Hi, I've been working on a 3-way system and trying to work out its sensitivity.
Is it as simple as taking the lowest sensitivity driver (99.2db woofer) and subtracting the effect of the crossover (2nd order lowpass)?
I have the following components:
HF - Beyma CP21/F on a 4th order HP
MF - JBL 2445J w/2308A horn on a 3rd order BP
LF - Eminence Delta Pro 12A on a 2nd order LP
Is it as simple as taking the lowest sensitivity driver (99.2db woofer) and subtracting the effect of the crossover (2nd order lowpass)?
I have the following components:
HF - Beyma CP21/F on a 4th order HP
MF - JBL 2445J w/2308A horn on a 3rd order BP
LF - Eminence Delta Pro 12A on a 2nd order LP
Last edited:
Yes, just like chain strength depends on the weakest link, speakers efficiency is dictated by the speaker with lowest efficiency.
I assume you will pad the mid and tweeter to match the woofer and make it flat.
I assume you will pad the mid and tweeter to match the woofer and make it flat.
Yes, both the midrange and tweeter have been padded down to match the woofer.
And yes pro 12A haha
And yes pro 12A haha
I have the ported enclosure tuned between 40 and 50hz (it's been a long time so I don't remember) but I have a subwoofer I can turn on when I want the extra LF information. Otherwise it's just enough bass for my small apartment.
Edit: 56hz... based on the last file i had open in WinISD lol
Edit: 56hz... based on the last file i had open in WinISD lol
Last edited:
Then your system is effectively as efficient as your woofer. With an active crossover efficiency is section by section.
The crossover shouldn't subtract any sensitivity (just a small secondary amount for the resistance of a series inductor on the woofer). However if your system has baffle step you may see a 6dB reduction.
Realistically closer to 95dB, but that's still very efficient. You won't need more than a few watts for home use.
The SPL of a system depends on the frequency; as you can tell on the datasheet.
send it 50Hz tone @2.83V and the SPL is different from a 300Hz tone or 1Khz tone.
Look at the datasheet of that driver and see where it’s 99dB @ 2.83V —> it’s 1Khz in this (and most cases, as is the standard)
But when you use it as a woofer you wanted to play up to 2 or 4 hundred Hz.
Now look again at the datasheet.
what’s the sensitivity there? It’s more like 96.5dB
The second issue is manufacturers measurements and stated sensitivity not in your box or cabinet, it’s usually on a very large baffle. The standard is a (large)
IEC baffle, but unless your speaker is mounted on the same badge the frequency response will not be the same.
Your speaker on a smaller baffle will suffer a droop in the lower frequencies as the lower frequencies have larger wavelength which will wrap around the sides and back of the baffle.
You can experience a loss of up to 6dB at these lower frequencies and an enhancement of up to 2-3dB in the upper mids.
What you really ought to do is to compensate for these baffle step losses; so your system sounds balanced.
Once you’ve got your crossover sorted out, the system should be measured via anechoic (or quasi-anechoic) methods.
The system sensitivity is then best specified at a frequency eg. 1Khz (old standard) or over a frequency range eg. 200-400Hz or 300Hz -3000Hz with @2.83V drive level.
depending on the amount of baffle step compensation your sensitivity will be between 90.5dB to 96.5dB
send it 50Hz tone @2.83V and the SPL is different from a 300Hz tone or 1Khz tone.
Look at the datasheet of that driver and see where it’s 99dB @ 2.83V —> it’s 1Khz in this (and most cases, as is the standard)
But when you use it as a woofer you wanted to play up to 2 or 4 hundred Hz.
Now look again at the datasheet.
what’s the sensitivity there? It’s more like 96.5dB
The second issue is manufacturers measurements and stated sensitivity not in your box or cabinet, it’s usually on a very large baffle. The standard is a (large)
IEC baffle, but unless your speaker is mounted on the same badge the frequency response will not be the same.
Your speaker on a smaller baffle will suffer a droop in the lower frequencies as the lower frequencies have larger wavelength which will wrap around the sides and back of the baffle.
You can experience a loss of up to 6dB at these lower frequencies and an enhancement of up to 2-3dB in the upper mids.
What you really ought to do is to compensate for these baffle step losses; so your system sounds balanced.
Once you’ve got your crossover sorted out, the system should be measured via anechoic (or quasi-anechoic) methods.
The system sensitivity is then best specified at a frequency eg. 1Khz (old standard) or over a frequency range eg. 200-400Hz or 300Hz -3000Hz with @2.83V drive level.
depending on the amount of baffle step compensation your sensitivity will be between 90.5dB to 96.5dB
Last edited:
Excellent post from tktran303 above, except I think he may have misread the dB scale on the Eminence spec sheet - the sensitivity up around 200-400Hz is still up around 98dB.
I could be wrong but the nature of the question suggests to me that the OP isn't using a xo simulation program like XSim for eg to design your speaker? Which really is absolutely necessary for decent results especially with the inherent difficulty of a 3-way.
For example, one other thing that affects sensitivity in a 3-way that hasn't been mentioned yet is the way the LP xo on the woofer sometimes will create a bump in the woofer SPL down somewhere around 80-200Hz, depending on the driver's Fs and the size of the peak. And you are only going to know if this is going to happen if you do the proper sims.
And it's important because in a nutshell, the speaker sensitivity will be determined by the woofer level down around 150-200Hz after baffle step loss has been taken into account and after seeing how the woofer responds to its xo filter and whether or not you want to take advantage of that LF bump (if it occurs) or if you want to squash it with an impedance flattening notch filter.
Here's an example with some random driver (note the vented box response has also been added to the spec sheet FR):
Black = spec sheet FR --> sensitivity at about 150Hz is just shy of 96dB
Blue = spec sheet FR + baffle diffraction --> sensitivity at about 150Hz is about 92dB
Green = spec sheet FR + baffle diffraction + a 2nd order LP filter, notice the LF bump of about 3dB --> sensitivity at about 150Hz is back up to about 96dB but the LF response is no longer flat, ie. F3 has been effectively diminished by ~4dB.
Orange = spec sheet FR + baffle diffraction + a 2nd order LP filter + an impedance flattening notch filter --> sensitivity at about 150Hz is now back down to about 92dB and the response is now flat and so F3 remains about the same.
I could be wrong but the nature of the question suggests to me that the OP isn't using a xo simulation program like XSim for eg to design your speaker? Which really is absolutely necessary for decent results especially with the inherent difficulty of a 3-way.
For example, one other thing that affects sensitivity in a 3-way that hasn't been mentioned yet is the way the LP xo on the woofer sometimes will create a bump in the woofer SPL down somewhere around 80-200Hz, depending on the driver's Fs and the size of the peak. And you are only going to know if this is going to happen if you do the proper sims.
And it's important because in a nutshell, the speaker sensitivity will be determined by the woofer level down around 150-200Hz after baffle step loss has been taken into account and after seeing how the woofer responds to its xo filter and whether or not you want to take advantage of that LF bump (if it occurs) or if you want to squash it with an impedance flattening notch filter.
Here's an example with some random driver (note the vented box response has also been added to the spec sheet FR):
Black = spec sheet FR --> sensitivity at about 150Hz is just shy of 96dB
Blue = spec sheet FR + baffle diffraction --> sensitivity at about 150Hz is about 92dB
Green = spec sheet FR + baffle diffraction + a 2nd order LP filter, notice the LF bump of about 3dB --> sensitivity at about 150Hz is back up to about 96dB but the LF response is no longer flat, ie. F3 has been effectively diminished by ~4dB.
Orange = spec sheet FR + baffle diffraction + a 2nd order LP filter + an impedance flattening notch filter --> sensitivity at about 150Hz is now back down to about 92dB and the response is now flat and so F3 remains about the same.
I was going to say 92-93, but didn't want to burst any bubbles. 😀
Of course not everyone does BSC (and you'll hear it). With some room gain you might be at 95, which is still high efficiency.
Of course not everyone does BSC (and you'll hear it). With some room gain you might be at 95, which is still high efficiency.
Well i figure it’s more than 91 at the very least… because my zen triode clone amp will go much louder than I am comfortable with without breaking up on these speakers, but definitely cannot say the same for my JBL 4307’s rated at 91, which break up in the upper end of “normal” listening levels on the same amp.
The main reason I want to know is just to be able to decide if i’m missing something by not having an off-the-shelf high efficiency speakers like the Klipsch Heresy which is 99db
I did design the crossover using software such as XSim AND vituixcad using response and impedance measurements i did in REW. Final measurement was bang-on to the simulation, however I did later play with the MF and LF padding to taste from flat. I didn’t notch out the 150-200hz bump, but I really like that bump, it’s where toms and snare membranes hit! From what I remember chose this woofer because it doesn’t have any of of the nasty 2k peaks that some other PA woofers do. This was an old project that I pulled out of storage because I decided to play with tubes, and all the sudden, high efficiency speakers became more relevant to me.
The main reason I want to know is just to be able to decide if i’m missing something by not having an off-the-shelf high efficiency speakers like the Klipsch Heresy which is 99db
I did design the crossover using software such as XSim AND vituixcad using response and impedance measurements i did in REW. Final measurement was bang-on to the simulation, however I did later play with the MF and LF padding to taste from flat. I didn’t notch out the 150-200hz bump, but I really like that bump, it’s where toms and snare membranes hit! From what I remember chose this woofer because it doesn’t have any of of the nasty 2k peaks that some other PA woofers do. This was an old project that I pulled out of storage because I decided to play with tubes, and all the sudden, high efficiency speakers became more relevant to me.
Glad to hear I was wrong and that you used XSim and measurements to design your speaker.
I'm working with one of my own speakers right now and I am preferring to more or less flatten the bump above 100Hz but keep another one centered around 80Hz. It really punches the kick drum to my ears. But that's one of the beauties of diy - you can tune your speaker to please your own individual tastes.
I'm working with one of my own speakers right now and I am preferring to more or less flatten the bump above 100Hz but keep another one centered around 80Hz. It really punches the kick drum to my ears. But that's one of the beauties of diy - you can tune your speaker to please your own individual tastes.
One of my friends is a music professional (ie. full time work is in music) and he said something I won’t forget
We were sitting in a pub or nightclub having a drink with him, and he pulls out his tablet and with a twirl of his finger cuts out all the bass, for 3 seconds, and then puts it back.
“Flat speaker are better than non flat, but you still need to EQ for the room and audience”
We were sitting in a pub or nightclub having a drink with him, and he pulls out his tablet and with a twirl of his finger cuts out all the bass, for 3 seconds, and then puts it back.
“Flat speaker are better than non flat, but you still need to EQ for the room and audience”