or Connect
AVS › AVS Forum › Audio › DIY Speakers and Subs › Will this Octo 18 build plan be suitable?
New Posts  All Forums:Forum Nav:

Will this Octo 18 build plan be suitable?

post #1 of 69
Thread Starter 
Looking to build an Octo 18 sub setup for a large space. The room is very large and open to entire first floor. 17' ceilings throughout. I have tried a ported approach using JTR Captivators (2 passive subs) driven by a Crown K2, and while they were great for HT use, I didn't care for their musicality on some program material. I would like to try the multiple sealed 18" approach utilizing the Dayton HO-18. I wanted to do 2 cabinets with 4 forward facing drivers in each up front. Off of the rear sub output of my Yamaha RX-A3010 AVR I will run my 2 current DIY subs (each with a 12" & 15" Dayton Reference HF driver and a 15" PR) off of the Crown K2. I picked up two of the CV5000 amps to put 5000 watts (or whatever it really puts out) into each of the quad 18" cabinets. I was considering building a LT circuit for the new 18" sub cabs because of the possible lack of room gain from the large open area. Do I need to make that decision now in choosing a cabinet volume or would 3.5 cu. ft. per driver work regardless? Would 1250 watts per driver be enough to handle the demands of a moderate LT circuit (target F3 of 16-20 Hz) and would the driver have enough Xmax to make it work? Is this entire approach feasible?
post #2 of 69
Yes.

I approve of this build. Do it! biggrin.gif
post #3 of 69
3.5 cubic feet per driver x4 is a good size with what is likely somewhere around 4-5k real power per amp. that is pushing the drivers pretty hard, but it should all hold together. i assume that you have read the review on data-bass.com of that driver. of course, the power line to the room needs to be considered. a 15 amp line isn't going to be sufficient. the drivers have sufficient excursion to eq them however you like. it just comes down to some form of protection. in your case, you are talking about relatively small enclosures to hold excursion down as the amp goes into clipping. that is a reasonable approach, but ideally you would test the excursion and power incrementally just to be sure. sounds like a killer build. ~126db @20hz in model 2pi space assuming no compression and full power is what winisd is showing using ricci's measured driver t/s parameters. since that driver will be in the ballpark of q=0.707, a full blown L/T isn't needed, a simple shelf filter is all you will need.
Edited by LTD02 - 3/30/13 at 5:58am
post #4 of 69
I went from a pair of ported caps to 8 of they Dayton 18's in the past few months and never looked back. Low end is great, and the mid-bass is simply killer. I'm running just under 3.5 cubes per driver with the same kind of power going to them (LG 14K clone) with no problems. Also I would remove or repurpose your old subs rather than trying to combine/integrate them with the new setup.

Just because you have a large room doesn't mean you will get 0 room gain though. What are your room dimensions?
post #5 of 69
Quote:
Originally Posted by LTD02 View Post

3.5 cubic feet per driver x4 is a good size with what is likely somewhere around 4-5k real power per amp. that is pushing the drivers pretty hard, but it should all hold together. i assume that you have read the review on data-bass.com of that driver. of course, the power line to the room needs to be considered. a 15 amp line isn't going to be sufficient. the drivers have sufficient excursion to eq them however you like. it just comes down to some form of protection. in your case, you are talking about relatively small enclosures to hold excursion down as the amp goes into clipping. that is a reasonable approach, but ideally you would test the excursion and power incrementally just to be sure. sounds like a killer build. ~126db @20hz in model 2pi space assuming no compression and full power is what winisd is showing using ricci's measured driver t/s parameters. since that driver will be in the ballpark of q=0.707, a full blown L/T isn't needed, a simple shelf filter is all you will need.

I've tried to address this before, but folks love WinISD, even when Josh posts the actual freakin' response and they're starin' right at it. cool.gif

The driver in 3.5 cubes will give a Qtc response of 1+, not .707. An L/T circuit doesn't give a hoot what WinISD or any theory says, it will only do what you ask it to do, and if you ask it to transform a .707 Qtc, that's exactly what it will do.

Here's the actual measurement from Josh with a Qtc of .707 and a Qtc of 1.0 overlaid. Which of the 2 does the actual response more closely reflect?



The +10dB L/T shown is designed to affect a Qtc of 1.0, not .707. IOW, a specialized transform circuit vs 'a simple shelf '. The result is preferable, and will give the OP his desired new F3 in the high teens.

IMO, a +10dB L/T is what will be required in that room, which means that you'll basically be running the subs +10dB hot for all effects below 20 Hz when you're running the system calibration flat. This is important to remember when playing soundtracks with heavy <20 Hz content and 'running the sub hot' and needs to be carefully considered in the design phase.
Edited by bossobass - 3/30/13 at 8:46am
post #6 of 69
"I've tried to address this before, but folks love WinISD, even when Josh posts the actual freakin' response and they're starin' right at it.
The driver in 3.5 cubes will give a Qtc response of 1+, not .707."

here is what i said, "since that driver will be in the ballpark of q=0.707, a full blown L/T isn't needed, a simple shelf filter is all you will need."

here is the measured response of the driver:



http://www.data-bass.com/data?page=system&id=89&mset=96

that is definitely in the ballpark of q=0.707.

the 18ho doesn't have major inductance problems like many other drivers, so the winisd model gets you close enough. room effects will be more significant than the last bit of model tweaking. ;-)
post #7 of 69
what makes the two graphs a little confusing is the scale changes.

here is the single sweep that bb posted along with the power response series that i posted.

they are the same.

post #8 of 69
Thread Starter 
Thanks for the assistance...it's nice to get feedback from the experts.
Yes I have read about it on data-bass. That's why I was counting on it having a bit more Xmax than the manufacturers spec of 12.75mm to make it possible to do an LInkwitz transform circuit or a negative gain shelf filter. I have provided the amps with 3 separate 10 gauge runs, each on a 30 amp breaker. An additional 20 amp line runs the receiver and ancillary equipment.
The room dimensions are about 20 x 20 x 17' ceilings, but the whole thing is wide open to the entire first floor with high ceilings and about 18,000 cu ft. total.
Once I nail down the proper cabinet size, which it sounds like is about 3.5 cu ft per driver, is to take the numbers crunched by Bassbox 6 program and enter them into the LT spreadsheet from the Elliot Sound Products website and build the appropriate circuit. However when I model the HO18 using Ricci's measurements in Bassbox 6 in a 3.5 cu ft box with heavy damping it gives me a QTC of .51 and an F3 of 49 Hz. What am I missing?
post #9 of 69
Thread Starter 
OK, ran the numbers again in Bassbox with no fill in the box and got Qtc of 0.618 and F3 of 42.25Hz. That's closer to what you guys have been saying but still quite a way off from the 1+ number stated above.
post #10 of 69
sounds like you are already to go.

i would suggest holding off on the l/t circuit until you are up and running and have some idea of what effect the room will have on the response. you may not need as much l/t boost as you might think based purely on a groundplane 2pi space model (winisd/bassbox).

the minor differences in q are likely down to some of the advanced parameters. winisd uses ql=10 and qa=100.

with settings of ql=1 and qa=10, i get qtc=0.608, which is pretty much what you have so i'd guess that is the difference.

edit 1:
here is what winisd help file has to say about those two parameters:

Under "advanced->", you can find controls for box losses. There, you can see what effect the each loss type will have on your box design (Smaller the value, greater the loss). There are three types of losses:

Ql, leakage losses. These are produced by leaks in enclosure or in driver itself. Generally, this is most dominant loss type in vented boxes. Typical value is 5-20. This is quite impossible to predict before building the actual box. For reasonable quality box, WinISD pro uses Ql of 10 by default.
Qa, absorption losses. These are produced by losses in enclosure. Any stuffing will increase the absorption loss. With no stuffing inside the box, 100 is typical. Heavily stuffed enclosure has this about 3-5.

edit 2:
pretty mid boggling how some of this 'stuff' can be modelled.

ricci posted measurements of a stuffing experiment on his site here:

http://www.data-bass.com/data?page=content&id=79

changing qa in winisd seems to capture the change in response and impedance measured in the article.
Edited by LTD02 - 3/30/13 at 7:34pm
post #11 of 69
Thread Starter 
I would suggest holding off on the l/t circuit until you are up and running and have some idea of what effect the room will have on the response. you may not need as much l/t boost as you might think based purely on a groundplane 2pi space model (winisd/bassbox).

Agreed, I just wanted to check with you guys to see if you think that, with the room size and power available, if 3.5 cu ft is the best enclosure size to build which would work whether I use a L/T or not.
Also wanted to mention that the goal for the cabinet is to use oak veneered plywood, finished in black stain or polyshades to match my other cabinets where you still see the grain. I plan on using oak solids on all edges. Dowel bracing. Double thick front baffle. Pretty standard stuff. Should end up something like Popalock's front stage subs, minus the triple front baffle, which I'm guessing is needed due to the SI 18's high surround - longer excusion for grill clearance. The extra strength doesn't hurt either I'm sure but don't know if I need that.
post #12 of 69
Quote:
Originally Posted by LTD02 View Post

what makes the two graphs a little confusing is the scale changes.

here is the single sweep that bb posted along with the power response series that i posted.

they are the same.


Silly me! After building some 100 L/T circuits, all I had to do was ignore the actual response (and Siegfried's advice and spreadsheets) and build a 2nd order shelf. rolleyes.gif

Compressing and stretching the scale does not change the Qtc the L/T will see, nor does it change the data in my post. Going from a 4.2 cubes box to a 3.5 cubes box will alter it further.

Quote:
Originally Posted by Linkwitz Riley View Post

OK, ran the numbers again in Bassbox with no fill in the box and got Qtc of 0.618 and F3 of 42.25Hz. That's closer to what you guys have been saying but still quite a way off from the 1+ number stated above.

The Qtc the L/T will see is around 1.1 to 1.2, NOT .61. There's a big difference and it will show up in both the naked and the in-room responses, which you'll just have to deal with after the fact anyway. I don't use smoothing EQ and this is just one reason I don't have to.

Really, why model or measure at all if you're going to say a 1.1 Q is the same as a .61 Q? You might as well just grab a handful of components and solder them on the board wearing a blindfold.

The only reason I went to the trouble to post the graphs earlier is because we do have the actual response from a reliable tech, and the difference between it and the modelers is apparent when it comes to building a L/T.

Toss the BB P6, Unibox and WinISD. They're only good for the inane discussions you read here on the subject of late and pretty much nothing more.

Worst case, just build the sub and post EQ the response. I'd rather see you do that than waste time and $$ building the wrong L/T circuit.
Edited by bossobass - 3/31/13 at 8:48am
post #13 of 69
Dupe post...
post #14 of 69
"The Qtc the L/T will see is around 1.1 to 1.2, NOT .61."

why would that be the case? if stuffing the enclosure reduces the qtc to 0.61 that would seem to be what the l/t would see. or, is what you are saying is that in your guesstimation the enclosure that models as 0.61 will measure as 1.1-1.2 and that is what the l/t will see?

"Toss the BB P6, Unibox and WinISD. They're only good for the inane discussions you read here on the subject of late and pretty much nothing more."

now, that would be silly!
Edited by LTD02 - 3/31/13 at 11:23am
post #15 of 69
Quote:
Originally Posted by LTD02 View Post

"The Qtc the L/T will see is around 1.1 to 1.2, NOT .61."

why would that be the case? if stuffing the enclosure reduces the qtc to 0.61 that would seem to be what the l/t would see. or, is what you are saying is that in your guesstimation the enclosure that models as 0.61 will measure as 1.1-1.2 and that is what the l/t will see?

"Toss the BB P6, Unibox and WinISD. They're only good for the inane discussions you read here on the subject of late and pretty much nothing more."

now, that would be silly!

This is what you'd get doing it your 'modeled way vs the way it should be done:



This isn't something that's debatable. You're free to see it however you wish, but offering people who are actually going to go through the effort and expense incorrect information is what's silly.
post #16 of 69
oh, i see, you are assuming that the stuffing does nothing to change q. fail.
post #17 of 69
Oh you two. rolleyes.gif
post #18 of 69
Quote:
Originally Posted by Scott Simonian View Post

Oh you two. rolleyes.gif


lol biggrin.gif
post #19 of 69
here is what i was talking about earlier.

the blue line is a model of the measured response.

the yellow line is a shelf filter to bring up the bottom end.

the target bottom end is the green line, -3db at 20hz in this example.

the red line is the new response with a bottom end that is -3db at 20hz.



edit: it is kind of difficult to see the red line, but it is essentially tracing right underneath the green low end target curve.
Edited by LTD02 - 3/31/13 at 9:11pm
post #20 of 69
Quote:
Originally Posted by LTD02 View Post

here is what i was talking about earlier.

the blue line is a model of the measured response.

the yellow line is a shelf filter to bring up the bottom end.

the target bottom end is the green line, -3db at 20hz in this example.

the red line is the new response with a bottom end that is -3db at 20hz.



edit: it is kind of difficult to see the red line, but it is essentially tracing right underneath the green low end target curve.

Yes, I understand what you were talking about. Your red line is shown in my graph above (the bottom graph of the 2). The top graph is the correct L/T for this sub, not the shelf you suggest. Seems like you ignore what I've posted and just keep pitching the flawed approach, but scale your red trace onto my graph and you'll see they're essentially the same.

If your aim is to build a L/T circuit for this sub, you would not use .61 as the Qo. If your aim is to throw a 2nd order shelf at the problem, that's a whole different story in which the Qtc is irrelevant.
post #21 of 69
Quote:
Originally Posted by Scott Simonian View Post

Oh you two. rolleyes.gif

Yeah, I know... we should be talking about cracked sheet rock and blurry vision, but we just can't help ourselves. rolleyes.gif
post #22 of 69
Thread Starter 
Bossobass,

If I elect to go the L/T route with this driver and a 3.5 per driver enclosure (unless you recommend a diiferent size), can you please provide the numbers below that you feel would be most appropriate for this driver based on your experience in building these circuits in the past? Thanks!

f(0) = ? Hz
Q(0) = ?
f(p) = ? Hz
Q(p) = ?
post #23 of 69
Quote:
Originally Posted by bossobass View Post

Yeah, I know... we should be talking about cracked sheet rock and blurry vision, but we just can't help ourselves. rolleyes.gif

Lol! C'mon, Dave. wink.gif I would just like to see you two not do the snippy back and forth sometime. smile.gif

How is everything going, btw? We are missing you on the DB forums.
post #24 of 69
"If your aim is to build a L/T circuit for this sub, you would not use .61 as the Qo. If your aim is to throw a 2nd order shelf at the problem, that's a whole different story in which the Qtc is irrelevant."

ok. that's all i was getting at when i mentioned the shelf filter could be used to bring up the bottom end.
post #25 of 69
Quote:
Originally Posted by Linkwitz Riley View Post

Bossobass,

If I elect to go the L/T route with this driver and a 3.5 per driver enclosure (unless you recommend a diiferent size), can you please provide the numbers below that you feel would be most appropriate for this driver based on your experience in building these circuits in the past? Thanks!

f(0) = ? Hz
Q(0) = ?
f(p) = ? Hz
Q(p) = ?

Start with Fo = 35, Qo = 1.2, Fp = 20 Hz and Qp = .71. Shoot for +10dB of boost, no more.

The spread sheet will only let you work within the laws of physics as there is a relationship between the amount of boost and Qo to QP, so just make sure that the k value is >0. If you drop below 0, there will be negative values for the components, which is, of course, impossible. k value can be 0.01, but must be >0.

If you get a <0 k value, change the Qo to a slightly lower number and check the k value. After some experimentation, you'll see what the allowable parameters are and you can fine tune the final response from there. Post a screen cap of the final solution if you get a chance.

Quote:
Originally Posted by Scott Simonian View Post

Lol! C'mon, Dave. wink.gif I would just like to see you two not do the snippy back and forth sometime. smile.gif

How is everything going, btw? We are missing you on the DB forums.

I know that, but it is what it is. I'm just trying to prod LTD into actually building a sub and testing real world tweaks so that he'll get out of the grid and into the real world. His advice will step up a notch, which is a good thing.

Been upgrading the rack, so no SL caps until I'm finished. Going with a new signal chain to see if I can get flat to 3 Hz before the roll off and what that'll do to system requirements. I'm also finalizing the AntiHorn satellites system (mains, CC and surrounds). Pretty freakin' awesome so far... info thread to follow. cool.gif

Quote:
Originally Posted by LTD02 View Post

"If your aim is to build a L/T circuit for this sub, you would not use .61 as the Qo. If your aim is to throw a 2nd order shelf at the problem, that's a whole different story in which the Qtc is irrelevant."

ok. that's all i was getting at when i mentioned the shelf filter could be used to bring up the bottom end.

Yes, I knew that, but it's confusing to people reading who don't know that. The thing to do is to post the result of your advice along with that advice (like you eventually did). If I could get away with a shelf, I would do it, but it isn't the optimal solution.
post #26 of 69
Quote:
Originally Posted by bossobass View Post

I'm just trying to prod LTD into actually building a sub

Ah! We all have tried. wink.gif That's right, John. I think it's about time you had your own build thread! biggrin.gif
Quote:
Originally Posted by bossobass View Post



Been upgrading the rack, so no SL caps until I'm finished. Going with a new signal chain to see if I can get flat to 3 Hz before the roll off and what that'll do to system requirements. I'm also finalizing the AntiHorn satellites system (mains, CC and surrounds). Pretty freakin' awesome so far... info thread to follow. cool.gif

Oh, awesome! Definitely interested to see what you come up with wrt the new signal chain and new speakers. smile.gif
post #27 of 69
lr, based on ricci's measurements, a single l/t cannot eq that response to be flat because a single l/t cannot be used to model it.

out of curiousity, i started with ricci's measured results (blue line) and changed the cab size down to 3.5 cubic feet with stuffing (red line).

with a target flat response that is -3db at 20hz (black line), you can get pretty close by adding this l/t (the purple line):

f0 = 36hz
q0 = 0.90
fp = 22hz
qp = 0.707

when the l/t (purple line) is added to the model response (red line), the result (green line) is within +/-0.5db of the target curve from 100hz down.

post #28 of 69
"Start with Fo = 35, Qo = 1.2, Fp = 20 Hz and Qp = .71."

adding bb's suggested l/t to the red response produces the brown response shown down 15db just so that it could be seen.

obviously that isn't going to work. the problem is the assumption about q and 1.2 is just off the charts too high.

post #29 of 69
How are you guys determining what the Q is? I know there's some debate on what the value is, nut what are you looking at to determine it? In my limited experience I've used Winisd to plug in the driver and box size and used that value.
post #30 of 69
there are two effects at work which makes it difficult to sort through wth.

the first is the q of the enclosure. that is what winisd can calculate based on the work of thiele and small.

the second effect is that of other driver effects. the largest of which is likely inductance. inductance changes up the shape of the frequency response.

if you believe that a subwoofer is a linear device, then the final q is simply the final frequency response.

the problem is that in actual measurements, a q that matches the bottom end rolloff does not model the high end rolloff (inductance effects) and a q that matches the top end rolloff does not model the low end rolloff appropriately.

and since a single filter can't be used to model both the peak and the rolloff, the response doesn't really have a single q
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: DIY Speakers and Subs
AVS › AVS Forum › Audio › DIY Speakers and Subs › Will this Octo 18 build plan be suitable?