Questions Re - Measuring MTM for Crossover-Modeling in VituixCAD:

[Efalegalo]

Hi All,

Working on my first DIY MTMs and wanted to begin by first confirming the measurement technique that should be employed.

I’ve read through the “Measurement Preparation with ARTA” document, but still have a couple of questions (included below). I think I’ll be okay measuring the tweeter. My question(s) mostly pertain to measuring the two 4ohm mid-woofer that will share the enclosure volume and will be connected in series for an effective 8ohm nominal load:

1) When taking fair field measurements, should the amplifier be only connected to one of the two mids?
a. If so: Would SPL not be artificially high considering the amp will be feeling a single 4ohm load?
b. If not: If the mic in on-level will one of the two mids, would the sound from the second woofer not impact the measured response?
i. Or is one supposed to take the measurement with the mic placed mid-way between the two mids? If so, and assuming that is where the tweeter is placed and assuming that is the design axis, do you indicate 0x,0y,0z for specifying the diver location in the program?

I think the document was quite clear that the impedance measurement shall be performed exactly as how one plans on making the final connection (e.g., in my case, with the mids connected in series). It was also clear that, when taking near-field measurements, both mids should be connected and an attempt should be made to muffle the sound of one of the two mids.

My understanding from reading the document above was that only one of the mids in the MTM needs to be measured (assuming symmetrical vertical/horizontal locations). If that is the case, then how does one correctly enter the x,y,x coordinates? I’m specifically concerned with the y coordinates here.

For example, if you really only measured the top woofer, the 10 degree positive vertical measurement would be different for the top woofer than the measured 10 degree positive vertical response for the bottom woofer. In my mind, measured 10 degree positive vertical response of the top woofer would match the 10 degree negative vertical response of the bottom woofer. Do you tackle this by reversing the file names? For example change “_ver_deg+10” to “_ver_deg-10”?

I’m sure I’m over complicating things. Just trying to understand how the software treats measured inputs.

[draki]

- Connect and measure one driver only. Measure at drivers' height (i.e. x,y,z all zero) ;

- In the simulator you will use two M drivers, so the total SPL will reflect this;

- Note: since the impedance is measured with both units in series (as it should) you will have to scale the impedance (serial of two) back down by a factor of 0.5 for each of the M drivers;

- Each unit is measured at its' level (axis) and then in the x/o schematic you define all x,y and z's as per actual layout on your baffle. Kimmo's preparation for measurement is actually very clear on this. By defining x,y and z in relation to the designing point (usually at tweeter level - meaning it has y=0) the program "knows" that the upper M has positive and the lower has negative angle inclination. Providing you have measured the drivers multi-axis, at 10 degrees step, as recommended;

[ergo]

I don't have the time to do a flow with screenshots today, but I'll try to describe process I've used for latest 2.5 way (it does differ a somewhat from Kimmos VCAD preparation doc)

*I'm measuring all drivers on design axis at 1m without moving the microphone as driver spacing is relative small (for this project mic between the mid and tweeter in vertical axis as that's where the ear height is sitting on the couch)
*I first set the amplifier output - in this case it was 1Vrms, so it was safe to measure tweeter without any protection capacitor. Best to play a test sweep with a midwoofer before connecting tweeter for test fo safety..
*Then I set the microphone gain, so that the loudest driver (usually the tweeter) does not clip the input
*Then measurements are started for tweeter - 0 degrees horizontal and then upto x off axis (best is until 180deg as VCAD expects)
*Then I shorted the lower woofer and connected upper woofer for measurements and repeated the 0deg and all off axis. Shorting the other woofer helps to work a bit against it becoming a passive radiator.
*Then I shorted upper woofer and measured 0deg and all off axis
*Then connected Tweeter and upper woofer in parallel (lower shorted) and measured the summed graph. This one will be used for setting the Z coordinate in VCAD later
*Then connected Tweeter and upper woofer and lower woofer in parallel and measured the summed graph. This one will be used for setting the Z coordinate in VCAD later
*As a last measurement step a nearfield measurement is done on one woofer and port is there is one.

With this all the measured data is collected. Now as VCAD has the nice 'Convert IR to FR' tool I use that to gate all the above and export the frequency response files as Preparation doc also guidelines.

In VCAD under options one should now make sure that the "Listening distance" parameter is set to 1000mm as we measured at 1m and this is where we want to check the simulation match with measured curve.

Next step will be to enter the tweeter data (freq. resp and impedance) and set the "Delay" vlaue on drivers tab so that the tweeter phase approaches 0degrees at 20kHz.
NB! the same delay value must be used for upper and lower woofer later also.

Now also the upper woofer data gets entered and the Y axis set to whatever is the real physical distancce of tweeter center to upper woofer center on front panel. This results in a summing curve. It is time to import the "Tweeter and upper woofer in parallel" measurement result as an overlay. There will be a small mismatch between measured and simulated curve. To fine tune it the upper woofer Z axis coordinate should be used for best match.

Once above is done I add the lower woofer data and Y coordinate. Again it results in summing curve of the 3 drivers in parallel. Again an overlay will be loaded - this time the "Tweeter and upper woofer and lower woofer in parallel". Now a Z coordinate of lower woofer is adjusted until the match is as good as possible between the simulated and measured curve. Maybe upper woofer coordinate needs a small nudge to one or the other direction too.

Below is an example image of the 3 driver summing and the overlay curve. Good match is found down to 250Hz. After achieving this there is higher confidence that the simulation will match reality.

I will then continue pretty much step by step as the preparation doc describes for Merging the low end nearfield responses, baffle step simulation and 1m far field measurement. Once done I will replace the short gated response files for upper and lower woofer with merged data. Tweeter data is correct as is already.

After this it is off to designing the Xover itself.

PS. Please see also Kimmo's recommendation below that is to measure each driver on it's own axis instead as recommended procedure!

[ergo]

Attached a wrong screenshot.. this one is the one I meant

[kimmosto]

Thanks guys for helping with these questions.

Maybe you've already noticed that ergo's list has few "extra" items and differences compared to recommendation. Quick comments for each group:

*Then I shorted the lower woofer and connected upper woofer ...
*Then I shorted upper woofer and measured 0deg and all off axis

Only one woofer can be measured if the others are identical.

*Then connected Tweeter and upper woofer in parallel (lower shorted) and measured the summed graph. This one will be used for setting the Z coordinate in VCAD later
...
*Then connected Tweeter and upper woofer and lower woofer in parallel and measured the summed graph. This one will be used for setting the Z coordinate in VCAD later

In VCAD under options one should now make sure that the "Listening distance" parameter is set to 1000mm as we measured at 1m and this is where we want to check the simulation match with measured curve.

Now also the upper woofer data gets entered and the Y axis set to whatever is the real physical distancce of tweeter center to upper woofer center on front panel. This results in a summing curve. It is time to import the "Tweeter and upper woofer in parallel" measurement result as an overlay. There will be a small mismatch between measured and simulated curve. To fine tune it the upper woofer Z axis coordinate should be used for best match.
Once above is done I add the lower woofer data and Y coordinate. Again it results in summing curve of the 3 drivers in parallel. Again an overlay will be loaded - this time the "Tweeter and upper woofer and lower woofer in parallel". Now a Z coordinate of lower woofer is adjusted until the match is as good as possible between the simulated and measured curve. Maybe upper woofer coordinate needs a small nudge to one or the other direction too.

Those can be done just like ergo did, but I never verify with parallel measurements and sum calculated with simulator, or recommend this procedure due to extra work. Another problem could be drivers; some drivers have so steep natural slopes that sum measurement and simulation don't give much sense. Skipping of this "extra" work requires that distance from mic to rotation center of driver is measured carefully with tape. 15 mm error in measurement distance of tweeter could cause significant trouble so take your time while setting positions of mic and speaker on rotation table to enable skipping of extra work.

*I'm measuring all drivers on design axis at 1m without moving the microphone as driver spacing is relative small (for this project mic between the mid and tweeter in vertical axis as that's where the ear height is sitting on the couch)

As ergo mentioned, this works directly with compact speakers, typically with small 2-way or 2.5-way if far field of lower woofer is not measured. But normally this is not possible with 3...5-way, horns, large radiators or big/complex constructions so simplest is to recommend that each driver model is measured at center elevation (driver Y = mic Y).

Next step will be to enter the tweeter data (freq. resp and impedance) and set the "Delay" vlaue on drivers tab so that the tweeter phase approaches 0degrees at 20kHz.

This is not needed anymore. Convert IR to FR has Far 1 button which sets timing of tweeter close to minimum phase. Reference time field allows adjustment with 1 us resolution. Show MP response before adjustment. Other drivers follow that because IRs are converter to FRs and exported to txt/frd with the same Reference time with the tweeter, assuming that all far field measurements are done at 1000 mm from baffle surface.

[ergo]

I have that "comparing the simulated sum to measured sum" habit from past when I mostly used LspCAD for the task. It is a bit of extra work indeed but somehow makes me more confident there was no silly mistake with driver polarity etc.

"This is not needed anymore. Convert IR to FR has Far 1 button which sets timing of tweeter close to minimum phase. " - aha yes. I did notice that once I've started using the Convert IR to FR I need to enter a much smaller value for the delay. But then indeed I could fine tune it already in that tool.

[Evil Twin]

A very lucid explanation and discussion of useful methods...

At some point, with enough theoretical knowledge and experience, you may find your own methodology and path.

The longer processes and additional work lessen the possibility of overlooking issues you may not be fully cognizant of. Also, measuring both drivers in a situation like this may not be needed, but then again, if the cabinet design and driver positioning is such the the drivers baffle loading differs, then their response may differ in subtle but possibly important ways.

When in doubt, use the more rigorous and detailed process until your level of experience supports the formation of valid judgements as to when it is safe or non-critical to simplify...

In general, Ergo is a cautious and meticulous developer, and this shows in his measurement processes.

[Efalegalo]

Thanks, Ergo/Draki,

To best of my ability, I would like to stick to exactly how Kimmo recommends taking measurements. I too used to develop my measurements as Ergo layed out (i.e. taking all measurements at the design axis).

To me, when reading the document, the confusion arises when it recommends that you only need to take measurements of one Mid (again assuming vertical/horizontal symmetry). This to me makes total sense and certainly helps cut down additional steps.

BUT - I'm still confused on how I then specify the mid driver locations on the baffle.

As an example, lets say:

1) Tweeter is the design axis (X=0mm, Y=0mm)
2) Enclosure is 300 cm tall and 200cm wide, tweeter at 150cm and 100cm (X=0mm, Y=0mm)
3) Mid 1 is at 225cm and 200cm (or X=0mm, Y=+75mm from design axis)
4) Mid 2 is at 75cm and 200cm (or X=0mm, Y=-75mm from design axis)

The document suggests I only needs to measure 1 of the Mid - lets say I choose the measure only the top one = Mid 1

In the horizontal plane, I can understand the measurements from Mid 1 will be identical to measurements for Mid 2.

However, when taking measurements in the verticle plane, I think the ver+15 measurement (just as an example) for Mid 1 would not be identical to Mid 2, because...

When taking the measurement of Mid 1 in the ver+positive direction, the frequency response impact resulting from the edge of the enclosure (top) will only be 75mm away.
When taking the measurement of Mid 2 in the ver+positive direction, the frequency response impact resulting from the edge of the enclosure (top) will be 225mm away.

Given the mirror locations, the measured response of the Mid 1 ver+15 should (in my mind) mirror the Mid 2 ver-15 position.

So, I don't think the predicted response will be accurate if I enter Mid 1 (x=0,y=+75mm) and Mid 2(x=0,y=-75mm).

I'm sure I'm missing something very basic?!!?

[morbo]

Just stopping in to say thank you to everyone contributing to this discussion, it is very useful to me and I'm sure others who are new to vituixcad or who are migrating to it from another tool. I will certainly be referencing this thread when I'm doing my own measurements and filter designs in a few weeks.

And not to be forgotten; huge thanks to Kimmo for making this software available to all

[kimmosto]

So, I don't think the predicted response will be accurate if I enter Mid 1 (x=0,y=+75mm) and Mid 2(x=0,y=-75mm).

I'm sure I'm missing something very basic?!!?

You have not missed anything. It's just question how much you are willing to use time and work for measurements to improve accuracy for example max. 0.6 dB in power response and 0.0 dB in axial and listening window average. Measurement method is already quite complex for majority so we are naturally interested in how many measurements and to what directions are needed to get "adequate" accuracy, and what kind of baffle is "symmetrical enough".
As preliminary spoiler I can say that I never measure vertical plane and negative horizontal angles if baffle is horizontally symmetrical and drivers are circular. No matter price tag of final commercial product. Rectangular radiators H!=W have to be measured in vertical plane of course.

Easiest method to evaluate what amount of measurements and directions are relevant producing "adequate accuracy" is using Diffraction tool. It's not accurate either, but gives interesting information for decisions, because possible error is generated by diffraction only (with circular radiators).

Here is comparison of extreme cases with your baffle dimensions: horizontal 0-180 (19 directions) of single (top) driver versus full dual plane data for both drivers separately (144 measurements). Angle step 10 deg. Sd=75cm^2 and edge rounding 10 mm. Thin overlay lines in Power&DI chart are full dual plane data (144 meas) and thick lines are half horizontal of top driver only (19 meas). Please ignore two top octaves because this is mid-woofer comparison.



Full data:


Hor 0-180 of top M only:


As we can see, max. error in power response is about 0.3 dB around 2 kHz. Polarmaps in vertical plane visualize where that difference is coming from.
Funny background detail is that measuring vertical plane of single driver only probably increases error compared to horizontal only, because MTM construction is vertically symmetrical and M drivers compensate each other.

Some cases are worse that this. For example tall and narrow box produces more hump (~0.6 dB) to power response at baffle step hump range because diffraction effects are not so diffused in vertical plane. Axial and listening window are okay. Most common reason to skip vertical plane measurement is that it's usually terribly difficult mechanically. This result does not increase motivation to measure it.

[kimmosto]

^Previous message is fixed too many times, but hopefully it's okay now. Hawk eyes have noticed that difference in DI curves at top octaves is a bit suspicious. I need to investigate what causes that.

[kimmosto]

I have that "comparing the simulated sum to measured sum" habit from past when I mostly used LspCAD for the task.

LspCAD uses the same coordinate and timing system than VituixCAD so basically measurement data should be captured with the same "rules" for both if target is to simulate also directivity in full 3D space. As far as I remember, IJ has not made any or at least as strict recommendations to produce measurement data for LspCAD as I have done for VituixCAD. Maybe he has not designed speakers with so clear directivity and power response perspective, or he is not as naive, stubborn and optimistic as I am, thinking that I could finally bury decades old bad tradition to measure and simulate to single spot only. That tradition is still clearly visible on discussion forums, and sometimes the same questions drop into my inbox, but I think this heavy and long boat has started to turn.
With this background, I will not give permission to measure at single spot other that two small drivers in compact group to keep instructions as simple and logical as possible. Frequency range and resolution targets, avoiding reflections and including diffractions related to IR time window is so that W and M should be measured at least as far from reflecting surfaces as T so lifting of speaker for W and M measurement to maximize gate time is one of the most logical things to remember and do.
Measuring at the same distance and rotating around center point of each radiator captures cone break-up and avoids scaling of SPL and timing i.e. locates measurement data to center point of radiator in every way. That is basic requirement needed to build virtual speaker with individual driver locations into the simulation, and isolate normal listening i.e. virtual mic distance setting from measurement data to allow 3D simulation to any radius.
Fortunately we don't have to measure all drivers at the same distance. If measurement distance varies, easiest is to set reference time in millimeters to Convert IR to FR or measurement window with REW. But then manual SPL scaling is also needed.

Quoted "comparing the simulated sum to measured sum" could be valuable to verify that measurement gear, software options and tape/laser have correct settings and speed of sound is expected 344-345 m/s. I'm not saying that it's bad. I just don't do it in my overwhelming self confidence

[ergo]

LspCAD uses the same coordinate and timing system than VituixCAD so basically measurement data should be captured with the same "rules" for both if target is to simulate also directivity in full 3D space. As far as I remember, IJ has not made any or at least as strict recommendations to produce measurement data for LspCAD as I have done for VituixCAD.

Your reasoning on measurements point makes sense and I shall try it out with the next project.

Ingemar (LspCAD) actually had something a bit similar, but it was a long time ago (2002)
The Ugly Duckling design doc
This is a step by step into to a 3 way project with LspCAD 5.x walking through the process step by step.
Back then LspCAD did not have the offaxis simulation yet. So indeed with progress meanwhile, I should also adjust my thinking according to new tools.
Ingemar also points out in the doc that this comparison to summed is rather to be 'pretty sure' about outcome and not as a mandatory step.

[Efalegalo]

Wow - Thank you Kimmo for the detailed response. Very helpful.

I'm glad I wasn't just losing my mind (lol). I understand now that taking all these additional measurements (beyond whats recommended in your Measurement directions) doesn't materially improve accuracy.

As a shortcut, if one was taking vertical measurements, can one not simply copy-paste (in Windows explorer) and just re-name the copied files with opposite direction indicator (i.e. change all the "+" to the "-", and vice versa? This should only take 2 seconds and will remove the need to take extra measurements of the second Mid. And follow this by specifying the correct driver location (i.e. Mid 1: x=0,y+75mm and Mid 2: x=0,y=-75mm in my example above).

I hope I'm not wasting anyone's time by asking such elementary questions. Most respondents here have actually built/designed speakers and have that practical knowledge. I, on the other hand, have only designed one simple 2-way, and thus can only approach this from a theoretical perspective. The genesis of my inquiry came from me researching the benefits of an MTM design and the importance of center-to-center spacing (between the mid and tweeter, and between the two mids) - and the overall ability of the MTM design to create a (relatively more) uniform response (compared to a typical two-way) in the verticle plane, or its ability to purposefully create a large frontal-lobe (with suckout/cancellation above and below the lobe) to minimize floor and ceiling reflections.

Laslty - I too would like to sincerely thank Kimmo for VituixCAD. I've learned more about speaker design from simply learning and trying understand the software than I have from any other source. Its just an incredible tool.

[kimmosto]

can one not simply copy-paste (in Windows explorer) and just re-name the copied files with opposite direction indicator (i.e. change all the "+" to the "-", and vice versa?

Sure, that is the fastest method to produce perfectly mirrored vertical data. I haven't tested which file utility would be easiest for this. Also VituixCAD could have sign swapping for vertical angles, but value might be quite low.

[Efalegalo]

Also VituixCAD could have sign swapping for vertical angles, but value might be quite low.

Agree - one can easily accomplish this on their own with minimal time/effort. Just glad to hear its a viable option.

[ergo]

I'm using Total Commander for such things (also free version will work), but perhaps there are easier tools

[kimmosto]

I've used TC almost daily for more than 20 years but Multi-rename maybe once or twice so I did not remember whole feature anymore Norton Commander was cool in the middle of 80s' so I had to make my own. Name was kc.exe of course 8)

[tcpip]

About MTM measurements: will it be good if we connect both midbass drivers the way we finally intend to (i.e. series or parallel) and take measurements of both drivers together, assuming the midpoint of the two driver centres as the acoustic centre of the "virtual midbass"? In other words, I'm suggesting that we treat the two midbass drivers as a single virtual midbass driver for measurement purposes, with its acoustic centre located at the midpoint of the two acoustic centres of the two physical drivers. Then we proceed to take vertical and horizontal measurements off-axis as we would do for a normal physical driver.

With this approach, any asymmetry with one of the drivers will be factored in into the final readings.

[kimmosto]

^Measuring together has also advatages, but measurement of single driver is recommended due to distadvantages:
- Cone break-up won't be captured without separate measurement from single driver or without adequate measurement distance which equals to axial 0 deg of single driver.
- Directivity of vertically lobing driver pair is captured as it is at the measurement distance. Simulation does not react to change of Listening distance in Options because pair is single driver in the crossover.
- Individual drivers travel in Z direction back and forth while vertical rotation causing travel of data back and forth in impulse response. This causes variation to effective gate length more than while measuring single driver. Some data of driver further from mic drops out of time window -> response is too smooth and drop early at LF. More room space and longer time window is needed to eliminate this.
- Cannot simulate mechanically adjustable modular constructions because measurement data is fixed for two drivers with certain positions.
- Cannot simulate 0.5 crossovers or attenuation and phase shifting of the other driver because driver pair must be single driver in the crossover.

Advantage(s):
+ Single vertical off-axis measurement sequence is okay assuming that measurement distance is adequate for simulation at normal listening distance.

[Evil Twin]

Wow - Thank you Kimmo for the detailed response. Very helpful.

I'm glad I wasn't just losing my mind (lol). I understand now that taking all these additional measurements (beyond whats recommended in your Measurement directions) doesn't materially improve accuracy.

As a shortcut, if one was taking vertical measurements, can one not simply copy-paste (in Windows explorer) and just re-name the copied files with opposite direction indicator (i.e. change all the "+" to the "-", and vice versa? This should only take 2 seconds and will remove the need to take extra measurements of the second Mid. And follow this by specifying the correct driver location (i.e. Mid 1: x=0,y+75mm and Mid 2: x=0,y=-75mm in my example above).

I hope I'm not wasting anyone's time by asking such elementary questions. Most respondents here have actually built/designed speakers and have that practical knowledge. I, on the other hand, have only designed one simple 2-way, and thus can only approach this from a theoretical perspective. The genesis of my inquiry came from me researching the benefits of an MTM design and the importance of center-to-center spacing (between the mid and tweeter, and between the two mids) - and the overall ability of the MTM design to create a (relatively more) uniform response (compared to a typical two-way) in the verticle plane, or its ability to purposefully create a large frontal-lobe (with suckout/cancellation above and below the lobe) to minimize floor and ceiling reflections.

Laslty - I too would like to sincerely thank Kimmo for VituixCAD. I've learned more about speaker design from simply learning and trying understand the software than I have from any other source. Its just an incredible tool.

A reminder - a very useful way to thank Kimmo is to donate. I frequently come across a new function or one I can use to solve a new problem, and I believe it is only fair to reward the creator...

[tcpip]

I agree with your points. I agree that my joint-measuring approach is only useful if the two drivers will be treated as a single driver in the crossover, therefore no chance of designing 0.5 type crossovers.

However, I see the following challenges if I measure the drivers separately:

• I'll need to cut a precise block of circular ply and block one hole when I measure the other driver. This blocking disc will have to be thick enough and not vibrate.
  
• my impedance measurements will be wrong, and it's not just a matter of scaling the data points. This is because I'll be measuring one driver into 30 litres whereas I'll be playing music with two drivers into 30 litres, therefore I presume the resonant peak in the two cases will be different. If I do separate drivers for SPL measurement and joined-drivers for Z measurement, can I simply mix and match like that?

Also, I did not understand your point about not being able to measure cone break-up if I do joint-driver measurement. I can clearly see cone breakups in the joint-driver measurements I have done, and the SPL curves match visually quite closely with manufacturer published data.

[draki]

If the MTM enclosure is internally divided in two halves (or made of two stacked separate boxes), then each M could be measured independently. Both acoustical FR and the impedance. That's what I do.

[Efalegalo]

A reminder - a very useful way to thank Kimmo is to donate. I frequently come across a new function or one I can use to solve a new problem, and I believe it is only fair to reward the creator...

Ditto - I donated just this last week.


Sent from my iPhone using Tapatalk

[tcpip]

If the MTM enclosure is internally divided in two halves (or made of two stacked separate boxes), then each M could be measured independently. Both acoustical FR and the impedance. That's what I do.

Yes, of course, that sort of internal design would completely solve my problems.

[kimmosto]

* I'll need to cut a precise block of circular ply and block one hole when I measure the other driver. This blocking disc will have to be thick enough and not vibrate.
* my impedance measurements will be wrong,

No. Blocking of the other driver is not needed (and not recommended) while far field measurement of single driver. Just do not connect it to amplifier. Wires could be shorted too.
Near field needs muffling of the other driver to damp mids because both have to be connected to amp. Also impedance measurement needs both without damping, and speaker located to free/normal listening position to get normal acoustic load.

I can clearly see cone breakups in the joint-driver measurements I have done

Sure in your case, but understand that it does not represent globally everything possible what I have to cover with my instructions. Cone break-up could be very narrow and 1m measurement distance with much larger drivers than 5" is different case.

Basically I don't care if you disagree and what you all do privately at home, but do not try to force me to explain same things 100 times here and there and e-mail.

[tcpip]

No. Blocking of the other driver is not needed (and not recommended) while far field measurement of single driver. Just do not connect it to amplifier. Wires could be shorted too.

Won't the passive driver then act as a passive radiator? Won't this affect the performance of the driver being measured?

Near field needs muffling of the other driver to damp mids because both have to be connected to amp.

How should I do the muffling in that case?

Basically I don't care if you disagree and what you all do privately at home, but do not try to force me to explain same things 100 times here and there and e-mail.

I was seeking clarification, not disagreeing. You sure have a strange way of handling clarifying questions. No one is forcing you to explain anything.

[kimmosto]

it does not represent globally everything possible

Height of the latest MTM part what I've designed is 780 mm. 10" mids and 230 mm tweeter so be confident that M is measured as single driver to avoid for example 14 deg error in axial direction.

[kimmosto]

Won't the passive driver then act as a passive radiator? Won't this affect the performance of the driver being measured?

Far field is merged at midrange typically 400...500 Hz where PR effects are not usually significant. Damping material in the middle of the box at T's elevation is worthy anyway reducing also leak from M to M. Damping layer could be quite heavy if the box is vertically symmetrical, preventing strong PR effect also at LF.

How should I do the muffling in that case?

Read instructions please.

I was seeking clarification

Hopefully you've got it. I have noticed tens of times that it's impossible (at least for me) to write instructions that everyone would read, understand and follow(~believe). Sorry if you had already read measurement instructions and it's not clear enough. But it did not look that because you asked damping of lower M and impedance meas so this looks unnecessary repeating from my point of view.