Let's tweak my crossover design! 2 x 12+6

[LowGrowl]

Hi! This is me again doing things i don't really know how to do.

Long story short: Before I was aware of all the Greenboy's fearful fenomena I got some drivers for a DIY project I had in mind.

Three things happened:
- I was able to build one single 12" thanks to the guidance of very nice TB'ers with very good results.

- I built a fEarful 12/6 as per greenboy's excelent plans. (enough said about those great cabs).

- I ended with 3 spare drivers from the first attempt.

So, I have decided to USE those drivers and try to make the best cab and filters for them.

So far

The drivers:
- DeltaLite-II 2512 (2 of them)
Fs = 37 Hz, Xmax = 4.9 mm, Xmech = 8.5 mm, Qes = 0.44, Re = 5.04 ohms, Le = 0.46 mH, Pe = 250 watts, Qts = 0.39, 1-W SPL = 99.9 dB.
- LA6-CMBR

The box:
Vented, very simmilar to fearfull 1212/6 but tuned and rezised to fit the deltalites (I'm specting the same flavor of mi first single 12 with I really liked)

Vb = 140.1 liters
Fb = 43 Hz
F3 = 51.06 Hz

NOW! the crossover design. Here´s where i need your help

I've read as much as I could across the net about crossover design and after a lot of trial and error with X-over 3 Pro (and a little cheating) this is the basic design I have at this moment:



Response (box simulation is included in the graph):


Looks ok?

What about this! Impedance:


No idea on this one!
Phase Response:


So, everything look ok to me but I'm just playing with a software It couldn't be that easy right?

Would you experts be so kind to point out any flaw? something to worry about?

Any advise will be highly appreciated.

Ray

[dhsierra1]

that should be a very nice rig, I know Will33 uses just that sort of setup but he biamps with a GK 800RB and I've used those as well without the 6" with very happy results

I'm just getting into passive xover design myself under the tutelage of someone who's been doing this for years, and unfortunately at the moment the only advice I can give you is that no matter what the modelling looks like, you're going to have to tweak it by ear after you decide on a design

That said, Duke LeJeune is the TB capo de tutti capos on passive xover design so if he doesn't chime in pretty soon I'd fire him a PM.

Good luck

[will33]

Yep, I play a very similair rig and quite like it, though passive XO's are not a strong suit of mine (yet)....one of the benefits of active biamping.


I can say you're right passing the mid up there 3rd order as it's closedback gives it a high fs. The summed frequency response and impedance plots look good to me. I don't know what's up with the phase response in the area of the filters there or if it's close, alright, or really needs improvement.

You are at the point I would just listen to it and make adjuatments, taking impedance measurements around the crossover region to make sure I wasn't doing anything detrimental.

The more parts you add to the filters, the more senaitive they get to tolerance, etc. I might try upping the cap value a little bit in the lowpass to help fight the woofers climbing response, and/or tweak the last cap in the highpass a little watching what that does to all your curve models.



You might be at a point where you just have to b
uy hard parts, and maybe a couple extras (values a little on wither side of the modwl predictions and play with them

[will33]

My biggest concern with messing with any of this is getting the impedamce wrong. A dip there would mean there's too much overlap. A dip that goes lower than that smooth spot in the woofers low mods could mean you've made a 3 ohm cab and are risking your amp.


There's a few guys here who know way more than me about this stuff. I'd wait for some of them to chime in before buying anything. From what I can tell, you're on the right track though. It may seem weird to have the filters set 200-300hz apart like that, but the FR and impedamce curve don't show anything bad. Maybe squeezing together more would help the phase plots line up?....just guessing there. The filter scheme labeled Butterworth is usually -3db at the corner, so squeezing them together would give you a 3db bump there.

[Arjank]

I'll start with a few tips for your crossover design.
-make sure the acoustic slope of the low and highpass are mirrored, though it allready looks promissing it's not optimal
-summation should be 6dB, in your simulation it's about 3-4dB which is an indication that the phase between the low/high-pass is not optimal.

Does that simulation programm of yours also take bafflestep into account?
I use Boxsim for my simulations, this program takes baffle-step into account which is important.
If the programm only uses the anechoic frequency response graph to calculate the response with the crossover the programm is IMO of not much use.

Edit: I see the programm let's you select BW, LR, Bessel(I guess), is it also possible to change the individual values? I never think in BW, LR or whatever, I want the drivers summation to be right acoustically. I don't give a damn what it does electrically (as long as I don't get a severe impedance dip).

[mystic38]

looking at the phase graph, for giggles i would reverse the mid driver...

[Arjank]

Originally Posted by mystic38

looking at the phase graph, for giggles i would reverse the mid driver...

You will not see a large dip in the frequencyresponse here because the summation is only 3-4dB(should be 6dB)
It's a good tip though for checking if the phase is ok.
Normally you would see a very steep dip around the crossoverpoint when you reverse polarity on one of the drivers.
Maybe I'll post some examples later today so show you what I mean.

[mystic38]

I was only being partly glib

i have designed and built a number of speakers over the years and usually stayed away from high order crossovers due to phase related issues.. in testing this speaker i would certainly do a test with the mid reversed.. phase errors are often (to me) more irritating in listening tests than some small dip in response...

it is important for the OP, as a new user of design software, (and assumedly not a electronics engineer) to NOT have blind faith in the graphs...so testing, and experimentation during testing, is to be encouraged, as is modification of the cabinet damping, bracing and stuffing.

Originally Posted by Arjank

You will not see a large dip in the frequencyresponse here because the summation is only 3-4dB(should be 6dB)
It's a good tip though for checking if the phase is ok.
Normally you would see a very steep dip around the crossoverpoint when you reverse polarity on one of the drivers.
Maybe I'll post some examples later today so show you what I mean.

[DukeLeJeune]

Putting on my guru robe, let me solemnly proclaim that, if you have a good crossover design program, you can predict in advance which polarity works best for the midrange driver nearly 50% of the time. So in other words, try it both ways, with program material (easier to judge), and the loudest polarity will be the correct one. The midrange cone may well be a significant fraction of a wavelength closer to the audience's ears than the woofer cone in the crossover region, but that is just one of many factors that can come into play.

Taking my guru robe off now... hmmm, it says "Holiday Inn" on the back...

[Arjank]

Originally Posted by DukeLeJeune

Taking my guru robe off now... hmmm, it says "Holiday Inn" on the back...

Wish I could take it off as easy as you Duke! Maybe it gets easier once I get a little older.

[LowGrowl]

Originally Posted by dhsierra1

that should be a very nice rig, I know Will33 uses just that sort of setup but he biamps with a GK 800RB and I've used those as well without the 6" with very happy results

I'm just getting into passive xover design myself under the tutelage of someone who's been doing this for years, and unfortunately at the moment the only advice I can give you is that no matter what the modelling looks like, you're going to have to tweak it by ear after you decide on a design

That said, Duke LeJeune is the TB capo de tutti capos on passive xover design so if he doesn't chime in pretty soon I'd fire him a PM.

Good luck

Thank you for the audvice... I'm going that way,
BTW I contacted Duke and he replied very fast!

[LowGrowl]

Originally Posted by Arjank

I'll start with a few tips for your crossover design.
-make sure the acoustic slope of the low and highpass are mirrored, though it allready looks promissing it's not optimal
-summation should be 6dB, in your simulation it's about 3-4dB which is an indication that the phase between the low/high-pass is not optimal.

Does that simulation programm of yours also take bafflestep into account?
I use Boxsim for my simulations, this program takes baffle-step into account which is important.
If the programm only uses the anechoic frequency response graph to calculate the response with the crossover the programm is IMO of not much use.

Edit: I see the programm let's you select BW, LR, Bessel(I guess), is it also possible to change the individual values? I never think in BW, LR or whatever, I want the drivers summation to be right acoustically. I don't give a damn what it does electrically (as long as I don't get a severe impedance dip).

Originally Posted by Arjank

You will not see a large dip in the frequencyresponse here because the summation is only 3-4dB(should be 6dB)
It's a good tip though for checking if the phase is ok.
Normally you would see a very steep dip around the crossoverpoint when you reverse polarity on one of the drivers.
Maybe I'll post some examples later today so show you what I mean.

Thank you Arjank, Yes the program let you change the individual values. I started with those alignments and with the LP so far from the HP trying to tame a nasty peak on the deltalites arround 1 to 2 KH. I've tried those drivers in a propper box without any filter and that peak is no pleasant at all. Finally i adjusted the low filter a little up to get a flatter response. I'll post the graphs in a minute.

And yes I will have to test and adjust by ear as everybody sugest. I knew it would mpt be so easy

[LowGrowl]

Originally Posted by DukeLeJeune

Putting on my guru robe, let me solemnly proclaim that, if you have a good crossover design program, you can predict in advance which polarity works best for the midrange driver nearly 50% of the time. So in other words, try it both ways, with program material (easier to judge), and the loudest polarity will be the correct one. The midrange cone may well be a significant fraction of a wavelength closer to the audience's ears than the woofer cone in the crossover region, but that is just one of many factors that can come into play.

Taking my guru robe off now... hmmm, it says "Holiday Inn" on the back...

Hey Duke!
I definitely going to do that testing, reversing the polarity of the midrange in the software gives a strange "s" shape in the crossover region and a mess in the phase response but I don't know if the program accounts for the kind of mounting of the drivers on the baffle.

Ray

[LowGrowl]

Hi!

thank you all for your suggestions! No doubt Talkbass is THE froum

I finally changed some values and got a nicer response. The impedance curve improved as well (flatter). As for the phase graph, well we don't really know right? hehe. well i'll try both polarities.

The new graphs:






It's not very different but it seems better.

I'm building the box right know and I will be ordering the electronic parts sometime soon.

Next question is:
Do you have any tip about debugging the thing? Some sort of measurement and adjustment empirical protocol? I know everybody is going to say "trust your ears", "If it sound right leave it that way" But, come on I'm a crosover design virgin!

Ray

[Arjank]

LowGrowl,

Though the overall response looks smoother there are a few things to consider.

1. the crossover-point is a bit on the high side for these drivers, it's better for the dispersion that the 12's quit doin' their job at approx 1khz. So try to push the crossover-point one octave down.
2. the summation is still not optimal, try to get it it to 6dB as close as possible.

Do you know if that programm of yours takes baffle-step into account? Is it freeware?
If so I can install it and fiddle around with the values to come up with an alternative crossover.

[farboozle]

the summation is still not optimal, try to get it it to 6dB as close as possible.

Can you elaborate on this? Are you looking at the summation at one frequency?

[Arjank]

Originally Posted by farboozle

Can you elaborate on this? Are you looking at the summation at one frequency?

Summation at the crossover point, in the above simulation the frequency is 2khz.

[LowGrowl]

Originally Posted by Arjank

LowGrowl,

Though the overall response looks smoother there are a few things to consider.

1. the crossover-point is a bit on the high side for these drivers, it's better for the dispersion that the 12's quit doin' their job at approx 1khz. So try to push the crossover-point one octave down.
2. the summation is still not optimal, try to get it it to 6dB as close as possible.

Do you know if that programm of yours takes baffle-step into account? Is it freeware?
If so I can install it and fiddle around with the values to come up with an alternative crossover.

Thank you very much for your interest

The program I'm using is X-Over 3 Pro, It's not freeware but you know... I does take baffle-step into consideration. however, In this simm It is disabled, however you can only turn it on/off, there's no way to enter the drivers position, baffle size, shape, etc. I' will run one simm with that option "on" to see what happens.

About the corossover freq, I was following the rule of thumb of crossin the mid at least on octave higher than it Fs (460Hz). Perhaps I'ts only a mith, I don't know

See here LA6_CBMR

So, what do you think would be the optimal crossing point, or the lowest freq that would be safe for the high pass?

Thanks again!
Ray

[LowGrowl]

So:

Originally Posted by LowGrowl

Hi!

Next question is:
Do you have any tip about debugging the thing? Some sort of measurement and adjustment empirical protocol? I know everybody is going to say "trust your ears", "If it sound right leave it that way" But, come on I'm a crosover design virgin!

Ray

Sorry for quouting myself, I'ts like liking your own status on facebook

[will33]

Octave higher than fs is 920 hz.

May have stray from the usual alignments/tweak part values, etc. to fight the 12's climbing response....and because the 6's impedance curve doesn't have a smooth spot in it of any appreciable bandwidth. Those 2
things will effect how the parts act on the actual drivers.