It's a matter of degree, Douglas. XSim is an excellent and easy to use tool. It also tells me that you don't advocate measuring or calculating sound power (not that things have to go that far by any means).
Generally I use XSim to get the crossover within a couple of standard values and then set the thing up, using REW to measure frequency response.
I also do a fair bit of listening as part of the process. I found out long ago that I can sometimes hear better... but I virtually always hear worse.
Oh come on. This is like saying a Shakespeare sonnet is nothing but consonants and vowels. Only TWO things that you put together... How hard can it be?
My opinion is this:
Whether it's developing a crossover by ear, from scratch or based on one before, or using Allen's method (also similarly described in a nice presentation by Dave Dal Farra on my web site HERE, near bottom of page) these are things best left to people with experience and knowledge. My fear is that a novice will get the whiff of "no measurements needed" and launch off of a cliff without a parachute. This is also why I feel that many DIYers who decide to try an open baffle system will fail. They don't have the appropriate knowledge or tools to know what they are doing. Points for jumping in and trying it I guess.
Even when you use manufacturer datasheet information combined with modeling and so on, why not just go that one extra step and get actual data on your speaker directly? The ability to measure your speaker is available to anyone these days.
So I always advocate measurements. Of course measurements must be done carefully, and the DIYer should also understand what a certain measurement is telling you, and what it is not telling you, e.g. a nearfield measurmeent or an indoor farfield measurement that also includes the room. These things take some time to understand I think. But there is no substitute for real data on your speaker. The rest is guesstimation to one degree of another, and a good and experiences guesstimator will likely be much better than a novice one.
Thanks for the clarification Charlie, I take your point. As you say learning how to do proper measurements and interpreting them correctly is no small matter in itself, so it depends on the OP, ie, he isn't necessarily guaranteed to get better results that way, Allen's method is just another option.
I trust you Douglas Blake, but could you be more specific about the "more than" compared to "A crossover is nothing but choices of freq, slope, filter pattern, and shelving volume"?
Is there some further parameter in a crossover amenable to human influence I'm not seeing?
There are various considerations that go into choosing those parameters. But mostly empirical questions that can't be addressed very well by simulation. You need to know - in your room and at your chair - how each of the drivers is performing. That's mostly a matter of getting a reading at your chair (with anywhere else in the room as a tolerable rough estimate).
Then you need some judgment about how to cut up the music band. My instinct is to have a mid-range that can on its own cover the whole wide middle octaves.... or if I have a largish dome tweeter, give it as many octaves as it can handle since those are good drivers.... and judgments of that sort.
In many years of fooling about with speakers, I have yet to hear any difference between one filter pattern and another on music.
B.
Last edited:
old- no dsp
LC Audio
L C Audio Technology / Series Filter
Acoustic Reality
The Acoustic Reality Series Crossover, AR-SXO
Maybe check how they tune the circuit, "How to fine-tune the ARsxo yourself", this might help.
or have a reference speaker you can compare with as you fine tune.
LC Audio
L C Audio Technology / Series Filter
Acoustic Reality
The Acoustic Reality Series Crossover, AR-SXO
Maybe check how they tune the circuit, "How to fine-tune the ARsxo yourself", this might help.
or have a reference speaker you can compare with as you fine tune.
Last edited:
Yeah, the little matter that all this interracts with the cabinet, the drivers themselves and can, if not tested, produce absolutely horrific outcomes.
For example, you might sit and decide to use 300 and 2000hz for your crossover points... but unless you allow for the drivers themselves you can end up with huge gaps in response or noise making overlaps. All these settings interract with the physics of the box and drivers, they are not chosen randomly or according to taste.
Last edited:
Instead of me suggesting the effects are trivial, can be anticipated, or apparent from benchmark FR runs I recommended, I'll just say, "How about posting some plots that demonstrate the effects you are talking about". (I won't even ask, as I really prefer, to see any evidence these effects are audible.)
I'm not arguing there are no interactions that a purist might name. But basically you can't null away all interactions because you don't have the tools to assess them or address them if you knew them nor might you actually hear them on music.
B.
I have a pair of speakers (Cambridge Audio SX-60) that have never sounded right since I bought them. They sounded weak, constrained and it seemed the woofer just wasn't breaking in.
Finally as a last act of desperation before taking them down to the trash bin I quickly hooked up a set of off the shelf second order crossovers. The difference was both immediate and profound. It was definitely audible and clearly visible in REW sweeps.
The new boards moved the crossover point from just over 1000hz to just under 2000, still low enough to avoid woofer breakup but now high enough that the tweeters weren't strained to reproduce mid-vocal sounds.
The frequency response even without baffle step now reaches down into the mid-30 hz range and rolls off gently above 15khz. Clarity and detail are both enhanced and best of all, no more chunka chunka from the port.
They aren't super hifi speakers by any stretch but after finalising the crossover replacement by fully disassembling the speaker, mounting the crossover, wiring with 16ga wire, removing some of the stuffing, putting neoprene gaskets under the woofers and spending some time tweaking the port they are finally an acceptable speaker that is actually pleasant to listen to.
The problem from the beginning was a bad crossover design, a poorly tuned port and skinny little internal wires.
Can I hear the difference on music? You bet your sweet bippy I can!
Finally as a last act of desperation before taking them down to the trash bin I quickly hooked up a set of off the shelf second order crossovers. The difference was both immediate and profound. It was definitely audible and clearly visible in REW sweeps.
The new boards moved the crossover point from just over 1000hz to just under 2000, still low enough to avoid woofer breakup but now high enough that the tweeters weren't strained to reproduce mid-vocal sounds.
The frequency response even without baffle step now reaches down into the mid-30 hz range and rolls off gently above 15khz. Clarity and detail are both enhanced and best of all, no more chunka chunka from the port.
They aren't super hifi speakers by any stretch but after finalising the crossover replacement by fully disassembling the speaker, mounting the crossover, wiring with 16ga wire, removing some of the stuffing, putting neoprene gaskets under the woofers and spending some time tweaking the port they are finally an acceptable speaker that is actually pleasant to listen to.
The problem from the beginning was a bad crossover design, a poorly tuned port and skinny little internal wires.
Can I hear the difference on music? You bet your sweet bippy I can!
Last edited:
Wow, I didn't expect so many reactions, as well as so many assumptions 🙂
I have a Umik1 and will be measuring.
My current - passive - DIY speakers are being "corrected" by a Raspberry Pi with Hifiberry DSP+ after doing RTA measurements in REW.
I think the "building an analog crossover without measurements" topic has a lot of sensible ideas, eventhough it may not be the end-all of crossover design.
I have all these tools at my disposition - REW, Umik1, DSP..
I was just wondering about how to get to a sensible base-level of crossover frequencies to work from 🙂
I mean, I've measured and "corrected" a set of passive speakers, but that's completely different from setting up / measuring individual drivers to find out the best way to cross them over.
I have a Umik1 and will be measuring.
My current - passive - DIY speakers are being "corrected" by a Raspberry Pi with Hifiberry DSP+ after doing RTA measurements in REW.
I think the "building an analog crossover without measurements" topic has a lot of sensible ideas, eventhough it may not be the end-all of crossover design.
I have all these tools at my disposition - REW, Umik1, DSP..
I was just wondering about how to get to a sensible base-level of crossover frequencies to work from 🙂
I mean, I've measured and "corrected" a set of passive speakers, but that's completely different from setting up / measuring individual drivers to find out the best way to cross them over.
Last edited:
Recommended procedure (by me, done several times)
1. Install all drivers in the box, wire them individully to amp/dsp channels
2. Measure each driver's on-axis and lateral 10, 30 and 60deg responses at 60-100cm distance (mic height at tweeter level for 2-way, between mid and tweeter for 3-way
- check impulse response and recognize fist reflectio (typically at 4-6ms after signal peak)
- set measurement IR gating appropriately (4-6ms)
- overlay on-and off-axis measurements of each driver to see top end resonances and directivity and low end roll-off. (woofer is trickier, you must temporarily use longer gating)
-decide where to put xo and what order to use
2. equalize each driver individually at least one octave past intended xo (on-axis or 10deg for waveguide tweeter)
3 set crossover parameters for tweeter and mid, measure both playing, set polarity and delay to get smooth summed response, Check also distortion, do not stress tweeter too much! Check off-axis response smoothness too!
4 set woofer-mid xo and timing (more difficult because of reflections)
5 measure and listen to the whole system on- and off-axis.
5 measure the whole thing
6 enjoy!
Prepare to use several months to test different crossovers and to find low bass eq and "house curve" that you like best!
1. Install all drivers in the box, wire them individully to amp/dsp channels
2. Measure each driver's on-axis and lateral 10, 30 and 60deg responses at 60-100cm distance (mic height at tweeter level for 2-way, between mid and tweeter for 3-way
- check impulse response and recognize fist reflectio (typically at 4-6ms after signal peak)
- set measurement IR gating appropriately (4-6ms)
- overlay on-and off-axis measurements of each driver to see top end resonances and directivity and low end roll-off. (woofer is trickier, you must temporarily use longer gating)
-decide where to put xo and what order to use
2. equalize each driver individually at least one octave past intended xo (on-axis or 10deg for waveguide tweeter)
3 set crossover parameters for tweeter and mid, measure both playing, set polarity and delay to get smooth summed response, Check also distortion, do not stress tweeter too much! Check off-axis response smoothness too!
4 set woofer-mid xo and timing (more difficult because of reflections)
5 measure and listen to the whole system on- and off-axis.
5 measure the whole thing
6 enjoy!
Prepare to use several months to test different crossovers and to find low bass eq and "house curve" that you like best!
Yeah so a few months have passed 🙂
I needed those months to assemble my test speaker, and more importantly my test amplifier.
By now I've built an amplifier with an ADAU1701 DSP hooked up to a 3x200W T-Amp board, so every driver "has its own amp" and I can play around with crossover / filter settings in the DSP.
I have a Umik1, I know how to use REW in a basic way (I've only used it for RTA room measurements so far..)..
I have the spec sheets of my drivers, and I can measure them.
I've also got a copy of Loudspeaker Design Cook Book and read it front to back,
and I've just received my copy of "Testing loudspeakers" by D'appolito.
Haven't started in it yet, but just skimming through it it seems like it currently might go over my head..
More to read, more to learn, more to experiment I guess 🙂
Still, the original thread I mentioned about "designing crossovers without measurements" gives a good starting point on crossover frequency selection.
From here on I'll just experiment and make lots of mistakes to learn from I guess 🙂
I needed those months to assemble my test speaker, and more importantly my test amplifier.
By now I've built an amplifier with an ADAU1701 DSP hooked up to a 3x200W T-Amp board, so every driver "has its own amp" and I can play around with crossover / filter settings in the DSP.
I have a Umik1, I know how to use REW in a basic way (I've only used it for RTA room measurements so far..)..
I have the spec sheets of my drivers, and I can measure them.
I've also got a copy of Loudspeaker Design Cook Book and read it front to back,
and I've just received my copy of "Testing loudspeakers" by D'appolito.
Haven't started in it yet, but just skimming through it it seems like it currently might go over my head..
More to read, more to learn, more to experiment I guess 🙂
Still, the original thread I mentioned about "designing crossovers without measurements" gives a good starting point on crossover frequency selection.
From here on I'll just experiment and make lots of mistakes to learn from I guess 🙂
Last edited:
First, buy a mic boom and locate your mic where your head would be and be able to relocate to same spot.
Test each driver at low power flat (needn't be the same for all but be careful with tweeters) and much wider band than you plan to use. Note FR and THD.
Then you can identify the band that each driver can handle well for your chair and the steepness that seems right for XO slopes.
A really compulsive builder would move the mic a few inches and see if all results are similar.
The rest is just details to get your house curve (no, nobody with good ears ends up flat).
B.
Test each driver at low power flat (needn't be the same for all but be careful with tweeters) and much wider band than you plan to use. Note FR and THD.
Then you can identify the band that each driver can handle well for your chair and the steepness that seems right for XO slopes.
A really compulsive builder would move the mic a few inches and see if all results are similar.
The rest is just details to get your house curve (no, nobody with good ears ends up flat).
B.
Last edited: