Hi everyone,
I have a big hole in my understanding of a topic. Let's say you want to try to achieve a 2 way, LR alignment with a given pair of speaker drivers.
You go to some calculator which gives you the capacitor and inductor values. Now, ignoring the drivers for this discussion (let's pretend they are ideal) the phases will match only assuming there is no distance (time delay) from the tweeter to the woofer, which is almost never the case.
You start with this ideal crossover, which looks pretty flat, dial in the delay and it all goes to pot. Your flat frequency response is now pretty wavy, and flipping the tweeter over only partially remedies it or makes it worse.
In my last two crossover designs I've gotten pretty lucky, but is there a canonical process for this, or a way to calculate an crossover with the time delay in mind from the start? Or is there a default circuit to move the 0 point for a driver one way or another?
Best,
Erik
I have a big hole in my understanding of a topic. Let's say you want to try to achieve a 2 way, LR alignment with a given pair of speaker drivers.
You go to some calculator which gives you the capacitor and inductor values. Now, ignoring the drivers for this discussion (let's pretend they are ideal) the phases will match only assuming there is no distance (time delay) from the tweeter to the woofer, which is almost never the case.
You start with this ideal crossover, which looks pretty flat, dial in the delay and it all goes to pot. Your flat frequency response is now pretty wavy, and flipping the tweeter over only partially remedies it or makes it worse.
In my last two crossover designs I've gotten pretty lucky, but is there a canonical process for this, or a way to calculate an crossover with the time delay in mind from the start? Or is there a default circuit to move the 0 point for a driver one way or another?
Best,
Erik
Last edited:
Hi Sreten, are you talking about the ladder delay network? Seems pretty complicated!
Erik
Ladder networks do work, but they are not really necessary. The ordinary way is to use assymetrical filters. David Smith (speakerdave) has covered this subject with a good example.
In order to justify the validity of doing this you could familiarise yourself with the group delay threshholds of audibility.
In order to justify the validity of doing this you could familiarise yourself with the group delay threshholds of audibility.
You need to physically offset the drivers. Typical woofer will be delayed by 25mm behind tweeter. Sometimes more. Designs that achieve time alignment uses several inches of offset sometimes. Look at Dunlavy SC-IV and Danley Synergy. There is no way an all pass delay will get you of order 1 millisecond delay. The easy way is to use DSP, the you can keep the flat baffle. That's what I do to implement the Harsch XO which requires typical 1 to 1.5ms delay for a low 400Hz xo. Sound needs to be delayed 13in for every millisecond so not practical from passive or even offset driver scheme.
Hi,
That's for LR2 and thoroughly discussed.
In the other designs asymmetry of the target acoustic
responses is used to compensate for driver offsets.
rgds, sreten.
Last edited:
That doesn't mean I can follow it. 🙂 I think I'll pass on that for now.
So, moving the xover points and q's of the each filter circuit? I can do that. 🙂 I thought of that but I thought that was cheating somehow.
Best,
Erik
Right, but for instance, my current project the cabinet is already built, but most of my designs would be no more comlicated than a Dayton kit box. 🙂
I can see how easy it would be in the DSP world.
I guess what I should have done was clarify in my questions that I wanted to alter my speaker level crossover to get proper phase matching at the crossover point. Like you mentioned, woofer's and mids often lag the tweeter by about an inch (paraphrasing) so I was curious how the big boys do this in their speaker level crossovers.
Thanks for all that,
Erik
Hi,
If it works its not cheating, your basically manipulating
the vertical response around the crossover region.
Due to the offset the ideal acoustic responses will not
sum ideally, so your basically making the best choice
of your real options, which is nothing like cheating.
rgds, sreten.
You're not moving the xover point. You are making the acoustic rolloff slopes asymmetrical to compensate for the Z-offset difference. The added phase delay of a 3rd order electrical on a tweeter with a 2nd order electrical on the adjacent woofer adds phase delay to the tweeter through the xover region.
Later,
Wolf
Cool! 🙂
I guess I thought it was cheating because I thought the pro's had a better weay, or there was some rule of thumb, like reduce filter Q by 0.05 per inch...
Thanks Sreten!
Erik
I guess I thought it was cheating because I thought the pro's had a better weay, or there was some rule of thumb, like reduce filter Q by 0.05 per inch...
Thanks Sreten!
Erik
Won't fly. This is frequency dependent, cross low enough and it no longer matters.
Reduce woofer filter influence by using a lower order filter (typical), increasing the woofer turnover frequency or vice versa on the tweeter.
You are making a tradeoff. If the response is increased at the crossover it can be dialled back by leaving the phase slightly out on axis.
- Status
- Not open for further replies.