My first project is a 4 way speaker...

[cochinada]

That looks absolutely fine and it's great that your hardware is compatible with making measurements!

One does not need to save the curves with ARTA what you need to do is export them as an frd file or similar for use with some simulation software. Saving simply saves the impulse itself so that you can load it up again within ARTA at a different time. The files you save are only usable by ARTA too, which isn't any use for other programs. Of course having the ability to save the impulse files is very useful but isn't necessary for most things.

I see... So next step is to find out how I can export the curves as FRD

I still don't understand exactly how OmniMic works and why it appears to be much more conservative. I guess I will open a post for that issue somewhere.

You just forgot to give me advice concerning exactly which drivers should I measure on the far field and at what distance (#228 at the end).

[TEK]

The problem is that I have a demo version that doesn't allow me to safe the curves

You are thinking about this the wrong way around. Someone is using a lot of their time and effort to create a product that is of great help to you. You can use 79 euro to support their continious work to create and improve features that yourself and other may enoy now and in the future. At the same time you also get some bonus features that you may enjoy.

So, there is no problem - only a easy way to show that you appricate their effort and that you want them to continue to develope the product

[cochinada]

You are thinking about this the wrong way around. Someone is using a lot of their time and effort to create a product that is of great help to you. You can use 79 euro to support their continious work to create and improve features that yourself and other may enoy now and in the future. At the same time you also get some bonus features that you may enjoy.

So, there is no problem - only a easy way to show that you appricate their effort and that you want them to continue to develope the product

The thing is I already purchased Omnimic. I'm just giving ARTA a try to find out if it suits me best or not. If it does it's another story but for now I cannot go out and buy every program out there just for testing.
First I need to understand what I'm doing wrong with Omnimic because for me it's not acceptable that it doesn't give you accurate results unless there is a good explanation for it.

[TEK]

The thing is I already purchased Omnimic. I'm just giving ARTA a try to find out if it suits me best or not. If it does it's another story but for now I cannot go out and buy every program out there just for testing.
First I need to understand what I'm doing wrong with Omnimic because for me it's not acceptable that it doesn't give you accurate results unless there is a good explanation for it.

Sorry, didn't catch that you already had Omnimic....

[cochinada]

Sorry, didn't catch that you already had Omnimic....

No worries .

There is a thread in other forum with more people asking the same question I have. Apparently they were happy with the answer someone gave them but I'm not. Unfortunately I'm not yet allowed to post there so I have to wait. :huh:

Meanwhile I'm trying to rearrange my living room to minimize reflections. Here is one change that as you can see removed one small bump at 3.1ms as well as some disturbances between 2.04 and 2.15ms:

Image not available

Image not available

[5th element]

WHen you have a frequency response on screen with arta simply click 'file > export' and pick ASCII. If you're displaying mag + phase it will auto export as a frd.

With regards to the drivers to measure - the ones closer to the floor or ceiling, the ones that give you the best, reflection free, results!

[cochinada]

With regards to the drivers to measure - the ones closer to the floor or ceiling, the ones that give you the best, reflection free, results!

What? :stupid:
The woofers that theoretically should give me the best, reflection free, results are perhaps W3 and W4 as W1 and W2 are too close to the floor.
For the mediums M2 should give the best results.
But then what? Should I use the same curves as M1=M2 and W1=W4 and W2=W3?

[---k---]

I've seen way too many threads of people having difficulty with gear to take measurements and using the various measurement software. It is getting better with newer some of the new gear and guides. But, it still isn't anywhere close to plug-and-play simple. That is unfortunate. But, I think this is the step that separates them men from the boys, so to say. ... You're in good company with your struggles.

I'm at the kiddie table and still sane.

[cochinada]

It's actually very frustrating as I'm just staring at the speaker in the middle of my living room in a chaos not going anywhere.

At this moment I'm even more confused and this is my understanding so far, or should I say misunderstanding?:

- The mic should be placed in one spot far enough for the drivers to properly integrate (whatever that means) in case the speaker is very large as mine.
- Then you will get the far field measure of each driver maintaining the proper phase relationship between all units and you can estimate the z-offset if at the same time you measure a different pair in parallel.
- You then take your mic (move it physically) and measure the near field for each, adding the diffraction and baffle step effects and splice the corresponding Far field with this curve to get the quasi-anechoic response of each driver.
- At last you would import this curve together with the measured impedance into the simulation program where you would also have to define all the offsets between the drivers.

I guess this is one method but I know there are others based on two-channel measuring for instance where apparently you don't need to keep the mic fixed at the same spot nor 'guessing' what the z-offset is.

Anyway the big question here is how does the crossover design program knows in what circumstances were the FRD files taken and adjust the calculations accordingly??? It has to make a difference if I measure them at 1m or 2m, on tweeter axis or off-axis, in the same fixed position or several...

[BobEllis]

I'm nowhere near an expert, but it might help if you don't think of finding the Z offset as guessing. Using 5th Element's procedure, it's really measuring. Think of adjusting the offset in the simulation as finding the right wrench for a bolt. You grab one that looks right, but it takes a few tries. The proper size is your measurement. That's all aligning the simulation and the measurements is.

You've got a complicating factor with a three way and multiple drivers, but the most critical z offset is mid to tweeter. Why not focus on that first? I'd take several far field measurements without moving the mike. Individual drivers (you can do a single mid or woofer for simplicity, or if you run them in parallel make the measurement on the tweeter axis), Mid and tweeter together, Mid and woofer together, all drivers together. If you measure with multiple drivers, be sure to simulate with multiple drivers or the predicted level mismatch will likely make finding z offset impossible.

"Properly integrated" means far enough that the measurement distance is far enough that the vertical distance between drivers is not a significant factor in their time of flight. Imagine measuring 5 cm from the tweeter. The arrival time of the pulse from the mids is affected much more by the vertical distance than the Z offset. The outer woofers even more so. Back out to 1.5 meters and the distance between the measurement point and drivers is much less affected by the vertical spacing of the drivers. The woofers are still more affected than the mids, but since the timing change between 1.5 M and a listening distance of say 2.5 M isn't much you get better results than in close.

I see 5th Element is reading this thread, so I'm sure you'll get some sound advice from him soon.

[5th element]

It depends on the design software as to how complex its own internal simulations are. In LspCAD you enter the drivers relative positions to one another, the diameter of the drivers pistons/cones and the mic position. It then simulates everything that is necessary to account for the changes that would occur as the mic distance is changed.

If you use 1 meter as your measurement distance this is mainly chosen because it allows you to get good data. Once your measured data matches up with the simulation, at the measurement distance, you can then go an increase the mic distance within the software out to your normal listening distance and design for driver integration from there.

Two channel measurement mode isn't that much of an advantage over single channel. On one channel you present the mic signal, on the other channel you present the signal going to the loudspeaker. As the program knows what the stimulus signal was like it can compensate for any differences present within the signal going to the loudspeaker (such as a 0.5dB droop in your DAC or something). It might also be able to compensate for any differences in the group delay between measurements because it would be aware of when the system actually outputs the signal. What two channel mode wont do is remove the requirement for keeping the microphone in the same place as it's the time of flight differences between the drivers that allows you to determine the correct phase.

I have always done this by keeping the microphone in one place but it is possible you could do it a slightly different way, although I am sure the accuracy of this would depend on how 'similar' the mic position actually ends up being for each measurements.

1) You'd place the mic on the listening axis and take a measurement of all the drivers connected in parallel as your reference.
2) You then move the mic up and down in position and take measurements directly on axis for all of the drivers individually.
3) You import all the data into the measurement program and as you're aligning all the individual driver measurements to the reference you should be able to get things to line up.

The trouble with this is that the individual drivers frequency responses will be different on axis to how they were during the reference (where most were measured off axis). The simulation program should be able to account for this fairly well, but what it cannot do is predict how your box diffraction is going to change the response with respect to being measurement at different angles. The overall trend should match up pretty well though.

The trouble with my method of keeping the mic always in one place is that you are already using the off axis response of some drivers so you don't want the simulation software to apply any additional compensation as it's already in place by virtue of the mic position. You get around this by setting the driver diameter to something really small. Now in reality this really doesn't matter, at least in my designs. This is because I use drivers no higher than where they would start to beam, so even if the program alters the response to account for any off axis variations it thinks are occurring, it does very little to affect the drivers response over the frequencies I will be using it to cover. The other reason is because if the off axis response is going to be crucial towards certain design aspects (such as getting a directivity match between two drivers) I will have taken detailed polar measurements separately and have already used them to this end.

In your design for example, measuring at 1 meter and on the tweeter axis would give you a direct on axis measurement for the tweeter and a few degrees off axis for the midrange drivers (probably 5-10degrees). The woofers on the other hand will be measured at quite an angle and at different angles for each. On the face of things you'd expect this to be quite a problem but in reality it will not be that way. The 12M at such a small angle will be behaving the same on axis and off axis up quite high in frequency. Given the proposed low xover frequency of less than 2kHz, it doesn't really matter what goes on with regards to the off axis response. If the loudspeaker program wants to add in a couple of extra dBs attenuation by 5kHz and a few more by 10kHz it's largely irrelevant to what's going on around the xover frequency.

Even though they are going to be measured at drastically wider angles the same is true for the 22Ws. They don't start to show off axis response variations until around 1kHz, which is much higher than the proposed xover frequency. The measurement software wont do anything to affect the response that low so you're good to go.

With regards to measuring the upper or lower drivers? This comes down to your room. If you have high ceilings then the upper drivers will work better as the first reflective surface will no doubt then be further away. Usually the floor is the first reflective surface one encounters when making measurements as we can easily get the loudspeakers more than ~1 meter away from the side walls.

Will the close proximity of the woofers to the floor affect the response in any way? Yes. It will add in boundary reinforcement and reduce the amount of baffle step compensation necessary. The effect reduces the further away the drivers get from the floor and increases again with distance to the ceiling. With such a large loudspeaker as the one you've got I would expect some trial and error to be required for finding the correct tonal balance. I'd shoot for around 4dB on the four 22Ws though.

Getting accurate data for simulation is often quite hard and requires a fair degree of lateral thinking to get right and even then something will be out by a little. By this I mean that making nice measurements is rather easy, but making ones that are meaningful, especially with big loudspeakers, can be a little bit of a challenge and the more factors you can account for and potentially remove from the equation, the easier the measurements will be to make and the better the end results will be.

For example, in my floor mounted bass drive units, in my FST coax three way, designing the crossover for the bass driver could have been a bit of a pain but with a bit of thinking it becomes a lot easier. For example when floor mounted, as I have, measuring baffle step (or trying to measure it would perhaps be more accurate) would be very difficult, but it's actually completely pointless. With the driver mounted as such no baffle step occurs. I don't need to try and measure it, the driver is mounted on the floor, it can only radiate sound upwards into a hemisphere, it cannot transition into fully spherical radiation as would normally happen, so no compensation is required.

The other trouble comes with trying to get nice measurement data for simulation its low pass xover. With the driver mounted so low down you immediately get a reflection off of the floor making things harder than you'd like. This isn't so much of a problem though as the 8" driver is flat, both on and off axis, up to 1kHz (cone edge surround resonance notwithstanding) and with a baffle width of 30cm, no diffraction issues are going to occur within this frequency range either. This means that its frequency response is actually a straight line up to and way beyond my intended xover frequency of 300Hz. As my target acoustic goal is a 4th order LW at 300Hz, this means I can apply a textbook filter and end up pretty much where I want to be with a high degree of certainty. Of course my textbook filter is DSP implemented so isn't affected by the drivers own impedance, but the basic idea is the same.

[cochinada]

8O Boy... I will have to read this a couple of times minimum and slowly digest it :??

Thank you both.

[cochinada]

Two channel measurement mode isn't that much of an advantage over single channel. On one channel you present the mic signal, on the other channel you present the signal going to the loudspeaker. As the program knows what the stimulus signal was like it can compensate for any differences present within the signal going to the loudspeaker (such as a 0.5dB droop in your DAC or something). It might also be able to compensate for any differences in the group delay between measurements because it would be aware of when the system actually outputs the signal. What two channel mode wont do is remove the requirement for keeping the microphone in the same place as it's the time of flight differences between the drivers that allows you to determine the correct phase.

Even if some things you wrote I cannot still grasp, to be completely honest, they make perfect sense but regarding the dual-channel measurements I just want to leave this note that someone with also immensely greater knowledge than me made:

"Mic position can be changed with 2-channel measurement by adjusting start point of time window. For example if you move mic 1 meter further, you should start time window about 1/344 s later. You just need to verify speed of sound using your measuring system to calculate exact time."

Moreover he also explained me this other advantage, I think:

"(two channel measurements).. Usually z-coordinate of acoustic center is not very interesting because it is included in phase response. All response measurements have the same timing reference if dual-channel measurement is used and microphone is kept at the same distance from mechanical reference point e.g. surface of front baffle and starting point of time window for impulse response is kept constant. Microphone is just moved vertically at the center point of driver under testing, and drivers are measured one at a time."

I hope he doesn't mind if I'm sharing it with this community. I tried also to keep it in scope.

<slightly off topic> By the way, I cannot get approval from TechTalk moderators in order to be able to participate in their forum where I would like to ask support for the OmniMic I bought. Does anyone also have or had the same problem? </slightly off topic>

Meanwhile I'm struggling with ARTA. I'm trying to calibrate a sound card but either the manual is wrong or I'm doing something wrong.
First step '1.5.1 Calibration of Soundcard Output Left Channel' goes smooth but '1.5.2 Calibration of Soundcard Input Channels' doesn't work as they say.

- To start with, as soon as I connect the left output to the left line input a red led turns on (clipping) and it's no use even if I reduce the gain control to the minimum.

- Second, if I press the button 'Generate sine (400Hz)' I have absolutely no input level at bottom peak-meters; zero, nada, niente, rien! At the same time button 'Estimate Peak Input mV' becomes disabled.

- To get some input level at bottom peak-meters I cannot press button 'Generate sine (400Hz)'. So, then it's possible to press button 'Estimate Peak Input mV'. But now the level on these peak-meters turns red and 100%. A message pops up saying signal is clipping.

- The only way to get some 'nice' readings is if I don't press the button 'Generate sine (400Hz)' and I disconnect the loop between the left output to the left line input. Then I can press button 'Estimate Peak Input mV' but it gives values of about 48000mv!

Does anyone have the same problem?

Thanks!

EDIT: I've opened a new thread just to keep this technical discussion in another topic.

[cochinada]

Meanwhile, and for the damping of the subwoofer boxes, I'm hesitating between 1cm as in the picture below or a double layer - 2cms. I wll use also the Muhwole to fill the inside.
One layer is already a pain in the *** so two layers are a double pain and if I can avoid that I would be much happier. What do you think? Would 1cm with the Muhwole be enough?

Image not available

[TEK]

I'm not sure what stuff the Muhwole is. I only find shaving brushes when searching for it
But the question is if there is much purpose of damping the inner walls off the bass section as the frequency that the driver operates on will probably below any cabinet resonance that you would stop by damping the walls...

On the Wavecor Ardent there were only stuffing in the bass chamber and no damping of the wall for this reason. The mid-chamber was heavy damped (Jon used led).

[cochinada]

I'm not sure what stuff the Muhwole is. I only find shaving brushes when searching for it

I misspelled.. it has two ll. ops: See this. Mine is all gray, so not exactly like in the picture.

But the question is if there is much purpose of damping the inner walls off the bass section as the frequency that the driver operates on will probably below any cabinet resonance that you would stop by damping the walls...

On the Wavecor Ardent there were only stuffing in the bass chamber and no damping of the wall for this reason. The mid-chamber was heavy damped (Jon used led).

Well, if it isn't I would be very happy

EDIT: http://www.data-bass.com/data?page=c...9#!prettyPhoto

So, my understanding is if one has internal reflections inside of the enclosure, damping and stuffing can help get rid of them or at least attenuate it. At the same time, the impedance and Qtc changes which will affect the frequency response, as can be easily simulated in so many programs out there like Unibox or WinISD just to cite two. For instance, in my case I have an internal volume of 70l and to achieve a target Qtc of about 0,7 I should have Qa around 30 considering a Ql of 30. Put it simply, this would mean to use 'some' stuffing/damping. From what I read a 5<Qa<10 means heavy stuffing/damping and 80<Qa<100 means very little. Still according to Unibox, 'walls covered' equate to Qa=20 and a heavy fill to Qa=5. So accepting this to be truth then I should not line the walls if I want to keep Qa>20 but changing Ql will also affect the response of course and so does the resistance in series... :roll:

[5th element]

"Mic position can be changed with 2-channel measurement by adjusting start point of time window. For example if you move mic 1 meter further, you should start time window about 1/344 s later. You just need to verify speed of sound using your measuring system to calculate exact time."

Well of course you can do this, but you'll need to measure the difference in mic distance to the mm if you do. And then of course there's the verifying the speed of sound part. I don't know about you but I'd much rather not be doing that.8O

Moreover he also explained me this other advantage, I think:

"(two channel measurements).. Usually z-coordinate of acoustic center is not very interesting because it is included in phase response. All response measurements have the same timing reference if dual-channel measurement is used and microphone is kept at the same distance from mechanical reference point e.g. surface of front baffle and starting point of time window for impulse response is kept constant. Microphone is just moved vertically at the center point of driver under testing, and drivers are measured one at a time."

Exactly, that's the point, the z-coordinate is included within the phase, so getting accurate phase data and inter-driver phase measurements is essential in being able to accurately align them. The actual Z point is academic, I'm not interested in it either, just the phase relationship between my drivers!

I can't really see why the above would be any different for a single channel measurement if the inherent system group delay was identical between each measurement. Two channel is necessary for the first thing above because everything is going to be timed from the point of the electrical impulse being measured, rather than some arbitrary undefined point with the single channel measurement, so adjusting the time of flight difference for different mic positions would presumably become quite difficult. But I can't see why two channel is necessary for the second paragraph. This is essentially saying what I said you could do back in my lengthy post. You do have to keep the mic distance identical between each measurement though. Moving it up and down is fine, just not backwards and forwards.

I hope he doesn't mind if I'm sharing it with this community. I tried also to keep it in scope.

No it's absolutely fine I am sure whoever it is would not mind, we are all wanting to learn, even me! I have specifically steered clear of adding additional complications and variables to the measurement setup as described, just so that they wouldn't, or rather couldn't, be a source of error.

From my point of view I have only seen 2 channel mode as a way of removing the frequency response errors inherent to the electronics in the measurement chain because I've not been wanting to add in any additional complexities to making measurements. (Like moving the mic up and down when it's not really needed or moving it anywhere and then measuring the difference in the distance to the baffle to the mm, deriving the speed of sound in my room and accounting it from the impulse start point). Yes it will work but it sounds like a big headache.

Meanwhile I'm struggling with ARTA. I'm trying to calibrate a sound card but either the manual is wrong or I'm doing something wrong.
First step '1.5.1 Calibration of Soundcard Output Left Channel' goes smooth but '1.5.2 Calibration of Soundcard Input Channels' doesn't work as they say.

I've never calibrated with ARTA as for me it's not been necessary - I know what the frequency of my electronics is like, so I cannot be of any help. If I get some free time I'll see if I can help.

[TEK]

came over this that also might be some interesting reading for you...

http://www.troelsgravesen.dk/measurements.htm

[cochinada]

Exactly, that's the point, the z-coordinate is included within the phase, so getting accurate phase data and inter-driver phase measurements is essential in being able to accurately align them. The actual Z point is academic, I'm not interested in it either, just the phase relationship between my drivers!

I can't really see why the above would be any different for a single channel measurement if the inherent system group delay was identical between each measurement. Two channel is necessary for the first thing above because everything is going to be timed from the point of the electrical impulse being measured, rather than some arbitrary undefined point with the single channel measurement, so adjusting the time of flight difference for different mic positions would presumably become quite difficult. But I can't see why two channel is necessary for the second paragraph. This is essentially saying what I said you could do back in my lengthy post. You do have to keep the mic distance identical between each measurement though. Moving it up and down is fine, just not backwards and forwards.

Sorry but I've read this more than 10 times and I still don't understand. You totally confused me with the "above" and "first thing above". :huh:

I've never calibrated with ARTA as for me it's not been necessary - I know what the frequency of my electronics is like, so I cannot be of any help. If I get some free time I'll see if I can help.

Thanks a lot! ARTA is a new thing to me and I'm basically learning as I go. I have the idea that calibrating the sound card would only reflect on the SPL level measured in the way to give more truthful values, but perhaps I'm completely wrong in my assumption. Secondly and most important, I don't how to work with dual channel measurement. What I mean is, I know what to do to select this mode but I don't know how to interpret the results and/or select the gate. For instance, there are these buttons 'Get', 'Zero', 'Inv' and 'Max' and I have no idea what to do with them. It would be nice if there was some tutorial out there...

[cochinada]

came over this that also might be some interesting reading for you...
http://www.troelsgravesen.dk/measurements.htm

Very interesting! I'm an usual reader of Troels site but somehow this mic section has eluded me. I was curious to see how much my mic setup would affect my measurements as the curves that Troels showed made me worried. However I didn't notice any changes by slightly adjusting the length of the mic, nor by placing for instance a small piece of foam behind it, although in parallel instead of perpendicular to the mic axis.

[5th element]

Moreover he also explained me this other advantage, I think:

"(two channel measurements).. Usually z-coordinate of acoustic center is not very interesting because it is included in phase response. All response measurements have the same timing reference if dual-channel measurement is used and microphone is kept at the same distance from mechanical reference point e.g. surface of front baffle and starting point of time window for impulse response is kept constant. Microphone is just moved vertically at the center point of driver under testing, and drivers are measured one at a time."

Is the above.

"Mic position can be changed with 2-channel measurement by adjusting start point of time window. For example if you move mic 1 meter further, you should start time window about 1/344 s later. You just need to verify speed of sound using your measuring system to calculate exact time."

Is the first thing above.

[cochinada]

OK. Now I understand

After much pondering I've decided to go with the dual channel measurement mostly because these speakers are really huge and since I can't raise the tweeter above 89cm and the medium above 1m, it's impracticable to place the mic at 2m or even 1.5m for proper integration of the drivers because I'm having too much reflections from the floor already before 3ms or so. As such my far field measurements would not extend low enough.

Anyway I will not go as far as calculating the speed of sound in my house as I will try to the best of my abilities to keep the mic at the same distance from the baffle surface and just move it up or down. I still have to decide if it's going to be 80cm or 1m. Then I would want to measure the responses off-axis and I already have a cunning plan for rotating the speaker that works well enough.

[cochinada]

After measuring at 80cm I've decided to measure again at 1m because due to the width of the baffle and also the 8" woofers diameter, 80cm is just not enough distance to be considered far field.
Since I wanted to measure off-axis responses and don't have a rotating platform I had to improvise...

Speaker at 60º. Mic pointing to W3...

Image not available

This pin is the rotating axis and it needs to be always in contact with the marker (there are two because the MTM is offset by 23mm to minimize diffraction)...

Image not available

And this is how I measure the angles...

Image not available

[cochinada]

Merging... I guess this is no exact science.

Take for instance the following curves for the tweeter:

Image not available

HF response (blue) is good down to 309Hz and LF (red) is good up to 4055Hz. However, as can be seen, there is much more 'information' on the HF above 1KHz than there is with the LF. What is the most correct? What should be the best central frequency for merging?

Here I've already scaled the LF and am considering 1200Hz with one octave of blending but am having second thoughts...

Image not available

What would you chose?

[5th element]

Typically you do not need to do any merging for the tweeters response. As you have seen the far field data is accurate down to around 300Hz, which is far below the intended xover frequency. Yes the two curves diverge below 300Hz, where the near field measurement shows its strength, but this will be significantly into the stop band of the tweeter once the xover is in place meaning it has little or no bearing on the optimisation of the xover components or values.

As you can see though the near field measurement is inaccurate towards high frequencies and it wont contain the effects of baffle diffraction either. The far field measurement on its own would be my single pick over any merging of the near field, as it should simply not be necessary for the intended range of the tweeter.

Nice work by the way on the off axis measurements with the front baffle positioned pivot, you've certainly got the right idea. And certainly building a turn table for a loudspeaker as large as this would be very difficult.

[cochinada]

Ok, thanks. One less headache for me.

But what about the medium (M2) response? With the far field measures I took they are valid down to 230Hz. Theoretically, what would be the minimum crossover point you would chose between it and the woofer with a LR2 for not needing to merge the medium response with the FF?

My original intention was something around 350-400Hz but W3 is pretty much flat up to 800Hz and also with low distortion. I didn't measure the latest yet but according to this it should be OK.

The medium has better distortion values still and so it would be nicer to extend it a little bit I guess. What are the other issues I have to consider here? Directivity? Power handling? ...?

[5th element]

Ok, thanks. One less headache for me.

But what about the medium (M2) response? With the far field measures I took they are valid down to 230Hz. Theoretically, what would be the minimum crossover point you would chose between it and the woofer with a LR2 for not needing to merge the medium response with the FF?

My original intention was something around 350-400Hz but W3 is pretty much flat up to 800Hz and also with low distortion. I didn't measure the latest yet but according to this it should be OK.

The medium has better distortion values still and so it would be nicer to extend it a little bit I guess. What are the other issues I have to consider here? Directivity? Power handling? ...?

Directivity on a driver basis shouldn't be an issue with any of the proposed crossover points between the woofer and the midrange drivers. All drivers should be operating as omnidirectional radiators so 200 or 800Hz, neither will matter.

Issues in the polar response due to the xover points chosen and the driver separation will come into affect though and given the loudspeakers height you are going to get lobes or points forming in the vertical off axis due to the separation of the bass drivers. This will create nulls in the output at specific frequencies and reduce the amount of energy being directed at the floor and the ceiling. Theoretically if everything was aligned correctly you could directly reduce the effect of floor bounce over the intended range of listening positions by positioning a null at the expected floor bounce frequency so that less energy will be reflected off of the floor to cause the null in the first place. This might be thinking about it too much as your xover point to the midrange drivers will also come into play with that and affect it more significantly, plus the large number of bass drivers, across their separation will help to combat floor bounce issues too.

With regards to the xover point itself and distortion/power handling. It's all a bunch of trade offs. Simulation will show you whether or not you are going to approach Xmax of the midrange drivers for a given SPL and allow you to choose a higher xover point, if necessary, for your design goals. Going LR2 dramatically increases the demands placed upon them vs LR4.

Some trade offs might be that at lower volume levels you could have lower distortion with a lower xover point, but at higher drive levels the distortion creeps up as the midrange drivers start to get pushed hard. At high SPLs it could be that the bass drivers distortion is actually lower than the midrange so opting for a higher xover point would result in lower distortion with the volume turned up.

Really though you need to run some simulations to see how hard the midrange drivers are going to be pushed, xmax vs SPL to know if this is going to be a problem. If you're only going to approach xmax at ridiculous SPL levels then I wouldn't worry at all about going with a lower xover point.

At the same time you might find you get better overall performance by choosing an LR4 xover and crossing lower. Still as your midrange and bass drivers have a considerable range of overlap between them any number of xover points could be chosen. From my point of view the xover point should be decided based on the vertical off axis performance and floor bounce considerations, then you see if you need to go LR2 or LR4 for decent performance.

[cochinada]

Thanks a lot! I certainly have a lot to simulate as soon as I finish the measurements. In the beginning of this project I did some simulations with the responses from manufacturers and got promising results with a 3rd order LP for the woofers and a 2nd order HP for the medium but with the real responses I have to start with a clean sheet and for sure will take note of what you wrote.

Meanwhile for what is worth it here are some distortion figures I just measured:

Tweeter...

Image not available

Medium...

Image not available

Woofer...

Image not available

Actually I think the medium and the woofer are very balanced and it's not a so clear win for the medium in my opinion.

[cochinada]

All measures are finished at last! It's not easy to rotate these beasts to measure different angles and moreover to do it 3 times (I actually must have done it a dozen times) but I was able to obtain the curves at the following angles:
0º, 10º, 20º, 30º, 40º, 50º, 60º, 80º, 100º, 120º, 140º, 160º

Tweeter...

Image not available

Medium...

Image not available

Woofer...

Image not available

This was a much more daunting process that I had foreseen and until now it was by far the most difficult task for me.
Defining the methodology that suited my particular speakers and room the best was not easy and demanded not only a lot of research but above all the help of all of you who were kind enough to show the interest to participate or read this thread and were patient enough to explain me many things that were new to me. I don't want to be unfair as many helped me a lot but some were willing to go that extra mile and I cannot thank you enough! As you might remember I opted to go for dual channel measurements for the reasons I already explained and so had to buy a suitable sound card for the purpose and drop Omnimic (on sale) to use ARTA instead. This gave me a lot of headaches too but eventually all ended well.

Finally I would like to thank the author of VituixCAD (Kimmo) who has been simply impeccable to bear with me and my endless questions that he surely finds too basic. I've learned a lot from him and if it wasn't for his guidance and amazing simulation program I would still be stuck almost for sure. There is much more information out there about measuring loudspeakers using a single channel vs. dual channel. In fact 5th element has wrote a wonderful tutorial explaining how to do it and for all who find this method suitable I vividly recommend it! In my case, there was none and Kimmo come to the rescue. With his permission I'm sharing what I believe is an invaluable guide for all of you who want to properly measure a loudspeaker using dual channel. Here it is:

http://kimmosaunisto.net/Software/VituixCAD/VituixCAD%20Measurement%20Preparations.pdf

This tutorial is based on my project and that makes me very proud for having a very small contribution to give to the community.

One final word about VituixCAD that I already mentioned a couple of times before: in my humble opinion this program does everything (or almost) you would need for simulating a loudspeaker. The only thing I would like to have is a way to know the watts dissipated on each component. Nevertheless this can be figured out easily with XSim so it's no big deal. I'm sure Kimmo would do it if he finds it brings added value. Anyway the program has so many features and extra tools like Enclosure, Merger, Calculator and Diffraction that you will struggle to find anything missing. In fact there are things I didn't even touch nor understand what they're really for. There is also this brilliant and intuitive GUI that enables anybody with none or very little experience to start designing their own crossover in the flash of an eye. You would think it would cost an arm and a leg if we think for instance about LspCAD or SoundEasy just to mention a few. Well... it doesn't! For the price Kimmo is asking I think it's a steal.

I don't have any personal gain here except to share it with you so here is the link if you're interested to learn more about VituixCAD:

http://kimmosaunisto.net/Software/Software.html#Installation

...and his webpage with some speakers he built:

http://kimmosaunisto.net/

P.S. now I have to learn all I can about Power response and Directivity Index.

[TEK]

Very happy to here that you are getting there.
I had to admit that after the first couple of rounds, and then the silent, I started to wonder if you had given up.
Glad to hear that you are not and that you are still on your way.

How is it, are you starting to get some kind of impression of how the speakers will perform? Do you think you are far away from opening a nice cold bottle with something good in it, sink down into a good chair, but your legs up and just enjoy the music?

[5th element]

Very nice work there! Your journey hasn't been easy but things are coming together now. It's nice to see that even though this is your first project you are taking the time to learn what is necessary to make the final result work properly. So many people build a cabinet, plonk some drivers in it and basically stop there putting some kind of third rate ill designed xover in place.

I don't think I mentioned the near-field, far-field measurements in my post above as I got side tracked on the polars. In my opinion you're going to want to merge near field data to the far field data of the midrange and woofer so you can correctly model the xover transfer functions down low. These need to incorporate the low end roll offs to be done properly. This does depend on how low you want to cross of course. If you're going to go with something like 500Hz and 4th order you could get away without as you'll have 24dB+ of accurate data down into the stop band, which realistically is plenty for a good xover design. That is providing there isn't anything significantly abnormal a little further out of band that you wouldn't expect (like a magnesium cone resonance or something).

[cochinada]

Very happy to here that you are getting there.
I had to admit that after the first couple of rounds, and then the silent, I started to wonder if you had given up.
Glad to hear that you are not and that you are still on your way.

How is it, are you starting to get some kind of impression of how the speakers will perform? Do you think you are far away from opening a nice cold bottle with something good in it, sink down into a good chair, but your legs up and just enjoy the music?

This project has demanded a lot of my resources so far, not only monetary but also psychological for strange as it may sound. My house has been in turmoil for quite a while and I'm not listening to music for ages, except in the car which is not the same thing by a long shot. I admit there was a point where I've questioned myself if I was going to be able to finish this and quitting crossed my mind for one millisecond but I'm one of those stubborn for the better and for the worse never give up guys so I moved on. Now I finally see the light at the end of the tunnel and I know I'm going to finish this and hopefully with a good result.

Very nice work there! Your journey hasn't been easy but things are coming together now. It's nice to see that even though this is your first project you are taking the time to learn what is necessary to make the final result work properly. So many people build a cabinet, plonk some drivers in it and basically stop there putting some kind of third rate ill designed xover in place.

Thanks! Indeed, specially for a guy without any experience except in assembling other's people kits which is like comparing seeing Naked and afraid in the comfort of your home to actually doing it.

Now I'm finally at the final stage, the one I always anticipated the most which is designing the xover. I will exhaust all variations I can imagine and some before stepping in. My goal now is to optimize the power response and have the flattest response possible at the same time, which is another very tough compromise. I need also to keep an eye on the directivity, specially vertical, which is always going to be the Achilles' heel on this project due to the topology I've chosen, amongst many other things.

I don't think I mentioned the near-field, far-field measurements in my post above as I got side tracked on the polars. In my opinion you're going to want to merge near field data to the far field data of the midrange and woofer so you can correctly model the xover transfer functions down low. These need to incorporate the low end roll offs to be done properly. This does depend on how low you want to cross of course. If you're going to go with something like 500Hz and 4th order you could get away without as you'll have 24dB+ of accurate data down into the stop band, which realistically is plenty for a good xover design. That is providing there isn't anything significantly abnormal a little further out of band that you wouldn't expect (like a magnesium cone resonance or something).

True. Me or Kimmo I should say, eventually merged both responses (as can be seen on his tutorial) since it's not decided yet what kind of filter or xover point is going to be used between the woofer and the medium so it's better to have the 'whole' curves available to play with.

[cochinada]

The best compromise for Power & DI so far (courtesy of Kimmo)...

Image not available

"The sound power response is the frequency response of the total radiated sound produced from the loudspeaker at different frequencies. This can be measured by measuring the loudspeaker in a reverberation chamber or calculated by measuring the loudspeaker in an anechoic chamber at many angles around a sphere. We do a total of 70 different measurements in the vertical and horizontal orbits (every 10 degrees) and then calculate the sound power response from those measurements.

Your impression of the sound quality of a loudspeaker in a room is based on a combination of the direct, early and late reflected sounds produced by the loudspeaker. Therefore, to accurately predict the loudspeakers' sound quality you need a lot of anechoic frequency response measurements that characterize the direct sound (on-axis//listening window frequency response curve, the early reflected sound(first reflection frequency response curve), and the late reflected sound (sound power response). We've shown that with this data, you can accurately predict the measured in-room response in a typical room between 200-300 Hz and 10 kHz. Below 300 Hz the room dominates what you hear (room resonances, solid boundary gain). Above 10 kHz or so, the absorption from air and room treatment will influence the late arrivals."

-Sean Olive

http://www.whatsbestforum.com/showthread.php?3277-Sound-Power-Response-vs-Frequency-Response

[cochinada]

Going back to the subwoofer boxes I've been busy and thinking about it again.

Take a look at this picture:

Image not available

So far I've covered the walls and used the Muhwolle to fill the red area (this is upside down)...

Image not available

Now I'm wondering if I should also fill the green area, just part of it or just stop here. The outside parts near the driver are very difficult to fill without the Muhwolle collapsing because I can't use much material for letting the driver breathe.

Any ideas, thoughts? suggestions? :roll:

[cochinada]

I felt like a bird doing it's nest...

Images not available

[TEK]

It looks very nice in there - like a place you would just crawl into and lay down :tired:
And with the size of those speakers it's almost so that there is enough room to do that as well :yesnod:

[cochinada]

And it's not finished yet. I have some scraps of material including Sonofil that I intend to use to the limit. For instance the top central wall is not covered as well as the internal sides of each baffle. This won't do. :nono:

[cochinada]

I don't know if it is possible to do much better than this. FR is not flat but any attempt to improve it simply ruins the Power response.
Also notice how the xover between medium and tweeter ends up being much higher than I would like. All attempts and simulations I've made with values in the range of 1800Hz just have terrible power response curves.

(courtesy of Kimmo)

Image not available

To make things even worse, my living room is also very bad acoustically speaking

[TEK]

Have you tried these filter in real life now?

Room is always a huge issue, and there is almost always a lot to gain by doing some kind of room treatment. Also the location of the speakers will have a lot to say in that manner.
But even in a "regular" living room you can do a lot of acoustics room treatment without having to leve the room looking like a sound studio.

You can for example add sound damping pictures. Either DIY variants or you can buy complete pictures. Those may actually look very nice. For example a big picture with a great nature foto, family photo or anything else you are found of may do a lot for your room, both visually and sound wise. And not only when listening to music, but in general you can get a room that is more pleasning to be in when it comes to noise, sound reflections and so on.
It's also possible to do different things with your roof, and done correctly it may look very nice and stylish.
The point is to add the room treatment in such a way that it is more looking like room design features than room treatment.
So, now you have your next projects layed out for you ;-)

[TEK]

-- dublicate post--

[cochinada]

Have you tried these filter in real life now?

Room is always a huge issue, and there is almost always a lot to gain by doing some kind of room treatment. Also the location of the speakers will have a lot to say in that manner.
But even in a "regular" living room you can do a lot of acoustics room treatment without having to leve the room looking like a sound studio.

You can for example add sound damping pictures. Either DIY variants or you can buy complete pictures. Those may actually look very nice. For example a big picture with a great nature foto, family photo or anything else you are found of may do a lot for your room, both visually and sound wise. And not only when listening to music, but in general you can get a room that is more pleasning to be in when it comes to noise, sound reflections and so on.
It's also possible to do different things with your roof, and done correctly it may look very nice and stylish.
The point is to add the room treatment in such a way that it is more looking like room design features than room treatment.
So, now you have your next projects layed out for you ;-)

I wouldn't mind to improve the acoustics in my room but the problem is how? The ideas you are suggesting are pleasant but I don't know how can I implement them.

Maybe these pictures will help to get an idea of my living room and its limitations.

View from above with kitchen behind the sofa (you will see why this is relevant)...

Image not available

View from above in 3D where you can see an opening to the kitchen in the wall opposing the left speaker. There is also this big glass window close to the right speaker (<1m) and a desk close to the left one (<1m).

Image not available

Here the opening is clearly visible. I have a vertical venetian blind separating the rooms...

Image not available

Wall behind the speakers with rack and tv and also 3 paintings.

Image not available

So, as you can see, not an easy project...:roll:

P.S. I have a projector hanging from the ceiling above the sweat spot (I'm always wondering when it will fall and kill me but that's irrelevant for now).

[TEK]

You have to know what you problem is before you can fix it.
You said "my livingroom is very bad acoustical".

What do experience that the problem is?

[TEK]

I see that your "speech transmission index" is excellent, I assume that is a good thing and that you room is not totally off.
I do not have enough experience to translate your graphs into actual issues that needs to be addressed. If you first locate for example a frequency area that should be handled, that echo is a problem and simular stuff you could start by taking actions to reduce the issues.
However, you DO live in a home and you DO have a quite small room - so it will not be perfect. But maybe you could address the most annoying issues.

[cochinada]

Well, that was also my thought but apparently 'excellent' is not always excellent. According to Kimmo, "Excellent resolution would give values 0.90-0.95 at midrange…lower treble." and I have between 0.7987 and 0.8773 but that's not the worst.
I have asymmetric channels as can be seen, so basically right ear and left ear will not be in 'harmony'. Both channels are bad between 3-6 kHz, but right has slightly more peaks and dips with also greater amplitude.

[TEK]

Hmm, that should be possible to do something about. 3-6 kHz is quite high, so that should not be that difficult to address.
I would assume some absorbents or diffusers would do the trick for you. Maybe someone else could chime in here with more knowledge about room handling?

I see that the whole right side is a large glass area. While your left side is a regular room. That would likely explain the different between the sides.
How much changes can you do?
A couple of small things that you could start to test.
1) You could move your TV and speakers to the other wall, rotating the room content 90 degrees. That will result in you having all the glass behind you, something that might not be ideal either. And you would loose your view when listening to music. So, I assume that would not be a ideal solution for you.
2) A much easier change would be to get some real heavy curtains and add them before your glass wall. You should of course mount them so that they can be opened and closed as much as you want to. Covering 2/3 of the glass wall would probably enough to get the desired effect, and then you could still be able to enjoy your view when listening to music. When not doing critical listening the curtains could be left open and you could maintain fully view.
What are the big circular columns behind the speakers? Bass traps or subwoofers?

Your speakers are also off center in your room. That will also add to the difference between the speakers. Especially when you have that AND the glass windows. If possible I would have tried to center them more - but I have a hard time seeing how you could be able to do that as long as you have that desk there. As you have very little space, maybe you should add furnitures that support that? If you replace that big desk with a detachable wall mounted desk instead you would be able to use more of the TV-wall and thereby center the TV and the speakers more in the room.
Behind the left speakers you have a lot of bookshelves, while behind the other one you have the bass trap/subwoofer. Not sure about the effect of the bass trap/subwoofer, but the bookshelves will work as a great diffusor. The bass trap/subwoofer probably not. This will also add even more to the different between the speakers.
Also, your right speakers is almost placed into a corner while the other speaker is not - that influence also, but probably more in the lower frequencies.

This is what I would have done:
- The shelf behind the chair. Get that out of the room (if you find some place to put it later on, that's fine, I don't know where).
- The shelf right inside the door. Move that to the other side of the room. Move the lamp and move the speaker a bit longer forward to get room for it.
- The second shelf from the door, move that towards the door.
- Turn the desk around and move it into the wall. Put the chair into the desk.
- Move the two shelf to the left of the tv into the corner on the right side of the tv. The larges shelf into the corner. You will have to move the bass trap/subwoofer. Align it with the other one.
- Move the TV bench to the middle of the wall
- Move the left speaker closer to the left wall.
- Move the right speaker further away from the right wall (the glass wall), so that the distance from the speakers to the walls are the same on both sides.
- Move the coach towards the left, so that one of the seats are in the center position between the speakers.
- Maybe move the coach a bit forward. However, the hole in the wall might allow the sound to flow out behind the coach, so that might work out OK.
- Get heavy curtains that you can have in front of the glass wall

I think that would be a good starting point. After doing that you should be a lot better of when it comes to balance between the speakers. You will probably also gain more by adding some diffusors or sound absorbents to your walls or roof to reduce flutter/echo and raise your speech index score even more. But you should probably first try to balance your room better.
All the points I have suggested above should be possible to try without damaging anything, doing any irreversible changes or add any cost - so it might be a good starting point.

Edit: Maybe you could move the desk in front of the glass wall? Giving you a nice view when working? That might work as well to balance the room out and get the speakers away from that glass wall.