The proper way to tune a vented design for SQ.

[camplo]

If you really believe that this is an issue and you seem to have decided on a sealed alignment then when it is finished make a passive conjugate network for it and remove the peak from the impedance entirely.

I am not aware of this method.

 

[fluid]

Have a look here for some inspiration

The making of: The Two Towers (a 25 driver Full Range line array)

 

[mark100]

I've never seen any simulation where the group delay on the low end is lower than above that. In your case It actually dips below the fairly steady level at around 200Hz. I don't think this is what group delay does. As I've said before, I think it should rise, not fall. The graph is probably not alles "group delay" for a reason So then what the hell is it?

The graph is definitely group delay

In simulations, the software 'knows' its time of reference that all calculations are built off of.
Not so with measurements, that must apportion whatever total delay is measured into a constant component and a frequency dependent component.
The constant component, loopback, time of flight, whatever you want to call the true constant delay between input and output that doesn't have any frequency dependent time in it...has to be spot on correct.
The delay apportionment process is very difficult to do accurately and repeatably when HF content is missing, due to the nature of linear data collection vs frequency.
If the constant delay apportionment isn't spot on correct, it will reflect itself in the phase and group delay plots being off some.
That's what I was trying to describe in the my previous post.

Sims don't have the problem of apportioning out constant delay.

Plus, sims usually include crossovers which overlay a downward sloping phase rotation across the spectrum, from low to high.
The rotation overlay can make it appear group delay always rises as frequency decreases.
Which isn't true.
The downward sloping overlay can mask what is really going on due to magnitude fluctuations.

Group delay is just the negative slope of the phase trace, so if phase turns upwards as frequency increases, group delay will decrease.
For example, picture a driver with a response that looks like a frown, having magnitude response that rolls off on both ends. (like any real world driver )
Where magnitude is flattest, in the middle of the response curve, that's where phase will be flattest too (and group delay close to zero)
It's phase trace, if properly timed, will be a smiley face, almost a mirror flip of the mag response over the X axis.
The thing to remember with phase and group delay, is that they represent relative time.
So where phase rises on the upper frequency end, group delay will decrease relative to the middle of the drivers frequency range.
A similar phase increasing phenom may be going in with the EQ notch that a linkwitz transform uses in addition to the low end boost. I dunno that for sure, but highly suspect.


Here's a good paper on group delay..http://www.cjs-labs.com/sitebuildercontent/sitebuilderfiles/GroupDelay.pdf

The whole "all time is relative" thing often gets mindbending, for me anyway..
Here's an article on negative group delay
Time Machine, Anyone? - Andor Bariska

 

[voodooless]

The graph is definitely group delay

Thanks for the info, very interesting.

But then there is still a dissonance between what a random group simulation shows us, what that measurement shows, and possibly what is perceived..

So what's up with that? If the measurements reflect your perception, then it would be useful to find out how the simulation can also be matched with this (just nevermind for now that the sample size for perception is 1).

 

[Sonce]

I am not aware of this method.

It is a simple RLC (all in series) in parallel to the woofer. Needs big L and C.

 

[camplo]

It is a simple RLC (all in series) in parallel to the woofer. Needs big L and C.

Never heard of this before...is this a popular thing? Expensive?

 

[camplo]

Have a look here for some inspiration

The making of: The Two Towers (a 25 driver Full Range line array)

Thank you sir, looking into it now.

 

[phase_accurate]

Never heard of this before...is this a popular thing? Expensive?

This is usually done when either your amp has problems with complex loads or when you are using a passive crossover (in order to get a purely resistive load on the filter).
And yes, if you want to do this at the fs of a beefy woofer it gets expensive - less so for midranges and tweeters.

Regards

Charles

 

[camplo]

I do this for the mid-range its going to make it sound very neutral?

 

[phase_accurate]

Depends on how you drive it. I doubt that it will have any influence when driven by an amp directly. But it definitely will if there is a voltage divider or passive crossover inbetween.

Regards

Charles

 

[voodooless]

I do this for the mid-range its going to make it sound very neutral?

It will depend on your amp mostly. If it has a low output impedance it will not matter to much. Generally the tube crowd would use such a Zobels to flatten the impedance as described earlier.

 

[camplo]

That makes sense.....


So I guess we can conclude that tuning lower...or even tuning to half of desired cutoff does increase sound quality on an objective but not so much on a subjective level.

Some of us are seemingly bound to one or the other, more so.

 

[fluid]

So I guess we can conclude that tuning lower...or even tuning to half of desired cutoff does increase sound quality on an objective but not so much on a subjective level.

It depends entirely on your goals and desires if doing either of those things will give a better result. I don't see how any conclusion was reached towards improved "sound quality" just different choices and trade offs. For some SPL rules all, some chase a perfect transient response, others low distortion, directivity or dynamics.