My cabinet conversion is nearing completion and I am getting ready for set up.
As a newbie, I am in danger of trying too many things at the same time and therefore getting confused. So I have to try to move in clear, simple steps, in the right order of priority, with clear objective, constraints and method for each.
So far I have the following plan which is limited by my current level of knowledge. I was mainly using this thread by @AllenB for guidance (although it goes more into passive filter design and is not specific to active filters) and I am now trying to make it more specific to active/DSP. I also used advice given in the build thread so far by @Busdriver02, @svp, @temp25, @Juhazi, @hifijim. I would like to build this guide so I can follow it myself and also make it available for others to use with DSP-based 3-ways. Would appreciate if experienced people can have a look and help me put it together.
As a newbie, I am in danger of trying too many things at the same time and therefore getting confused. So I have to try to move in clear, simple steps, in the right order of priority, with clear objective, constraints and method for each.
So far I have the following plan which is limited by my current level of knowledge. I was mainly using this thread by @AllenB for guidance (although it goes more into passive filter design and is not specific to active filters) and I am now trying to make it more specific to active/DSP. I also used advice given in the build thread so far by @Busdriver02, @svp, @temp25, @Juhazi, @hifijim. I would like to build this guide so I can follow it myself and also make it available for others to use with DSP-based 3-ways. Would appreciate if experienced people can have a look and help me put it together.
I see a crossover as primarily an acoustic endeavour, and where most of the time is spent before and during construction and revisions. The filters are just the bit at the end. My active and passive crossovers are fundamentally otherwise identical.
1. I wouldn't bother doing an initial spin until you've done the initial level setting. Just do a sweep of each driver at ~10 degrees off axis and lower the midrange and tweeter levels until they line up with the woofer section.
2. Then do a full spin on each driver individually.
3. Import into VCad and start modeling. You're looking to explore crossover point and subsequent off axis behavior (directivity). You can also build an initial set of filters, and driver delay.
4. Import all that into HFD (I just write it down or alt-tab between programs)
5. Measure again at 10 degrees off axis, make adjustments and fine tune.
6. Once happy, do a full full speaker spin just to verify and have a pretty picture to show off your work.
7. Profit.
2. Then do a full spin on each driver individually.
3. Import into VCad and start modeling. You're looking to explore crossover point and subsequent off axis behavior (directivity). You can also build an initial set of filters, and driver delay.
4. Import all that into HFD (I just write it down or alt-tab between programs)
5. Measure again at 10 degrees off axis, make adjustments and fine tune.
6. Once happy, do a full full speaker spin just to verify and have a pretty picture to show off your work.
7. Profit.
Thought: Do you have a balcony in your apartment?
EDIT: I'll just throw in what I'm thinking. You could place the speaker on your balcony, then rig up a series of clamps/poles to hold the microphone out past the railing. That way you could get better isolated measurements.
The clamps/poles I'm thinking of are commonly used in photography to mount flashes to things when setting up lighting for a shoot.
EDIT: I'll just throw in what I'm thinking. You could place the speaker on your balcony, then rig up a series of clamps/poles to hold the microphone out past the railing. That way you could get better isolated measurements.
The clamps/poles I'm thinking of are commonly used in photography to mount flashes to things when setting up lighting for a shoot.
Last edited:
How do I select which type of crossover to use and how steep the order should be? These posts talk about lots of iterations, but iterations of what? Trying to make acoustic response of each channel to match a certain slope and then the sum to be straight? If so, what slope then and why?
Yes, that post talks about adding to the existing responses to get to the named slopes, and that may take some fancy filter work.. but you still need to know what slopes to target.
To get you started, use the least slope needed to take acoustic problems down so you can't hear them. In the case of a woofer that could be breakup related, and I'm not talking about peaks. Those need their own attention. I'm talking about excluding the overall breakup issue where needed.
If I'm making sense, this means if you low pass a woofer at a lower frequency then you have more freedom to choose your slope, although you don't always have a choice.
On top of that, you want to exclude parts of the range that don't match in directivity. For tweeters, there's power handling to consider and there's delay to compensate.
In the end you reach a response and phase relationship between them.
To get you started, use the least slope needed to take acoustic problems down so you can't hear them. In the case of a woofer that could be breakup related, and I'm not talking about peaks. Those need their own attention. I'm talking about excluding the overall breakup issue where needed.
If I'm making sense, this means if you low pass a woofer at a lower frequency then you have more freedom to choose your slope, although you don't always have a choice.
On top of that, you want to exclude parts of the range that don't match in directivity. For tweeters, there's power handling to consider and there's delay to compensate.
In the end you reach a response and phase relationship between them.
Last edited:
Ok, so if there are problems (e.g. reach breakup, reach xmax, reach power handling limit, less than an octave away from driver response roll-off) just outside the XO F, you want a steep slope. Right? If the problems are further away, you can use less steepness - but only if needed. The need may come from ...? need to have a smoother directivity transition? That would only apply to the MF/HF, right? What would need a less steep slope in the LF/MF transition, say, in the 300Hz region? what else may need a less steep slope?
I have heard reasonably knowledgeable people (lifetime in loudspeaker industry) say: use 24db (4th order) or even better 48dB (8th order) slopes and XO F two octaves away from the nearest problem. If true, what considerations would need reducing the steepness?
I have heard reasonably knowledgeable people (lifetime in loudspeaker industry) say: use 24db (4th order) or even better 48dB (8th order) slopes and XO F two octaves away from the nearest problem. If true, what considerations would need reducing the steepness?
Last edited:
I see your point, yes, but it also depends on the configuration.
The fullrange crowd have a configuration which is a large woofer with a fullrange driver crossed at a few hundred Hz give or take. Many of them do that first order (by the traditional definition).
Only building active speakers here, but I was under the impression that a DSP can adjust values quite a lot more precisely than passive filters. Any truth to this?
With the speakers I build, I am able to achieve the same level of precision and the same level of filter detail either way.
I'm going to disagree and be the "just do it simple" guy. Don't worry about the actual acoustic slope. I'm not sure why anyone would care about a named slope shape once into digital. The "order" of the slope is a remnant of the math used to create them prior to digital.
Just use 4th order LR filters (double stack .707 Q filters). This maintains an in phase response through the crossover on both drivers.
Adjust the combined response with IIR filters (peaking and shelving) as required.
Adjust the crossover point to tune directivity behavior.
I don't think the drivers you've picked should require anything weird to get them to work. But, steeper slopes can allow you to push a crossover closer to areas you'd otherwise want to avoid.
Just use 4th order LR filters (double stack .707 Q filters). This maintains an in phase response through the crossover on both drivers.
Adjust the combined response with IIR filters (peaking and shelving) as required.
Adjust the crossover point to tune directivity behavior.
I don't think the drivers you've picked should require anything weird to get them to work. But, steeper slopes can allow you to push a crossover closer to areas you'd otherwise want to avoid.
I remember the first time trying to adjust my DSP, and very wrongly assumed that digital filters somehow "saved" a badly designed speaker. But no - a badly designed speaker is just that - a badly designed speaker. No filter is more or less a "correct" filter. A good filter - passive or active - is just the filter that is best suited to a given loudspeaker design/layout - to obtain a given goal. So if a given passive filter suits your speaker design great - then that solution will be great - I agree 👍
For me personally, it was just much easier to program values in a DSP, rather than wrestle with a ton of passive components, to achieve a given cross-over frequency or EQ - especially when fooling around with a lot of tricky drivers and designs.
It is also extremely rare that I've ever encountered a passive speaker that did not need some level of EQ. And with today's prices on electronics, it is then just easier to make the entire speaker active in the same get-go, and get a full package - IMO 🙂
A given electrical slope, only seems to work perfectly with a very flat driver initially. I paired a Satori MW13TX with a 26ADC in a waveguide at 2 kHz, which worked great, since both drivers went further. So the LR24 filter was added where both drivers were still linear. If the natural response of the driver has already started to fall, then a lower level slope should give you the right acoustical slope
OK, I can manage pretty much everything from this post, except: Adjust the crossover point to tune directivity behavior. How do I do that?
How do I see problems in directivity and how I correct them? and how do I know if a particular problem is correctable by crossover point/slope?
and I assume this only (or mainly) applies to the MF/HF (in a 3-way), right?
Last edited:
That makes sense! and is an actionable advice - it is a method. The electrical slope is the addition to the natural roll off (assuming they are at the same or near F). Does this mean you can/should have different electrical slope on different sides (if natural roll off of the drivers is different)? e.g. LR4 on one and LR8 on the other (still giving you same phases).
Last edited:
Great 😀
Try this too. When you applied the filter, then try and measure the speaker on-axis, then at around 10, 20, 30, 45, 60 and 90 degrees - at the same distance - horizontally. Turn the speaker where it is, so that the front baffle is the one only being turned but not moved forth or back. Imagine the rear of the speaker being pushed from side to side, but the center of the dome tweeter - seen directly from the op, straight down - is the pivot point.
By doing this, you should get a set of curves, that should give you a rough idea of how the dispersion change - which a good indicator whether you more or less matched the response of your midrange and tweeter, so that you as the listener hear them as mostly one unit.
When building a speaker, we need several driver units to cover the entire frequency spectrum, but making the filters right, should still "fool" us into the illusion that it is one single sound-source that we are listening to.