Hi Bill nice writeup. Most of the steps are portable between programs I believe. if you have the XSIM screenshot that you speak about in step 23 that would be great.
Another question i have is regarding the floor bounce. Having a large 3 way with the woofer close to the floor will augment the boundary gain and this will be picked up by the microphone. But how to retain this information if the near field will be merged with the farfield? Would the far field measurement retain enough of the boundary reinforcement - The near field would not have this data.
I know SoundEasy will add boundary reinforment sim to your baffle diffraction, you tell it how far the speaker is from floor, back wall and side wall and it makes some wiggles. Interesting to compare a woofer on the floor vs hiked up a couple feet.
If I recall Jeff's diffraction spreadsheet has some options for boundary reinforcement but were pretty basic. Can't speak for other baffle sims as I pretty much stick to SoundEasy these days, but maybe check what VituixCAD can do.
The real question is how much BSC to use if the woofer is on the floor. Is it 0, 3, or full 6dB? It's a tough call but without the extra boundary reinforcement included in your measured data I would lean towards 3-4dB over a full 6. The problem is that unlike baffle diffraction sim, boundary reinforcement isn't a smooth slope that's easily compensated for, it's a compounding reflection that makes the bass response pretty wiggly as I'm sure you know if you've ever taken in room measurements without the gating.
I agree db. John H once told me, and I have also found out for myself, that the low to the floor woofers create as many issues (muddy bass) as they fix. Juggling act like everything else in this great hobby of ours.
A little room EQ can really save the day to help with room modes at about 200Hz and below. And if you have the EQ you can compensate for any errors in BSC with it as well
The modern approach is to use multiple subwoofers in different room locations to even out the room modes. It does work, but I still like the idea of a big 3way over a smaller 2 way and subwoofers.
Except for WAF, I agree. Big woofer 3 ways have always sounded musically better to my ears. That said I loved my subwoofers in my car stereo systems and liked my subs when I had a 5.1 HT system. But for my pure music system no.
Another question i have is regarding the floor bounce. Having a large 3 way with the woofer close to the floor will augment the boundary gain and this will be picked up by the microphone. But how to retain this information if the near field will be merged with the farfield? Would the far field measurement retain enough of the boundary reinforcement - The near field would not have this data.
Short answer is yes. For crossover work, the woofer's far field measurement sets the woofer's level that includes the boundary dB gain. Floor bounce is a secondary arrival that you will gate out of the far field. When you blend the woofer's near field you will adjust the near field level to match the far field with the gain.
So I am trying some compare what I get from SE to what you get from VituixCAD or Jeff's diffraction spreadsheet.
In SE, you have x,y,x coordinates for the speaker placement, as well as a "reflection coefficient" to tell it how much influence the reflections have in the response. 1.0 would be 100% reflection, so I simulated with a coefficient of 0.5, or 50% reflection, 50% absorption.
In VituixCAD you have options to add in floor and side wall reflections, and it also has a "absorption" value in dB. I found that 3dB absorption here roughly matched up with the 0.5 coefficient used in SE. VituixCAD does not include the rear wall in its calculation.
I first verified that the diffraction simulation was fairly similar between SE and VituixCAD without the boundary reinforcement. VituixCAD is very smoothed, and I don't know where or if I can change that, but the results are otherwise similar.
I then simulated with boundary reinforcement added. I simulated with the same baffle, driver location, mic location, and same boundaries. I put the speaker at floor level, and 1.5m from the side wall. I set the rear wall boundary to zero in SE. The results are different, but also similar. VituixCAD shows a lot more ripple, but in general the response below 1kHz is of similar shape.
I then added in some rear wall distance in SE of 0.7m to see how much that affected things. You can see the results in the screenshot below.
SE:
red line - no reinforcement, just raw baffle step
green line - reinforcement added, zero distance from rear wall
blue line - reinforcement added, 0.7m from rear wall
VituixCAD:
faint brown line - no reinforcement, just raw baffle step, decent match with SoundEasy
blue line - reinforcement added
So with all that done, there's a few takeaways. VituixCAD I think is missing a large component which is the rear wall, and shows a lot more ripple than what I get in SE. I do get a lot of ripple in SE if I increase the reflection coefficient so that the walls are very reflective, but its rather smooth at the low levels used. Which leads to the next big question - what coefficient is right for your room? I will have to do some digging to see if there's a standard value of reflection vs absorption for a normal hollow gypsum wall and an insulated gypsum wall. This coefficient has great effects on the results.
I may try to re-do this test a little down the road with a real speaker and do some comparisons to real in-room measurements. Mind you the in-room measurements will include more than 3 walls but it is good to have some frame of reference to what the simulation does vs what I see in reality.
Later I will post back with some info of how Jeff's Diffraction spreadsheet compares, as well as a simulation in SE of what happens when you raise the speaker up off the floor.
Now, it lists an absorption coefficient of 0.1 for gypsum, one would think that the reflection coefficient that I would enter to SoundEasy would then be 0.9, but that wouldn't necessarily be correct. There are actually 3 components at play with any material, reflection, absorption and transmission. Now we all know that the drywall will transmit sound, and this factor will vary with frequency, which raises a flag as to the accuracy of such a simulation that relies on only a reflection coefficient or absorption value.in reality the reflection coefficient is likely to vary over frequency slightly.
The other flag raised with such a simulation is that the wall and floor are usually different materials, yet we enter a single coefficient for everything. I think these simulations may be good as a guide, but not the best tool as an accurate replication of real world situations. At the very least if the data input is correct we can get a good idea of the amount of BSC needed based on a floor level woofer, but don't read into the wiggles of the chart too much.
I tried entering the same data to Jeff's boundary simulator and it
provides similar results, however does not have any ability to adjust
the reflection coefficient so all results there assume a 100% reflective
surface, so it appears to over-exaggerate everything a bit.
Ok, last one for tonight. Here I took the same conditions as above, and plotted at 1ft increments. I adjusted the scale to 1dB/div instead of the 5db/div to really zoom in on the difference.
Red - floor level
Green - 1ft off the floor
Blue - 2ft off the floor
Purple - 3ft off the floor
It's surprising, that at 1ft all the gain above 200Hz is pretty much gone, and at 2ft all the gain above 150Hz is pretty much gone.
I don't know how much I believe the gain above 1kHz. If I lift the speaker 2 inches off the ground the gain above 1kHz goes away.
Anyway, back to the topic at hand, how much BSC should we use when the woofer is on the floor? If we make some assumptions, let's say our walls and floor average out to about 10% absorption, 20% transmission, and 70% reflection. Here's the comparison, here I lifted the speaker 2 inches off the ground to show the loss above 1kHz, but looking at the gain all the way up to 1kHz one would have to conclude that you still need 6dB of compensation since there is still a 6dB slope from 1kHz to 100Hz. Hmm...
Regarding room boundary reinforcement, baffle step, and overall room cabin pressure gain, the program I like to use is Jeff Bagby's Baffle Diffraction and Boundary Simulator (BDBS) ver. 1.3 found here:
The baffle diffraction portion of BDBS is the same as that found in the Blender program. The boundary simulator portion of BDBS uses the Allison room boundary reinforcement equations. If you place your woofer near the floor in the model, the program will show how the woofer's spl is boosted by the floor. Move the woofer higher, and you can watch how the floor boundary reinforcement effect on the woofers spl changes with frequency. It is a nifty little program, but, as dcibel points out, it does not have adjustments for wall or floor material absorption.
Attached is a comparison graph showing how the low frequency response of a speaker changes as you adjust the distance to the floor from 4" all the way up to 36". I kept the rear wall at 8 ft and the side wall at 12 ft so that you can see the floor boundary effect more clearly. If the woofer is 4" off the floor, then no BSC is needed. If the woofer is 24 to 36" above the floor, then you need to apply the full 6dB of BSC ( but only if the speaker is well away from the rear and side walls). If the woofer is 10" to 16" above the floor, then you probably need roughly 3dB of BSC.
These results are consistent with my experiences with my Kowaxial speakers. For those that attended InDIYana a couple years ago, when these speakers played almost everyone agreed that they were very "dark" sounding with "waaaay" too much BSC. This was because the woofer was way down near the floor and I, in error, was applying 4-6dB of BSC. I measured the design FRD's with the speaker on top of my 17" high DIY horizontal polar table. I should have done the measurements with the speaker placed on the floor. It makes a huge difference. Even though the FRD is bouncing all over the place below 400Hz, you can clearly see how the average spl below 400Hz changes as you move the speaker higher or lower.
Yes, and that would probably have worked out quite well in this case because the CX150 coaxial could have handled a lower xover with ease. The only reason I went with a higher xover was because I did not want to shell out the cash for the big inductors and caps that would have been required to do this. So this was both a mental and monitary mistake on my part. Live and learn.
Regarding room boundary reinforcement, baffle step, and overall room cabin pressure gain, the program I like to use is Jeff Bagby's Baffle Diffraction and Boundary Simulator (BDBS) ver. 1.3 found here:
The baffle diffraction portion of BDBS is the same as that found in the Blender program. The boundary simulator portion of BDBS uses the Allison room boundary reinforcement equations. If you place your woofer near the floor in the model, the program will show how the woofer's spl is boosted by the floor. Move the woofer higher, and you can watch how the floor boundary reinforcement effect on the woofers spl changes with frequency. It is a nifty little program, but, as dcibel points out, it does not have adjustments for wall or floor material absorption.
Attached is a comparison graph showing how the low frequency response of a speaker changes as you adjust the distance to the floor from 4" all the way up to 36". I kept the rear wall at 8 ft and the side wall at 12 ft so that you can see the floor boundary effect more clearly. If the woofer is 4" off the floor, then no BSC is needed. If the woofer is 24 to 36" above the floor, then you need to apply the full 6dB of BSC ( but only if the speaker is well away from the rear and side walls). If the woofer is 10" to 16" above the floor, then you probably need roughly 3dB of BSC.
These results are consistent with my experiences with my Kowaxial speakers. For those that attended InDIYana a couple years ago, when these speakers played almost everyone agreed that they were very "dark" sounding with "waaaay" too much BSC. This was because the woofer was way down near the floor and I, in error, was applying 4-6dB of BSC. I measured the design FRD's with the speaker on top of my 17" high DIY horizontal polar table. I should have done the measurements with the speaker placed on the floor. It makes a huge difference. Even though the FRD is bouncing all over the place below 400Hz, you can clearly see how the average spl below 400Hz changes as you move the speaker higher or lower.
Thanks for this info. I am doing the same thing right now, a 3way with a 10" woofer on the floor and mid and tweet some 40" up. I am going to take some real measurements this weekend but I wanted to play around with the simulations first to see what they did and then see if the boundary simulation has any resemblance of the real world measurements.
Sooooo, I got some measurements, still have to process them and run them through the Blender. But have a couple of questions. This is for a MTM with 4" woofers on a narrow 6.5" baffle.
This is the screenshot of it. The Blue is the NF port Response (with Both Woofers in Parallel), Red is the Upper Woofer NF. The green is from the mic at the speaker side, roughly at the center vertically and horizontally.
For PCD and Blending, the NearField with the Farfield, is the port response strictly required for the purpose of XO and BSC? It's good to have, but can it be done without?
Not for the response itself and the process of blending. From Jeff B's white paper on In Room Response:
Note: for dual ports, find the total port area of both ports and determine the effective diameter for the combined area. Now, take the measurement at only one port, find the ratio of the total port diameter to the cone and make the calculation and then add 6dB. So, for our example, if there had been 2 – 2” ports then the port output would be lowered by -7.96 + 6 dB = -1.96 dB. For dual woofers and a single port we will need to do it the other way around and subtract another 6dB from the port output.
Since this is a MTM, and I have measured one woofer NF, but measured the port with both the woofers, do I subtract 6DB from the port? Also, The distances to the cone and to the port are not exact. I was between 0.5-0.125 to the cone and just at the port entrance (plane of the port) - is this fine, or it needs to be more precise?
Another question:The next paragraph says how to determine the tuning frequency of the enclosure and port.
Here’s a tip – what is the best indicator of the tuning frequency (Fb) from this data? The peak in the port’s output usually appears a little higher in frequency than the actual tuning frequency, so it is not the best indicator. However, the deep notch in the woofer’s response will be right on the actual tuning frequency of the enclosure and port. Do a near-field measurement at your cone and you will find the tuning frequency very quickly.
From the Diagram in the earlier port - I don't see any deep notch on the woofer NF response (the red curve). Did something go wrong? The lower woofer is an almost exact overlay of the upper woofer NF.
The port response is not required but can be helpful. Easier to get started without the port.
The way I read Jeffs' paper you are correct. After figuring out the reduction based on the port diameter to the equivalent two woofer diameter, take another 6 dB.
Check the gating on your near field measurement and reduce any smoothing
Check the gating on your near field measurement and reduce any smoothing
the gating is below 20hz. This image shows the upper woofer with gating (blue) and the lower woofer Raw Response (red). Changed smoothening to 1/12 for both, though i am not sure it is being applied...
For PCD and Blending, the NearField with the Farfield, is the port response strictly required for the purpose of XO and BSC? It's good to have, but can it be done without?
Not for the response itself and the process of blending. From Jeff B's white paper on In Room Response:
Note: for dual ports, find the total port area of both ports and determine the effective diameter for the combined area. Now, take the measurement at only one port, find the ratio of the total port diameter to the cone and make the calculation and then add 6dB. So, for our example, if there had been 2 – 2” ports then the port output would be lowered by -7.96 + 6 dB = -1.96 dB. For dual woofers and a single port we will need to do it the other way around and subtract another 6dB from the port output.
Since this is a MTM, and I have measured one woofer NF, but measured the port with both the woofers, do I subtract 6DB from the port? Also, The distances to the cone and to the port are not exact. I was between 0.5-0.125 to the cone and just at the port entrance (plane of the port) - is this fine, or it needs to be more precise?
Another question:The next paragraph says how to determine the tuning frequency of the enclosure and port.
Here’s a tip – what is the best indicator of the tuning frequency (Fb) from this data? The peak in the port’s output usually appears a little higher in frequency than the actual tuning frequency, so it is not the best indicator. However, the deep notch in the woofer’s response will be right on the actual tuning frequency of the enclosure and port. Do a near-field measurement at your cone and you will find the tuning frequency very quickly.
From the Diagram in the earlier port - I don't see any deep notch on the woofer NF response (the red curve). Did something go wrong? The lower woofer is an almost exact overlay of the upper woofer NF.
Merging the port output is completely unnecessary for crossover design purpose, it's just a verification of your cabinet design. I don't even bother to try to merge it anymore, I just check my tuning and that's it.
So a few tips. First, I'd forget the math equation in Jeff's paper,
there is an easier way. Have a look at any simulation of a ported
speaker, plot the speaker output and the port output on top of each
other, like this:
What you see here is that as frequency decreases, the port output and driver output will have similar slope and converge at 0Hz. Simply adjust the port output amplitude to look like this at low frequency and you're done.
Now, looking at your measurements, I would say that the blue line and green line have similar slope at low frequency, but not red. When you leave a speaker disconnected, you now have a cabinet with a port and passive radiator, the cabinet response will not be the same as if both drivers are connected, so you will have to take that nearfield measurement with both speakers connected I think.
With the nearfield measurements, like John said, remove all gating and smoothing. It looks like you are still gating at something like 60ms, change that to 2000ms or whatever the maximum value is. You should see a hole in the driver response just like the simulation screenshot above.
Attached is a comparison graph showing how the low frequency response of a speaker changes as you adjust the distance to the floor from 4" all the way up to 36". I kept the rear wall at 8 ft and the side wall at 12 ft so that you can see the floor boundary effect more clearly. If the woofer is 4" off the floor, then no BSC is needed. If the woofer is 24 to 36" above the floor, then you need to apply the full 6dB of BSC ( but only if the speaker is well away from the rear and side walls). If the woofer is 10" to 16" above the floor, then you probably need roughly 3dB of BSC.
Playing around with some in-room measurements tonight, I think you may be right. With a driver on the floor, you basically want to aim for no BSC, however the response is a bit more complex than that. In my testing there will also definitely be a big dip in the reponse around 600-800Hz, but for a woofer crossover at 300-400Hz it's maybe not a great concern.
As a benefit, this makes for a smaller woofer inductor, so less cost, and in my design it becomes pretty borderline on the attenuation, I end up using absolutely no padding at all on the midrange, but that's ok with me, just a few less resistors to buy
ok, this now matches the white paper... This is both woofers connected, but only the upper woofer is measured... There is a slight variation in the response in the upper frequencies from the single woofer measurements (1200 is a bit more pronounced hump), but the lower frequencies match and the tuning dip is now evident.
See the Green plot - this has been raised by 20db to distinguish from the earlier ones - i tried to match the amp response as close to yesterday's measurement - but it won't be exact.
tried fiddling the responses up and down to make the ends of the port response and the woofer response from the last plot meet and pushed out the gating further out and this is what i get.
I'll try importing this to PCD and see what the summed response looks like.
PCD Summed Response - Do i take this and then merge with the Far filed? Why the does the response looks like it has built in BSC, rising from the mids to the lowers frequencies?
Comments
Another question i have is regarding the floor bounce. Having a large 3 way with the woofer close to the floor will augment the boundary gain and this will be picked up by the microphone. But how to retain this information if the near field will be merged with the farfield? Would the far field measurement retain enough of the boundary reinforcement - The near field would not have this data.
Ani, here is the step 23 screenshot.
Regarding room boundary reinforcement, baffle step, and overall room cabin pressure gain, the program I like to use is Jeff Bagby's Baffle Diffraction and Boundary Simulator (BDBS) ver. 1.3 found here:
http://audio.claub.net/software/jbabgy/BDBS.html I think this is one of the programs that dcibel is referring to.
The baffle diffraction portion of BDBS is the same as that found in the Blender program. The boundary simulator portion of BDBS uses the Allison room boundary reinforcement equations. If you place your woofer near the floor in the model, the program will show how the woofer's spl is boosted by the floor. Move the woofer higher, and you can watch how the floor boundary reinforcement effect on the woofers spl changes with frequency. It is a nifty little program, but, as dcibel points out, it does not have adjustments for wall or floor material absorption.
Attached is a comparison graph showing how the low frequency response of a speaker changes as you adjust the distance to the floor from 4" all the way up to 36". I kept the rear wall at 8 ft and the side wall at 12 ft so that you can see the floor boundary effect more clearly. If the woofer is 4" off the floor, then no BSC is needed. If the woofer is 24 to 36" above the floor, then you need to apply the full 6dB of BSC ( but only if the speaker is well away from the rear and side walls). If the woofer is 10" to 16" above the floor, then you probably need roughly 3dB of BSC.
These results are consistent with my experiences with my Kowaxial speakers. For those that attended InDIYana a couple years ago, when these speakers played almost everyone agreed that they were very "dark" sounding with "waaaay" too much BSC. This was because the woofer was way down near the floor and I, in error, was applying 4-6dB of BSC. I measured the design FRD's with the speaker on top of my 17" high DIY horizontal polar table. I should have done the measurements with the speaker placed on the floor. It makes a huge difference. Even though the FRD is bouncing all over the place below 400Hz, you can clearly see how the average spl below 400Hz changes as you move the speaker higher or lower.
Here is a pic of my Kowaxial speakers. Notice how close the woofer is to the floor. Xover was 400Hz. This was a recipe for mid-bass boom!!
This is the screenshot of it. The Blue is the NF port Response (with Both Woofers in Parallel), Red is the Upper Woofer NF. The green is from the mic at the speaker side, roughly at the center vertically and horizontally.
Not for the response itself and the process of blending. From Jeff B's white paper on In Room Response:
Since this is a MTM, and I have measured one woofer NF, but measured the port with both the woofers, do I subtract 6DB from the port? Also, The distances to the cone and to the port are not exact. I was between 0.5-0.125 to the cone and just at the port entrance (plane of the port) - is this fine, or it needs to be more precise?
Another question:The next paragraph says how to determine the tuning frequency of the enclosure and port.
Here’s a tip – what is the best indicator of the tuning frequency (Fb) from this data? The peak in the port’s output usually appears a little higher in frequency than the actual tuning frequency, so it is not the best indicator. However, the deep notch in the woofer’s response will be right on the actual tuning frequency of the enclosure and port. Do a near-field measurement at your cone and you will find the tuning frequency very quickly.
From the Diagram in the earlier port - I don't see any deep notch on the woofer NF response (the red curve). Did something go wrong? The lower woofer is an almost exact overlay of the upper woofer NF.
The way I read Jeffs' paper you are correct. After figuring out the reduction based on the port diameter to the equivalent two woofer diameter, take another 6 dB.
Check the gating on your near field measurement and reduce any smoothing
the gating is below 20hz. This image shows the upper woofer with gating (blue) and the lower woofer Raw Response (red). Changed smoothening to 1/12 for both, though i am not sure it is being applied...
See the Green plot - this has been raised by 20db to distinguish from the earlier ones - i tried to match the amp response as close to yesterday's measurement - but it won't be exact.
I'll try importing this to PCD and see what the summed response looks like.
Does the baffle step diffraction show up in measurements or it needs to be simulated? Any way to compare the sim with the actuals from the far field?