I have been pondering building some three ways speaker, using PA drivers which have been mentionned numerous times on this forum (15pr400,10pr320 and hf108).
I have a minidsp shd with Dirac, but I only have four channels, so I will have to add a passive crossover somewhere.
I used SPL trace from Faital and here is what I can do with all-active :
yet, the SHD only has two channels, so I thought I could plug one channel on the woofer amp and the other one on the mid-high amp. Here is what I did with vituixcad:
This doesn't look so good 🙁 I suspected replacing digital XO with electrical ones would mess it, especially when factoring driver impedance in.
Is this the right approach for a three way when only two DSP channels are available? What looked easy enough with a digital crossover is getting a lot harduous when adding some passive elements in the signal chain.
Thank you for your advice 🙂
-
I have a minidsp shd with Dirac, but I only have four channels, so I will have to add a passive crossover somewhere.
I used SPL trace from Faital and here is what I can do with all-active :
yet, the SHD only has two channels, so I thought I could plug one channel on the woofer amp and the other one on the mid-high amp. Here is what I did with vituixcad:
This doesn't look so good 🙁 I suspected replacing digital XO with electrical ones would mess it, especially when factoring driver impedance in.
Is this the right approach for a three way when only two DSP channels are available? What looked easy enough with a digital crossover is getting a lot harduous when adding some passive elements in the signal chain.
Thank you for your advice 🙂
-
Yes, if you only have enough active crossover outputs to biamp, then generally putting the woofer on its own amp & running the mid & high passive is best.
A couple of general thoughts on this kind of modelling...
1: remember to run the traced curves through the diffraction simulator before using them in the main crossover simulator, unless you're planning on building the whole thing into the wall of course.
2: don't just input nominal crossover points & slopes into a calculator & expect them to work straight away, those kind of preset filter blocks usually don't account for impedance as you've guessed. You'll need to adjust component values to get the final slopes you need. VituixCAD does have an optimiser built in which can be useful if you don't know what to start adjusting yourself.
3: for level matching try and use the relative amp gains to match low and mid, and only use resistive pads on the HF. It looks like you've started along that idea but not got it fully adjusted yet.
A couple of general thoughts on this kind of modelling...
1: remember to run the traced curves through the diffraction simulator before using them in the main crossover simulator, unless you're planning on building the whole thing into the wall of course.
2: don't just input nominal crossover points & slopes into a calculator & expect them to work straight away, those kind of preset filter blocks usually don't account for impedance as you've guessed. You'll need to adjust component values to get the final slopes you need. VituixCAD does have an optimiser built in which can be useful if you don't know what to start adjusting yourself.
3: for level matching try and use the relative amp gains to match low and mid, and only use resistive pads on the HF. It looks like you've started along that idea but not got it fully adjusted yet.
Thank you David!
I have played a bit and I think I have a better result :
While this looks ok on simulation, is this thing actually feasible in reality or am I getting way over my head? I haven't accounted for directivity yet as I don't have spl trace for the faital drivers off axis.
Also, the HF108 is horn loaded, and I am planning on a LTH102 which is a 60x40, is there a way to simulate the spl trace of hf108+horn?
Thanks!
Luc
I have played a bit and I think I have a better result :
While this looks ok on simulation, is this thing actually feasible in reality or am I getting way over my head? I haven't accounted for directivity yet as I don't have spl trace for the faital drivers off axis.
Also, the HF108 is horn loaded, and I am planning on a LTH102 which is a 60x40, is there a way to simulate the spl trace of hf108+horn?
Thanks!
Luc
Definitely better!
If you take the time to get it right, you certainly can get a valid system together like this, though I think you still have a few things to cover.
Using the diffraction simulator will get you a first order approximation of the off axis responses for the 15 & 10, that's one of the really handy things about it.
Also, given you're biamping, I wouldn't bother looking at the impedance for the 15 which is what seems to be showing at the moment, but you'll definitely need to keep an eye on it for the mid & high section, so make sure you have the impedance traces loaded for those.
If the manufacturer's data for that HF108 was measured on that horn then you can at least get the on axis response just like for the 10 & 15, but you'd really need to buy one and measure it to get the off axis right, as the diffraction simulation would not be able to take account of the directivity of the horn.
If you take the time to get it right, you certainly can get a valid system together like this, though I think you still have a few things to cover.
Using the diffraction simulator will get you a first order approximation of the off axis responses for the 15 & 10, that's one of the really handy things about it.
Also, given you're biamping, I wouldn't bother looking at the impedance for the 15 which is what seems to be showing at the moment, but you'll definitely need to keep an eye on it for the mid & high section, so make sure you have the impedance traces loaded for those.
If the manufacturer's data for that HF108 was measured on that horn then you can at least get the on axis response just like for the 10 & 15, but you'd really need to buy one and measure it to get the off axis right, as the diffraction simulation would not be able to take account of the directivity of the horn.
I did play a bit this morning with directivity. I was able to use real measurements from audioxpress of a HF108R on LTH102, from 0 to 60.
I also used the diffraction simulator on the 15 and 10.
I have the impedance traces for the 10 and compression driver loaded, but they don't show up.
Here is what I get, it looks too good to be true...
I wonder if I should worry about the dip of the in room response at 1.5k?
The 10 inches may be too big and I should use an 8 inches with my 1.8k crossover?
I also used the diffraction simulator on the 15 and 10.
I have the impedance traces for the 10 and compression driver loaded, but they don't show up.
Here is what I get, it looks too good to be true...
I wonder if I should worry about the dip of the in room response at 1.5k?
The 10 inches may be too big and I should use an 8 inches with my 1.8k crossover?
Hmm, I haven't played with modelling a hybrid active-passive system like this, I wonder if you may need to temporarily mute the traces for the 15 to get it to show the impedance of the combined mid-high section, or perhaps (again temporarily) take out the active stages before the passive section.
You'll always get some kind of narrowing of the pattern as you transition from a direct radiating mid to a horn loaded HF, you might play with the crossover frequency and slope to see if you can get it a little more gradual if you want.
It looks like you have got the mid starting to narrow from 800Hz on up, so you may already have achieved the smoothest transition you can with this combination of drivers.
Using an 8" would actually make for a more pronounced change in the off axis response as a smaller driver will naturally have a wider coverage pattern as you come up in frequency towards the crossover point.
You'll always get some kind of narrowing of the pattern as you transition from a direct radiating mid to a horn loaded HF, you might play with the crossover frequency and slope to see if you can get it a little more gradual if you want.
It looks like you have got the mid starting to narrow from 800Hz on up, so you may already have achieved the smoothest transition you can with this combination of drivers.
Using an 8" would actually make for a more pronounced change in the off axis response as a smaller driver will naturally have a wider coverage pattern as you come up in frequency towards the crossover point.
So, I am tackling the enclosure now. What is the consensus on the midrange chamber? It should be sealed, intuitively.
Should I aim for the perfect alignment for the driver? 16,6 liters in the case of this 10 inches. Or is bigger an option? Smaller?
luc
Should I aim for the perfect alignment for the driver? 16,6 liters in the case of this 10 inches. Or is bigger an option? Smaller?
luc
It will be about the combined alignment, so the box response doesn't have to be perfect, just something you can work with. Also conside the impedance that will result from the box.
Member
Joined 2003
All that is required is to put the power amplifier after the active blocks. Not sure why the gain/delay block A2 is in the middle, but move it to the end of the active portion and you should see the impedance seen by the amp block in the impedance chart.
A "hybrid" approach can work very well, I've done just this myself for a 3-way using 2 channel Hypex DSP plate amps, however the rest of this simulation using traced manufacturer data is a bit of a wasted effort IMO. Acoustic delay between drivers, and directivity of course, especially the horn needs to be taken into account, real world measurements are the best way to accomplish this, easily done if one follows the measurement instructions for VituixCAD.
Have fun!