Anyone want to reverse engineer this crossover for me?

[eperson]

http://www.ascendantaudio.com/Arbiter%20TM%20crossover%20schematic.pdf

Any of the forum "brains" with more electronics knowledge than I do want to take a shot at reverse engineering this crossover? Looking for crossover freq, order, and any other info that is important to know like attenuation, etc.
I know a little of what is going on, but dont have enough knowledge to really figure out the details of this design.
Thanks!

[cjd]

Off the cuff, it's funky on the tweeter for sure.

Woofer is nothing particularly fancy. Looks vaguely like it might be following a LR3 if such a thing existed.

I'm just throwing it at stock perfect response and impedance @ 8ohm.

C

[eperson]

Oh yes, the woofer is 4 ohm and tweeter is an XT25 which I believe is also 4 ohm.

[eperson]

And yes, it looks like a 3rd order on the woofer, with an extra resistor thrown in. The inverted phase on the tweeter is especially interesting on the tweeter side, since it looks much like a 4th order.

[eperson]

Anyone?

Jon, I'm looking in your direction :lol:

[Dennis H]

Nobody's going to be able to help you without measured curves from the drivers. We know the tweeter XO has 4th order electrical response and the woofer XO is 3rd order electrical. That doesn't tell us much about the combined acoustical response without knowing what the drivers are doing -- frequency response curve, impedance curve, baffle dimensions, etc.

[eperson]

Well that is getting into the actual design of the crossover. I am just looking for what the electrical characteristics of the circuit are.
To start off with, what is the crossover point? What role are the resistors playing, attenuation only?
I dont have much of an FR plot for the woofer, and while I assume the vifa xt25 is a standard unit, I cant be sure. Im not looking to analyze the design on this, just looking for info on the circuit itself.

[eperson]

Rough specs on the woofer here: http://www.ascendantaudio.com/poly_6.5_page.html
XT25, well you know about that already.

[Paul H]

Well that is getting into the actual design of the crossover. I am just looking for what the electrical characteristics of the circuit are.
To start off with, what is the crossover point? What role are the resistors playing, attenuation only?
I dont have much of an FR plot for the woofer, and while I assume the vifa xt25 is a standard unit, I cant be sure. Im not looking to analyze the design on this, just looking for info on the circuit itself.

No one's trying to be difficult, but the crossover point for a circuit is dependent on the speakers frequency response and impedance curves, so it isn't simple.

Those resistors placed after the parallel portions of the circuit are different. On the tweeter, they will increase the crossover frequency with a slightly different roll-off rate than the inductors alone would provide, and change the phase response. I'm not sure why these were used - maybe the designer couldn't get the phase between tweeter and woofer to align without using them?

Paul

[Dennis H]

Well that is getting into the actual design of the crossover.

Uh, yeah. Otherwise, what's the point?

I am just looking for what the electrical characteristics of the circuit are.

They are irrelevant (and the question is impossible to answer) without considering the drivers. But you've already been told the answer at the simplest level -- 4th order highpass and 3rd order lowpass. Frequencies -- who the hell knows without the driver curves? Phase response -- who the hell knows? You are asking seemingly easy questions that don't have any easy answers. Can't get there from here.

[cjd]

That's not actually a straight-up 4th order on the tweeter - that, or the crossover sim from FRD isn't working with it quite right. It peaks UP to ~6000Hz, drops to another blip/shelf -20dB at 2200 or something. I'm running these numbers off the top of my head from memory, so they're likely not right.

C

[eperson]

Here's what I have to go on:

http://www.ascendantaudio.com/Images/Poly6.5_SPL_impedance.gif

http://www.partsexpress.com/pdf/264-555.pdf

Im not trying to get the exact specifications really, even just a rough idea where this is being crossed at would help. Also any idea why the phase on the tweeter is reversed?
Im not trying to be difficult either, but I imagine that there is some way to figure out what each circuit does just by the values of the components, correct? I mean, I can figure out what basic crossover components I need for a basic xover with just a freq. and impedance of the drivers. Why isnt it possible to figure it out in reverse, at least roughly(within 500hz or so), given that the drivers are operating in a frequency range that suits them and dont have much SPL variation through the crossover region that has to be compensated for, which is a logical assumption given the graphs.
I dunno, I guess being an engineer, I dont always worry that I dont have every little piece of info before starting in on a problem, I have to make reasonable assumptions and get an initial answer, then go back and do a sanity check to see if the answer is reasonable. I cant just throw up my hands and say "oh shit, I dont have every little detail, so even attempting to find a solution is a waste of time"

[Dennis H]

I dont have every little detail

You're still not getting it. What you are calling "little details" are the fundamental specs that everything else flows from. Without them you are pi**ing in the wind.

[eperson]

What details do you need to be able to figure it out?

Can the crossover be modeled in LSPCad, or is more info needed?

[ThomasW]

or is more info needed?

Reread post #6

The crossover in most component sets is commonly cheap, usually using only a couple of parts. Our crossover is far from common, and actually offers a very advanced filter. It is a hybrid elliptical crossover that allows the tweeter to be crossed over very low without power handling problems. The tweeter actually receives a 60 dB per octave, 10th order dropoff until its output can no longer be heard. This allows for a crossover point of 2000 Hz

Statement about the Arco, same tweeter , different woofer....

Since Ascendant publishes this info about one of their systems, I bet an email would get similar info about any of their systems...

[eperson]

Yea good luck with that, Ive emailed Chad numerous times about different things. I never got a response on the theory behind his(?) design. The only response I get is when I want to buy something, then of course money talks

But nevertheless, I thought some people on here could give a bit more insight into the design, and unbiased opinion about its characteristics. Maybe I'm looking at it too simply, but either tell me Im way off *** base and where, or just humor me for a second. Im just here to learn.

Lets just look at the lowpass side. I give it an input, and it attenuates the signal above a certain frequency at a known slope, correct? And the high pass filter does the same thing on the opposite side, albeit at a different slope. Where do the lines cross?
Cant this be done simply in LSPCad?

Edit by moderator to remove profanity, DON'T make me do it again!

[jdybnis]

To get the info you want, simulate the crossover and drivers. Plug the info into speaker workshop, soundeasy, lspcad, etc. The info you need is the drivers' impedance and frequency response, in addition to the circuit diagram you provided.

Without the impedance numbers you can't simulate the circuit to get the electrical transfer function. Without the frequency response, he electrical transfer function won't give you the acoustic transfer function. Assuming a flat frequency response to the drivers' natural rolloffs will probably get you within +-1 octave of the actual number.

Google for the FRD consortium. There is a tool on their site for converting an impedance and frequency response plot into an frd file that can import into your sim-software. If you can get the frd files for the drivers. I'll run the sim for you in Soundeasy.

[eperson]

Thanks Josh, thats what I needed to know. Ill put together some rough frd files when I get home. I believe can probably figure it out myself with the FRD crossover designer spreadsheet.

For a very rough estimate, couldnt you just put in a file that had constant spl of 88db for the entire freq. range for both drivers, and a constant 4ohm impedance curve for both drivers? I think this is what Ive been getting at the entire time, but everyone wants perfect numbers Yes, I know this will give me a wrong answer, but its better than the nothing I have currently.

[cjd]

I did that, and it doesn't help answer the questions you are asking particularly.

If I were to guess, I would put a crossover as likely in the region of ~2000Hz. The tweeter network response is funky. The woofer response is fairly standard. All been said before.

C

[eperson]

I did that, and it doesn't help answer the questions you are asking particularly.

If I were to guess, I would put a crossover as likely in the region of ~2000Hz. The tweeter network response is funky. The woofer response is fairly standard. All been said before.

C

Yea, that would have been my guess as well. Anyone know if its possible to model those extra resistors in parallel with the FRD spreadsheet? I guess Ill find out tonite when I can mess around with this for awhile.

[Mark Seaton]

Im not trying to be difficult either, but I imagine that there is some way to figure out what each circuit does just by the values of the components, correct? I mean, I can figure out what basic crossover components I need for a basic xover with just a freq. and impedance of the drivers. Why isnt it possible to figure it out in reverse, at least roughly(within 500hz or so), given that the drivers are operating in a frequency range that suits them and dont have much SPL variation through the crossover region that has to be compensated for, which is a logical assumption given the graphs.

You need to do some homework on how the varying impedance (with frequency) of a loudspeaker interacts with a crossover. You can't plug in a nominal 4 or 8 Ohm load and expect the response to look like the real thing. The inductance of the tweeter and its resonance would at minimum have to be approximated (possible to do, but not without knowledge of the driver). Even so, it's still a guestimate. The answers you have been given are about as much as can be confidently said.

[eperson]

Yea, Im aware of it, I am going to play around with some of the FRD stuff tonite and see if I can get some more information that I can use.

[Marzen]

SPL Trace from FRD might allow you to generate your own text data file for import into your sim software.

[eperson]

Well I got a chance to make some frd and zma files for these drivers, and those look pretty good for the most part, at least they are as close as the graphs I have to work from. I think that I have narrowed down the crossover point to right around or just above 2k, but that is about all the good I can get out of the program. Trying to model this crossover with the FRD crossover designer, the combined response graph comes out looking terrible. Plus, Im not sure if I can implement the parallel resistors in each leg of the circuit into the simulation accurately. I can put those values into the spreadsheet, but Im not sure if the program is implementing them correctly in the circuit or not. Either way, the tweeter part of the crossover just models terribly. I dont understand why the tweeter cant be rolled off at a regular 3rd or 4th order, in my experience the xt25 is very nice and forgiving to cross over.
The woofer is at least decent as far as the crossover part is concerned, and the design of that part of the circuit seems fairly logical.
I can give anyone that wants them the frd and zma files if they would like to play around a bit.

[jdybnis]

Email them to me and I'll give it a shot. Email me at gmail. I've got the same user name there as I do here.

[eperson]

Sent, thanks for the assistance. The graph is going to look a bit scary when you put in the crossover data I believe

[jdybnis]

Here it is. My first guess is that the woofer data is more reliable since chad uses a modified XT25. Maybe I have the tweeter efficiency too high. The crossover point could be around 4k and the entire response slopes down 12dB from 100Hz to 20kHz? Or the maybe the tweeter response is totally off and crossover point is around 1.5k to make the response flat above 400Hz. That sounds reasonable for a home speaker. This is for a car right? I wouldn't rule anything out.

[peterS]

Here it is. My first guess is that the woofer data is more reliable since chad uses a modified XT25. Maybe I have the tweeter efficiency too high. The crossover point could be around 4k and the entire response slopes down 12dB from 100Hz to 20kHz? Or the maybe the tweeter response is totally off and crossover point is around 1.5k to make the response flat above 400Hz. That sounds reasonable for a home speaker. This is for a car right? I wouldn't rule anything out.

car i believe

[cjd]

That's about what I expected to see on the tweeter.

Odd isn't it?

C

[eperson]

The crossover is for home theater. I dont know that the tweeter response is right either, just making assumptions based on a normal xt25. I know there are a few versions of them. Thats roughly the response I got, but it looked worse because I couldnt implement the resistors correctly I dont think. I dont like the woofer crossover at all, it starts rolling off much too soon in my opinion.

[jdybnis]

The crossover probably includes baffle step compensation. And don't forget the box. Play around with ported enclosures tuned somewhere around 40hz and the low end response might flatten out.

[eperson]

Yes, I am going to run a 0.75 cu ported box, tuned somewhere just south of 40. Yea, I guess the baffle step might flatten out that midrange response a bit, and maybe the tweeter isnt exactly like I have it modelled and it might have a dip where that hump is that is being corrected for. Either way, ill just have to measure the response when the drivers come in next week and deal with any issues in the response later.

[alexnick]

A little bit off-topic: Which is the most accurate crossover? The transient perfect one?

[cjd]

A little bit off-topic: Which is the most accurate crossover? The transient perfect one?

A correctly designed one.

Define accurate. The transient perfect guys will of course say theirs is. Those of us that find high order slopes and funky filters useful will suggest they simply enjoy the distortion that introduces, because finding drivers that actually cover the full spectrum they're being asked to play without distortion artifacts well within the audible range in these instances is a task... perhaps impossible. In the end, who knows.

C

[eperson]

Anyone ever heard of an xt25sc50?

[jdybnis]

pic?

[eperson]

Looks like any other xt25, just with the small mounting flange. Never heard of that particular model number before though, or cannot even find mention of it once on the internet. So like much of this project, I have limited information.

[eperson]

http://206.13.113.199/ncdiyaudio/mark/Testing/logic_xt25/logic_xt25neo.htm

This is probably the closest I can get to the right graph. Description matches pretty well, Logic by Vifa and the small faceplate.

[alexnick]

Quote:
Originally Posted by alexnick
A little bit off-topic: Which is the most accurate crossover? The transient perfect one?


A correctly designed one.

Define accurate. The transient perfect guys will of course say theirs is. Those of us that find high order slopes and funky filters useful will suggest they simply enjoy the distortion that introduces, because finding drivers that actually cover the full spectrum they're being asked to play without distortion artifacts well within the audible range in these instances is a task... perhaps impossible. In the end, who knows.

C

Let's say that we have ideal speakers in their frequency range, which is the most accurate xover?(phase, linearity,off-axis/on-axis response,IM, etc.)

[cjd]

If everything is perfect, no crossover is going to be the best. Single driver with perfect linear response full spectrum, no distortion artifacts, no anomalies...

Once again - in their frequency range, metal cone drivers with steep filters are hard to beat. On the flip side, they're pretty unlistenable outside their ideal range. Floppy mushy cones tend to introduce "pleasant" distortion everywhere, but are far less miserable outside their ideal range. It's not anything of an absolute here - your question is pretty directly looking to have an answer to one of the big trade-off decisions that a designer has to make.

C

[alexnick]

cjd, I'm still wondering whether a 2-way system with a steep filter is better than a 3-way one designed for/with a transient-perfect xover?!

[Mark Seaton]

Alexnick... That one is a question and debate for the ages. Both have trade offs once you have to deal with real devices.

[cjd]

Well, what about a 3-way with steep filters!

Alexnick... That one is a question and debate for the ages. Both have trade offs once you have to deal with real devices.

Yup. What I've been (trying to) say(ing).

C

[eperson]

Well I got them up and going, at least one of the two towers so I could test it out. Seems to be all ok, although I am interested to see what they sound like when I get them both done, and measure the response curve. They seem to be pretty power hungry at least