have to let the drivers roll off where they kind of want to, but as an initial starting point that should be adequate. Since you have a fairly wide range for xo frequency and because you aren't going to be pushing these too hard, you don't need anything too steep so 2nd order acoustic should be fine. That'll usually keep the xo parts count and costs down as well compared with going with steeper LR4 slopes, or 4th order acoustic.
*Note finally that xo acoustic slopes order and xo electrical order are not necessarily the same thing. The acoustic order refers to the actual slope of the FR and can be achieved with whatever order of electrical parts that gets the job done and can vary depending on what the natural roll-off of the driver is and where you choose your xo point. The electrical order is just the number of series and parallel inductors and capacitors that are used in the actual xo filter.
1a) On-axis FR: Driver Resonance Frequency (Fs) -> Dickason states it this way. At a minimum, keep the xo frequency 2 x Fs and better at 3 to 4 x Fs. This rule of thumb (ROT) is generally most important for tweeters. So for the RST28, Fs is 710Hz which is fairly low or what we might call a pretty robust tweeter, so your Fmin would be 1420Hz and better yet up at ~2100Hz. Except this ROT doesn't take the xo slope into account. So I find it better to express it in dB. So keep the driver Fs a minimum of ~30dB down from the fundamental, maybe more if you're going to be pushing it hard. So if the acoustic slope is 2nd order (12dB/octave), Fmin is going to need to be about 2.5 x Fs or about 1800Hz but if you go 4th order acoustic* (24dB/octave), your Fmin only needs to be a little bit above 1420Hz.
In this case with a 2-way, the max SPL is going to be limited to a single woofer's output which won't be as loud as say your tower or CC 3-ways, so in this case you won't be asking too much from it so it could be fairly low if desired.
1b) Look for the Bad Stuff: Look at the on-axis FR and identify any problems you want to keep out of the passband or will have to pay special attention to. So this is mostly about cone breakup and resonant peaks but also other frequency peaks or possible problems. Nothing to note with the tweeter but we might take notice of the 5kHz or so resonant peak to the woofer although with a paper cone instead of the metal one, this resonance isn't quite as bad but you may want to do something about it anyways since it's simple and only takes 1 extra small value part or sometimes 2 (so not expensive to implement at all).
2) Off-axis ROT: Since the total sound you hear in-room is a combination of the on- and off-axis responses, you want to consider the off-axis response too when picking your xo point. Generally you want to keep the xo point below the frequency at which the woofer's response at 45 degrees (or the lower frequency driver's response) falls 3dB below the fundamental.
Looking at your rear A 45 degree measurement compared to the on-axis FF measurement, I'd probably say you want to keep Fxo below about 3000Hz.
You might also want to look at the tweeter's off-axis response and see what that looks like too. A narrow vs a wider dispersion pattern may mean you set the tweeter level slightly higher or lower in SPL. Here with the RST28, looking at the driver spec sheet, the 45 degree off-axis response is actually still really quite loud, so a fairly narrow dispersion pattern which is a good thing for HT, so we might might want to attenuate the on-axis response just a touch more than usual, although ROT #26 is of course that your ears are always going to be the final arbiter in all things xo in the end.
3) Harmonic Distortion: Just generally, you'll want to keep the worst offender, the 3rd order harmonic distortion as low as you can in the drivers' passbands. Have a look in REW at your drivers' FF measurements and hit the distortion tab. With these drivers I don't think you're going to find much affecting your decision based on the other factors but have a look anyways.
4) Cone Resonance Behavior: And lastly, a Cumulative Spectral Decay (CSD) or waterfall plot can let you know about other cone resonances that you want to avoid or treat depending on the situation. Again REW will do this for you although I have to admit that I've been a little lazy and haven't quite figured out how to get the program to display its waterfall plots in a meaningful format like Zaph or others have presented with different measuring programs. Not to worry here anyways as I'm pretty sure you're not going to find any problems here with the paper cones and the low Fs tweeter you've selected.
All of that to say that for your 2-ways, I would choose 2nd order acoustic Linkwitz-Riley (LR2) slopes with a xo target somewhere around 2500Hz. That xo point is not set in stone, sometimes you just have to let the drivers roll off where they kind of want to, but as an initial starting point that should be adequate. Since you have a fairly wide range for xo frequency and because you aren't going to be pushing these too hard, you don't need anything too steep so 2nd order acoustic should be fine. That'll usually keep the xo parts count and costs down as well compared with going with steeper LR4 slopes, or 4th order acoustic.
*Note finally that xo acoustic slopes order and xo electrical order are not necessarily the same thing. The acoustic order refers to the actual slope of the FR and can be achieved with whatever order of electrical parts that gets the job done and can vary depending on what the natural roll-off of the driver is and where you choose your xo point. The electrical order is just the number of series and parallel inductors and capacitors that are used in the actual xo filter.
And here's my Xsim set-up tutorial....
1. Start by loading your drivers, label them, derive phase if needed and input the proper delay.
2. Then move to the Frequency Response graph.
- Under 'Curves', turn off the System phase and set it as black.
- Set the tweeter as red and turn on the phase.
- Set the woofer as blue and turn on the phase.
- Turn on the LP filter (driver only) and set it to grey. I just turn 1 target on at a time to try and keep the chart a little less busy.
- (For 3-ways, set your mid to purple and turn on phase.)
And yes, I have a thing about color consistency. It's a hold over from starting with PCD in which the driver curve colors are fixed and it just makes reading the FR graph easier. At least for me anyways. lol
3. Under Scale, set Min Frequency to 10Hz and the Vertical Center to a value that allows you to see what the drivers are doing down at least 30 or 40dB below the fundamental. You'll want to be at about 70dB to 80dB here.
4. Because I use target filters from Response Modeler and because those are set at 90dB, you lastly need to adjust the SPL level of your curves to match the target curve levels. And the key here is...... set your woofer curve SPL level at about 200Hz to 90dB by increasing the amp wattage.
Note that #'s 2 and 3 (Curves and Scale) revert back to default every time you re-open XSim, so repeating those 2 sequences every time you use the program becomes necessary.
So after all of the above your XSim for your rears should look more or less like the attachment below. That's before any xo work. 😀 I can attach that if you like, just let me know.
1. Start by loading your drivers, label them, derive phase if needed and input the proper delay.
2. Then move to the Frequency Response graph.
- Under 'Curves', turn off the System phase and set it as black.
- Set the tweeter as red and turn on the phase.
- Set the woofer as blue and turn on the phase.
- Turn on the LP filter (driver only) and set it to grey. I just turn 1 target on at a time to try and keep the chart a little less busy.
- (For 3-ways, set your mid to purple and turn on phase.)
And yes, I have a thing about color consistency. It's a hold over from starting with PCD in which the driver curve colors are fixed and it just makes reading the FR graph easier. At least for me anyways. lol
3. Under Scale, set Min Frequency to 10Hz and the Vertical Center to a value that allows you to see what the drivers are doing down at least 30 or 40dB below the fundamental. You'll want to be at about 70dB to 80dB here.
4. Because I use target filters from Response Modeler and because those are set at 90dB, you lastly need to adjust the SPL level of your curves to match the target curve levels. And the key here is...... set your woofer curve SPL level at about 200Hz to 90dB by increasing the amp wattage.
Note that #'s 2 and 3 (Curves and Scale) revert back to default every time you re-open XSim, so repeating those 2 sequences every time you use the program becomes necessary.
So after all of the above your XSim for your rears should look more or less like the attachment below. That's before any xo work. 😀 I can attach that if you like, just let me know.
Attachments
Hey Chris!
Man I've been waiting to get back to this! I have had my Mothers 60th birthday on Thursday and family has been over ever since.
But I'm now ready to dive back into the xo sims!
Thank you for you tutorial and all. As always, much appreciated!
Do you mind attaching the target curves you have for me? I could do it, but with re-navigating and figuring out Response Modeler again and working out small kinks there, it would be very much appreciated to have that and move on with the sims!
In the mean time I'm going to be reading the sections on my books I got about xo design and my old school notes, more of a refresher to get the big picture and the nitty gritty back in my head.
Man I've been waiting to get back to this! I have had my Mothers 60th birthday on Thursday and family has been over ever since.
But I'm now ready to dive back into the xo sims!
Thank you for you tutorial and all. As always, much appreciated!
Do you mind attaching the target curves you have for me? I could do it, but with re-navigating and figuring out Response Modeler again and working out small kinks there, it would be very much appreciated to have that and move on with the sims!
In the mean time I'm going to be reading the sections on my books I got about xo design and my old school notes, more of a refresher to get the big picture and the nitty gritty back in my head.
Last edited:
Darn family. They get in the way of everything..... 😱 (just kidding)
Attached are the target curves, LR2 at 2500Hz. I will repeat though that they are just a suggestion, just a starting point in other words.
I might add that with the surround speakers, getting a perfectly flat response isn't perhaps quite as important as with your other speakers. Or to put it another way, I don't think I'd be throwing a whole slew of xo parts at the surround speakers trying to get the flattest response possible. Guess that's a not so subtle hint as you work your way through it.
Maybe post what you come up with and then I'll give you some feedback. Unless you want any more input to start. Just let me know.
Attached are the target curves, LR2 at 2500Hz. I will repeat though that they are just a suggestion, just a starting point in other words.
I might add that with the surround speakers, getting a perfectly flat response isn't perhaps quite as important as with your other speakers. Or to put it another way, I don't think I'd be throwing a whole slew of xo parts at the surround speakers trying to get the flattest response possible. Guess that's a not so subtle hint as you work your way through it.
Maybe post what you come up with and then I'll give you some feedback. Unless you want any more input to start. Just let me know.
Haha in the way, but worth it! 
Thank you for that, and yeah definitely just a starting point. But good for me to have a reference per-say.
Also, that is a good point about not trying to get it perfectly flat. Since the surrounds are only 2-way for starters, but also since they are also only surrounds. While not insignificant, not as important as the mains.
Thank you for that, and yeah definitely just a starting point. But good for me to have a reference per-say.
Also, that is a good point about not trying to get it perfectly flat. Since the surrounds are only 2-way for starters, but also since they are also only surrounds. While not insignificant, not as important as the mains.
Yes I loaded the blended NF + FF files I did for the rear woofer A. With the added simmed box response. Is that just the FF measurement in your attachment?
Yeah it was just the FF, I put in the blended response after posting that since of course that made much more sense.
But your blended response doesnt have that dip in the tuning frequency like mine does. Yours just rolls off nice looking like. What did you do there?
But your blended response doesnt have that dip in the tuning frequency like mine does. Yours just rolls off nice looking like. What did you do there?
Response Blender also allows you to splice on the LF box response to the NF measurement. Essentially that is adding in the response of the port, which the NF measurement doesn't include. Not absolutely essential but it does give a more complete picture.
Oh yeah thats right, I forgot you had done that.
So, now that I got the hang of setting up the XSim project. What do you suggest is next?
Like, what do you want to see from me next?
I think you want crossover points for each speaker. As far as the slope tho, I will look at them more but I'm thinking they will all end up being LR2. Possibly a 4th order if the steeper slope is necessary.
But tbh right now Im kinda stuck with finding the xo points. In school we kinda just looked at the FR's and chose a point that was reasonable with popular xo points and also what looked right.
It seems like there might be more to it than that tho haha.
Right now it seems like within a reasonable frequency range, I kinda just pick a point.
I will also be looking into "The Bad Stuff" more. But when I was doing that earlier I was getting lost in it cuz the responses are not exactly flat and have multiple peaks in different places. So idk if I should take note of them all or what.
I know you explained a process, but tbh I'm still just a little lost. Or more so, I feel like there is something I am missing, sorry 😕
I've been inputing the process for blending and AC's into my notebook today up until now.
Idk if I told ya but I keep a record of almost every step of the way in this project. So when I go to build another, I have exact documentation of what I have already done. Haha it takes a little time to cut out the screenshots and tape them in with then detailed explanations. But it helps me retain information much better and also then easy to go back and find what I need to do!
But anyways, I'm back to the xo stuff😎
So, now that I got the hang of setting up the XSim project. What do you suggest is next?
Like, what do you want to see from me next?
I think you want crossover points for each speaker. As far as the slope tho, I will look at them more but I'm thinking they will all end up being LR2. Possibly a 4th order if the steeper slope is necessary.
But tbh right now Im kinda stuck with finding the xo points. In school we kinda just looked at the FR's and chose a point that was reasonable with popular xo points and also what looked right.
It seems like there might be more to it than that tho haha.
Right now it seems like within a reasonable frequency range, I kinda just pick a point.
I will also be looking into "The Bad Stuff" more. But when I was doing that earlier I was getting lost in it cuz the responses are not exactly flat and have multiple peaks in different places. So idk if I should take note of them all or what.
I know you explained a process, but tbh I'm still just a little lost. Or more so, I feel like there is something I am missing, sorry 😕
I've been inputing the process for blending and AC's into my notebook today up until now.
Idk if I told ya but I keep a record of almost every step of the way in this project. So when I go to build another, I have exact documentation of what I have already done. Haha it takes a little time to cut out the screenshots and tape them in with then detailed explanations. But it helps me retain information much better and also then easy to go back and find what I need to do!
But anyways, I'm back to the xo stuff😎
So in terms of choosing the target xo point:
- if you use LR2, then the lowest point would be about 1800Hz or so (2.5 x tweeter Fs)
- at the high end, the off axis response of the woofer suggests staying below at least 3000Hz
- looking at the woofer FR, you've also got a resonance peak at about 5000Hz that you also want to keep suppressed and so I might try to keep the xo point at least an octave below that, so about 2500Hz, although you can treat this problem pretty easily with a special notch filter
- still looking at the woofer FR, you are right, you have a fairly large peak to deal with centered around 900Hz followed by a trough centered around 1600Hz which are both going to be somewhat problematic and could be dealt with in different ways.
- same thing with the tweeter, in the passband you've got a kind of wide but small peak in the response centered up around 8kHz which is right up around the sibilance zone. For me that would be something to look at with the front speakers but not something I would worry about with the rear speakers. Unless of course my ears tell me otherwise.
- you might want to look at the harmonic distortion of each driver too, but I'm pretty sure that isn't going to add any thing important with these 2 drivers.
So for LR2 slopes, I'd say pick a xo point in between about 1800Hz and 2500Hz. I just happened to choose 2500Hz. But 2000Hz or 2200Hz could be good as well.
But any of them are just a starting point for your sims. The important thing is to get in there and start adding components together and see how the drivers are going to respond and what the summed response looks like. And if there are any particular problems that show themselves. And if so, then start the sim again with the added info from your 1st attempt. I will almost always try out at least 4 or 5 different versions of the xo before I settle on something I want to try out. Be sure to save each version btw.
So for example, it is very typical in a 2-way to have some kind of mid peak in the mid/woofer response. Basically that's from the baffle diffraction effects. And typically, you have 2 choices on how to treat it. You can keep increasing the values of your LR2 filter until the top of the peak gets lowered to the appropriate level. But that might not end up resulting in the smoothest response. Maybe the xo point might need to be lowered in such a scenario sometimes. Or the other way to deal with it is to lighten up on the LR2 filter and then use a series notch filter wired in parallel to flatten the peak. Both methods are valid and you should try both.
But I think the most important thing for you is to just start playing with XSim. You're going to learn the most here by actually engaging in the process. Then post your 1st result and I can give you some feedback. I'm assuming some level of comfort with xo components and filters though, but if that's not quite the case, let me know and I can give you some more details on that score too.
- if you use LR2, then the lowest point would be about 1800Hz or so (2.5 x tweeter Fs)
- at the high end, the off axis response of the woofer suggests staying below at least 3000Hz
- looking at the woofer FR, you've also got a resonance peak at about 5000Hz that you also want to keep suppressed and so I might try to keep the xo point at least an octave below that, so about 2500Hz, although you can treat this problem pretty easily with a special notch filter
- still looking at the woofer FR, you are right, you have a fairly large peak to deal with centered around 900Hz followed by a trough centered around 1600Hz which are both going to be somewhat problematic and could be dealt with in different ways.
- same thing with the tweeter, in the passband you've got a kind of wide but small peak in the response centered up around 8kHz which is right up around the sibilance zone. For me that would be something to look at with the front speakers but not something I would worry about with the rear speakers. Unless of course my ears tell me otherwise.
- you might want to look at the harmonic distortion of each driver too, but I'm pretty sure that isn't going to add any thing important with these 2 drivers.
So for LR2 slopes, I'd say pick a xo point in between about 1800Hz and 2500Hz. I just happened to choose 2500Hz. But 2000Hz or 2200Hz could be good as well.
But any of them are just a starting point for your sims. The important thing is to get in there and start adding components together and see how the drivers are going to respond and what the summed response looks like. And if there are any particular problems that show themselves. And if so, then start the sim again with the added info from your 1st attempt. I will almost always try out at least 4 or 5 different versions of the xo before I settle on something I want to try out. Be sure to save each version btw.
So for example, it is very typical in a 2-way to have some kind of mid peak in the mid/woofer response. Basically that's from the baffle diffraction effects. And typically, you have 2 choices on how to treat it. You can keep increasing the values of your LR2 filter until the top of the peak gets lowered to the appropriate level. But that might not end up resulting in the smoothest response. Maybe the xo point might need to be lowered in such a scenario sometimes. Or the other way to deal with it is to lighten up on the LR2 filter and then use a series notch filter wired in parallel to flatten the peak. Both methods are valid and you should try both.
But I think the most important thing for you is to just start playing with XSim. You're going to learn the most here by actually engaging in the process. Then post your 1st result and I can give you some feedback. I'm assuming some level of comfort with xo components and filters though, but if that's not quite the case, let me know and I can give you some more details on that score too.
Thank you!
Writing everything out is actually helping tremendously.
I'm sorry for you having to basically repeat what you have said before.
I was just unable to juggle everything at once for a couple hours there. But I just gotta put it all down on paper.
For right now, Im taking into account the points you have said and knowledge I know and I am using REW to look at the responses, I can just see the FR's so much better there. Then I will get back to you on the xo points I have picked and why.
But in the mean time; you had told me to turn on phase in XSim for the responses. I am wondering where I should take into account phase. That has always, since school, been the thing that I havnt quite got a grip on exactly yet.
Writing everything out is actually helping tremendously.
I'm sorry for you having to basically repeat what you have said before.
I was just unable to juggle everything at once for a couple hours there. But I just gotta put it all down on paper.
For right now, Im taking into account the points you have said and knowledge I know and I am using REW to look at the responses, I can just see the FR's so much better there. Then I will get back to you on the xo points I have picked and why.
But in the mean time; you had told me to turn on phase in XSim for the responses. I am wondering where I should take into account phase. That has always, since school, been the thing that I havnt quite got a grip on exactly yet.
You have 3 main goals in your xo design:
1. Create an acceptable summed response. Plus or minus 3dB is widely held as sufficient since 3dB is considered perceptually as a just noticeable difference. I would say that's your minimum though. Plus or minus 1.5dB would be better for speakers of higher quality. Of course, if you have a preference away from a flat response, then that's ok too. Gently falling in the HF's or a smiley face are some people's other preferences for their summed response.
2. Keep the impedance amp friendly. For an 8ohm nominal amp, keep the minimum impedance above about 6ohm. For a 4ohm amp, keep the minimum above 3ohm.
3. Align the driver phase in the region that the drivers overlap, aka the xo region. This is why looking at the system phase tells you practically nothing, so turn that one off. So basically just keep your eye on the driver phase and try to line them up together as best you can on each side of the xo point. Pic 1 below shows acceptable phase alignment. Very frequently the spot where the phase wraps from top to bottom (ie. where it goes straight vertical) ends up close to the xo point but this is not necessarily always the case. If you flip the polarity of each driver, you get the result in pic 2. So exactly the same thing, just expressed differently. So here you can see that the drivers' phase cross pretty much at the xo point and then diverge above and below it. It would be better actually if they aligned more closely for a greater frequency range but sometimes, you just have to take what you can get. Pic 3 demonstrates the reverse null when you only flip the polarity on 1 driver and is usually a decent indication of good phase alignment. Deep is good but so is wide too. The wider the null, then the closer the alignment is for a wider range of frequencies.
In terms of a 2-way on a flat vertical baffle, usually going 3rd order electrical on the tweeter and 2nd order on the woofer will get the phase in the right neighborhood. Then it's up to the right combination of components and component values to get them as close as possible while still keeping goals 1 and 2 above also in the acceptable category. But things vary. Sometimes reversing the polarity of 1 of the drivers is needed too. Sometimes the same electrical orders will work too. I've got two 2nd order filters with the same polarity in the example attached, so it's hard to know what's needed. It's mostly a matter of trial and error to see what works, at least for me anyways, but it doesn't usually take too much effort to quickly see what's going to work and what isn't.
1. Create an acceptable summed response. Plus or minus 3dB is widely held as sufficient since 3dB is considered perceptually as a just noticeable difference. I would say that's your minimum though. Plus or minus 1.5dB would be better for speakers of higher quality. Of course, if you have a preference away from a flat response, then that's ok too. Gently falling in the HF's or a smiley face are some people's other preferences for their summed response.
2. Keep the impedance amp friendly. For an 8ohm nominal amp, keep the minimum impedance above about 6ohm. For a 4ohm amp, keep the minimum above 3ohm.
3. Align the driver phase in the region that the drivers overlap, aka the xo region. This is why looking at the system phase tells you practically nothing, so turn that one off. So basically just keep your eye on the driver phase and try to line them up together as best you can on each side of the xo point. Pic 1 below shows acceptable phase alignment. Very frequently the spot where the phase wraps from top to bottom (ie. where it goes straight vertical) ends up close to the xo point but this is not necessarily always the case. If you flip the polarity of each driver, you get the result in pic 2. So exactly the same thing, just expressed differently. So here you can see that the drivers' phase cross pretty much at the xo point and then diverge above and below it. It would be better actually if they aligned more closely for a greater frequency range but sometimes, you just have to take what you can get. Pic 3 demonstrates the reverse null when you only flip the polarity on 1 driver and is usually a decent indication of good phase alignment. Deep is good but so is wide too. The wider the null, then the closer the alignment is for a wider range of frequencies.
In terms of a 2-way on a flat vertical baffle, usually going 3rd order electrical on the tweeter and 2nd order on the woofer will get the phase in the right neighborhood. Then it's up to the right combination of components and component values to get them as close as possible while still keeping goals 1 and 2 above also in the acceptable category. But things vary. Sometimes reversing the polarity of 1 of the drivers is needed too. Sometimes the same electrical orders will work too. I've got two 2nd order filters with the same polarity in the example attached, so it's hard to know what's needed. It's mostly a matter of trial and error to see what works, at least for me anyways, but it doesn't usually take too much effort to quickly see what's going to work and what isn't.
Attachments
Alright here are the xo points I have thought of and the reasons behind them.
Lemme know whatcha think! 😀
Fyi, all the screenshots are in order of my list. Also, did my best to help you out and color code the responses like you wanted for XSim.
First off I will include the Fs of the tweeters and mids based off the impedance measurements. As well as the frequency at which the 45deg response falls below 3dB from the fundamental.
Fs:
- Tweeters: ~800Hz
-Mids: ~150Hz
45deg fall off point:
- Rear A&B Woofer: ~3kHz
- Tower A&B Woofers: ~3.5kHz
- Tower A&B Mids: ~4.5kHz
- CC Woofer: ~3.5kHz
- CC Mids: ~1.5kHz
Tower A: Crossover points at -> 600Hz and 2.5kHz
I have picked 600Hz because:
- It is more than 3 x Fs of the mids
- It is way below the the 45deg fall of point of the mids and the woofers
- At 500Hz the woofer is at 88.5dB and at 700Hz the mids are at 88dB. So a xo point at 600Hz seems like a good point that wont be too drastic.
I have picked 2.5kHz because:
- It is more than 3 x Fs of the tweeter
- It is below the 45deg fall off point of 4.5kHz for the mids
- Looking at the mids at the point, it looks like a natural point at which to cut off the mids FR.
Tower B: Crossover points at -> 600Hz and 2.5kHz
The reasoning is the same for Tower A. Plus, the two towers should have the same xo points.
Rear A: Crossover point at -> 3kHz
I have picked 3kHz because:
- It is 3.75 x Fs of the tweeter
- It is right at the 45deg fall off point of the woofer. So, I know you said keep if below. But dont know if keeping it at the point would matter too much more than say if the xo point were to be at 2.5kHz. I'm good with a xo point at 2.5kHz as well.
- A xo point at 3kHz would naturally follow the slope of the woofer at 3k. Since the filter would actually start filtering at about ~2.5kHz. It would be a smooth roll-off, not the biggest point but something worth saying I guess.
Rear B: Crossover point at -> 3kHz
Same reasoning as for Rear A. Again, I'm good with either 3kHz or 2.5kHz. Whatever you think is best Im good for haha
Center Channel: Crossover points at -> 500Hz and 2.5kHz
I have picked 500Hz because:
- It is more than 3 x Fs of the mids
- It is below the 45deg fall off point of the mids and woofers
- The woofers at 500Hz start to dip about -7dB and then it goes right back up. Basically a roller coaster for the woofers after 500Hz. But, the mids at 500Hz up to 1kHz are pretty flat. So, it looks like a really good point to crossover those two drivers.
I have picked 2.5kHz because:
- It is more than 3 x Fs of the tweeter
- Again the mids at 2.5kHz start to slope down and it seems like a natural point to filter it down.
Some additional thoughts I have on the CC upper xo point:
- The xo point of 2.5kHz is far above the 45deg fall off point of the mids. B ut theres no way the xo point would be at 1.5kHz for the mids to tweeter. Thats just too low in my opinion.
- A xo point at ~4kHz also wouldnt seem like a terrible idea. Since the tweeter seems to "behave" better after 4kHz (less wobbly and rollercoaster like). Thought it would be compromise since the mids have that dip from 2.2kHz - 3.2kHz.
Tomorrow I will also be pointing out the bad things and peaks or such that could be fixed.
Anyways, I hope I did alright! 🙂
I am pretty comfortable with crossover components and much more with filters. But when it comes to crossovers, it seems like there are so many different approaches people take. So, I can try it myself and then bounce it off you, or if you want to give a rundown of what you do and post it. Im good with either, up to you!
Lemme know whatcha think! 😀
Fyi, all the screenshots are in order of my list. Also, did my best to help you out and color code the responses like you wanted for XSim.
First off I will include the Fs of the tweeters and mids based off the impedance measurements. As well as the frequency at which the 45deg response falls below 3dB from the fundamental.
Fs:
- Tweeters: ~800Hz
-Mids: ~150Hz
45deg fall off point:
- Rear A&B Woofer: ~3kHz
- Tower A&B Woofers: ~3.5kHz
- Tower A&B Mids: ~4.5kHz
- CC Woofer: ~3.5kHz
- CC Mids: ~1.5kHz
Tower A: Crossover points at -> 600Hz and 2.5kHz
I have picked 600Hz because:
- It is more than 3 x Fs of the mids
- It is way below the the 45deg fall of point of the mids and the woofers
- At 500Hz the woofer is at 88.5dB and at 700Hz the mids are at 88dB. So a xo point at 600Hz seems like a good point that wont be too drastic.
I have picked 2.5kHz because:
- It is more than 3 x Fs of the tweeter
- It is below the 45deg fall off point of 4.5kHz for the mids
- Looking at the mids at the point, it looks like a natural point at which to cut off the mids FR.
Tower B: Crossover points at -> 600Hz and 2.5kHz
The reasoning is the same for Tower A. Plus, the two towers should have the same xo points.
Rear A: Crossover point at -> 3kHz
I have picked 3kHz because:
- It is 3.75 x Fs of the tweeter
- It is right at the 45deg fall off point of the woofer. So, I know you said keep if below. But dont know if keeping it at the point would matter too much more than say if the xo point were to be at 2.5kHz. I'm good with a xo point at 2.5kHz as well.
- A xo point at 3kHz would naturally follow the slope of the woofer at 3k. Since the filter would actually start filtering at about ~2.5kHz. It would be a smooth roll-off, not the biggest point but something worth saying I guess.
Rear B: Crossover point at -> 3kHz
Same reasoning as for Rear A. Again, I'm good with either 3kHz or 2.5kHz. Whatever you think is best Im good for haha
Center Channel: Crossover points at -> 500Hz and 2.5kHz
I have picked 500Hz because:
- It is more than 3 x Fs of the mids
- It is below the 45deg fall off point of the mids and woofers
- The woofers at 500Hz start to dip about -7dB and then it goes right back up. Basically a roller coaster for the woofers after 500Hz. But, the mids at 500Hz up to 1kHz are pretty flat. So, it looks like a really good point to crossover those two drivers.
I have picked 2.5kHz because:
- It is more than 3 x Fs of the tweeter
- Again the mids at 2.5kHz start to slope down and it seems like a natural point to filter it down.
Some additional thoughts I have on the CC upper xo point:
- The xo point of 2.5kHz is far above the 45deg fall off point of the mids. B ut theres no way the xo point would be at 1.5kHz for the mids to tweeter. Thats just too low in my opinion.
- A xo point at ~4kHz also wouldnt seem like a terrible idea. Since the tweeter seems to "behave" better after 4kHz (less wobbly and rollercoaster like). Thought it would be compromise since the mids have that dip from 2.2kHz - 3.2kHz.
Tomorrow I will also be pointing out the bad things and peaks or such that could be fixed.
Anyways, I hope I did alright! 🙂
I am pretty comfortable with crossover components and much more with filters. But when it comes to crossovers, it seems like there are so many different approaches people take. So, I can try it myself and then bounce it off you, or if you want to give a rundown of what you do and post it. Im good with either, up to you!
Attachments
Yes you need to set hard targets in order to create the target xo curves, but I think I need to reiterate that these targets are not set in stone. It's actually better to think in terms of a range of xo frequencies than a fixed point and it's really important to remain flexible. Frequently it is going to be the drivers themselves which dictate where the xo point is going to be. And sometimes it only takes about 5 minutes in XSim to realize that despite your best efforts, the xo point is not going to want to go exactly where you think it should.
In terms of the xo point's relationship with Fs, I'm also going to reiterate that's it's better to think in terms of dB down instead of just 2 or 3 times Fs. Slope matters in other words and where you put the xo point changes things too because the filter roll-off combines with the driver's natural acoustic roll-off differently depending on where that xo point is. And there is always a resonance compensation filter of 1 kind or another that can help to bring Fs down where it needs to be if necessary.
Plus you are mostly only concerned with Fs attenuation levels for tweeters. A midrange driver can actually be run as a little woofer which means that there is no real problem running it down to and past its Fs, either in a sealed or a ported box. So that factor becomes less important for a mid in a 3-way.
I didn't actually want to even get to the 3-way xo's until you had the 2-way successfully completed because of the inherent complexity of the 3-way. So for example, for the mid to woofer xo range in a 3-way, I'll also be looking closer at the following:
- where the harmonic distortion starts increasing in both the mid and the woofer;
- where the natural roll-off of the mid starts and whether or not that roll-off might get a little boost from the xo filter;
- whether or not the woofer LP filter is going to create unwanted peaking down near the woofer Fs (both of these last 2 things you can only determine by actually working on the drivers in the xo program btw);
- the cone excursion of each driver at lower frequencies and how much intermodulation distortion the woofers are likely to suffer from in the higher frequencies.
Probably the most important of those in your case is where the mid naturally rolls off, so 500-600Hz is probably about right. What I might attempt is to take it about as low as it can go while still maintaining a flattish summed response. For a more coherent front soundstage, I think I would probably try to use the same xo points for the L&R's as well as the CC too.
So my suggestion really is to just start by working on the rear xo. Put some components together and see what the drivers are telling you and what you can come up with. And then we'll proceed from there.
In terms of the xo point's relationship with Fs, I'm also going to reiterate that's it's better to think in terms of dB down instead of just 2 or 3 times Fs. Slope matters in other words and where you put the xo point changes things too because the filter roll-off combines with the driver's natural acoustic roll-off differently depending on where that xo point is. And there is always a resonance compensation filter of 1 kind or another that can help to bring Fs down where it needs to be if necessary.
Plus you are mostly only concerned with Fs attenuation levels for tweeters. A midrange driver can actually be run as a little woofer which means that there is no real problem running it down to and past its Fs, either in a sealed or a ported box. So that factor becomes less important for a mid in a 3-way.
I didn't actually want to even get to the 3-way xo's until you had the 2-way successfully completed because of the inherent complexity of the 3-way. So for example, for the mid to woofer xo range in a 3-way, I'll also be looking closer at the following:
- where the harmonic distortion starts increasing in both the mid and the woofer;
- where the natural roll-off of the mid starts and whether or not that roll-off might get a little boost from the xo filter;
- whether or not the woofer LP filter is going to create unwanted peaking down near the woofer Fs (both of these last 2 things you can only determine by actually working on the drivers in the xo program btw);
- the cone excursion of each driver at lower frequencies and how much intermodulation distortion the woofers are likely to suffer from in the higher frequencies.
Probably the most important of those in your case is where the mid naturally rolls off, so 500-600Hz is probably about right. What I might attempt is to take it about as low as it can go while still maintaining a flattish summed response. For a more coherent front soundstage, I think I would probably try to use the same xo points for the L&R's as well as the CC too.
So my suggestion really is to just start by working on the rear xo. Put some components together and see what the drivers are telling you and what you can come up with. And then we'll proceed from there.
I want to add just one more thing in terms of where to cross that I was perhaps remiss in ignoring for your 2-way. And that is that it may in fact be helpful for you to look at the harmonic distortion of the drivers.
So the following is just an example but I'm going to use older versions of the drivers that you are using so the comparison might be fairly close.
Pic 1 below is the harmonic distortion for the older RS150S-8 as measured by Zaph. Just in case you need a refresher, the 2nd harmonic is an octave above the fundamental (let's say it's the low E on a guitar) so the 2nd is the next E up the fretboard, the 3rd harmonic is I believe seven notes up (fundamental x 3) so now that's the following B, the 4th is another octave up from the fundamental so another E, the 5th is another 7 notes up and so on and so forth. So from that you can infer that the 3rd harmonics in particular as well as 5th (usually lower in SPL though) are the least desirable since they are actually different notes whereas the 2nd and the 4th are actually at least still the same note as the fundamental but just higher in tone, although still unwanted.
When taken all together, I tend to think of the harmonic distortion of a driver as producing a chord instead of just the single pure note that was recorded. Pretty close to an E bar chord up at the 7th fret with the bottom E open in my example. An oversimplification perhaps but you get the idea.
So again looking at the 1st pic, we can see that the 2nd harmonic is actually fairly low and fairly flat which is pretty good, but that both the 3rd and 5th distortion products start rising from about 300Hz. Now pic 2 is for the old RS28A-4. Here the 3rd harmonic starts rising with the lower frequencies this time starting at about 4kHz.
So now let's compare them keeping in mind where to cross between the 2. At about 2kHz, the 3rd harmonic for the woofer is down at about -55dB while for the tweeter it's down at about -65dB. So a 10dB difference which equates to about twice as loud. It's not until about 1kHz or so that the 3rd harmonic for both drivers is down at about the same level, about -60dB. So essentially the tweeter is the cleaner of the 2 drivers above about 1kHz in terms of distortion.
That is of course not the only factor we have to consider but in a 2-way with these drivers, I would probably try to get as much out of the tweeter as possible without overly stressing it out, or using too many components either. That would mean pushing the lower limit of the xo frequency target range down closer towards 2kHz instead of 3kHz, especially when that means better off-axis performance from the woofer too.
These are not your drivers of course, but it may in fact be helpful for you to go back into your rear speaker measurements in REW and take a look at the harmonic distortion as well.
Hope that helps.
So the following is just an example but I'm going to use older versions of the drivers that you are using so the comparison might be fairly close.
Pic 1 below is the harmonic distortion for the older RS150S-8 as measured by Zaph. Just in case you need a refresher, the 2nd harmonic is an octave above the fundamental (let's say it's the low E on a guitar) so the 2nd is the next E up the fretboard, the 3rd harmonic is I believe seven notes up (fundamental x 3) so now that's the following B, the 4th is another octave up from the fundamental so another E, the 5th is another 7 notes up and so on and so forth. So from that you can infer that the 3rd harmonics in particular as well as 5th (usually lower in SPL though) are the least desirable since they are actually different notes whereas the 2nd and the 4th are actually at least still the same note as the fundamental but just higher in tone, although still unwanted.
When taken all together, I tend to think of the harmonic distortion of a driver as producing a chord instead of just the single pure note that was recorded. Pretty close to an E bar chord up at the 7th fret with the bottom E open in my example. An oversimplification perhaps but you get the idea.
So again looking at the 1st pic, we can see that the 2nd harmonic is actually fairly low and fairly flat which is pretty good, but that both the 3rd and 5th distortion products start rising from about 300Hz. Now pic 2 is for the old RS28A-4. Here the 3rd harmonic starts rising with the lower frequencies this time starting at about 4kHz.
So now let's compare them keeping in mind where to cross between the 2. At about 2kHz, the 3rd harmonic for the woofer is down at about -55dB while for the tweeter it's down at about -65dB. So a 10dB difference which equates to about twice as loud. It's not until about 1kHz or so that the 3rd harmonic for both drivers is down at about the same level, about -60dB. So essentially the tweeter is the cleaner of the 2 drivers above about 1kHz in terms of distortion.
That is of course not the only factor we have to consider but in a 2-way with these drivers, I would probably try to get as much out of the tweeter as possible without overly stressing it out, or using too many components either. That would mean pushing the lower limit of the xo frequency target range down closer towards 2kHz instead of 3kHz, especially when that means better off-axis performance from the woofer too.
These are not your drivers of course, but it may in fact be helpful for you to go back into your rear speaker measurements in REW and take a look at the harmonic distortion as well.
Hope that helps.
Attachments
Last edited: