• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Active Curved Beyma 15MC700nd/AE TD12M/TPL150H

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
This project started 4 years ago, but lots of things happened in between and i was only able to complete the speakers today :)

Design

Here the speaker dimensions, as designed 4 years ago:

cNXrgVr.png


Drivers

- Beyma 15MC700nd: X < 350Hz
- AE TD12M: 350 < X < 1500Hz
- Beyma TPL-150H: 1500Hz < X < 20000Hz

(Crossover points will be dictated by real measures, but should be around those values).
Crossover will be Active, using a Monacor DSM-48LAN.


Boxes

Here the empty boxes, before stuffing and final coating:

rF3ufrP.jpg


HwG8YlG.png



ys8GSst.jpg


Final Speakers
(sorry for the low image quality, i'm a really bad photographer)

Grills off:

fA3r7pv.jpg


Grills On:
RDa1lXc.jpg


One full speaker weight around 60kgs!

Some details

The backplate for the TPL-150H was removed, leaving the felt protection:

wbHDf85.png


Drivers were stuffed with a mixture of:
- 4mm Bitumen pads glued to the walls
- Wool felt glued to the bitumen pads
- One high density rockwool 70 kg/m³ wall at the middle of the box
- Low density "fluffy" rockwool 30 kg/m³
- Raw wool

0j9VUrB.jpg


8EmvT4r.jpg


GIWJilT.jpg



PmNEn7B.jpg



Initial Driver Measurements
(just for checking everything was ok)

A nearfield measure for both TPL-150H boxes (measured indoor, with ceiling at 1.8m at the measurement position):

z6kxrRq.jpg



A nearfield measure for both Beyma 15MC700nd boxes (measured indoor, with ceiling at 1.8m at the measurement position):
MnhgscE.jpg


(i accidently removed the measures for the AE TD12M, but both tracked perfectly as well).
You can find detailed measurements for the AE TD12M on drivervault.

Finally, a nearfield measure for all drivers (measured indoor, with ceiling at 1.8m at the measurement position):
PFP1rTh.jpg


Next step

I'm now waiting to receive a 1200mm x 610mm x 3mm MDF board, so i can draw the exact positioning for the 0º to 180º measurement using the Ground plane technique, to get the required measures to use with VituixCAD for crossover development.

I'll be requiring your help for fine tuning the crossover once i have all the required measures, in order to obtain the best result possible :)
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
Finally, the 3mm MDF board and a 6m cable have arrived, the weather seems finally fine, so measurements are planned for tomorrow.
I've draw the 0º to 180º positions for each measurement, so it would be easy to place the speaker in the exact position:

WgyT07t.png
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
@fluid, a quick question: for this speaker, which point should i choose as the reference axis?

OuAvK2n.png


1 - The center of the tweeter?
2 - The point between the Tweeter and the Mid?

Based on the e CTA-2034A standard, it seems that 1) should be used, but i would like to confirm it before start measuring :)
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
There is no right or wrong answer. Wherever you want the reference axis to be for the design. As the AMT will have limited vertical directivity you would want your ears to be closer in line with it unless the speaker is intended to be tilted. If in doubt I would say pick 1.
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
There is no right or wrong answer. Wherever you want the reference axis to be for the design. As the AMT will have limited vertical directivity you would want your ears to be closer in line with it unless the speaker is intended to be tilted. If in doubt I would say pick 1.

Interesting :)

Actually, when i designed the cabinets, the dimensions were chosen so that the AMT center (position 1), would be in a straight line to my ears, in my seated position at MLP.
Funny thing is, i never sit totally upright when i use the system, so in normal use, i'm semi-inclined, thus my ears are actually in a straight line not with position 1), but in line with position 2) :)

I see that some speakers, for example, Buchardt A500, according to Erin's review, have the "reference plane in this test is between the tweeter and the midwoofer".
Using 2) would also allow the use of a lower angle for ground plane measurements, as if using 1) i'll need a 12.8º angle, and if i use 2), i only need a 9.4º angle (based on this calculator). This would be useful, as due to the speakers size and weight, a lower angle is safer.

Based on the above, do you think position 2) might be a better option for reference axis?
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
The Beyma AMT's have really narrow directivity high in frequency so you would need to take that into consideration. You can make a CAD drawing to see where the radiation pattern from the speaker will cover your listening position. You may choose the lower point as the reference position but need to tilt the speaker down slightly to cover your listening position correctly.

The other thing to consider is how Vituix will interpret the data and measuring in a way to ensure that there aren't errors introduced to the crossover simulation from how the speaker is measured. The reference point can be changed in Vituix but the data needs to be acquired properly to make that work.

Measuring a completed speaker all in one go is different than measuring the speaker drivers separately for crossover simulation. You might be best to ask Kimmo directly over at the HTGuide thread because using full range ground plane isn't really covered in the Vituix documentation from what I can see.
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
As planned, i took all the required measures, for each driver.
It has painful, temperature was too hot, boxes are too heavy, and it took ages!

I used the AMT center as the reference axis.


Following VituixCad instructions for exporting measures:
"Remove all measurements from the list.
Open far field measurements of tweeter. Select axial response “ hor 0”. Select Impulse graph tab. Set Y-axis unit to
%. Click IR windows button to adjust gating. Set Left window Tukey 0.25, length 2.0 ms. Right window Tukey 0.25.
Adjust length so that time window ends in the beginning of the first reflection. Should be 3-5 ms. Click Apply
Windows to All and Save mdat".

Now, when i import all the drivers to VituixCad, the responses look different, as if they weren't gated:

uu1nwSe.png



For example, this is the gated response for the AETD12M, which looks completely different than what is presented in VituixCad:

OFXiWON.jpg



Am i doing anything wrong?

Can anyone please have a look in the VituixCAD project, and check if i'm doing anything wrong?

Here the final MDAT projects for each driver: https://we.tl/t-EfIxckx7o1
- Beyma15Measures.mdat
- TD12MMeasures.mdat
- AMTMeasures.mdat

Here the VituixCAD project, as well as the exported responses:
- https://we.tl/t-aFvWbl3PtG

Thank you in advance !!!
 

3ll3d00d

Active Member
Joined
Aug 31, 2019
Messages
207
Likes
173
If I export from the AMT mdat then I see data that does not match the txt files in your vituixcad so I think the Q is how exactly did you export & what did you export? are you sure the shared MDAT is the source of those files?

why did you measure such a limited frequency range for each driver btw?
also the mdat as shared does not have the window applied exactly as stated in the instructions you followed (the left window was left at 125) & the TD12 file has this oddity in the axial measurement

* Note: ; Delay -7,592.6911 ms (-2,604.293 m, -(8544 ft 3 in)) relative to Loopback from 2: Out 2 to 2: In 2 with 7,600.0000 ms (2,606.800 m, 8552 ft 6 in) timing offset

which seems strange

how did you decide where to put the window? the blip at 6.5ms is an obvious reflection to my eyes and quite possibly the one at 3ms is also but the stuff before it seems like some mess with the current window point not being so obvious to me

1654550456029.png
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
@3ll3d00d , thank you for having the time to look into it.

If I export from the AMT mdat then I see data that does not match the txt files in your vituixcad so I think the Q is how exactly did you export & what did you export? are you sure the shared MDAT is the source of those files?
I've probably made a mistake.


why did you measure such a limited frequency range for each driver btw?

I wanted to guarantee i eliminated all reflections to have valid data for the frequency ranges for each driver:
15 => 50Hz to 300 (measured from 50 to 800Hz),
12 => 300 to 1500 (measured from 250 to 1800Hz),
AMT => 1500 to 20000 (measured from 800 to 20000Hz


also the mdat as shared does not have the window applied exactly as stated in the instructions you followed (the left window was left at 125) & the TD12 file has this oddity in the axial measurement

* Note: ; Delay -7,592.6911 ms (-2,604.293 m, -(8544 ft 3 in)) relative to Loopback from 2: Out 2 to 2: In 2 with 7,600.0000 ms (2,606.800 m, 8552 ft 6 in) timing offset

which seems strange

how did you decide where to put the window? the blip at 6.5ms is an obvious reflection to my eyes and quite possibly the one at 3ms is also but the stuff before it seems like some mess with the current window point not being so obvious to me

I was going back and forth between VituixCad instructions, and this post, in the "designing an active 2 way speaker crossover", where the Impulse response is shifted before exporting the measures:

index.php


When doing this, the Window Ref time changes to a negative value, while on VituixCad instructions, its stated to use 0.

I'm lost on this, don't know which one to use :(

Do you think i've made some mistake in the measures, and should remade them?
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
I haven't had a chance to look at your files but it seems like you used a loopback of some kind. If it was a hardware loopback with an analogue mic then there should be no need to adjust the timing offset other than removing the fixed time of flight from the speaker to microphone distance which can be done in the measurement window as shown here.
VituixREW.png

Estimating IR delays on polar measurements is fraught with danger as it relies on the peak of the IR which is not always in the same place when you rotate the speaker.

Can you explain how you set up the measurements and what you did from a loopback and timing perspective?

The comb filter like dips suggest that there is a timing/phase issue between the measurements.

Also it is better to measure as close to full range as you can for the drivers and not cut them off as there can be interactions between them far outside the crossover points depending on the slopes being used.
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
Can you explain how you set up the measurements and what you did from a loopback and timing perspective?

The equipment used for measurements:

- Behringer UMC404HD
- Behringer ECM8000
- HPA D500
- Laptop

Connections:
  1. Mic to Interface => XLR to XLR cable
  2. Laptop to interface => USB cable
  3. Loopback => Main R OUT to Input 2 => XLR to XLR cable
  4. Interface to Amplifier => Main L OUT to Amplifier Channel 1 => XLR to XLR
shl6jhh-png.49499


The Soundcard was previously configured.

Regarding the "loopback and timing perspective", i've used the same variables as you did above, except i used a time offset corresponding to 2.6m, which was the distance i set up the microfone (i used a distance higher than 2m, to decrease the angle for the Ground plane measures):

fnhbma3.png



Here the setup in place:

o3N0NrK.jpg



I took the measures for the 15" without the angle, as it was impossible to put the 15" in place safely.
Erin's tests showed no difference for tilted measures in the frequency the 15" will play, so i decided to play it safe, even if some other variables won't be as reliable (potentially phase or distance between drivers (??), etc).
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
I'm lost on this, don't know which one to use :(

Do you think i've made some mistake in the measures, and should remade them?
One strategy that I think should work is to make the first on axis measurement without a timing offset. Then do the estimate IR delay and choose shift and update timing offset. Any subsequent measurements should then have the same time of flight offset included straight away.
 

3ll3d00d

Active Member
Joined
Aug 31, 2019
Messages
207
Likes
173
I was going back and forth between VituixCad instructions, and this post
Just use the vituixcad instructions, they work fine. You will notice that vituixcad adds tails to the response when the measurement is band limited like you have, I doubt the resulting phase response will be useful and the tails added are inaccurate anyway.

Using a single on axis measurement to get the actual offset as posted above is the way to go though just entering the mic distance is fine too. You can always adjust the offset later in vituixcad anyway.

Are you measuring each individual driver ground plane? If so, how do plan to get accurate delays for each driver? Will you measure the completed driver on axis with each driver separately and then use this to tweak per driver delay in vituixcad along with entering the relative positions of the drivers on the baffle?
 
Last edited:

3ll3d00d

Active Member
Joined
Aug 31, 2019
Messages
207
Likes
173
btw it is not clear how you determined the window position, it looks too short to me on the right side.
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
Are you measuring each individual driver ground plane? If so, how do plan to get accurate distances for each driver?

The idea was to measure with the full speaker.

Unfortunately, when using the ideal 9.9º angle (for a 2.6m mic placement), it was impossible to have the full speaker placed, so, for the AMT and 12, i measured this way:
o3N0NrK.jpg


This guaranteed accurate distances between the AMT and the 12".

For the 15", i had to put the speaker fully vertical:
A6eEZ5s.jpg


This may not have a fully accurate distance between the 15 and the other drivers, but i assume the difference would be small (probably assumed wrong).

it is not clear how you determined the window position, it looks too short to me on the right side.

I think by now, its clear that i somehow messed the measurements, and the best course of action is to repeat them (as painful as it is :( ).

Although using the tilted approach would be the best way to guarantee accurate measures, Erin's tests showed that this impacts mostly a small region around 2Khz to 4Khz.
index.php


I'm tempted to use a no tilt approach, even with the 2Khz to 4Khz, as it would guarantee accurate distances in the crossover areas (around 300Hz and 1500Hz).
Besides, that error will be easily corrected once Dirac is applied.

So, in a nutshell:

1 - Make the first on axis measurement without a timing offset (using the AMT).
Then do the estimate IR delay and choose shift and update timing offset, and leave it as is for all the next measures.

2 - Measure using a wider FR for each driver:
- 15" => 50 to 1000Hz
- 12" => 50 to 3000Hz
- AMT => 800 to 20000Hz.

Am i missing anything?
 

3ll3d00d

Active Member
Joined
Aug 31, 2019
Messages
207
Likes
173
I can't think of a reason not to sweep the bigger drivers full range. The lower bound to protect the AMT seems quite high too though I can't say I know how low it goes, I would try to go further personally.

How are you rotating it given that tilt? It looks tricky to do accurately. I don't understand the AMT measurement approach, it's just mounted a short distance off the floor. Why not just measure it on a stand? It seems like you have plenty of space to get perfectly good measurements in that way. Personally I would be relatively unconcerned about measurements of the woofer for crossing at 300Hz, actual response will be dominated by the room anyway so you are mostly just getting the overall level right. I would put my time into getting the T and M right and then worry about the W. I would also suggest getting a working speaker is a priority vs perfect measurements, i.e. work through the entire process end to end and resolve any gremlins as you go, once you understand the whole process, repeating it with improved measurements is easy (or at least easier)
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
I can't think of a reason not to sweep the bigger drivers full range. The lower bound to protect the AMT seems quite high too though I can't say I know how low it goes, I would try to go further personally.

I see, i'll do full sweeps for the 12" and 15", and the AMT from 300Hz, audioxpress has measured the Beyma TPL-200, and they measured it up to 300Hz.


How are you rotating it given that tilt? It looks tricky to do accurately.

Manually.
The MDF board bellow the speaker has the exact positions for each angle from 0º to 180º:

Lk78h92.png

I used a laser distance meter that also measures angles to guarantee that each iteration used the exact 9.9º angle.

I don't understand the AMT measurement approach, it's just mounted a short distance off the floor. Why not just measure it on a stand? It seems like you have plenty of space to get perfectly good measurements in that way.

I wasn't able to find a secure way to elevate the full speaker up to a reasonable height, moreover guaranteeing that the axis of rotation is correct.
When using a stand, the axis of rotation becomes the middle point on the stand, and not the center of the speaker.

I conducted a set of experiments between regular far field measurements and Ground plane, and it showed an almost perfect match between them.
For example, my last trial using a XTZ 99.26 speaker:

RWwBgV7.jpg


As such, its way easier to place the speaker on the floor, and rotate it to match the positions on the MDF board.


Personally I would be relatively unconcerned about measurements of the woofer for crossing at 300Hz, actual response will be dominated by the room anyway so you are mostly just getting the overall level right. I would put my time into getting the T and M right and then worry about the W. I would also suggest getting a working speaker is a priority vs perfect measurements, i.e. work through the entire process end to end and resolve any gremlins as you go, once you understand the whole process, repeating it with improved measurements is easy (or at least easier)

Yeah, i fully agree, i just want a working speaker, even if it isn't perfect.
Nevertheless, it seems that these measures contain some errors, so i'll take what was pointed me here, correct it, and redo the measures until the end of the week.
I don't mind redoing everything until i get it right, the only annoyance is the weight and size of the speakers, and have to move them from floors :(

Once again, thank you for having the time to look into this!
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
Another thing to consider: i just realized that on Erin's Ground plane measures, he has the tweeter as the driver near the ground:
pxMaaQH.png


Based on @3ll3d00d suggestion on focusing on getting AMT + 12" right, and the above, it would simplify measures a lot, as i would only need to use a 3.9º angle for 2m, instead of a 12.8º.

So basically measure the AMT + 12" from 0º - 180º, using a 3.9º angle, having the AMT near the ground:

wg1QvRK.png


And then measure the 15" in the speaker normal orientation from 0º - 180º (and it should be possible to tilt the speaker with the 3.9º angle)
fA3r7pv.jpg


Opinions? :)
 

3ll3d00d

Active Member
Joined
Aug 31, 2019
Messages
207
Likes
173
I believe the height of the driver under test relative to the height of the microphone off the ground drives the upper frequency limit of the accuracy of the measurement which basically means get the driver under test as close to the ground as possible. It means you definitely want the AMT close to the ground. I would measure the mid the same way in that case rather than as a combined unit and work out z axis offsets separately later.
 
OP
M

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
Are you suggesting to measure each driver on the floor, isolated from the others?
That would simplify measurement by a long mile.

What would be need to be adjusted in VituixCad so this could work?
 
Top Bottom