Hello to everyone!
It´s a funny thing with DIY... you can get really addicted... 😀 After finishing my very first speaker, the FH MkIII and shortly after XRK´s phenomenal FAST TL floorstander (which is my best speaker ever, it outperforms many many high-priced commercial speakers) I thought, maybe I should - no, in fact I definitely want to - design my first really DIY speaker.
I ask the professionals at the beginning to be lenient with me 🙂
The design will be a ported box, with the Dayton Audio RS225-8 woofer (a great drive) and the Dayton Audio DC28F-8 tweeter. At this point thanks to XRK for supporting my first steps and for the recommendation of the simulation software.
After collecting the data and feeding the simulator I got this as the first (I mean, tenth or twentieth 😀) result:
The circuit:
Simulated frequency response:
Simulated impedance curve: the impedance minimum is @5,25 Ohms
Simulated impulse response:
Simulated step response:
As I wrote the box should be a ported one with a volume of 49.4 liters, the port will have a cross section of 50 cm2 (about 7.7 in2) and a length of 29.6 cm (11.65 in).
The width of the baffle will be 26 cm (10.23 in), the CTC (the shortest possible) 17 cm (6.7 in).
With that, f3 should be @30 Hz and the sensitivity @87 dB.
So, that´s it 🤔
At the end I would like to ask you if this even makes sense or am I completely wrong... 😛
Many thanks in advance.
It´s a funny thing with DIY... you can get really addicted... 😀 After finishing my very first speaker, the FH MkIII and shortly after XRK´s phenomenal FAST TL floorstander (which is my best speaker ever, it outperforms many many high-priced commercial speakers) I thought, maybe I should - no, in fact I definitely want to - design my first really DIY speaker.
I ask the professionals at the beginning to be lenient with me 🙂
The design will be a ported box, with the Dayton Audio RS225-8 woofer (a great drive) and the Dayton Audio DC28F-8 tweeter. At this point thanks to XRK for supporting my first steps and for the recommendation of the simulation software.
After collecting the data and feeding the simulator I got this as the first (I mean, tenth or twentieth 😀) result:
The circuit:
Simulated frequency response:
Simulated impedance curve: the impedance minimum is @5,25 Ohms
Simulated impulse response:
Simulated step response:
As I wrote the box should be a ported one with a volume of 49.4 liters, the port will have a cross section of 50 cm2 (about 7.7 in2) and a length of 29.6 cm (11.65 in).
The width of the baffle will be 26 cm (10.23 in), the CTC (the shortest possible) 17 cm (6.7 in).
With that, f3 should be @30 Hz and the sensitivity @87 dB.
So, that´s it 🤔
At the end I would like to ask you if this even makes sense or am I completely wrong... 😛
Many thanks in advance.
Hi,
When you say "collected the data", did you measure responses in a prototype box with that baffle size, or just use Dayton's standard frd files?
If the latter, they won't include the effects of bafflestep or edge diffraction, which can significantly change the frequency response of both drivers.
I don't know if XSim has the ability to add in those effects or not, but VituixCAD certainly does.
Also, I don't know if the HF will be happy crossed over that low - I know that's about where Dayton claim it can be used but at only ~2/3 octave above resonance it'd probably need to be a steeper filter to avoid it sounding stressed.
The traditional rule of thumb, especially if not using steep filters, is to cross at least an octave above resonance to help avoid that particular issue.
Last suggestion for now, it might be worth turning on the phase trace for each driver to check they're reasonably well matched at least for an octave or so each side of the crossover.
HTH,
David.
When you say "collected the data", did you measure responses in a prototype box with that baffle size, or just use Dayton's standard frd files?
If the latter, they won't include the effects of bafflestep or edge diffraction, which can significantly change the frequency response of both drivers.
I don't know if XSim has the ability to add in those effects or not, but VituixCAD certainly does.
Also, I don't know if the HF will be happy crossed over that low - I know that's about where Dayton claim it can be used but at only ~2/3 octave above resonance it'd probably need to be a steeper filter to avoid it sounding stressed.
The traditional rule of thumb, especially if not using steep filters, is to cross at least an octave above resonance to help avoid that particular issue.
Last suggestion for now, it might be worth turning on the phase trace for each driver to check they're reasonably well matched at least for an octave or so each side of the crossover.
HTH,
David.
A simulation is only good as the data used. If you simply entered into XSim frd and zma files found in parts-express site, then your simulation is wrong. This is because in that case you missed baffle step and diffraction effects, and the path difference between the drivers.
Another thing: the crossover point seems pretty good for providing a constant directivity between an 8" bass to a tweeter, but the tweeter chosen is not really suitable to the task. You took advantage of the natural roll off of the tweeter, but at the end you don't provide adequate protection to it, you need a more robust tweeter and a higher electrical order network. This is what Zaph wrote about this tweeter: "The popular "silkie" returns with a redsign and hopefully some consistency. Response is relatively smooth but harmonic distortion is average to poor. It will work crossed LR4 at 2kHz or higher. Not the best value in the under $20 range. Only 3 screws on a flange this big is a bad idea. Tested February 2006."
I won't comment for now about the box and tuning, but F3 doesn't have any acoustical merit, F6 and F10 have.
Ralf
Another thing: the crossover point seems pretty good for providing a constant directivity between an 8" bass to a tweeter, but the tweeter chosen is not really suitable to the task. You took advantage of the natural roll off of the tweeter, but at the end you don't provide adequate protection to it, you need a more robust tweeter and a higher electrical order network. This is what Zaph wrote about this tweeter: "The popular "silkie" returns with a redsign and hopefully some consistency. Response is relatively smooth but harmonic distortion is average to poor. It will work crossed LR4 at 2kHz or higher. Not the best value in the under $20 range. Only 3 screws on a flange this big is a bad idea. Tested February 2006."
I won't comment for now about the box and tuning, but F3 doesn't have any acoustical merit, F6 and F10 have.
Ralf
Hi plott from hernals in vienna,
in addition to what has been said already, you should probably simulate the response of the bass reflex enclosure. That will also affect impedance response. And give you lower bass.
If you want to cross the tweeter so low you might consider a waveguide, supporting the tweeter at low frequencies.
Edit: for an example of crossing a waveguided tweeter low see http://www.donhighend.de/?page_id=2660
And one more edit, for more explanation of big woofer+waveguide concept regarding donhighend's "big yellow taxi": http://www.donhighend.de/?page_id=1461
in addition to what has been said already, you should probably simulate the response of the bass reflex enclosure. That will also affect impedance response. And give you lower bass.
If you want to cross the tweeter so low you might consider a waveguide, supporting the tweeter at low frequencies.
Edit: for an example of crossing a waveguided tweeter low see http://www.donhighend.de/?page_id=2660
And one more edit, for more explanation of big woofer+waveguide concept regarding donhighend's "big yellow taxi": http://www.donhighend.de/?page_id=1461
Last edited:
Thank you very much David and Ralf,
it helps very much.
Yes, the data is what is provided by Dayton Audio, I thought this is okay for the first simulation.
Well, the next step will be to choose another tweeter, measure the driver to get real frd-data, and redesign the xo in order to cross one octave above the resonance, I don't like stressed/screaming mids/highs.
Cheers
it helps very much.
Yes, the data is what is provided by Dayton Audio, I thought this is okay for the first simulation.
Well, the next step will be to choose another tweeter, measure the driver to get real frd-data, and redesign the xo in order to cross one octave above the resonance, I don't like stressed/screaming mids/highs.
Cheers
Greetings from Leopoldstadt, stv 🙂
Many thanks, yes, the effect of the enclosure is another thing, I will take everything in consideration. Step by step 🙂
Many thanks, yes, the effect of the enclosure is another thing, I will take everything in consideration. Step by step 🙂
No, this is not OK for a first or whatever crossover simulation. You need to simulate what the drivers are doing on the chosen baffle, and if you only have the infinite baffle responses you need to alter them adding baffle step and diffraction. There are tools for that.
And you need to provide an adequate path difference between the drivers. If you left the default 0, then the simulation is simply wrong, as a 1“ difference is a lot for a 2KHz (or lower) crossover point.
As to the bass simulation, you can work on it later, as what happens below 100Hz doesn’t have any effect on a 1+ KHz crossover.
Measuring in box drivers is always the best method, but a good simulation will let you choose the drivers before buying them.
Ralf
PS: you should modify the default mid point for the XSim FR graph, 75dB or less should be a good value
And you need to provide an adequate path difference between the drivers. If you left the default 0, then the simulation is simply wrong, as a 1“ difference is a lot for a 2KHz (or lower) crossover point.
As to the bass simulation, you can work on it later, as what happens below 100Hz doesn’t have any effect on a 1+ KHz crossover.
Measuring in box drivers is always the best method, but a good simulation will let you choose the drivers before buying them.
Ralf
PS: you should modify the default mid point for the XSim FR graph, 75dB or less should be a good value
I would look for a better tweeter than the DC28F: it's OK. I've used it in Curt Campbell's Tritrix speakers and enjoy them. However, the dome material seems to attract every bit of dust, mozzies, midges and fluff that's around and the connection terminals are a bit flimsy. Also, the DC28 is a 'good' tweeter and the RS225 is highly regarded as a very high quality, high value woofer, so there might be a quality mis-match.
There are stock shortages at the moment, but something like the aluminium dome Peerless DA25 - used in Wolf's "Zingers" speakers with an 8" woofer - might work. I've used it in a small two way with a Peerless mid and liked it very much. It was, of course, someone else's design as I know b- - -r all about designing crossovers and don't have measuring equipment. So I've no idea how to match it with the RS225, I'm afraid.
Geoff
There are stock shortages at the moment, but something like the aluminium dome Peerless DA25 - used in Wolf's "Zingers" speakers with an 8" woofer - might work. I've used it in a small two way with a Peerless mid and liked it very much. It was, of course, someone else's design as I know b- - -r all about designing crossovers and don't have measuring equipment. So I've no idea how to match it with the RS225, I'm afraid.
Geoff
My 2ct: use a waveguided tweeter. The end result of matching the directivity of the tweeter to that of the woofer will pay off a lot more than picking the tweeter itself. And although it might seem a bit more difficult to design, you’d have to measure the acoustic output anyway, so that eases that part of the job.
I think RST28F is more well matched. And if you use a waveguide you can get a nice result. Trick is make the box and then measure response of drivers in the box. Then do the sim. Measure woofer only, tweeter only, woofer plus tweeter in parallel as three data sets without physically touching speaker or mic. Then use acoustic interferometry to determine acoustic offsets down to the mm. If you can predict the woofer and tweeter in parallel accurately (down to the last wiggle), your sim will be correct 98% of the time. Need to measures the impedance in the box for each driver and rename as Zma file.