TOPIC:

I've been building a set of 4-way stereo speakers and am getting close to needing a crossover solution. My requirements are toslink input, 8-channel analog out, and FIR filtering to implement a linear phase crossover. My source  (Yamaha wxc-50) has volume control in the digital realm, so volume control at the crossover is not necessary. I was originally planning on 2 2x4hd units, as that's all I could find that meet my need. This morning I stumbled across the Flex Eight and thought I had found my ideal solution...  Upon reading the specs, it looks like the FIR filtering is only on the 2 input channels and the outputs are IIR only, can someone confirm this? A linear phase crossover cannot be implemented unless FIR is available on all the output channels, so it's an important detail.

No, a linear-phase crossover can certainly be implemented using only FIR capability in the input section.  (In fact, that's the preferable way to achieve it.)
Use normal PEQ IIR corrections on the individual outputs to make their responses minimum phase.  Then, use a FIR input/global section to create an "unwrap" response to linearize the entire system.  You can use a tool like Rephase to generate that.

Of course, the usual qualification always exists.  The linear-phase response can only exist at one point in the measurement space.....best case. 

Dave.  
 

Thank you, Dave. I suppose I need to keep reading up on linear phase crossovers. I was under the impression that you wanted to linearize and match the phase response of each driver independently so that when combined there would be a perfect alignment in the audible overlap region. Is the problem with that approach that the delays required of each driver would not be equal at every frequency, thus making perfect alignment in the overlap region virtually impossible? Sorry, like I said, I'm still learning about all this stuff.

EDIT: So I went back and read this article  rePhase FIR Tool (minidsp.com)  and it sounds like what you're talking about is phase correction of a loudspeaker and what I'm talking about is a linear phase crossover. Is there a source somewhere that describes the pros/cons between the 2 approaches? Thanks again.

wesharrison wrote:

EDIT: So I went back and read this article  rePhase FIR Tool (minidsp.com)  and it sounds like what you're talking about is phase correction of a loudspeaker and what
I'm talking about is a linear phase crossover. Is there a source somewhere that describes the pros/cons between the 2 approaches? Thanks again.

I don't consider those to be two different things.

There are a couple of different ways to approach this objective.
However, with the Flex Eight you are limited to FIR corrections in the input portion only.

So, in that case, you'd implement your system crossover using the standard functions in each branch.....Xover, PEQ's, delays, gain, etc, etc.  When finished you should have (ideally) a flat amplitude response but with the phase response/distortion associated with the type of crossovers you used.  As an example, maybe you used LR24 crossovers at 100Hz and 3000Hz.  You'd then use a tool like rephase (Filters Linearization tab) to generate a global correction response which you'd apply in the input section of the Flex Eight.

Assuming everything happens as planned, you'd now have a system that exhibits both flat amplitude response and flat phase response.....ie, linear-phase.  

Dave.
 
 

Thanks again, Dave.

I was/am under the impression that manipulating the phase response of individual drivers affects their relationship to one another and that manipulating phase response of a system does not.

For example, I believe you can minimize comb filtering by independently manipulating the phase response of two drivers and correcting the global phase response of a system would have no effect on that artifact if it existed before the correction. Am I incorrect here? Again, I'm just learning here and trying to visualize the sine waves in my mind.

As I said, there are a couple of different ways to approach this objective.
But.....you queried about the Flex Eight.  In that case you only have one approach.  

That said, I don't really consider correcting phase distortion to be a priority objective in speaker design.  There are myriad other factors higher on the priority list, in my opinion.

Dave.

Thank you, Dave.

I suppose the real question I'm trying to answer is whether to purchase 2 2x4HD units or one Flex Eight unit to implement stereo 4-way active crossovers. It seems the Flex has better analog stage than the HD, but 2 HD's means double the processing power for 75% of the cost. Any thoughts there?

To me, it's not even a close call.  The Flex Eight is the superior scheme in all aspects.
The only thing we don't know is regarding long-term reliability of the Flex device.  The 2x4HD has been a long-lived device that most folks have used very successfully.  Hopefully the Flex units can do the same.

Dave.

Thank you for keeping the conversation going Dave. By "scheme", do you mean the global phase correction versus individual driver phase correction that we have previously talked about, or are you seeing some hardware layout differences?

Both units use the same processor. I'd think having one processor per four channels would be more flexible than one per eight. Seems the allocation of filtering capabilities between input and output channels is more of a software design than physical limitation, though I'd be stuck with whatever plug-ins are available...

No, I'm talking about the overall "scheme" regards user interface, volume control, etc, etc.
Sans the missing streamer capability, (easily added externally) it seems to check all the boxes for an integrated 4-way crossover/EQ setup that would function as a friendly control unit for an advanced system.
I don't consider the FIR capability/architecture differences between the two units to be an important factor.  But, I don't use FIR filtering capabilities on any of my systems since I don't believe it's necessary.  You might. 

Dave.