This is a simulation of a flat face 30mm tweeter (no waveguide) in this cabinet. Tapered bevels would be used, so the midrange would "see" a 25mm bevel, the tweeter would "see" a 50mm bevel.
This simulation uses idealized driver responses where each driver is a perfect piston. It demonstrates the feasability of this architecture. I use this process to design the baffle spacing.
This simulation uses idealized driver responses where each driver is a perfect piston. It demonstrates the feasability of this architecture. I use this process to design the baffle spacing.
Jim,
For those following along with your detailed explanations, appropriate use of pictorial elements and overall excellent documentation, how did you go about deciding-
A) baffle dimensions prior to selection of drivers? (a lot of people starting out in this hobby get excited about fancy shiny drivers, and thus select drivers first (understandably))
B) Decide on what SPL capability is "enough" (a lot of people starting in his hobby haven't had an opportunity to build a speaker, so a 90dB concept vs a 93dB concept seems somewhat arbitrary. (understandably))
PS. VituixCAD2 is a powerful and flexible tool, don't you think?
I send Kimmo a Christmas present every year for this software that he gives to us for free... I also wish I had the time to record how to us it. You know, like recording mouse movements over the UI, whilst talking over it (voice over). If this was done, ala Khan Academy style, many more hobbyists can benefit from this powerful tool.
For those following along with your detailed explanations, appropriate use of pictorial elements and overall excellent documentation, how did you go about deciding-
A) baffle dimensions prior to selection of drivers? (a lot of people starting out in this hobby get excited about fancy shiny drivers, and thus select drivers first (understandably))
B) Decide on what SPL capability is "enough" (a lot of people starting in his hobby haven't had an opportunity to build a speaker, so a 90dB concept vs a 93dB concept seems somewhat arbitrary. (understandably))
PS. VituixCAD2 is a powerful and flexible tool, don't you think?
I send Kimmo a Christmas present every year for this software that he gives to us for free... I also wish I had the time to record how to us it. You know, like recording mouse movements over the UI, whilst talking over it (voice over). If this was done, ala Khan Academy style, many more hobbyists can benefit from this powerful tool.
ABEC simulation of an SB26 in an Augerpro waveguide 127mm wide 78.49mm high.
The edge of the waveguide is 30mm down from the top of the cabinet, so 5mm from waveguide edge to roundover start. This doesn't include the mounting flange in the real thing so might have to go a little lower. FRD files for vituix attached and a screenshot showing two filters to get a pretty good LR4 at 2500.
Because the data ends at 1500Hz Vituix gets a bit crazy with the interpolation below that. I don't think it will be much of an issue for the task at hand.
The edge of the waveguide is 30mm down from the top of the cabinet, so 5mm from waveguide edge to roundover start. This doesn't include the mounting flange in the real thing so might have to go a little lower. FRD files for vituix attached and a screenshot showing two filters to get a pretty good LR4 at 2500.
Because the data ends at 1500Hz Vituix gets a bit crazy with the interpolation below that. I don't think it will be much of an issue for the task at hand.
Especially in an active system, you basically don't need a volume (read; can be very minimal), since there is barely any cone excursion, there no piston load.
(in a passive system it can help with the impedance for the passive crossover)
Obviously, as always, up to certain limits, check your simulations.
Especially max power on the driver, since electrical power is inversely proportional to the acoustic volume (acoustic power)
Most 4 inch mids are fine with 1-2 liters or so.
(in a passive system it can help with the impedance for the passive crossover)
Obviously, as always, up to certain limits, check your simulations.
Especially max power on the driver, since electrical power is inversely proportional to the acoustic volume (acoustic power)
Most 4 inch mids are fine with 1-2 liters or so.
Until I select a mid driver, I won't spend much time optimizing the midrange enclosure. The design I have now is 3.3 liters internal volume, and with 1/2" wall thickness, it consumes 5 liters of the space in the main cabinet.
Yes it could get smaller I am sure.
It is a great tweeter, but as you said, it is about $100 now. It would severely squeeze the midrange budget down to about $20. With this new 10" woofer, I might be forced to let the budget grow a bit, but I am still shooting for a combined midrange+tweeter cost of $150/side.
I will answer the easy one first. How did I decide how much SPL capability was enough? I listened to my main system at volume levels I considered realistic, and I used an SPL meter to evaluate one speaker at 1m. I also played pure tones at 40 – 80 Hz and evaluated what I thought would be the minimum SPL I would want for realistic low frequency reproduction. A 40 Hz peak of 96 dB was almost enough. The same peak at 99 dB was more than enough.
I started to write down the process I went through over the last couple of weeks, and soon realized it is so iterative that describing it in a linear progression becomes incomprehensible.
It is a very iterative process. I use VituixCad diffraction tool to simulate the baffle diffraction effects for each driver. Designing the baffle means the height and width of the baffle, the size and spacing of the drivers, and the size/shape of the edge bevel/radius. At this step you will assume the drivers are "perfect pistons", and the crossover will be a hypothetical filter using idealized slopes and eq's.
I use Excel to manage the dimensions of the cabinet design. I try to design a box volume and box shape that incorporates your baffle design, and also provides the correct box volume for the woofers. The excel worksheet takes into account the wall thickness to calculate external dimensions. It calculates driver edge distance, and the center-to-center distances for the drivers. The worksheet also calculates the ratio of the major internal dimensions, and this allows me to avoid integer multiples… for example a cabinet internal dimensions of 10” x 20” x 30” (ratio of 1 : 2 : 3) would be undesirable because standing waves could add together. Internal dimensions of 9.25” x 21.25” x 30.75” (ratio of 1 : 2.3 : 3.3) provides the same volume but would be much preferable.
The process starts with some assumptions. The first assumption is the size of the drivers to be used, so that the minimum width and height of the baffle can be defined. I assume a cabinet wall thickness. I assume an initial cabinet volume target, even though I do not yet know which woofer I will use.
Next is a long period of iteration, where changes in cabinet size/shape are defined in Excel (to make sure it is geometrically possible), then evaluated in VituixCad diffraction tool to get a polar response, and simulated in VituixCad crossover tool to check for acceptable power response and directivity index. The process is repeated many times. Various woofers will require different cabinet volumes. The size of the bevels/roundovers will affect the positioning of the drivers, and the spacing of the drivers will affect the directivity index.
This thread goes into details about how to evaluate a good diffraction performance from a driver/baffle combination.
https://www.diyaudio.com/community/...ations-with-ideal-drivers.380658/post-6880731
particularly after post#180.
j.
Here is the VituixCad Simulation using the data provided by @fluid. This is an ideal piston representation of a 10 inch woofer, a 4 inch mid driver, and an ABEC simulation of an SB26 in an Augerpro waveguide 127mm wide 78.5mm high. I allowed Vcad to optimize the response a bit.
It looks really good.
Now here is the response of the non-waveguide option, again with some Vcad optimization. This is an ideal piston representation of a 10 inch woofer, a 4 inch mid driver, and a 30 mm tweeter. When looking at this, it is important to keep in mind that the actual response of any tweeter above ~ 6k is going to deviate from an ideal piston. The actual driver I select will determine what the dispersion, directivity, and response is above that point.
Well this looks really good too... The main difference obviously is the directivity from 1k - up. The waveguide option has a DI of 5 db at 1k, and it rises smoothly all the way to 20k. The bevelled box non-waveguide option has a DI of 5 db at 1k, and it stays constant until about 6k, and from there it rises according to the inherent directivity of whatever tweeter is chosen.
Thank you @fluid and @augerpro.
j.
It looks really good.
Now here is the response of the non-waveguide option, again with some Vcad optimization. This is an ideal piston representation of a 10 inch woofer, a 4 inch mid driver, and a 30 mm tweeter. When looking at this, it is important to keep in mind that the actual response of any tweeter above ~ 6k is going to deviate from an ideal piston. The actual driver I select will determine what the dispersion, directivity, and response is above that point.
Well this looks really good too... The main difference obviously is the directivity from 1k - up. The waveguide option has a DI of 5 db at 1k, and it rises smoothly all the way to 20k. The bevelled box non-waveguide option has a DI of 5 db at 1k, and it stays constant until about 6k, and from there it rises according to the inherent directivity of whatever tweeter is chosen.
Thank you @fluid and @augerpro.
j.
I think this combination would make a really nice low cost 3-way.
Tweeter: Dayton ND25FW-4 waveguide tweeter ($22)
Mid: SB acoustics SB12PAC25 ($37)
Woofer: MCM 55-5670 ($40)
However, considering the electronics cost of an active 3-way, it makes more sense to use better drivers... but in a passive system these drivers would be interesting... I might do it sometime.
Seems like a fine system to me.
Personally would use that GRS sub with more oomph in it I think.
But after a while those things are kinda splitting hairs at some point.
Also fine candidate for a hybrid active-passive filter.
Last edited:
It does look well controlled but the waveguide holds directivity far better than the mid so there is still something of a mismatch. A smaller waveguide could bridge the gap between the two so there is more of a progressive rise in the horizontal DI.
I have no idea if this is of any real importance though.
There is also the option of cutting/routing a slight waveguide into the front baffle to assist the bevels and avoiding the need for anything to be 3D printed.