I was obsessing over this last week. It seemed to be then, and up until a few minutes ago, that while the Linkwitz-Riley LR4 crossover has both drivers in phase at the crossover frequency, because each has 180 degrees shift from opposite directions, as you move away from the crossover frequency, the phase would unwind in different directions, so they would not always be in-phase at every frequency.
This is wrong, I finally figured tonight. First of all, Linkwitz says specifically that both drivers are always in phase for LR4. At every frequency. (And I should believe him, of course.)
Second of all, it is true, the driver phase either wind up more (as you attenuate) or unwind back to zero (as you increase to max). However, when you are moving UP from the crossover frequency, this means the HP is "unwinding" and the LP is "more-winding", and the HP winds toward leading, and the LP winds toward lagging, so when one is winding and the other is unwinding they are actually moving in the same direction because they wind in different directions.
I used to think the polarity reversal on one side of an LR2 was the magic that made the in-phase-at-every-frequency work. Because of that polarity reversal, I reasoned wrongly, the phases move in the same direction as you move away from the crossover point in either direction. But that's just a compensation for the crossover frequency itself, and it doesn't change the way the phases on each driver change as you move away from the crossover point.
Along with all this, I believed there were "in-phase" connections of LR4 and "out-of-phase" connection of LR4 that would work even better because less group delay. I read that somewhere about some crossover, but it was clearly not about LR4. I believe another JAES author (Bullock?) wrote about a series of all pass crossovers. Linkwitz himself stumbled upon another one (the Dueland) which might actually be better than LR4 because more gradual phase changes.
Somewhere I think I read about delaying the LR4 Highpass by "360 degrees" presumably at the crossover frequency, to account for the fact the leading vs lagging of the two drivers. Now I can't seem to find that suggestion. I didn't try that much seriously before (I tried going in that direction a bit, and results were weird enough to make me think about doing a more serious investigation).
Well if that works for LR4, what about doing LR2 using 180 degrees of delay, rather than an actual reversed polarity?
This is wrong, I finally figured tonight. First of all, Linkwitz says specifically that both drivers are always in phase for LR4. At every frequency. (And I should believe him, of course.)
Second of all, it is true, the driver phase either wind up more (as you attenuate) or unwind back to zero (as you increase to max). However, when you are moving UP from the crossover frequency, this means the HP is "unwinding" and the LP is "more-winding", and the HP winds toward leading, and the LP winds toward lagging, so when one is winding and the other is unwinding they are actually moving in the same direction because they wind in different directions.
I used to think the polarity reversal on one side of an LR2 was the magic that made the in-phase-at-every-frequency work. Because of that polarity reversal, I reasoned wrongly, the phases move in the same direction as you move away from the crossover point in either direction. But that's just a compensation for the crossover frequency itself, and it doesn't change the way the phases on each driver change as you move away from the crossover point.
Along with all this, I believed there were "in-phase" connections of LR4 and "out-of-phase" connection of LR4 that would work even better because less group delay. I read that somewhere about some crossover, but it was clearly not about LR4. I believe another JAES author (Bullock?) wrote about a series of all pass crossovers. Linkwitz himself stumbled upon another one (the Dueland) which might actually be better than LR4 because more gradual phase changes.
Somewhere I think I read about delaying the LR4 Highpass by "360 degrees" presumably at the crossover frequency, to account for the fact the leading vs lagging of the two drivers. Now I can't seem to find that suggestion. I didn't try that much seriously before (I tried going in that direction a bit, and results were weird enough to make me think about doing a more serious investigation).
Well if that works for LR4, what about doing LR2 using 180 degrees of delay, rather than an actual reversed polarity?
No comments:
Post a Comment