Healthy critique is always welcome 🙂 I must expand on this a bit, I should have chosen the wording more carefully. I was thinking in the context of designing a crossover, actually the whole system. I have noticed there is million ways one can make the on-axis response flat, or any single axis for that matter. The difference of various crossover filters and that of the physical design is mainly everywhere else than on any particular single axis, which is just one data point of many that make the whole output. If the rest of the 3D balloon is ignored deliberately it feels silly to me and in this context a single axis response is meaningless, it is the whole picture that matters eventually.
Any designer should do themselves a favor to measure and simulate the whole acoustic response of the whole system because that is what they are designing. It will show any issues the system contains and crossover is just one opportunity to affect the end result while the physical construct operating in acoustic domain is responsible for the rest, the most. While playing with few designs and the data in VituixCAD as well as some simulated scenarios I've noticed the crossovers are really trivial to optimize for any particular physical construct (measurements) and actually what defines the quality must be in the physical construct, which affects the measurements! Here is the part where resonances and diffraction and other construction related issues raise their head and makes the engineer ponder the compromises taken and are eventually the only evil left on the way to better sound.
From this perspective, head on my own butt, it is sometimes hard to see what other people try to achieve, so sorry about that 😀
For far field listening I put my coins on the power response over on-axis but actually the DI is most important since it pretty much tells how smooth either can be in relation to the other. CTA-2034 standard puts number 12% on the listening window response and 88% for the off-axis (power and early reflections both 44%) to approximate sound at listening position in an average living room listening situation and I'm going with that until know better. Also inline to what Toole says 😉 Virtually any non-coincident speaker, whose measurements I've seen on the web, have a dip on the power at xo so it seems to be no issue. Exception is the "kimmosto c-c rule" stuff but that is another topic altogether and not very relevant for the nearfield monitor in this topic. I haven't yet made or read about a listening test for the power dip in xo vs. no dip to make an opinion if there is difference.
Last edited:
Vous pouvez le faire en Français aussi pour moi (j'habite en Wallonie, le part Francophone de Belgique). But i don't know if the forum likes that french...
I was looking for this if it's possible, and a "perceuse colone" is a drillpress in English 😉. But i don't have that in my workshop neighter. I can do that at an other place still.
Sorry to keep hijacking your thread, Waxx, with off topic discussion...
Well it is true that the CTA-2034 standard does say that. The standard is intended to represent an "average room", to give speaker designers a standardized target which will satisfy the maximum number of consumers. It is based loosely on Toole and Olive's work, but my reading of the Toole/Olive material leads me to believe that the listening window is quite a bit more important than 12%. Toole never assigns a specific weighting, probably because there was a lot of scatter in his test data. Peoples preferences varied quite a bit, and the correlations were loose. No surprise there... it's a little like asking people which Beatles song is their favorite.
The sound power response is a good representation of the percieved timbre of a speaker in an "average room". All rooms are different, however, and my room must be quite different. When I measured my Room Curve by averaging 10 different responses, the net average did not look like the sound power curve or the PIR curver... it very closely resembled the listening window response. I encourage everyone to make an averaged room curve measurement of their speakers in their room. Once you know that, you don't need 36 measurements to feed into a software to tell you what the PIR is... you already know what the real room response is...
Yes I have also noticed that power-dip in many speakers, and I followed Kimmo's design strategy discussions with great interest.
A couple of months ago, I had a discussion with Fluid about two different DSP crossovers I was evaluating. Both had a similarly flat on-axis response. One had the common -3 dB dip in power response (+3 dB rise in DI) and the other had a much flatter power response with no dip (a smooth DI curve). I was pondering why these two responses sounded so similar to me. Most of the time I could not tell them apart. Fluid pointed out a study by Lipshitz and Vanderkooy. They found that power response dips can be quite severe without them necessarily being audible, whereas power response peaks were more easilly detected. Toole makes a similar comment in his book.
Here is a link to Fluid's post (is this guy amazing?)
https://www.diyaudio.com/forums/multi-way/366347-active-satori-textreme-17.html#post6785249
This makes sense to me, and like any good scientist, I assign more weight to evidence which agrees with my theory, and tend to dismiss evidence which goes against my theory 😉 . I say that jokingly, but it is something we all have to be aware of and guard against.
This is a great discussion, and it almost deserves its own thread. I always learn a lot from these kinds of discussions. Thanks !
J.
Well it is true that the CTA-2034 standard does say that. The standard is intended to represent an "average room", to give speaker designers a standardized target which will satisfy the maximum number of consumers. It is based loosely on Toole and Olive's work, but my reading of the Toole/Olive material leads me to believe that the listening window is quite a bit more important than 12%. Toole never assigns a specific weighting, probably because there was a lot of scatter in his test data. Peoples preferences varied quite a bit, and the correlations were loose. No surprise there... it's a little like asking people which Beatles song is their favorite.
The sound power response is a good representation of the percieved timbre of a speaker in an "average room". All rooms are different, however, and my room must be quite different. When I measured my Room Curve by averaging 10 different responses, the net average did not look like the sound power curve or the PIR curver... it very closely resembled the listening window response. I encourage everyone to make an averaged room curve measurement of their speakers in their room. Once you know that, you don't need 36 measurements to feed into a software to tell you what the PIR is... you already know what the real room response is...
Yes I have also noticed that power-dip in many speakers, and I followed Kimmo's design strategy discussions with great interest.
A couple of months ago, I had a discussion with Fluid about two different DSP crossovers I was evaluating. Both had a similarly flat on-axis response. One had the common -3 dB dip in power response (+3 dB rise in DI) and the other had a much flatter power response with no dip (a smooth DI curve). I was pondering why these two responses sounded so similar to me. Most of the time I could not tell them apart. Fluid pointed out a study by Lipshitz and Vanderkooy. They found that power response dips can be quite severe without them necessarily being audible, whereas power response peaks were more easilly detected. Toole makes a similar comment in his book.
Here is a link to Fluid's post (is this guy amazing?)
https://www.diyaudio.com/forums/multi-way/366347-active-satori-textreme-17.html#post6785249
This makes sense to me, and like any good scientist, I assign more weight to evidence which agrees with my theory, and tend to dismiss evidence which goes against my theory 😉 . I say that jokingly, but it is something we all have to be aware of and guard against.
This is a great discussion, and it almost deserves its own thread. I always learn a lot from these kinds of discussions. Thanks !
J.
Last edited:
I'm not sure if you are assigning those numbers to their importance to the sound if you are I think that is not their intention.
What those weightings attempt to do is give a good indication of what the speaker will measure like in a steady state in an average room.
The Listening Window is meant to best reflect the direct sound of the speaker i.e. what it will sound like. By averaging those measurement's within a small window, some diffraction and other positional effects are averaged out. In that sense the actual 0 degree or listening axis alone shouldn't be the only target.
The overall sound power and PIR is still important and the DI is a good indication. The best balance may well be a compromise between the on axis and overall soundpower or PIR, and that can change depending on the type of speaker and intended listening distance.
It is very possible to have a flat listening window and a nice PIR and Soundpower that is smooth but not with every speaker concept.
A reference to much of where this came from is in
Devantier, A. (2002). “Characterizing the Amplitude Response of Loudspeaker Systems,”113th Convention, Audio Eng. Soc. , Preprint 5638
and a good collection of parts that doesn't need AES access was posted by Juhazi elsewhere
https://www.edn.com/loudspeakers-objective-evaluations-part-2-measuring-the-essential-properties-of-loudspeakers/
Reading between the lines the Listening window is the sound of the speaker unless the directivity goes wild just outside it.
Have you ever tried to make a Moving Mic Measurement? That is a very quick way to get the same thing and is very repeatable. If you do it would be interesting to compare as a steady state measurement should tilt down more with a rising directivity design.
I'm not sure I should comment 😉
Hehe, yeah thanks both of you hifijim and fluid for the thoughts. I think this is the core of speaker design amids the nice software. Knowledge how things translate to sound in particular situation. I admire your knowledge on the literature, I'm working mostly on the thought process with very little literature behind and I really should read more.
Yeah actually me talking about sound power is symptom from the fact that the power is derived from the full data and the full data is actually the thing here. With it, the software is able to plot all these various graphs for power, DI, early reflections, predicted in room response and what not. When the data is available and the designer knows how to interpret it to achieve results they are after it becomes really powerful combination.
Measurements at the listening spot are very useful to verify what the system actually delivers but it really doesn't reveal what the response is made of. I mean if there is an issue it is really hard to translate it to some specific aspect from the in room response alone. If one has the full 360 data available on screen the cause for dip might be apparent, be it in the listening window or early reflections or just combination of early reflections. There might be possibility to adjust the issue in the crossover or it might turn out the speaker position could be better or the construct should be changed somehow. Most of these are easily tested within the simulator if the full data is there. I think this is very powerful, visualization of the invisible helps the brain a lot to solve any issues and not just for the crossover but the whole system.
Words of wisdom 😀 kudos
Nice, thanks bringing up the limitations of the home measurements! Also the resolution is not the best <1kHz due to windowing, some errors can happen on merging far and near field responses and what not. I must admit I've really struggled to improve the measurement and preparation process to get better results, better correlation of the visuals to the sound and measured in room response. At the beginning I had to adjust the sound quite a lot by ear but lately the process has been more consistent. Part of this might be difference between the DSP and settings I have used in VCAD and partly due to errors in data used for the sims. I suspect the simulator being a math crunch is as accurate as the data fed into it so it is not exactly walk in the park. And, the takeaway is that still it would be an approximation of the reality and should be remembered when interpreting the simulation.
To the core, the measurements and simulated graphs show the full system and what the concept is capable of and the person looking at it determines if it is suitable for the situation / goal in mind.
Alright, lets keep polluting on some other threads shall we? 😀
Yeah actually me talking about sound power is symptom from the fact that the power is derived from the full data and the full data is actually the thing here. With it, the software is able to plot all these various graphs for power, DI, early reflections, predicted in room response and what not. When the data is available and the designer knows how to interpret it to achieve results they are after it becomes really powerful combination.
Measurements at the listening spot are very useful to verify what the system actually delivers but it really doesn't reveal what the response is made of. I mean if there is an issue it is really hard to translate it to some specific aspect from the in room response alone. If one has the full 360 data available on screen the cause for dip might be apparent, be it in the listening window or early reflections or just combination of early reflections. There might be possibility to adjust the issue in the crossover or it might turn out the speaker position could be better or the construct should be changed somehow. Most of these are easily tested within the simulator if the full data is there. I think this is very powerful, visualization of the invisible helps the brain a lot to solve any issues and not just for the crossover but the whole system.
Words of wisdom 😀 kudos
fluid said:
Nice, thanks bringing up the limitations of the home measurements! Also the resolution is not the best <1kHz due to windowing, some errors can happen on merging far and near field responses and what not. I must admit I've really struggled to improve the measurement and preparation process to get better results, better correlation of the visuals to the sound and measured in room response. At the beginning I had to adjust the sound quite a lot by ear but lately the process has been more consistent. Part of this might be difference between the DSP and settings I have used in VCAD and partly due to errors in data used for the sims. I suspect the simulator being a math crunch is as accurate as the data fed into it so it is not exactly walk in the park. And, the takeaway is that still it would be an approximation of the reality and should be remembered when interpreting the simulation.
fluid said:
To the core, the measurements and simulated graphs show the full system and what the concept is capable of and the person looking at it determines if it is suitable for the situation / goal in mind.
Alright, lets keep polluting on some other threads shall we? 😀
Last edited:
Wel, it's interesting, so this kind of polution is not that bad.😀
But on this. You got the theory and the reality and those are not always lining up perfectly. Budget, physical and practical limits are also into play. If that wasn't everybody would be in an acousticly perfect room with the speakers set up like it should. But the reality is that most don't, they are limited to the space they have, and the speaker (or general audio setup) should be adapted to that.
In my case, a small room (3.5x3.5x2.5m with one stone wall (in front of me) and 3 gyproc walls with a lot of insulation in. Floor is stone with a carpet (soon at least) and the ceiling is wood/gyproc with a lot of insulation. My desk is standing against the stone wall.
In that case, dispertion off axis is not that important, and should not be to wide or the room comes into play to much. I'm sitting on axis all the time on a fixed height (on a chair) about 50cm from the speaker. So i'm still wondering why you think i should have a waveguide, especially if i'm working with a tweeter that is very good (flat) off axis on it's own and can go way lower than i will cross it (see pictures from troels his measurements (so both argument to get a waveguide are not really relevant. Add to that that the budget is limited, just like the tools i have to make this speaker. I just think it's over the scope of the project to be honest and are very tempted to it like i originally planned. I would only change the crossover to a higher order after i measured the drivers in the cabinet, and for the rest stick to the original plan.
But still interesting discussion mostly here. I learned a lot.
Attachments
I think that you dismiss the off axis response of the speakers too quickly and your preference in commercial monitors suggests that well controlled directivity is something you like. Think about the length of wavelengths as frequency goes down, you have no choice but to hear the room at some point even when you are in a strong direct field.
I don't want to discourage you and whatever constraints you have are up to you to work within.
Not that there is anything wrong with XSim but Vituix will show you much more about the design of the speaker and how your crossover driver selection and baffle design interact with each other. An amount of time spent here can get you much closer to a good speaker even if you just trace manufacturer responses and simulate with idealised drivers. Most of what is measured in terms of directivity comes straight from the dimensions of the drivers and cabinet.
I don't know if the DXT-MON is within your budget or the other version with the Monacor waveguide but it will take a lot of effort to get better than these.
Salut Waxx,
On se tutoie stp? 😉
Pour la defonceuse: jette un oeil a cette video ( tu peux zapper les premieres 5m 50s):
La défonceuse : les bagues à copier - YouTube
Hi Waxx,
We can use you rather than You please?
For the router take a look at this video ( you can skip the tirst 5mn 50s)
On se tutoie stp? 😉
Pour la defonceuse: jette un oeil a cette video ( tu peux zapper les premieres 5m 50s):
La défonceuse : les bagues à copier - YouTube
Hi Waxx,
We can use you rather than You please?
For the router take a look at this video ( you can skip the tirst 5mn 50s)
I don't do kits for my own anymore, designing is at least half the fun and i want to learn. So no option.
That is even with a waveguide and controlled directivity the case i think. I see what a waveguide can do good on that, but it's also a matter of cost/effor vs benefit issue for me. I'm sure i'm going to study this more, but then for speakers that are not that close nearfield.
I know from my experience that a flat baffle is in reality less an issue than a lot think here. My favorite monitors, the ATC's don't have a waveguide neighter for their tweeters. And a lot of speaker i like (that are not always monitors) idem.
I use both, i know xsim the best, but i always double check in virtuixcad. But virtuixcad needs me to measure the drivers in the cabinet to be real accurate (i can't really use calculations for the phase and time alignment in the crossover there), and i'm not in that phase of the project yet.
This is not totally true for Atc, some doesn't have waveguides others do have them : the biggest - monitors- one have waveguide within the faceplate of tweeters they use but it is not as visible as typical wg as they xover high in freq, and their design philosophy ( ka: 2 within the bandwith of each transducers).
Last edited:
I meant more as design inspiration to what is possible at a reasonable price in a similar form factor.
I'm fairly sure sure kimmo has put the calculated minimum phase option back in even though he doesn't like it.
This is the Heismmann one for the Monacor Waveguide
3D printed adapter plate | Waveguide - tweeter
The tweeter dome and surround needs to fit inside the waveguide. You could make something similar out of two 2mm thick sheets of something or whatever the thickness needs to be to have it match up properly.
3D printed adapter plate | Waveguide - tweeter
The tweeter dome and surround needs to fit inside the waveguide. You could make something similar out of two 2mm thick sheets of something or whatever the thickness needs to be to have it match up properly.
wel, that is what i try to do, i see the frontplate of the SB26 tweeter is 3mm thick, so i'm going to adapt the design to that, but for the rest it looks like i had the same id. I did also enlarge the cutout for the dome to the size of the visaton hole (38mm)
I have the tweeter here so i did measure it in reality, not on the drawings.
I have the tweeter here so i did measure it in reality, not on the drawings.
Attachments
If you don't find a way to do this by yourself (or a workshop around you) there is still another option: Schaeffer apparenbau.
I've had them to make multiples racks front plate in aluminium for my gear and some plate in perspex ( plexiglass) for mics and loudspeakers.
It is in no way cheap but the work is top notch quality wise.
I see that there are many 3d printshops that work from online. You just need to send the stp file. Some have a good reputation.
So it's a matter to adjust my design to perfect before i send it. But it ads another 40€ or so to the price of the project. + i need to buy those waveguides. So maybe it's better to just let them print some good waveguides for this. And i also wonder what would be the best material to use to print these waveguides.
So it's a matter to adjust my design to perfect before i send it. But it ads another 40€ or so to the price of the project. + i need to buy those waveguides. So maybe it's better to just let them print some good waveguides for this. And i also wonder what would be the best material to use to print these waveguides.
This will be a very cool project, but I wonder why you don't spend $1000 for room acoustics in this case.
As you may know, many proven commercial monitors sound just fine especially after minor EQ correction in proper room. I have used Barefoot, Dynaudio, JBL, Genelecs, Yamaha etc., then all are perfectly usable monitors after EQ in well acoustically controlled room with proper positioning at the end of the day. I don't say they sound the same and I'm sure every audiophile has his own preference, but as long as the goal is better mixing environment, those similar sized monitors work the same in my experience.
As you may know, many proven commercial monitors sound just fine especially after minor EQ correction in proper room. I have used Barefoot, Dynaudio, JBL, Genelecs, Yamaha etc., then all are perfectly usable monitors after EQ in well acoustically controlled room with proper positioning at the end of the day. I don't say they sound the same and I'm sure every audiophile has his own preference, but as long as the goal is better mixing environment, those similar sized monitors work the same in my experience.
Last edited:
John Krutke used another approach that may fit the bill here. Essential is that you adjust the throat size to that of the dome plus surround.
Who is saying i'm not going to do something about that? But i'm very limited in options as this is a rental house and a protected +400 years old historical farm where changing the building is very limited by rules and laws. The layout is also a bit strange trough that as the owner had to work arround that.
And i don't think they sound the same. I worked in studio's, and still use Yamaha and Genelec monitors in the same radiostudio every week. And those yamaha's (HS80 or something like that) are crap compared to the Genelec's (840's).
And why is everybody trying to steer me away from diy, isn't this a diy audio site?
