2-way crossovers

[89-300ce]

I'm thinking of building my own crossovers. I'd like to run the 511b horn/BMS 4550 driver out of phase with the 2235 or ME150H and use a baffle step on the woofer to time align both drivers at 800Hz. I kinow I should use as steep a slope as possible when trying to time align this way, but too steep a slope will show the crossover more. I'm thinking of a second order Bessel.

Will Zilch's HF attenuation for the 811b/BMS 4550/N-800F (1200Hz) but adjusted for 8 ohm drivers work?

Is this a steep enough slope to not get out of phase cancellations as I move away from the 800hz crossover point?

Is the Bessel an appropriate curve to integrate the horn and cone?

Jorg

[Zilch]

http://audioheritage.org/vbulletin/s...687#post163687

I believe you'll find its asymmetry, highpass being 1200 Hz, lowpass 800 Hz, performs part of the compensation, which is completed by the final parallel trap.

The serial notch knocks down a 19 khz spike characteristic of BMS drivers on these horns.

Compensation portion constructed and retested here:

http://audioheritage.org/vbulletin/s...77&#post165377

Some JBL ~800 Hz LF filters:

http://manuals.harman.com/JBL/HOM/Te...L200B%20ts.pdf

http://manuals.harman.com/JBL/HOM/Te...S3100%20ts.pdf

http://www.jblproservice.com/pdf/Net...%20Network.pdf

http://manuals.harman.com/JBL/HOM/Te...ummit%20ts.pdf

[89-300ce]

Thanks Zilch,

I'm trying to work my way through this.
Is there a freeware calculator that you can enter component values and get frequency? I can only go from frequency to components and it's a pain to work it backwards.

Jorg

[Mr. Widget]

Is there a freeware calculator that you can enter component values and get frequency?

Here is a calculator that does your basic textbook filters... easy to use and it works pretty well. It isn't exactly what you are looking for, but you can probably make it work for you.

http://ccs.exl.info/calc_cr.html


Widget

[89-300ce]

Here is a calculator that does your basic textbook filters... easy to use and it works pretty well. It isn't exactly what you are looking for, but you can probably make it work for you.

http://ccs.exl.info/calc_cr.html


Widget

Yes, thanks, I can make it work. It's just a pain. My numbers never seem to come out exact to the schematics. The coil and cap values never seem to match the network schematics. For the HF driver for instance. The cap determines the Hz and the coil the slope? So if the cap is 1200Hz and the coil number doesn't match the calculators it means they used other than a standard 16db slope?

Jorg

[Zilch]

The slope is fixed by the topology.

The Q is variable, and chosen in combination with frequency for the specific design.

The "textbook" transforms are rarely optimum for any given system.

They are merely points on continua, by this perspective....

[speakerdave]

I'm thinking of building my own crossovers. . . .

To begin to get a sense of what's involved from an informed perspective, read this:
http://sound.westhost.com/lr-passive.htm

David

[89-300ce]

The slope is fixed by the topology.

The Q is variable, and chosen in combination with frequency for the specific design.

The "textbook" transforms are rarely optimum for any given system.

They are merely points on continua, by this perspective....

So I should use tested values like in the JBL 800Hz LF crossovers in the links provided.

Won't the 800Hz lowpass and 1200Hz highpass create a dip at crossover ?

Jorg

[Zilch]

Won't the 800Hz lowpass and 1200Hz highpass create a dip at crossover?

It depends upon the response characteristics of the drivers, horns, and cabinets used. What ultimately matters is the acoustic response, as provided by the elements in combination.

Run the numbers on the 846A and 846B crossovers. What frequencies do you come up with, LF vs. HF? How can that work?

Read up the link Dave just posted.

Referring back to earlier discussions, you need measurement gear to do this, and experience using it, unless you're willing to live with replicating previously proven designs by others, which you've now discovered are not textbook.

You can't just throw drivers and crossovers in boxes and expect an optimal outcome without doing the requisite design engineering, which may involve multiple iterations.

What's my approach to this? Remember that I'm the Quick and Dirty DIY guy, and others do it differently. Make your best educated guess at what will work, based upon what has worked for others. At some point, you have to put down the books and calculator, recognizing that you may have to pick them up again, many times. Build it, measure it, listen to it, refine it as you learn more, trying other options. It's never actually done, but you may enjoy it all the while....

[89-300ce]

So the journey is the reward?

Would the combination of L200 LF filter ( if I go 2235 ) with the N800-8k HF filter and your compensation be a good start point for this project?

If I found I needed to cross the HF lower I would have to modify the notch filter to extend the attenuation down?

I would keep the zobels as they are?

What would you recommend as a starting place for hardware and software test equipment for this DIY journey?

I have a 200w Sumo and a 50w Musical Fidelity. Might it make more sense to spend the test equipment money on a bi-amp setup? This bi-amp would be flexible enough to accomodate other two way's I might try to mess up down the road? Rod Elliott's article sure makes that seem attractive.

Jorg

[speakerdave]

Active crossovers (low-level or line level) do avoid some of the problems passive crossovers (post-amp, or high-level) have interacting with the electrical characteristics of the drivers, but off-the-shelf they are generic or textbook in the sense that they do not take into account the true acoustic performance of the real world drivers in question, because they assume flat response in the pass band and at least an octave into the stop band. That is almost never true, especially in a two-way.

Please understand that I am not presenting myself as an expert on this, but rather someone who is one-half step, perhaps, ahead of you in this learning process.

You could do everything Rod Elliot suggests and it would still only be an educated guess unless you have some acoustic test equipment (like CLIO and a suitable room to use it in) to provide yourself with a feedback loop on the acoustic side of the transducers.

No one can really answer questions about the interaction of specific drivers and existing crossovers unless they have examples on hand (that perform the same as yours) and are willing to do the testing, a very time consuming process. But generally, for example, the low pass filter for the woofer is designed in relation to the acoustic performance of the driver in the high pass section, so it will not work to take the low pass from one system that used your woofer and the high pass from another system that used your tweeter and splice them together, even if they were designed around the same crossover frequency. That would at best be a starting point that would need to be evaluated with acoustic testing of the driver combination you are using.

Generally speaking it is better to copy an already engineered system.

For me the only way all this is worthwhile is if you are trying to DIY a very expensive SOTA system. Of course there the risks are even greater because of the cost for drivers. But, nothing ventured . . . .

That is, at least for now, the state of my apprentice understanding on this.

As for using standard electronic crossovers--can it be done? Of course; it has been done many times, and people have often found them satisfactory.

You might find it informative to read the manual for the 5235 JBL electronic crossover.

http://www.jblproservice.com/pdf/Vin...5%20manual.pdf

David

[4313B]

Thank you for the post David! Very nice!

[Zilch]

So the journey is the reward?

Indeed.

Would the combination of L200 LF filter ( if I go 2235 ) with the N800-8k HF filter and your compensation be a good start point for this project?

It's L200B, actually, an important distinction.

I posted several 800 Hz crossovers so that you could see they were all different in detail. You'll have to start with your own "best guess." I have the L200B LF filter running a system using 2235H here, yes. Is it optimal? I don't know. Everything is "developmental" here. I'm not there yet.

If I found I needed to cross the HF lower I would have to modify the notch filter to extend the attenuation down?

Not likely. Look at the BMS factory curves. They're down at 1 kHz already. The only way to know is to measure.

I would keep the zobels as they are?

The zobels are often not textbook, either, as they are sometimes used to fine-tune the filters. That's also an empirical determination.

What would you recommend as a starting place for hardware and software test equipment for this DIY journey?

You need a high-resolution RTA, at least. I started with a Behringer DEQ2496 ($300,) and still use it. I found its built-in EQ with AutoEQ function convenient for development work, telling me how much to correct, and where. There's also computer-based ones like TrueRTA.

You also need Woofer Tester 2 or Pro.

I have a 200w Sumo and a 50w Musical Fidelity. Might it make more sense to spend the test equipment money on a bi-amp setup? This bi-amp would be flexible enough to accomodate other two way's I might try to mess up down the road? Rod Elliott's article sure makes that seem attractive.

Yes. I'm using a Behringer CX3400 ($130) right now. Measurement HF amp is Sonic Impact Super T-Amp, LF a pair of 40-watt JBL 6210. For listening, it's different setups.

[4313B]

So the journey is the reward?

Indeed.

Sometimes I wonder if some of you guys are using crack pipes that are long past their expiration dates.

Try listening to some finished product sometime...

[JBLRaiser]

Quote:
Originally Posted by 89-300ce
So the journey is the reward?

Indeed.
Yes Grasshopper.