So I'm in the process of an OB "2.5" way build. Here is the parts list:
Here is my concern. For the crossover frequency between the bass and midrange, I think I am OK- I will probably do something in the 300-500Hz range, well above the midrange's Fs/low frequency (~90Hz). But for the coaxial...... strangely, the low end of the tweeter (4.5kHz) is well above the beaming frequency of the midrange (~3.4kHz I think). The Fs of the tweeter (2.8kHz) is below that beaming frequency, and the tweeter does have decent SPL below its rated range but I don't want to push the crossover to say, 3k and potentially damage the driver.
So the TL;DR is, what frequency should I crossover the midrange/tweeter at? To be fair I haven't noticed anything super bizarre about the sound; I have the drivers in a test baffle now crossed over at like 150 and 5k I believe. The coax sounds a little lifeless/dull with the plain DSP crossover I have vs the passive crossover it had in the PE box so I guess I can study the original crossover to see what frequencies they chose. Here is that schematic:
What do you think? Am I overthinking this? Subjectively my only complaint is that the coax sounds a little lifeless. I will take some measurements and play with the EQ for taste I guess.
- Amps: Tinysine TSA7800B (2.1 way, TWS, DSP, 1 per channel- currently testing single channel with a ZOUDIO 4ch)
- Bass driver: Goldwood GW-212/4
- Midrange: Dayton CX120-8 (4" driver)
- Tweeter: Dayton CX120-8 (3/4" tweeter)
Here is my concern. For the crossover frequency between the bass and midrange, I think I am OK- I will probably do something in the 300-500Hz range, well above the midrange's Fs/low frequency (~90Hz). But for the coaxial...... strangely, the low end of the tweeter (4.5kHz) is well above the beaming frequency of the midrange (~3.4kHz I think). The Fs of the tweeter (2.8kHz) is below that beaming frequency, and the tweeter does have decent SPL below its rated range but I don't want to push the crossover to say, 3k and potentially damage the driver.
So the TL;DR is, what frequency should I crossover the midrange/tweeter at? To be fair I haven't noticed anything super bizarre about the sound; I have the drivers in a test baffle now crossed over at like 150 and 5k I believe. The coax sounds a little lifeless/dull with the plain DSP crossover I have vs the passive crossover it had in the PE box so I guess I can study the original crossover to see what frequencies they chose. Here is that schematic:
What do you think? Am I overthinking this? Subjectively my only complaint is that the coax sounds a little lifeless. I will take some measurements and play with the EQ for taste I guess.
It's hard to offer an opinion because we can't hear it or see any measurements. The datasheet suggests to be a 3k crossover would be appropriate but this is the acoustic target and may not be what electrically you will need to program the dsp to achieve this. Do you have a measurement mic? Time gated measurements are applicable in this range.
Some 12in coaxials cross at 1,5 - 1,6 kHz with nice results, so I dont see a problem if 4in crosses at 4,5 kHz.
I have no experience with DSP, so I cant give you advice how to "copy" a passive x-over.
I have no experience with DSP, so I cant give you advice how to "copy" a passive x-over.
Parts-Express has the FRD and ZMA files online. It looks like you are using XSim, so you can play around with that.
There a a good bit of overlap between woofer and tweeter, which should make things easier.
There a a good bit of overlap between woofer and tweeter, which should make things easier.
If you're worried about overthinking this, I probably won't help here, but is it really just a 4" radiating piston? Or does it have a 3/4" hole in the middle?
In which case maybe you could model it (on a single plane) as two surfaces, each width a width of ~41mm, and a center-to-center spacing of ~39.7mm. Quarter wavelength of 39.7mm = 2161Hz if I've understood correctly, so you might expect some lobing effect from here up. Obviously all these sums assume flat surface which it isn't, so I wouldn't pay much attention to them 🙂
In which case maybe you could model it (on a single plane) as two surfaces, each width a width of ~41mm, and a center-to-center spacing of ~39.7mm. Quarter wavelength of 39.7mm = 2161Hz if I've understood correctly, so you might expect some lobing effect from here up. Obviously all these sums assume flat surface which it isn't, so I wouldn't pay much attention to them 🙂
All I did was build the schematic provided and use the files for the driver. I have that driver so the files were on my computer. I used 0 degree frd.Please check to see if i made a mistake
That's what Dayton sent lol. I think I will just set the XO at 3k and cross my fingers.
In some ways phase seems a little less critical with a coax. However, as always it requires more attention when done without measurement, as is the case with many DSP crossovers.
Unintended amounts of phase shift can affect higher order modes due to a malformed wavefront.
It might be difficult to tell clearly by listening. Fortunately with a coax you can get a usable crossover result from simpler measurements or simulations by just seeing that the tweeter and woofer remain in phase. There's no need to be exact at this stage.
Unintended amounts of phase shift can affect higher order modes due to a malformed wavefront.
It might be difficult to tell clearly by listening. Fortunately with a coax you can get a usable crossover result from simpler measurements or simulations by just seeing that the tweeter and woofer remain in phase. There's no need to be exact at this stage.
That's another possibility...
How are you level matching the woofer/mid/tweeter? Could you post your SigmaStudio crossover diagram?
What is BSC? Did you end up going with a different driver, and if so what do you recommend? I am going to have to put this project on hold until the new year.
I've used those coaxials crossed over at 5kHz (and at 500Hz to a woofer) and they sound great. Sure, the directivity doesn't match there, but everything is a compromise.
Point taken, but the issue is really the purpose of using a coax when the directivity being that much off. The benefits aren't really there.
Hmm, for polar matching and assuming the woofers are ~pistonic to the Vc frequency, which judging by how closely its TL modes maintain flatness, then the woofer's 2.5" = 4380 Hz = sqrt(47.44*4380) = ~456 Hz XO and the coax woofer's 1" = ~4311 Hz, making this the end of it matching the tweeter using traditional theory, so from this point on it's about finding a measured overlap (if any), but the ~4311 Hz isn't enough lower to be a power handling problem unless this is a ~max peak SPL and/or a high average SPL app.
They're free...
I will be honest... I don't understand much of this. Where are you getting these dimensions and formulas from? In any case I will try these frequencies. I don't plan to run these super loud so I feel like I can get away with dubious XO frequencies but 4.3kHz seems well within the margin of safety.
Sound is 'round' and expands exponentially, so choosing matching polar responses of two drivers is done based on a frequency point of the driver, so regardless of driver size, if the frequencies match, so does the polar response, though of course baffles, any nearby boundaries ultimately modify it, but it's an easy way to choose XO points as at least a good starting point without fancy measurements.
Anyway, the math is the mean between two frequencies with the 1st being between the driver's Fs and the VC's dia., the theoretical BW of a point source driver and if too low, then calculate the mean of the 1st XO point and VC, etc., till you get up to where it needs to be for whatever reason.
I assume there's some math that allows one to start (much) higher up in lieu of a bunch of iterations, but my math skills are vestigial at best and had to be shown what little I do know by numerous math mentors.
Anyway, the math is the mean between two frequencies with the 1st being between the driver's Fs and the VC's dia., the theoretical BW of a point source driver and if too low, then calculate the mean of the 1st XO point and VC, etc., till you get up to where it needs to be for whatever reason.
I assume there's some math that allows one to start (much) higher up in lieu of a bunch of iterations, but my math skills are vestigial at best and had to be shown what little I do know by numerous math mentors.