Hi, if you test the baffle in a diffraction simulator you'll notice any flat baffle around the drivers makes diffraction effects appear, the more baffle area the more ripple. This relates to wavelengths, anything between wavelengths of the driver beaming and baffle support will diffract and cause interference. Basically if one wants to avoid diffraction roundovers must start immediately next to the transducer(s) but any roundover is better than none.
How much this is a concern it is up to you of course. I wanted to demonstrate this so that you are aware. This is one aspect to better cheap commercial products, have less diffraction related interference.
Attached are diffraction simulations of ideal 25mm tweeter on baffle from your PDF. Attachments are roundovers of 0, 18mm, 36mm and 72mm. As seen on the attachments when the roundover radius approaches the width of flat baffle around the drivers the diffraction ripple starts to disappear. Also, difference of 0 and 18mm radius roundovers is not too much because the baffle is so wide (most of the ripple is at wavelengths greater than 18mm).
Attachments
Last edited:
I did check that also in Virtuixcad. The perfect speaker would have a sphere, but practicality an so is also into play. And i think the main thing is have no sharp corners. and the positioning of the speakers won't be perfect neighter, so the wall behind it will also cause a bit ripple and so. To do that i would have to make the front very thick which will drive up the costs to much and blow my budget with the actual wood prices.
But all speakers are colouring that is true. And it's good you show that (altough i was aware of that). It's a matter of compromises (all speakers are compromised, even the best ones) and i think my design is less compromised than what i can buy in shops for my budget.
And @ hifijim: i did check a 4th order crossover (first run) and it should work without getting too expensive on parts. I may do that... but only after i measured the speaker. But this test sim with published graphs shows that it's a good id.
But all speakers are colouring that is true. And it's good you show that (altough i was aware of that). It's a matter of compromises (all speakers are compromised, even the best ones) and i think my design is less compromised than what i can buy in shops for my budget.
And @ hifijim: i did check a 4th order crossover (first run) and it should work without getting too expensive on parts. I may do that... but only after i measured the speaker. But this test sim with published graphs shows that it's a good id.
Attachments
Last edited:
Alright, It wasn't immediately obvious you had done diffraction sims. I'm very much into thinking through compromises and you seem to think so too, hence interest in this 🙂 I hope you don't mind me posting another experiment.
I have never done speaker like this so I thought it it was nice opportunity to checkout what is the rationale behind the crossover you've chosen.
Here are again ideal drivers on the baffle in your PDF plan. First two attachments are ideal 3rd order butterworth and 4th order LR filters at 1800Hz. It seems this xo point is a bit too high directivity wise for the woofer and here the BW3 seems to be clearly better in terms of overall smoothness of both listening window and power response than 4th order LR filter. I tried to experiment lower xo points and below 1000Hz, where both drivers radiate omni, there is not much difference between the variants. I think the BW3 was best I could do around the 1.8kHz mark of all the filter variants available.
Next few attachments show we can better the LR filter response by adjusting either xo point to have some overlap. But, I don't know how all this would affect actual sound especially near field. Alsow the BW3 response could be bettered by slightly varying the other filter frequency.
I'm not sure what information there is to harvest from this other than you really should build and measure the whole thing to get the best possible xo circuit simulated for the speaker. I didn't check what your analog xo actually does, it might be well optimized but I don't see any of the off axis data so I suspect it isn't optimal yet unless very lucky. Based on the experiment I would have chosen smaller woofer instead to allow the high crossover point and have nice off-axis response and made a three way to have the bass as well but again like you said it would be another set of compromises and I'm mostly playing with the graphs in simulator not with real sound here so anything goes. BW3 filter at 1800Hz is not too bad although a bit narrow vertical lobe at the xo.
ps. I've done fair amount of recording and mixing as hobby and the worst problem with monitors/monitoring situation has been the sound is not stable, eg. the sound changes when the head moves and this I have found really frustrating. Trying to EQ something (at high freq) and the sound depends on the position of the head 😀 Having KH120 monitors for nearfield work and would be interested on something more stable but now after thinking your project I should try better positioning first, especially the vertical listening axis changes a lot if the monitors are not carefully positioned to minimize variation on the vertical axis with head movement.
I have never done speaker like this so I thought it it was nice opportunity to checkout what is the rationale behind the crossover you've chosen.
Here are again ideal drivers on the baffle in your PDF plan. First two attachments are ideal 3rd order butterworth and 4th order LR filters at 1800Hz. It seems this xo point is a bit too high directivity wise for the woofer and here the BW3 seems to be clearly better in terms of overall smoothness of both listening window and power response than 4th order LR filter. I tried to experiment lower xo points and below 1000Hz, where both drivers radiate omni, there is not much difference between the variants. I think the BW3 was best I could do around the 1.8kHz mark of all the filter variants available.
Next few attachments show we can better the LR filter response by adjusting either xo point to have some overlap. But, I don't know how all this would affect actual sound especially near field. Alsow the BW3 response could be bettered by slightly varying the other filter frequency.
I'm not sure what information there is to harvest from this other than you really should build and measure the whole thing to get the best possible xo circuit simulated for the speaker. I didn't check what your analog xo actually does, it might be well optimized but I don't see any of the off axis data so I suspect it isn't optimal yet unless very lucky. Based on the experiment I would have chosen smaller woofer instead to allow the high crossover point and have nice off-axis response and made a three way to have the bass as well but again like you said it would be another set of compromises and I'm mostly playing with the graphs in simulator not with real sound here so anything goes. BW3 filter at 1800Hz is not too bad although a bit narrow vertical lobe at the xo.
ps. I've done fair amount of recording and mixing as hobby and the worst problem with monitors/monitoring situation has been the sound is not stable, eg. the sound changes when the head moves and this I have found really frustrating. Trying to EQ something (at high freq) and the sound depends on the position of the head 😀 Having KH120 monitors for nearfield work and would be interested on something more stable but now after thinking your project I should try better positioning first, especially the vertical listening axis changes a lot if the monitors are not carefully positioned to minimize variation on the vertical axis with head movement.
Attachments
Last edited:
I see what you try to do, but i don't find how i can define the offset of the woofer in that program (and you did not include it). In Xsim, which i use for years now with good result, it can be added in the section of the woofer (but must be calculated manually which is not an issue for me), but it seems virtuixcad does not allow that.
I do know that the crossoverpoint could be a bit lower with higher order crossovers than the 1800Hz but also that directivity only really becomes an issue above 2kHz with this woofer. Below 2kHz off axis the loss is less than 3dB which should be not hearable. I know it's a relative big midwoofer for such frequency, but the phase plug it has does work in reducing beaming and push the usable frequency range up.
I do know that the crossoverpoint could be a bit lower with higher order crossovers than the 1800Hz but also that directivity only really becomes an issue above 2kHz with this woofer. Below 2kHz off axis the loss is less than 3dB which should be not hearable. I know it's a relative big midwoofer for such frequency, but the phase plug it has does work in reducing beaming and push the usable frequency range up.
Last edited:
Do you mean Z-offset? That is included in the measurements if you do the measurements like they should be done and then the Z offset of each driver is left to zero in the program. The Y and X coordinates are set against rotation and reference axis which is usually the tweeter axis. The sims I posted are ideal drivers, flat disks on the baffle, I set -166mm for the woofer Y coordinate but everything else is 0 since the diffraction sim results were "measured" on axis of each of the drivers like they should be done in real world as well (both drivers are on same rotation axis, X is the same). I chose to use the tweeter axis as reference axis, which means the tweeter Y coordinate is left to zero and the woofer is below this. The power of all this is that you can freely select the reference axis you want to design against! Now I'm able to select what horizontal and vertical angle the graphs show.
You could measure both drivers on the same mic position, for example at the tweeter axis, and then you would leave the Y coordinate of the woofer 0 as well because the Y offset is already in the measured response. The coordinates map the measured responses to the "reality". You can bend the rules to your liking and where you need the resolution most in comparison to time spent doing the measurements. It should all work fine as long as the coordinate system is understood and used to reflect the measurement setup. It is all in the manual and not difficult at all after the first a-ha moment 🙂 If you lose the timing on the measurements, then you need to specify all in hand but this is very tedious and prone to errors. Recommended way is to use the two channel measurement method that preserves time information.
All the coordinates are found/set in the crossover window, by clicking on the respective driver symbol. Hope it helps 🙂
ps. yeah if you get good results with on axis measurements only you can do it. But especially designing for far field listening the on axis response is almost meaningless in comparison to power response for example so I would not recommend it to anyone nowadays when we have free software and cheap measurement gear capable of showing much much more closer approximation of the actual response.
You could measure both drivers on the same mic position, for example at the tweeter axis, and then you would leave the Y coordinate of the woofer 0 as well because the Y offset is already in the measured response. The coordinates map the measured responses to the "reality". You can bend the rules to your liking and where you need the resolution most in comparison to time spent doing the measurements. It should all work fine as long as the coordinate system is understood and used to reflect the measurement setup. It is all in the manual and not difficult at all after the first a-ha moment 🙂 If you lose the timing on the measurements, then you need to specify all in hand but this is very tedious and prone to errors. Recommended way is to use the two channel measurement method that preserves time information.
All the coordinates are found/set in the crossover window, by clicking on the respective driver symbol. Hope it helps 🙂
ps. yeah if you get good results with on axis measurements only you can do it. But especially designing for far field listening the on axis response is almost meaningless in comparison to power response for example so I would not recommend it to anyone nowadays when we have free software and cheap measurement gear capable of showing much much more closer approximation of the actual response.
Last edited:
The thing is, i don't have the cabinets yet, so i can't measure. But i will when i have them and watch this in virtuixcad next to xsim
A 3th order is also feasable. So i have a lot of choices. It's only harder to match phase in 3th order (but doable).
But like said, it's prestudies we do on crossovers now, and you give me usefull tips on to look at what to get good results, so thanks. Let's first build that cabinet and mount the drivers in so i can measure them like it should.
A 3th order is also feasable. So i have a lot of choices. It's only harder to match phase in 3th order (but doable).
But like said, it's prestudies we do on crossovers now, and you give me usefull tips on to look at what to get good results, so thanks. Let's first build that cabinet and mount the drivers in so i can measure them like it should.
Attachments
Yeah my sims don't include impedance or the real world acoustic response so it is not too accurate and something else might work mucho better. It was interesting to see the BW3 filters were bit better looking than LR4 in this case though. Anyway, it is pretty easy to dial in right components with good measurement data in the VCAD. Best wishes for your project!
I've read that, but i can't use the augpro waveguide because i don't have the tools to make it nor the skills. And i can't cut a whole like that with a router. If you know a good round waveguide that is not to big to srew up the alignment for this tweeter, and that i can buy i may consider it. But i only found that eliptical one for the SB26 tweeter.
And sloping the baffle is also not fit for this project, it's limited in budget and wood is expensive right now. I can't afford to use 2 or 3 layers of plywood for that.
And sloping the baffle is also not fit for this project, it's limited in budget and wood is expensive right now. I can't afford to use 2 or 3 layers of plywood for that.
Waxx, maybe you already know this one and discarded it (or it doesn't accept your tweeter of choice ) but have you looked at visaton wg148 waveguide?
Limited in budget and yet surpassing the cheaper than ready made offerings? To be honest, I have my doubts. That is only hardware wise.
Next thing is: are you capable of making 0-180 degree dual channel measurements to feed VituixCad? If not, try to master that first of all, before embarking in any new project.
The Heissman projects offer the best of worlds for the non -super experts.
Next thing is: are you capable of making 0-180 degree dual channel measurements to feed VituixCad? If not, try to master that first of all, before embarking in any new project.
The Heissman projects offer the best of worlds for the non -super experts.
Sorry us being pessimistic but it looks like you can't get full potential of the drivers with the current plan. Even though the drivers are good them selves it is the complete system that defines the quality at the end. Getting good system is hard work, hence notion it is difficult to match commercial offerings with the cost.
The tweeter has such a big face plate on it the roundovers would have to be r > 3cm even with no baffle around the tweeter faceplate. You can of course make the thing without without big roundover/slants but the diffraction kind of prevents the performance happening. In a no roundover/slant manufacturing situation one would have to use waveguide or run the tweeter freestanding (a tweeter without faceplate).
Anyway, if you are already set to build it as is then by no means go ahead 🙂 It is possible that the sound is fine enough for the purpose and if you have an itch to try better it some day you could make another box with the roundovers / slants.
Hopefully I didn't kill all the fun! What ever you do you should have fun doing it I think.
The tweeter has such a big face plate on it the roundovers would have to be r > 3cm even with no baffle around the tweeter faceplate. You can of course make the thing without without big roundover/slants but the diffraction kind of prevents the performance happening. In a no roundover/slant manufacturing situation one would have to use waveguide or run the tweeter freestanding (a tweeter without faceplate).
Anyway, if you are already set to build it as is then by no means go ahead 🙂 It is possible that the sound is fine enough for the purpose and if you have an itch to try better it some day you could make another box with the roundovers / slants.
Hopefully I didn't kill all the fun! What ever you do you should have fun doing it I think.
Nobody killed the fun until now i think and comments are critical but no dismissing in my view, what Waxx asked for.
I keep on thinking the initial Waxx idea is not bad at all.
Yes without waveguide and (big) roundovers there will be issue but i've made enough mix on less than stellar monitors of this same kind i wouldn't loose sleep over the initial plan if the box is better than what is offered by entry level commercial offer.
Waxx being a sound engineer i'm sure he have the skills and gear to perform enough meaningfull measurements and how to interpret them too.
After all, albums mixed with ns10 are plentyful... and if we had to review the design they wouldn't go past the initial idea ( not even the drawing board i think!). 🙂
I keep on thinking the initial Waxx idea is not bad at all.
Yes without waveguide and (big) roundovers there will be issue but i've made enough mix on less than stellar monitors of this same kind i wouldn't loose sleep over the initial plan if the box is better than what is offered by entry level commercial offer.
Waxx being a sound engineer i'm sure he have the skills and gear to perform enough meaningfull measurements and how to interpret them too.
After all, albums mixed with ns10 are plentyful... and if we had to review the design they wouldn't go past the initial idea ( not even the drawing board i think!). 🙂
Well, it seems that some here only want perfect monitors and can't see passed the goal of the project. I called it a "monitor like speakers" for a reason. I see how crappy cheap "monitors" are made, and how much erros they contain. I don't try to do better than the Genelecs or ATC's of the monitor world, but than the KRK or the Dynaudio's or the Yamaha's. They don't have roundovers or waveguides, use very cheap drivers and crossovers and are build of thin mdf or chipboard. And my budget is also limited to what they cost.
I am open for critics, like on the use of a waveguide and on the crossover. But saying it's impossible with these drivers or a diy is ********. Many kits are out there (i could make one of those) that are also not that refined and do better. But i more like to design myself and learn from comments here. So...
I am open for critics, like on the use of a waveguide and on the crossover. But saying it's impossible with these drivers or a diy is ********. Many kits are out there (i could make one of those) that are also not that refined and do better. But i more like to design myself and learn from comments here. So...
Last edited:
Keep going, Waxx. Your design has a lot of potential. All designs are flawed, the goal is to minimize the flaws to the point that we can live with it.
I made a two-way passive design last year for my basement workshop. I was not expecting high performance, and I re-used a 25 year old tweeter. The result was actually quite good, and I don't think I could match it for less than $800 retail.
Request Peer review of crossover
I made a two-way passive design last year for my basement workshop. I was not expecting high performance, and I re-used a 25 year old tweeter. The result was actually quite good, and I don't think I could match it for less than $800 retail.
Request Peer review of crossover
Waxx,
Isn't a 'Fablab' available around you?
They most often have 3d printers now. And given you are open to share your work this could be compatible with the spirit of those labs.
Even if they can't print a whole WG like Augerpro's one, an adapter plate ( as suggested by Markbakk's post 28) could be easily printed and adapted to round WG like the Visaton or WG300.
It could even be done with aluminium within a local mechanical lab ( it may be cheap even). Even some technical high school could do it, like some dedicated to mechanic ( i worked in one as supervisor during my study and they happily did some parts for microphones i worked on at the time).
Isn't a 'Fablab' available around you?
They most often have 3d printers now. And given you are open to share your work this could be compatible with the spirit of those labs.
Even if they can't print a whole WG like Augerpro's one, an adapter plate ( as suggested by Markbakk's post 28) could be easily printed and adapted to round WG like the Visaton or WG300.
It could even be done with aluminium within a local mechanical lab ( it may be cheap even). Even some technical high school could do it, like some dedicated to mechanic ( i worked in one as supervisor during my study and they happily did some parts for microphones i worked on at the time).
Last edited:
There is non near, but some are on the other side of the province (50min drive). But it's not only the Augerpro print that is an issue, also the routing of the hole in the plate. I don't have access to CNC nor can make that kind of cuts with a router myself. That's why i'm looking for a round waveguide. The making of it i can maybe do at certain places, but i need a design that works in my case. I'm still looking arround for solutions for that.
With respect, tmuikku, I must disagree with your statement that the on-axis response is almost meaningless. I would argue that above 1 kHz, the first arrival sounds at your ears, i.e. the on-axis, is the most important. Floyd Toole discusses this in section 18.2 of his book. As Toole points out, if sound power was the only thing that really mattered, you could rotate your speakers 90 degrees or 180 degrees and it would not matter, they would sound the same... clearly they do not.
A good goal for a speaker is a smooth flat frequency response on-axis, and a smooth gently decreasing frequency response in sound power. But if I have to make a compromise, I will sacrifice some smoothness in the sound power curve.
Waxx - I am not convinced you need a waveguide. A cone/dome speaker will often have a -3 dB drop in power response near the crossover. Especially if 2nd or 4th order crossovers are used. If a 3rd order crossover is used, the natural 90 degree phase lag between mid and tweeter can compensate for this and you can get a smooth on-axis and sound power response. But this will depend greatly on the location and size of the diffraction hump. In any case, a small -3 dB drop in power response over an octave is not very significant. A peak in sound power would be much more concerning.
The closer you sit to your speakers, the more important the on-axis sound becomes. The more absorbant your room is, the more important the on-axis sound becomes.
Yes it makes sense about the need for a round waveguide. There is solution to route non regular shape but i'll save you for a description in english by myself... 😀
For an adapting plate ( for a round wg) no need for a cnc, a lathe and a vertical drill should be enough (une perceuse colonne- don't know the exact english term for this tool).
If you choose to try it you could even thread them yourself if on aluminium (if you are not used to do this practice a bit before attempting on the final part though).
Anyway any mechanic workshop have the tools to do this for what i've seen.
For an adapting plate ( for a round wg) no need for a cnc, a lathe and a vertical drill should be enough (une perceuse colonne- don't know the exact english term for this tool).
If you choose to try it you could even thread them yourself if on aluminium (if you are not used to do this practice a bit before attempting on the final part though).
Anyway any mechanic workshop have the tools to do this for what i've seen.
Last edited: