ok, my apologies, i am using my laptop now to read the posts. I now also read the note about the difference (red curve) i somehow failed to see on my phone.
This is the problem i have: when mountd in a box and measuring at 5mm from middle of cone, i measure no infuence whatsover from the baffle etc.
When increasing the measuring distance the baffle effect increases as well (as well as the room) , obviuosly up to some distance then it will stay even.
I miss this aspect in the sims shown.
Here a sim in VituixCAD of my GAYA2 baffle and upper woofer, using a on-axis measurement of the actual woofer ( 1m ):
As shown the difference is strongest at the low frequency, some difference up to about 1.5 kHz Basically measurement distance in this case should be at least 1000mm if conditions allow.
Note that this is a sim with a flat disk radiating, diameter is Dd, so limiting its omnidirectional radiation uward in frequency, and some rectangular volume behind the baffle. See tick box " Open baffle" not ticked. If this it ticked the difference is quite much more.
I also used some edge radius, value based on experience.
By the way: 300mm is the closest value VituixCAD will allow. And from about 25Hz downward it is a straight line , a choice by Kimmo (maker of VituixCAD).
If i do not use a driver measurment, just the disk, i can export that response and use it f.i. in Acourate:
A note: this shows the baffle width, the box width is 330mm (chopped elliptical shape crossection ), which would shift the effect a little bit lower in frequency. The cross-over for woofer is 433 Hz by the way.
As i am used to the baffle step is about 6-7 dB, in OP's sim it is much more - 9dB - or some reason.
This is the problem i have: when mountd in a box and measuring at 5mm from middle of cone, i measure no infuence whatsover from the baffle etc.
When increasing the measuring distance the baffle effect increases as well (as well as the room) , obviuosly up to some distance then it will stay even.
I miss this aspect in the sims shown.
Here a sim in VituixCAD of my GAYA2 baffle and upper woofer, using a on-axis measurement of the actual woofer ( 1m ):
As shown the difference is strongest at the low frequency, some difference up to about 1.5 kHz Basically measurement distance in this case should be at least 1000mm if conditions allow.
Note that this is a sim with a flat disk radiating, diameter is Dd, so limiting its omnidirectional radiation uward in frequency, and some rectangular volume behind the baffle. See tick box " Open baffle" not ticked. If this it ticked the difference is quite much more.
I also used some edge radius, value based on experience.
By the way: 300mm is the closest value VituixCAD will allow. And from about 25Hz downward it is a straight line , a choice by Kimmo (maker of VituixCAD).
If i do not use a driver measurment, just the disk, i can export that response and use it f.i. in Acourate:
A note: this shows the baffle width, the box width is 330mm (chopped elliptical shape crossection ), which would shift the effect a little bit lower in frequency. The cross-over for woofer is 433 Hz by the way.
As i am used to the baffle step is about 6-7 dB, in OP's sim it is much more - 9dB - or some reason.
First look at the time domain graph. This graph shows a superposition of 5 different puls responses. These are calculated responses from the described point source setup assessed along different axial distances. They consist of an initial pulse at time 0 and a second negative pulse at times where the sound pressure front reaches the baffle edge, thus experiencing a geometry change from half- to full space. This makes the sound pressure drop by half. Have a look at the matching step response:
Have a closer look now e.g. at the black and the blue pulse response.
The black pulse response is labeled 8.7 which is an index for a measuring distance of 8.7 * 34.3m corresponding to the listening distance of 3m.
The blue pulse response is labeled 4.0 corresponding to a practical measuring distance of 4.0 * 34.3m == 137cm.
Of course, the time for the pressure front to travel from center to the edge of the baffle is always the same. But observed from different distances, the second (negative) impulse, induced by the baffle edge, comes and is seen from different angles. Center, baffle edge and obersvation point form a triangle The larger the observing distance, the smaller the delta between kathete and hypotenuse. So at infinity the delay becomes min = D/2. And in the fictive center of the baffle the delay grows to max. D.
Consequently ...
The black pulse response (3m) shows a delay of 0.514ms for the 2nd pulse.
The blue pulse response (1.37m) shows a delay of 0.531ms for the 2nd pulse.
Now switching into the frequency domain.
To everyone of the 5 impulses there is a corresponding frequency response, and you see all of them oscillating between -6dB and +3.5dB along the frequency axis. This forms the "band" of sinefomred jaggies in the upper frequency range.
Now have a look at the graph labeled 4d, and focus on the black and blue frequency response again. Black is the frequency response measured at 3m, blue the one at 4d = 137cm.
The red graph shows the delta (=mismatch) of SPL between both the black and blue responses. It is accordingly labeled "Diff4-6" which means curve 4 (blue) - curve 6 (black)
So watch the blue, the black and the red (=mismatch) graph. In this example measuring at 4d = 137cm and at 300cm, the delta of SPL between both measuring locations and on the frequency scale will be
mismatch<<0.5dB below 1kHz
mismatch -0.8dB @ 1.75kHz
mismatch +1.2dB @ 2.12kHz
mismatch -2.0dB @ 3.63kHz
mismatch +2.3dB @ 4.02kHz
mismatch -3.1dB @ 5.52kHz
...
mismatch +5.8dB @ 11.7kHz
...
Therefore, a filter generated from this measurement at 137cm would not provide satisfactory results for a listening distance of 3m. If there was a 1:1 relation between theory and practice. Which there is not, nota bene.
The other graphs show the red mismatches correspondingly for the observation distances of 1d, 2d and 6d.
Yes! All is about learning by doing, and maybe DIY is one of the last resorts to be on personal paths. And once again, I talk about a theoretical model only. And maybe about it's implications for the practice. But I will not dig deeper into the swamp of psychoacoustics, individual perceptibility, objectivness vs. subjectiveness, usefulness vs. arts for arts sake and the like.
@Daihedz
You (and others) might find this a good read, particularly the section on measurement distance:
https://audioxpress.com/article/measurements-for-loudspeaker-modeling-files
You (and others) might find this a good read, particularly the section on measurement distance:
https://audioxpress.com/article/measurements-for-loudspeaker-modeling-files
The question i have given the document Charlie shared, is how much is the result off when measuring at say 1m versus 2.4m or 4.7m as the example in the article. (Sorry for editing twice, phone edit not easy)
Thank you for the interesting link. I was not aware of this writing. At first, I was confirmed that a PA professional would start from a quite severe point: Take the diagonal of the baffle, e.g. the biggest applicable measure. This is a hard criterium. I would have intuitively chosen something like the arithmetic or the geometric mean of both length and width. But ok, long live the diagonal as the most demanding starting point.
And then comes the rules of thumb, 3 fold the diagonal (80cm = 240cm) as a minimum measuring distance for "accurate data" up to a limit frequency of 5kHz, and 5.9 times the diagonal = 470cm for a limit at 10kHz. I was very eager to compare the "accuracy" claim with my theoretical workout. Surprize, surprize !!!
Transferring these values into my graphing scheme results into the following graphs:
Time domain:
2.40m / 5kHz rule of thumb (BD*3.0):
4.7m / 10kHz rule of thumb (BD*5.9):
BD_Inf stands for a theoretical observation point in infinity, because it is PA. In practical terms of graphing resolution, infinity starts at 10m or so.
Look at that, now. Both variants for the thumb rules end up with the first occurrence of worst-case maximum mismatches with a deviation of whooping +-9dB.
I had expected some mismatch, but certainly not this result. So either the rules of thumb are insufficient (hard to beleive because it's the pro's practicioner's rules), or the practical reality is much, much more benign than the results of my theoretical musings (I humbly tend to beleive in this latter case). This points to the evidence that my modelling is a mere theoretical outline of worst-case which will not be encountered in real-life. Nice to know ...