I’m designing three-way “dual woofers” speakers format. The woofers used are 8 Ohms nominal resistance. I wonder when designing the crossovers in the software, should I use “two 8 Ohms drivers in parallel” or “a single 4 Ohms driver”?
In theory, they should have difference on SPL of the woofers. That is one will produce extra 6dB at woofers’ operating range, while the other will not. But, what else?
In theory, they should have difference on SPL of the woofers. That is one will produce extra 6dB at woofers’ operating range, while the other will not. But, what else?
Are the woofers identical ? If so, the net impedance would be half of what it was previously.
However, I think you need to use the 'two 8 Ohms drivers' case.
However, I think you need to use the 'two 8 Ohms drivers' case.
One other difference would be the resulting radiation pattern.
Last edited:
Yes, they’re identical.
Could you please explain the difference in radiation pattern?
Interesting! Both cases. But, in the simulation software I used, VituixCAD2, it seems to focus on driver’s resistance (Re) only for simulating the frequency response—there’s no driver’s diameter required to fill.
It is learning process.
Drivers are simulated on baffle in full space.
Since drivers are impossible to mount physically on top of each other.
The result phase difference needs accounting for since center to center
spacing is gonna be what it is in real life.
In short summary it will be ideal if mounted as close as possible.
Sim needs to know positions of actual drivers.
You can leave microphone in same place/ use the listening position
for all driver measurements of each individual driver in diffraction tool.
Or you can measure them each on center and go through the painful process
of figuring out correct XYZ offset coordinates.
Why people do this dont know, but either way the position of each driver
needs to be correct.
Figure out listening position on baffle usually tweeter.
Then leave the mic there for all drivers. So dont have to practically
kill yourself figuring out XYZ data.
But would need done depending how you use diffraction tool.
I design baffle only in sketchup so I know exact driver positions
and make sure the outer diameters can physical fit and not overlap.
Otherwise for flatest response find the offset on the baffle.
Or just make them all a line for" looks" and deal with designing
a crossover that fixes non ideal mounting positions.
which wont be dead center on the baffle, wont be nonsense
skinny baffles either.
Impedance Data is ZMA file or Text file.
With 8 ohm impedance curve loaded for each driver.
When you connect them it will be correct lower impedance
that results.
No need to calculate with external calculator.
Use library of filters and then change parameters
in edit fields to match.
Simple scroll impedance with mouse wheel as well
since yes the crossover point wont actually be 8 ohms.
Drivers are simulated on baffle in full space.
Since drivers are impossible to mount physically on top of each other.
The result phase difference needs accounting for since center to center
spacing is gonna be what it is in real life.
In short summary it will be ideal if mounted as close as possible.
Sim needs to know positions of actual drivers.
You can leave microphone in same place/ use the listening position
for all driver measurements of each individual driver in diffraction tool.
Or you can measure them each on center and go through the painful process
of figuring out correct XYZ offset coordinates.
Why people do this dont know, but either way the position of each driver
needs to be correct.
Figure out listening position on baffle usually tweeter.
Then leave the mic there for all drivers. So dont have to practically
kill yourself figuring out XYZ data.
But would need done depending how you use diffraction tool.
I design baffle only in sketchup so I know exact driver positions
and make sure the outer diameters can physical fit and not overlap.
Otherwise for flatest response find the offset on the baffle.
Or just make them all a line for" looks" and deal with designing
a crossover that fixes non ideal mounting positions.
which wont be dead center on the baffle, wont be nonsense
skinny baffles either.
Impedance Data is ZMA file or Text file.
With 8 ohm impedance curve loaded for each driver.
When you connect them it will be correct lower impedance
that results.
No need to calculate with external calculator.
Use library of filters and then change parameters
in edit fields to match.
Simple scroll impedance with mouse wheel as well
since yes the crossover point wont actually be 8 ohms.
Last edited:
So basically yes use 2 woofers
Since each baffle simulation will have different response
even though matching drivers.
You will quickly see resulting peak or " baffle step"
in fullspace simulation.
So whatever frequency that happens at.
I know you like wide baffles
so it wont be that bad and offset the driver
to one edge for flatter response.
You use second woofer on lower crossover point
which will help fill in the full space losses
below the baffle step.
the higher frequency woofer is mounted
as close as possible to midrange.
The lower woofer will be larger distance from
midrange but wont be a problem and cause cancellations
because that is woofer used for lower crossover frequency
to fill in losses.
Since each baffle simulation will have different response
even though matching drivers.
You will quickly see resulting peak or " baffle step"
in fullspace simulation.
So whatever frequency that happens at.
I know you like wide baffles
so it wont be that bad and offset the driver
to one edge for flatter response.
You use second woofer on lower crossover point
which will help fill in the full space losses
below the baffle step.
the higher frequency woofer is mounted
as close as possible to midrange.
The lower woofer will be larger distance from
midrange but wont be a problem and cause cancellations
because that is woofer used for lower crossover frequency
to fill in losses.
Let me explain the root of this thread.
I have been reverse engineering the commercial speakers. They’re all three-way with dual woofers format; ADS L910, L1530, L1590, and L1590/2.
Only L1590/2 has relatively ‘flat’ frequency response. Yet, I didn’t say it has the best sound because I prefer those non-flat FR on older models.
I observed every model has equipped with moderate resistors in its midrange and tweeter’s circuits, except the L1590/2. The L1590/2 is the only model that uses no resistor on tweeter’s circuit and only a 2-Ohm resistor on midrange’s. As a result, the L1590/2 is the only model that has different sound from its predecessors.
I tried to add some attenuation of about ‘6dB’ to the midrange and tweeter of the L1590/2, in the simulation. And it was surprising me when the FR curve went similar to those its predecessors.
In all, I thus wonder if the designers did something differently when they designed the L1590/2. Since the difference found involved 6dB deviated between woofer and mid/high. I suspected they might have different setting parameter in their device/computer. And I also suspect it might be the ‘Re’ of the woofers. As mentioned earlier, they’re two possible settings; 8//8 = 4 Ohms, or a single 4 Ohms.
I have been reverse engineering the commercial speakers. They’re all three-way with dual woofers format; ADS L910, L1530, L1590, and L1590/2.
Only L1590/2 has relatively ‘flat’ frequency response. Yet, I didn’t say it has the best sound because I prefer those non-flat FR on older models.
I observed every model has equipped with moderate resistors in its midrange and tweeter’s circuits, except the L1590/2. The L1590/2 is the only model that uses no resistor on tweeter’s circuit and only a 2-Ohm resistor on midrange’s. As a result, the L1590/2 is the only model that has different sound from its predecessors.
I tried to add some attenuation of about ‘6dB’ to the midrange and tweeter of the L1590/2, in the simulation. And it was surprising me when the FR curve went similar to those its predecessors.
In all, I thus wonder if the designers did something differently when they designed the L1590/2. Since the difference found involved 6dB deviated between woofer and mid/high. I suspected they might have different setting parameter in their device/computer. And I also suspect it might be the ‘Re’ of the woofers. As mentioned earlier, they’re two possible settings; 8//8 = 4 Ohms, or a single 4 Ohms.
Do you have the driver parameters for these old commercial speakers with which to perform a simulation? The choices made in designing a speaker in those days were significantly different to today. Large speaker size was more tolerated, wide baffles were more accepted compared to the narrow towers common today, amplifiers were less powerful raising the advantage of higher efficiencies compared to today, records had little low bass information lessening the advantage deep bass extensions compared to modern digital recordings, the modest tweeters of the day crossed to mids at substantially higher frequencies than today which had a range of knock on effects, etc... all of which means a close examination of vintage speakers can be as misleading as it is informative for a modern speaker design despite the physics remaining the same.
The commonly used linear model for a driver only has a few parameters with the resistance of the coil being one of them. It is easy enough to change 8 to 4 in a simulation program and compare but in the real world such a change would also change the moving mass, Bl factor, coil inductance,... The parameters tend to come in sets optimised for particular types of speakers.
The commonly used linear model for a driver only has a few parameters with the resistance of the coil being one of them. It is easy enough to change 8 to 4 in a simulation program and compare but in the real world such a change would also change the moving mass, Bl factor, coil inductance,... The parameters tend to come in sets optimised for particular types of speakers.
I don’t have driver parameters of those commercial speakers except the Re of each driver. And I don’t know why it’s important to have them since the VituixCAD2 requires to fill simply ‘Re’ and it can generate frequency & impedance responses, transfer function, and etc.
It isn't possible to simulate the response of a driver without the values of a set of driver parameters. I am not familiar with your software but at a guess it has a set of parameters stored for 4 and 8 ohm versions of your driver and the name plus 4 or 8 specifies which to use in the simulation.