Can anyone help with Alignment?

[3ll3d00d]

The crossover points I have chosen are ~1.2khz and 6.5khz.

nice and flat btw I used an 8 cycle fdw here (no particular reason, just from eyeballing the response)

if I were you, I'd do the following

1) work out why the tweeter droops
2) get a timing reference
3) find a space that gives you a longer reflection free zone
4) measure using the above and verify that you can sum in software to produce a result that matches an acoustic sum (you don't need any EQ for this)

1 2 & 4 are mandatory, 3 is preferred but if you don't have such a space then not the end of the world

only once you have done all this then

5) take measurements on and off axis (personally I do 7.5-15-22.5-30-37.5-45 when getting a rough idea, 15-30-45 is fine but once you are taking a few then it's only a question of another 15-30mins to get a few more angles so in for a penny, in for a pound.....)
6) load it into vituixcad and away you go

 

[Robbie010]

You may correct me, but the only crossover I can see ist between Woofer an Tweeter at around 3+ khz? and the mid is running free....

The woofer has a 12db Low Pass applied at 1,200hz

The midrange has a High Pass at 1,200hz and a Low Pass at 6,500hz, both 12db

The Tweeter has a 12db High Pass at 6,500hz.

I believe I need to implement steeper slopes to try and avoid the woofer and tweeter interacting at 3khz, that isn’t an intended crossover point.

 

[Robbie010]

4) measure using the above and verify that you can sum in software to produce a result that matches an acoustic sum

Can you please elaborate on this?

I’m trying to understand what is meant by checking the crossover Sum?

 

[Salt]

Last attempt:
put a
low pass (LP) filter on the woofer with 300 Hz/24 dB LR4 (4 th order),
high pass (HP) filter at same 300 Hz with same LR 4 (Linkwitz Riley 4 th order/24 dB) on the mid
and
LP on to mid LR4 at 2.6 kHz,
same 2.6 khz LR4 HP to tweeter.

Then report measurements of single transducer's and allover response.

 

[3ll3d00d]

I’m trying to understand what is meant by checking the crossover Sum?

Measure each individually, sum them using trace arithmetic (if using rew) or as shown in vituixcad

Measure them all playing together in rew

Those two (computed, acoustic) should match exactly

If they don't, you are doing something wrong so fix it before spending a chunk of time measuring n drivers m times

 

[Flaesh]

300 Hz is very low for HF144. And 2.6kHz is very low for HF144 also.
The length of XT1464 is 24 cm.

 

[Salt]

It's not for final alignment but we could see if OP does put things correct in xover.

 

[Robbie010]

More tweaking today. I applied 24db crossover slopes instead of 12db and re-measured each driver at the same 50cm distance with a timing reference, here is my results:

Separate sweeps overlayed:

All Driven:

Impulse Response:

In order to time align the drivers do I now apply appropriate delays to the Midrange and Tweeter to align their peaks or the start of their rise?

 

[levimax]

More tweaking today. I applied 24db crossover slopes instead of 12db and re-measured each driver at the same 50cm distance with a timing reference, here is my results:

Separate sweeps overlayed:

View attachment 480911

All Driven:

View attachment 480913

Impulse Response:

View attachment 480914

In order to time align the drivers do I now apply appropriate delays to the Midrange and Tweeter to align their peaks or the start of their rise?

View attachment 480915

Looking much better! I am waiting to hear what answers you get to your time alignment question. I have seen this answered many different ways by many different experts. All have great rational with huge experience and all give different answers.

 

[3ll3d00d]

align their peaks or the start of their rise?

Rise though seems like your tweeter and mid are inverted? The tweeter seems quite messy in fact that one might be trickier. Post the mdat ?

 

[Robbie010]

Rise though seems like your tweeter and mid are inverted? The tweeter seems quite messy in fact that one might be trickier. Post the mdat ?

Attached.

 

[3ll3d00d]

Attached.

I meant all the measurements to see how they fit together

 

[Robbie010]

I meant all the measurements to see how they fit together

 

[Salt]

More tweaking today. I applied 24db crossover slopes instead of 12db and re-measured each driver at the same 50cm distance with a timing reference, here is my results:

Separate sweeps overlayed:

View attachment 480911

All Driven:

View attachment 480913

Impulse Response:

View attachment 480914

In order to time align the drivers do I now apply appropriate delays to the Midrange and Tweeter to align their peaks or the start of their rise?

View attachment 480915

Low to mid crossover looks pretty fine, but the result gives a dip, so first try to revert polarity of the mid (as it also dips at crossover to the tweeter, where not run at same level and not at -3 dB each).

 

[3ll3d00d]

and the all driven sum?

just looking at the freq response as presented, the woofer is doing too much so the sum of woofer and mid will be elevated (you can see it if you load same in rew, tell it time align and then do a vector sum)

 

[Keith_W]

I'll show you a few things.

This is the sum of woofer, mid, and tweeter from the measurements as you posted. You can clearly see cancellation at the XO points.

So I inverted the polarity of the mid driver. You can see that it has improved the cancellation point slightly at the W-M XO but worsened the M-T XO.

So I inverted the polarity of mid AND tweeter. Or you could invert the polarity of the woofer alone and leave mid and tweeter untouched. This produces the best result so far, but it is still far from ideal.

Using REW's Alignment Tool, I selected tweeter and midrange driver. Slide around the "fine delay adjustment" slider and watch the graph. You will see that at one point, the cancellation disappears. When it does, click "aligned copy" for M and T, then use A+B to sum them.

This is the result of "time alignment" of the midrange to the tweeter. The dip is gone. You will need to repeat this exercise for W-M, i'll leave that to you.

This is how I do my time alignment:

1. In the impulse response view, align the START of the impulse of the two drivers you wish to align. Use the tweeter as the reference. Adjust the mid and the woofer to align to the tweeter and note these values. The purpose of this step is to get rid of egregious delays which are > 1 cycle.
2. Because there may still be phase cancellation even after drivers are aligned, we need to also align the phase to remove cancellation at the XO point. I use REW's Time Alignment tool as shown.
3. Dial these values into your DSP and take a verification measurement.

(Edit): note you can set the gain of these curves in the Alignment tool. I prefer to set them by adjusting the volume of the amps. You may/may not have this ability. How you choose to do it is up to you.

 

[Robbie010]

I'll show you a few things.

View attachment 481034

This is the sum of woofer, mid, and tweeter from the measurements as you posted. You can clearly see cancellation at the XO points.

View attachment 481035

So I inverted the polarity of the mid driver. You can see that it has improved the cancellation point slightly at the W-M XO but worsened the M-T XO.

View attachment 481036

So I inverted the polarity of mid AND tweeter. Or you could invert the polarity of the woofer alone and leave mid and tweeter untouched. This produces the best result so far, but it is still far from ideal.

View attachment 481037

Using REW's Alignment Tool, I selected tweeter and midrange driver. Slide around the "fine delay adjustment" slider and watch the graph. You will see that at one point, the cancellation disappears. When it does, click "aligned copy" for M and T, then use A+B to sum them.

View attachment 481039

This is the result of "time alignment" of the midrange to the tweeter. The dip is gone. You will need to repeat this exercise for W-M, i'll leave that to you.

This is how I do my time alignment:

1. In the impulse response view, align the START of the impulse of the two drivers you wish to align. Use the tweeter as the reference. Adjust the mid and the woofer to align to the tweeter and note these values. The purpose of this step is to get rid of egregious delays which are > 1 cycle.
2. Because there may still be phase cancellation even after drivers are aligned, we need to also align the phase to remove cancellation at the XO point. I use REW's Time Alignment tool as shown.
3. Dial these values into your DSP and take a verification measurement.

(Edit): note you can set the gain of these curves in the Alignment tool. I prefer to set them by adjusting the volume of the amps. You may/may not have this ability. How you choose to do it is up to you.

I've searched everywhere in REW and cannot find the Alignment Tool anywhere..........?

 

[Robbie010]

Got it!

Christ they made that hard to find.........

 

[Keith_W]

You're not the only one who had trouble finding it I did too. But now I know it's there, it's not so difficult. It's certainly easier than finding tools in my house, that's for sure.

Just a little note about taking the verification measurement. For a large speaker, measuring all the drivers too close will result in parallax error. Not to mention, you may be off-axis for some drivers. I prefer to take my time alignment verification measurements from further away ... usually the MLP. Your speakers aren't too large. I would probably verify the time alignment from 2m away.

 

[Robbie010]

Noted regards distance.

Here are, the combined Sums with delays applied: