Please Help With a Three Way Crossover

[rlabomb]

Hey Everyone,

I am currently building my first three way speaker system and am basing it on a design from the late fifties. The speakers will use a vintage CTS 15 inch ceramic magnet woofer with an early Trio 8 inch mid/woof and probably a B-G planar transducer for the highs. I have built a few two ways and regularly use military spec hermetically sealed capacitors for my crossovers. They aren't cheap but are very smooth and clean sounding. I want to create the simplest three way crossover possible. I definitely prefer a minimum of components. The 15 will only be used to 500 Hz while the eight will go up to a max of 2500 Hz. Any help with crossover design would be greatly appreciated. Thanks in advance.

--Rlabomb

[BlueWizard]

Well, really a 3-way crossover is just two 2-way crossovers.

Here is a place with a 2-way and 3-way crossover designer -

2-Way XO -
http://www.lalena.com/Audio/Calculator/XOver/Help.aspx

3-Way APC XO -
http://www.lalena.com/Audio/Calculator/APCXOver/

Another 2- Way crossover designer -

http://www.apicsllc.com/apics/Misc/filter2.html


to use two 2-way crossovers -

Calculate the low crossover point and this will give you the values for the woofer and the highpass of the mid.

Calculate the high crossover point, and this will give you the low pass for your Mid and the high pass for your tweeter.

Just a few thoughts.

Steve/bluewizard

[sreten]

Hi,

Those calculators are basically completely
wrong and not the simplest way to do it.

If the drivers are not matched for sensitivity L-pads may be needed.

There have been some interesting posts on minimal series 3-way
crossovers, by Andy Graddon I think, but the search function here
is useless and the links to his pages no longer work.

/sreten.

[BlueWizard]

Sreten,

with all due respect, and I do respect you; your knowledge and experience are infinite compared to mine.

But...how have you helped?

At least I pointed the guy to some resources to actually get the job done.

You claim -

"Those calculators are basically completely wrong and not the simplest way to do it."

But again...no explanation and no alternate resources. And really...what could be simpler than entering the speaker values and getting the L and C values in return.

One of the links calculates Linkwitz-Riley, Butterworth, and Bessel filters. Am I really to believe that those calculation are utterly and completely wrong?

If they are wrong, then point the person to some that are right.

Again, with all due respect...how have you helped?

Steve/bluewizard

[Andy G]

The 3-way series x-o stuff is no longer on my web site due to the limited amount space given by my isp. Also time for upkeep and update is an issue.

(If anyone want the series x-o pages, feel free to email me, I've still got them somewhere)

If you can post the frequency response curves of the drivers , I can see if a simple 3-way series might work, and maybe give some ideas for some starting point values.

[Andy G]

Originally posted by BlueWizard
Sreten,

with all due respect, and I do respect you; your knowledge and experience are infinite compared to mine.

But...how have you helped?

Sreten is correct. these things are "text book" cross-overs, which only work with "text book" drivers.. and I don't think there are many drivers that come close to being that !!

Many some of the Peerless nomex, which are flat as, but most drivers have issues that need to be addressed or avoided.

You CANNOT just take random drivers and apply text book calculations and hope to achieve much more than mediocre, but more likely........ yuck !!!

[sreten]

Originally posted by BlueWizard
Sreten,

with all due respect, and I do respect you; your knowledge and experience are infinite compared to mine.

But...how have you helped?

At least I pointed the guy to some resources to actually get the job done.

[very likely done wrong I'm afraid]

You claim -

"Those calculators are basically completely wrong and not the simplest way to do it."

But again...no explanation and no alternate resources. And really...what could be simpler than entering the speaker values and getting the L and C values in return.

[does simple mean it is correct ?]

One of the links calculates Linkwitz-Riley, Butterworth, and Bessel filters.
Am I really to believe that those calculation are utterly and completely wrong?

[they are theoretically correct and practically near pointless]

If they are wrong, then point the person to some that are right.

[it is not that simple]

Again, with all due respect...how have you helped?

[I could not find AG's stuff on the simplest possible that might work]

[FWIW i regard saying do not do that as helping]

Steve/bluewizard

Hi, well AG's chimed in so that might help, / sreten.

[BlueWizard]

Thanks AG for the response, and thanks Sreten for not taking offense, or at least, not showing offense. Again, it was said ...with all due respect and a great deal of respect is certainly due.

But, I think we might be up against the $1,000 response to a $100 question.

This person made it clear he wants a basic straight forward 3-way crossover. Since he is using what I assume is an old CTS woofer, he likely doesn't have and can't get the specs on it.

So, his choices are work off of generic values or test the speakers to find the T/S parameters and get a frequency response graph.

Which brings up the next question, does this person have a computer, design- modeling- testing-software, and all necessary gizmos and gadgets (all of which cost money, and take time to learn) to test for the parameters? And how much time and money is he willing to spend to get all these gizmos and gadgets necessary to do detailed testing. The basic Dayton woofer tester (WT3) is only about $100 or so, is he willing to spend that much to get it perfect or more perfect?

I think the first step is to make a determination of resources. Which I admit I didn't do in this case any more than you. Then tailor the response to the person's skill and resource level, though with encouragement to move both skill and resource upward.

If he has the material resources, then he should certainly use them. But if not, I think the response should be tailor to the resources he does have.

You say generic calculation won't produce more than mediocre results. But 'mediocre' relative to what? Mediocre relative the the absolute level of perfection you are striving for, or mediocre relative to common low to mid priced consumer speakers?


I think, as a first step, we need to question the selected crossover points. Did this person select these because he has some knowledge that they are workable with these speakers, or did he pick them because they seemed like cool values.

So, more than anything, I think questioning the crossover points is the first step.

The low crossover seems reasonable at 500hz. Certainly that is well within the working range of the 15" woofer, and likely within the working range of the 8" MidBass. But in a 3-way, the workable frequency span is pretty wide.

The high crossover could be a problem though. First, do we have some assurance that the 8" MidBass is functioning well at 2,500hz? Not just functioning, but functioning well. For most midbass, 2,500hz is pushing pretty close to the upper limit.

Next, the B-G planar transducer, are we sure the 2,500hz crossover points is well within the working range of this tweeter?

At this point it might be nice to remind people that rated or stated frequency response is pretty much meaningless. Just because a spec sheet says 2,000 - 20,000hz doesn't mean the speaker is functionally usable across that range. Likely you would have to have a significantly higher crossover than 2,000hz to safely and effectively use this speaker.

So, we need some answers from the original poster.

First is some assessment of his skill and resources. Exactly how much equipment does he have and how much work is he willing to do to optimize the crossovers for these specific speakers?

Next, we need some assurance, that the given speakers (model numbers would be ideal) can actually functionally operate in the selected range.

Once we determine what crossover can be made, then we can move on to how it is made.

Does that sound like a reasonable approach?

EDITED:

Oops, forgot this link-

Elliot Sound Products - Passive Crossover Network Design
http://sound.westhost.com/lr-passive.htm

A pretty good primer for anyone starting out.

Also, here is a link to a discussion on Crossover design with links to other discussions on crossover design.

How do you design a crossover??????

Scroll down to post #14 for more links.


Here is another crossover discussion that goes on for several pages -
3way XO help greatly appreciated!



Steve/bluewizard

[sreten]

Hi,

$1000 answer for a $100 question ? My first one ?

What seems reasonable to you, might not seem reasonable to me ...

Though the questions that need to be asked are starting .....

How old is the trio 8" ? designed for a tweeter or not perhaps ?

I do find crossover primers incredibly dull and misleading.

Getting the original poster to understand to a degree what is going
on is the only way to get near the simplest 3-way possible that is
also not very likely very wrong.

/sreten.

[Andy G]

my guess is that the simplest way would be to grab some electrolytics caps and some variable L-pads, use a cap size for protection designed for around 4k on the BG, and a cap on the 8"so it doesn't confuse the bottom end.
L-pads on both to balance them. Maybe put a coil on the woofer to roll it off at its top end.

It is likely that these older style drivers will actually work quite well that way.

Trying to design higher order x-os for drivers with unknown frequency response is a mugs game, and not worth the drama.

Either do the full measurement thing or go 1st order protection and hope !!.