So now we have this:
We're going to setup the off axis target slope next (in room) and then the optimizer will allow us to automatically tune the EQ to get our desired response, balancing on axis and off axis response.
So, in the optimizer window, we're going to select predicted in room and modify target:
You can now modify the pink target line in the CTA-2034 window.
So now that we have targets to work with, we can start adding EQ filters to get this squared away. This will also give the optimizer the things it needs to work with to tune. Unlike in REW, this tool does not add it's own filters, it can also be used to tube a passive crossover.
So just looking at it, I'd like to add a peaking filter around 1100hz to knock down that bump. And I'd like to lift the top end from about 10khz and up, so I'll use a high shelf there.
These are the two filters I added:
Now let's setup the optimizer.
First we need to tell it which values is can alter to try to achieve the target that we gave it. This is done by checking the box under Opt on the individual crossover filters/functions.
This is the high shelf I put on the tweeter. I've checked all three fields.
In the crossover sketch, the green highlighting shows which fields I'm going to let the optimizer alter. For now, I'm going to leave the crossover out of it, and just alter the two bands of EQ and the midrange and tweeter levels.
Ok, final setup: The relative weighting between reference axis and predicted in-room
I've selected 25% reference axis, and 75% in-room. There's no right/wrong here, it's voicing preference.
But stick with me. Click on the optimize button.
What you get won't be exactly the same each time you run it, it's iteratively adjusting the parameters to achieve your target. So if you have different starting points, it can vary a bit in the output.
Here's what I just got:
That's not what I was envisioning with the tweeter. So I'm going to move the high pass filter, and adjust the tweeter level and then re-run the tool.
Re-run:
Much better. But I don't like the way the speaker response rises slightly across the entire FR range. So I'm going to go back to the optimizer and adjust my in-room target to slope down a bit more. I moved that value from -0.5dB/oct tilt to -0.7dB/oct, and re-ran again.
I'm pretty happy with that.
We're going to setup the off axis target slope next (in room) and then the optimizer will allow us to automatically tune the EQ to get our desired response, balancing on axis and off axis response.
So, in the optimizer window, we're going to select predicted in room and modify target:
You can now modify the pink target line in the CTA-2034 window.
So now that we have targets to work with, we can start adding EQ filters to get this squared away. This will also give the optimizer the things it needs to work with to tune. Unlike in REW, this tool does not add it's own filters, it can also be used to tube a passive crossover.
So just looking at it, I'd like to add a peaking filter around 1100hz to knock down that bump. And I'd like to lift the top end from about 10khz and up, so I'll use a high shelf there.
These are the two filters I added:
Now let's setup the optimizer.
First we need to tell it which values is can alter to try to achieve the target that we gave it. This is done by checking the box under Opt on the individual crossover filters/functions.
This is the high shelf I put on the tweeter. I've checked all three fields.
In the crossover sketch, the green highlighting shows which fields I'm going to let the optimizer alter. For now, I'm going to leave the crossover out of it, and just alter the two bands of EQ and the midrange and tweeter levels.
Ok, final setup: The relative weighting between reference axis and predicted in-room
I've selected 25% reference axis, and 75% in-room. There's no right/wrong here, it's voicing preference.
But stick with me. Click on the optimize button.
What you get won't be exactly the same each time you run it, it's iteratively adjusting the parameters to achieve your target. So if you have different starting points, it can vary a bit in the output.
Here's what I just got:
That's not what I was envisioning with the tweeter. So I'm going to move the high pass filter, and adjust the tweeter level and then re-run the tool.
Re-run:
Much better. But I don't like the way the speaker response rises slightly across the entire FR range. So I'm going to go back to the optimizer and adjust my in-room target to slope down a bit more. I moved that value from -0.5dB/oct tilt to -0.7dB/oct, and re-ran again.
I'm pretty happy with that.
I forgot to add this: These FR exports from REW are from your latest measurements, and I used a frequency dependent window at 14 cycles instead of gating. There is no smoothing in VCAD.
If you have time, I'd like to see what this sim looks like in VCAD. I adjusted gains to match the SPL level in your sim. Both are virtually the same on axis, but I crossed lower.
Woofer LR4 @ 300
Mid BW4 @ 240 and BW4 @ 1900
Tweeter BW3 @ 2500
EQ
W -2.5 @ 265 Q=3
M +1.5 @ 600 Q=1
M -2 @ 1200 Q=1
T -1.5 @ 8500
Levels
W 0
M -7
T -8
Attachments
Last edited:
Thanks for doing this. I see a difference between 2k and 5k, but I don't really know what to make of it.
I did not know about this option. Am I right to understand this is using up the excess sensitivity of the tweeter (as it is attenuated anyway) by adding a suitably sized resistor in series and that would reduce the amplitude of some of the peaks and dips on the tweeter response?
If so, how is the resistor sized and what damping effect should be expected? How effective is this method?
It's something sometimes done with passive x-overs. The parallel resistor will damp the tweeter resonance. The frequency response does not change, but would be a few dB lower, and then turned back up actively. The idea would be possible better tweeter performance near resonance, so it might subjectively sound better, if you were to cross low.
I'm not an expert on polars, but I'm thinking my low 2k crossed sim looks rather good off axis.
I'm not an expert on polars, but I'm thinking my low 2k crossed sim looks rather good off axis.
Last edited: