This is normal of waveguide/horn response, the Faital has a little bit of Tractrix to it, so the on on axis is a little flatter than it would be in something designed for more constant directivity.
Some relevant information I posted here
https://www.audiosciencereview.com/.../revel-m105-copy-diy-build.29465/post-1030095
I don't see the value in a 1.3m measurement with the speaker on the floor that close to other boundaries. It will not give either an accurate anechoic or in room response.
Thank you for pointing this out. I had actually noticed this small wiggle but thought that its impact is relatively less due to its lower amplitude in the impulse response plot.
I compared the different plots at 4.4ms,10ms, and 13ms. Here are the frequency responses corresponding to above gate durations overlaid on top of each other.
impulse response plot showing low amplitude reflections upto about 13ms
All frequency response plots seem to be close to each other especially above 1kHz and even down to about 600Hz. I was more interested in the above 1kHz response in this case to learn more about the horn's response.
In this measurement scenario, the horn's center is 103cm up from floor, 180ish cm down from the roof, 200cm away from one sidewall and 250+cm away from the other side wall.
I was thinking that the relatively low amplitude of the reflections from boundaries and their impact on the above 1kHz range response was due to the amount of sound absorbing material i had put on the floor, the relatively larger distances to other reflecting boundaries, and the relatively high directionality of the horn ( due to which, above 1kHz, radiation to larger off axis angles are significantly attenuated compared to a wide radiating mid+tweeter type design) and hence creating only low amplitude reflections to arrive at the mic
Thank you fluid
I have gone through the post. I will go through the white paper also..
How much distance would be a good distance to measure for this speaker?
I had been hearing that greater than 3-4 x baffle width is a good distance. This is why i tried to take 130cm distance measurement which is greater than 4 x baffle width (30cm). With the floor being the closest boundary (103cm) out to 90 degree off axis angle both vertically and horizontally, i thought having significant amount of absorption might reduce the floor reflection. The roof is about 1.8m up from the center of the horn and the sidewalls are atleast 2m away from the centre of the horn.
When i try to do angular measurements next, first, in this same room (if it doesn't work out, i will have to take the speakers upto the roof of the building), i was thinking to get the centre of each driver to about 120-130 cm up from the floor with significant amount of sound absorption material put on the floor. I was planning to do measurement with mic tip about1.5m away from the baffle.
No absorption on the floor would not work at lower frequencies. However, if you're interested only in the response of the horn then a mattress on the floor should do. You could also use similar material to enclose the speaker from the sides, so as to increase the ratio between the direct and reflected sounds. Putting the speaker on a platform (bed) would get it closer to the ceiling.
To get into the far field of the speaker the rule of thumb is often given as 3 times the longest dimension of the cabinet.
https://www.prosoundtraining.com/20...e distance,convenience – any distance will do.
When designing your own speaker consider the listening distance too, it may be convenient to measure at that distance or pick another one and then change the reference distance in Vituix later. It wasn't the distance so much as the orientation that I thought was not much use but that depends on how far you can get away from boundaries.
For the horn measuring at an equidistant point between floor and ceiling without absorption would maximize reflection free time, but you could contrast with the absorbent as you had it and the horn lower to the ground to get the longest gate you can. 4.3ms is not too bad but more would be better.
For the woofers you could try a ground plane measurement as you have tiles and it looks like quite a big room. With a tilt of the right angle this can be quite accurate at high frequencies too.
Measuring the woofers as you had it set up just seems odd to me, the absorbent is in the way of the bottom woofer more than the top and it's neither anechoic or in room, and to me representative of nothing useful either for crossover design or tonal balance.
As the woofer box is separate from the top you can measure the top box (or currently just the waveguide) alone about at 1 meter distance. Such measurements would have some error due to woofer box missing (or you could use it as a stand if you wish) but it is heck of a lot easier process to rotate small construct and measure at one meter than try find large space and measure big one higher up. Although accuracy would perhaps be better.
There is many chances to make mistake with the measurements and with the data before it is loaded into VCAD so take your time, enjoy the process and current results. With DSP you can always tune up the xo later, make new more accurate measurements if you find real world results don't match the sims.
There is many chances to make mistake with the measurements and with the data before it is loaded into VCAD so take your time, enjoy the process and current results. With DSP you can always tune up the xo later, make new more accurate measurements if you find real world results don't match the sims.
Thanks fluid. Now I get the idea.
I had tried doing ground plane measurements in the past in this room. I will try to do it for this cabinet also and see if I am able to do it, especially the tilting part. Given that the cabinets are about 35+ kgs in weight, I dont know how stable it will be after tilting, but I will try.
Yes the woofer measurements are not useful with the setup I had taken yesterday. As you pointed out, the absorbing material was almost directly in front of the bottom woofer and the bottom woofer was very close to the floor. My intention was to get some useful data, if possible for the horn. This is why I tried this setup. Next time I do measurements, I will keep all these aspects in mind..
Thanks tmuikku..
My plan was to study/compare/avoid the error in horn's response without considering the effect of the cabinet (at least in the horizontal plane measurements) with the setup I had taken yesterday. But I will refine the set up and see what changes it brings..
Sure. I understand that there is lot to learn in this process of taking at least reasonably accurate/useful measurements. I am enjoying and learning a lot..
The tilt is usually not that much, couple of wooden wedges and timber should get you thereGiven that the cabinets are about 35+ kgs in weight, I dont know how stable it will be after tilting, but I will try.
http://www.mh-audio.nl/Acoustics/Groundplane.html
For distances of measuring often 1m or 2m will be chosen for the ease of referencing the sensitivity back to a standard value.
Gotta mention quickly something as you now start to tune the DSP. I saw the pictures of the speakers and they point straight out. Have you experimented with toe-in yet? There is possibility to toe-in the speakers quite alot, 45 degrees to cross in front of listening spot if you can have the standard stereo equilateral triangle setup (with 60deg angle from speaker to ear, or 30 degrees depending where you reference) which makes listening axis about 15 degrees off-axis. Now while tuning crossover you should take the listening angle into account and make that the smoothest perhaps compromising the on-axis 0 degree response some. Top of VituixCAD main window there is reference angle box where you can adjust this and the graphs will show response at the angle.
You could try and search mr. Geddes writings on the subject. Advantages of such setup are numerous including stabilizing the phantom center and reduced early reflections. Also, it should maintain good imaging while still have some envelopment, big sweetspot and all but you might like it as you have it currently as well so this is just friendly pointer to experiment with toe-in and take the listening angle into account while tuning the crossover.
99% of loudspeaker setup pictures in the forum are like yours, not sure why, perhaps people haven't tried anything else or perhaps they found out it was good one. This is slow paced industry and perhaps everyone just doing what the granpa did without second guessing I encourage to experiment with this, crucial thing to make the speaker work with the room. The speaker is probably having smooth DI, smooth horizontal off-axis response so it will sound good at any listening angle, especially if you tweak the filters the listening angle in mind. Not all speakers can do this and they would miss all the benefits.
You could try and search mr. Geddes writings on the subject. Advantages of such setup are numerous including stabilizing the phantom center and reduced early reflections. Also, it should maintain good imaging while still have some envelopment, big sweetspot and all but you might like it as you have it currently as well so this is just friendly pointer to experiment with toe-in and take the listening angle into account while tuning the crossover.
99% of loudspeaker setup pictures in the forum are like yours, not sure why, perhaps people haven't tried anything else or perhaps they found out it was good one. This is slow paced industry and perhaps everyone just doing what the granpa did without second guessing
I encourage to experiment with this, crucial thing to make the speaker work with the room. The speaker is probably having smooth DI, smooth horizontal off-axis response so it will sound good at any listening angle, especially if you tweak the filters the listening angle in mind. Not all speakers can do this and they would miss all the benefits.
Last edited:
Thanks tmuikku.Gotta mention quickly something as you now start to tune the DSP. I saw the pictures of the speakers and they point straight out. Have you experimented with toe-in yet? There is possibility to toe-in the speakers quite alot, 45 degrees to cross in front of listening spot if you can have the standard stereo equilateral triangle setup (with 60deg angle from speaker to ear, or 30 degrees depending where you reference) which makes listening axis about 15 degrees off-axis. Now while tuning crossover you should take the listening angle into account and make that the smoothest perhaps compromising the on-axis 0 degree response some. Top of VituixCAD main window there is reference angle box where you can adjust this and the graphs will show response at the angle.
You could try and search mr. Geddes writings on the subject. Advantages of such setup are numerous including stabilizing the phantom center and reduced early reflections. Also, it should maintain good imaging while still have some envelopment, big sweetspot and all but you might like it like you have it as well so this is just friendly pointer to experiment with toe-in and take the listening angle into account while tuning the crossover.
99% of loudspeaker setup pictures in the forum are like yours, not sure why, perhaps people haven't tried anything else or perhaps they found out it was good one. This is slow paced industry and perhaps everyone just doing what the granpa did without second quessing
I will definitely try out the toe-in suggestion. Currently the speakers are just kept 1.2m apart. I sit about 2.5m away from the speakers. I just kept them straight (without toe in) for taking pics and getting it aligned with mic for some measurements.
However, I definitely felt that 1.2m separation is too short a distance between the speakers. I also noticed that everything seemed to be happening in the space between the speakers and it feels like not much sound is coming from the sides to the left and right of the speakers. Maybe due to the high directivity of the horn unlike the conventional speaker setups i have heard in the past, not enough radiation is reaching the walls for some reflections or something. Once I have some reasonably good measurements of the drivers in the box, I will definitely try out keeping the speakers further away and toeing them in.
Yeah I thought so something like this might be going on Yeah if at all possible experiment with speaker positioning as well as the toe-in. Listening distance to a speaker should be roughly the same as distance between the speakers. Distance from speakers to walls will affect a lot as well like the listening position, especially low mids and high bass there might be nasty dip resulting from boundary interference and low bass suffers room modes. If at all possible make everything symmetric left and right, if left speaker is 50cm from wall put right speaker 50cm of wall as well. It take some time to find optimal position for your room and possibilities. Take your time
Yeah if at all possible experiment with speaker positioning as well as the toe-in. Listening distance to a speaker should be roughly the same as distance between the speakers. Distance from speakers to walls will affect a lot as well like the listening position, especially low mids and high bass there might be nasty dip resulting from boundary interference and low bass suffers room modes. If at all possible make everything symmetric left and right, if left speaker is 50cm from wall put right speaker 50cm of wall as well. It take some time to find optimal position for your room and possibilities. Take your time
Last edited:
I started taking polar measurements and got till this point, measuring the top woofer. I took upto 50 degrees off axis from 1 metre away
I was happily cruising along taking measurements. Then disaster struck. REW started crashing again and again after a sweep. Then finally when i resolved it, now loopback is not being detected whatever i do. Dont know if it is sound card issue, or some software issue. I am also getting errors with messages mentioning some "Java null pointer exception " something.
Tried a lot to rectify it but to no avail. Maybe i need to take a break and come back to it again. I may even need to change my laptop since its keyboard is malfunctioning now..
On the good side, I am able to get upto a clean 6ms gate time before significant reflections arrive at mic using a combination of absorption material and positioning of the speaker in the room.
Hopefully, i get to complete this soon.
I was happily cruising along taking measurements. Then disaster struck. REW started crashing again and again after a sweep. Then finally when i resolved it, now loopback is not being detected whatever i do. Dont know if it is sound card issue, or some software issue. I am also getting errors with messages mentioning some "Java null pointer exception " something.
Tried a lot to rectify it but to no avail. Maybe i need to take a break and come back to it again. I may even need to change my laptop since its keyboard is malfunctioning now..
On the good side, I am able to get upto a clean 6ms gate time before significant reflections arrive at mic using a combination of absorption material and positioning of the speaker in the room.
Hopefully, i get to complete this soon.
Last edited:
After a real lot of hard and boring work, this is the best i could do regarding taling measurements
Measurement session in progress
Top woofer nearfield
Top woofer farfield polar
Rosso65CDNT on Faitalpro LTH142 horn polar
I can't take far field measurements of the bottom woofer. Tried a lot. But couldn't. Problem is I am unable to lift the cabinet high enough now to take measurements with center of the bottom woofer facing the mic at about 1m height. Anyway I am going to cross it over around 350Hz eventually. I may try to take a ground plane or nearfield measurement for it. And I will have to do some vituixcad adjustment for it.
Will try designing the crossover now
Measurement session in progress
Top woofer nearfield
Top woofer farfield polar
Rosso65CDNT on Faitalpro LTH142 horn polar
I can't take far field measurements of the bottom woofer. Tried a lot. But couldn't. Problem is I am unable to lift the cabinet high enough now to take measurements with center of the bottom woofer facing the mic at about 1m height. Anyway I am going to cross it over around 350Hz eventually. I may try to take a ground plane or nearfield measurement for it. And I will have to do some vituixcad adjustment for it.
Will try designing the crossover now
Last edited:
I am in the same situation. I am unwilling to lift my system high enough to get a far field measurement of the lower woofer. Our situations are very similar.
I feel that if the near field measurements of the upper and lower woofer match, and they do for me, I am willing to make the design assumption that the far field measurements of the upper and lower will be close enough. I use the upper woofer far field response to merge with the lower woofer near field response. This means the upper and lower woofer merged responses are very close to the same.
This is not ideal, perfect, not a best practice, but it is good enough for design purposes, in my opinion.
First attempt at crossover..
This is to get an idea about what can be done with this.. The vertical polars maybe wrong. I will try to study and use some document related to same that FM decibel had posted long back in VituixCAD thread and try to see what can be done. As next step i will try to simply the signal processing and the filter blocks involved.
This is to get an idea about what can be done with this..
The vertical polars maybe wrong. I will try to study and use some document related to same that FM decibel had posted long back in VituixCAD thread and try to see what can be done. As next step i will try to simply the signal processing and the filter blocks involved.
I am curious why you use two active amp stages on your horn, one with +14 dB gain, the other with -7.8 dB gain. I have not seen this technique used before. Is there some advantage to it?
I also noticed your "y dimensions" on your woofers are -34 mm and -64 mm. Perhaps you meant -340 and -640 ?
I also noticed your "y dimensions" on your woofers are -34 mm and -64 mm. Perhaps you meant -340 and -640 ?
I just added that 14dB amp block to model that difference in sensitivity between the compression driver and the paralleled woofers.. Just modeling to see what can be done with that delta.
It is just a dummy block. I will measure and adjust amp volume to compensate the sensitivity delta later when i am ready to apply the crossover on the system Other blocks are the actual frequency response shaping blocks.
@jcga ,@hifijim : Thank you very much for pointing this out the 34mm and 64mm issue. It is a mistake on my part. Just because i am used to the 'cm' convention, I mistakenly put the numbers there..
I will adjust the numbers and work on it again