I am planning on building my first own Speaker which should be a 2-way.
I designed this Crossover in VituixCAD and am pretty happy with the curve, just wanted to have experts have a look on it and if there are any major problems or things I have overseen because designing this seemed relatively easy
Screenshot by Lightshot
Drivers are the Dayton DSA175-8 and the Dayton DC28F-8
I designed this Crossover in VituixCAD and am pretty happy with the curve, just wanted to have experts have a look on it and if there are any major problems or things I have overseen because designing this seemed relatively easy
Screenshot by Lightshot
Drivers are the Dayton DSA175-8 and the Dayton DC28F-8
It looks like you have some peaking in your bass filter, I'd try putting some resistance in series with the shunt cap. Try starting with maybe 2 or three ohms and moving up or down to see how it affects things.
Also it looks like you haven't applied baffle step to the drivers measurements, which will change things fairly dramatically as well. Could be fine if you plan on having the speakers hard up against the wall but if away from the walls will need considering.
Finally without seeing the individual drivers curves (post crossover) it is a bit hard to tell what the actual crossover point is, but it seems from the phase curves it is 3rd order accoustic, The specs for the tweeter say it is usable down to 1300 Hz so may be ok...
Tony.
Also it looks like you haven't applied baffle step to the drivers measurements, which will change things fairly dramatically as well. Could be fine if you plan on having the speakers hard up against the wall but if away from the walls will need considering.
Finally without seeing the individual drivers curves (post crossover) it is a bit hard to tell what the actual crossover point is, but it seems from the phase curves it is 3rd order accoustic, The specs for the tweeter say it is usable down to 1300 Hz so may be ok...
Tony.
Looks good. I think you have allowed for 3dB bafflestep compensation given the woofer is nominally 88dB above bafflestep. I'd be worried you're crossing the DC28F a little too low... but listening test will tell.
What Z offset did you use on the driver tab for the DS175?
Are you targeting an LR acoustic response? if so flip the polarity of one of the drivers (select it then Ctrl-I). I don't think you've quite got phase tracking right yet. A shaping resistor as (shunt in series with the 2nd order capacitor on the woofer) will help bring up the response and can be used to tailor phase tracking to match
What Z offset did you use on the driver tab for the DS175?
Are you targeting an LR acoustic response? if so flip the polarity of one of the drivers (select it then Ctrl-I). I don't think you've quite got phase tracking right yet. A shaping resistor as (shunt in series with the 2nd order capacitor on the woofer) will help bring up the response and can be used to tailor phase tracking to match
Here's an existing project which uses those drivers:
BR-2 | Parts Express Project Gallery
They use the Dayton BR-1 cabinet but you could easily make your own
Of course, I haven't heard them but the designer is well regarded
HTH
Geoff
BR-2 | Parts Express Project Gallery
They use the Dayton BR-1 cabinet but you could easily make your own
Of course, I haven't heard them but the designer is well regarded
HTH
Geoff
Not sure what's meant by 'unreliable' in the above comment, but in a general sense if I have already bought some drivers and would like to make a listenable speaker, using an existing, well-designed project should be a good way to go. It's worked for me several times.
Very cute dog in the photo!
Geoff
Very cute dog in the photo!
Geoff
I usually tend to agree on the "leave it to the experts" recommendation, but then how are novices going to learn xover design!? You already have the parts. Are you able to measure SPL curves? Are you ready for a journey? Go for it!
That said, your xover does not look awfully wrong at all. The woofer/tweeter transition looks okay.
The "hump" from 100 to 300 Hz looks like a measurement artifact to me. How did you get these data?
Also, you need to understand that the on-axis SPL curve is just one of the many things to consider in a loudspeaker design. The on-axis SPL curve will not tell you the full story of how your design will sound. To get a better understanding about "how does it sound" you'll need to look at the off-axis SPL response, too. Vituix has some powerful built-in tools for this. In particular, Vituix calculates the directivity index and the power response, which will be more useful than just the on-axis response alone. You'll have to get SPL curves of the raw drivers (no xover) at off-axis angles, both horizontally and vertically, at fixed angular intervals (10° to 15° or so), covering the full 4pi sphere. That's a bit of work, but will give you much more useful data.
In the end, what really works to learn xover design is to do some test builds in order to hear what you see on the computer screen. Then start changing things by going back and forth between the test builds, measurements, and computer models.
That said, your xover does not look awfully wrong at all. The woofer/tweeter transition looks okay.
The "hump" from 100 to 300 Hz looks like a measurement artifact to me. How did you get these data?
Also, you need to understand that the on-axis SPL curve is just one of the many things to consider in a loudspeaker design. The on-axis SPL curve will not tell you the full story of how your design will sound. To get a better understanding about "how does it sound" you'll need to look at the off-axis SPL response, too. Vituix has some powerful built-in tools for this. In particular, Vituix calculates the directivity index and the power response, which will be more useful than just the on-axis response alone. You'll have to get SPL curves of the raw drivers (no xover) at off-axis angles, both horizontally and vertically, at fixed angular intervals (10° to 15° or so), covering the full 4pi sphere. That's a bit of work, but will give you much more useful data.
In the end, what really works to learn xover design is to do some test builds in order to hear what you see on the computer screen. Then start changing things by going back and forth between the test builds, measurements, and computer models.
Parts Express' product pages have the off axis data for both drivers, so it should be straightforward to load that information into Vituix.
It should be interesting to compare what you get from Vituix with the BR-2; of course those speakers were designed using measured data but it's always interesting to compare sims with final results. I've done that with all my DIY projects and it's an instructive exercise.
Geoff
It should be interesting to compare what you get from Vituix with the BR-2; of course those speakers were designed using measured data but it's always interesting to compare sims with final results. I've done that with all my DIY projects and it's an instructive exercise.
Geoff
...and these data are useless for the purpose of designing an xover for a given speaker.
What is needed is measurements of the drivers IN THE BOX / ON THE BAFFLE on question.
1) Dayton Audio's frequency responses are measured at near field below 450 Hz and far field at 1 m above that. Driver is probably located in quasi infinite baffle or very large baffle, but size and volume are unknown. Frequency response data is not valid for crossover design of any boxed speaker because diffraction effects including baffle loss and directivity changes to different off-axis angles are missing. Without quite advanced post processing data is valid for designing flush/wall mounted speakers i.e. half space designs only.
2) BR-2 crossover design looks really bad. Axial response has hump where tweeter+baffle combination is not yet directive and mid-woofer is already low-passed. Also BSC of woofer looks excessive and directivity of cone+baffle combination is not compensated in axial response.
I would call those as the most classical sound balancing errors in 2-way designs. Designing crossover with simulator using axial response only is very poor method i.e. whole design is unreliable. Measuring with USB mic unable to capture timing does not increase credibility.
But all this is quite normal in diyaudio.
Before we get into yet another (heated) discussion about dos and donts of speaker design for rookies: So you want to design your own speaker from scratch!
I wouldn't for a moment suggest designing a speaker just using makers' data, but I still think it's interesting to have a play with it and see how the results compare to the real thing, measured in box/baffle - which is indeed the only way to do this stuff properly. For example, sims are are a good way to 'guesstimate' the effects of changing crossover components.
For each of my DIY speakers (five at the moment, all other peoples' projects, properly measured and well designed) I've made an Xsim model and compared those results with the real, in-box, measured responses.
The differences are indeed striking, as expected.
Geoff
For each of my DIY speakers (five at the moment, all other peoples' projects, properly measured and well designed) I've made an Xsim model and compared those results with the real, in-box, measured responses.
The differences are indeed striking, as expected.
Geoff
Last edited:
Lists about available tools with some subjectively selected vintage favorites by native English are important in diyaudio to maintain status quo:
1) Design speakers without measurements.
2) Design crossover with Dayton Audio's data.
3) Measure on-axis in a single point with the cheapest or some American USB mic and design XO with on-axis responses only.
4) Recommend previous three items endlessly - especially for beginners. Don't care about Toole or professional designers writing here because any diy is better than scientific excuses and commercial products.
Sim vs real World
I don't really want to go too off-topic or stir things up, but here's my attempt at showing the differences between a sim using just makers' data (the first graphic) and real world, based on Paul Carmody's excellent Classix II speakers (second graphic). The sim uses the same XO parts; the second graph is from Paul's website.
Clearly, if you tried to make the sim itself look decent in terms of FR and built that, you would end up with a disaster.
Geoff
I don't really want to go too off-topic or stir things up, but here's my attempt at showing the differences between a sim using just makers' data (the first graphic) and real world, based on Paul Carmody's excellent Classix II speakers (second graphic). The sim uses the same XO parts; the second graph is from Paul's website.
Clearly, if you tried to make the sim itself look decent in terms of FR and built that, you would end up with a disaster.
Geoff
Attachments
Last edited:
I forgot to ask the OP whether you already have the drivers - rather a basic question.
There are some projects which use the similar sized, but different spec Dayton DA175: these designs are now a little old, but the "Karma Indignia" and "Silver Als" bookshelves used the DA175 and the DC28F and were well received, as far as I recall.
However, the DSA175 has almost twice the "Xmax" (linear excursion) and is more efficient by around 3-6dB than the DA175, so the crossovers for either speaker wouldn't work properly with it.
The DSA looks like a nice driver, strange that there are so few projects out there.
Geoff
There are some projects which use the similar sized, but different spec Dayton DA175: these designs are now a little old, but the "Karma Indignia" and "Silver Als" bookshelves used the DA175 and the DC28F and were well received, as far as I recall.
However, the DSA175 has almost twice the "Xmax" (linear excursion) and is more efficient by around 3-6dB than the DA175, so the crossovers for either speaker wouldn't work properly with it.
The DSA looks like a nice driver, strange that there are so few projects out there.
Geoff
Just wondering how much of the thread you actually read Kimmo. Unfortunately I never have done the latter posts which should cover doing real measurements. But perhaps before you judge too harshly you should have a look at this post.
So you want to design your own speaker from scratch!
The idea of that thread is an accessible stepping stone for people who want to start on the journey of getting something better. Think back to when you first started, how long has it taken you to reach the level of proficiancy that you have? How does the fact that I have written this in English (the only language I know, diminish it's worth?
How is the thread ignoring Toole and professional designers? Especially when I put this clearly in the first post:
Even with the experience I have, doing proper polar measurements for designing with off axis measurements I have found quite challenging. Sure if you want the best crossover you can make this is important, but you need to work up to this level! You can't expect someone to go from zero experience to expert in one step
Tony.
Couldn't agree more. Explaining electro-acoustics can be hard work, and we're not always doing well to educate newbies. Learning this stuff is hard, too. All I can recommend to novices is to read the books, read the explanations by the experienced guys here on diyAudio, and try to understand the physics!
Getting off-axis measurements works exactly the same as on-axis, so I don't think working with off-axis data is more "challenging". All you do is to rotate the speaker at fixed angular increments and take measurements at each step. It's a bit of work to get it done, but hey, it's a hobbby! If you do it once, just push through it and think about it too much. If you do it over and over, think about building an automated rotating table that will do the work for you. Just pay attention to where you set the rotation axis, as this will be important for the data processing.
No matter if you're a novice or a pro, the off-axis data really does help a lot when it comes to designing speakers and xovers. In contrast to a single on-axis measurement, the combined data taken around the speaker will be much more meaningful in terms of "how will it sound like?".
I agree on it's utility, I hope more people do it, but it does get rather specialised when you make the best of it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.