Crossover phase help

Yep, I am sure you do.

I was not talking about crossover points, I was commenting on adding all-pass filters to linearise the phase response ... its the phase response of the xover+driver you have to correct, not the xover alone. 

All-pass filters can in theory be inserted into any or all of the Xover bands.

However usually at the outputs so as to compensate also for component tolerance variations and others such as individual speakers, etc, to a greater extent.

Makes a big difference to the transient performance of a system when properly set up.

Have a look here:
http://www.edn.com/electronics-blogs/living-analog/4375814/All-pass-filter-phase-shifter
Not the only web-page on this by a long way..... but does cover the 'basics' and some of the design maths.... together with basic cascading of them

Mik

Edited by audiomik - 03 May 2015 at 6:41am

RobinMatrix wrote: Before you get too carried away with all-pass filters to linearise the phase do please remember one key point, something that is continually forgotten by many: The phase and frequency response you want to get right are *NOT* the response of the corssover ... but the response of the SYSTEM ... that means crossover, amp and drivers.  It is completely pointless adding all-pass and making subtle tweeks to get the electrical response of the filter right when the response is going to be the product of the filter, the amp and the driver ... you can make the sum of the crossover perfectly flat and linear phase ... but the actual response of the system will be pants, because you have to take into account the drivers. The simplest illustration of this is you design a perfect 12db an octave roll off for the HF, and you position the FF3 point at 4Khz .. the HF driver itself also drops off below 4KHz at around 6dB an octave so the actual response (and phase shift) is related to an 18dB/octave slope, not 12dB so your theoretical corrections are wrong. And yes, because of the Bode relationship,  whenever you get a change in the frequency response, you get a change in phase .. so if you were wondering if the drivers changes in response where it falls off will produce a phase shift, yes they will. And finally, there is the delay/phase shift from driver positioning ... you can add all-pass filters and get the electrical response right, but could well be making the acoustic response worse, as you are adding delay to something that needs delay removing ... Ever wondered why many passive crossovers have a single pole filter on the bass and a two pole on the HF?  They often have the same slope on the HF and LF sides of the curve simply because the designer is using the HF roll off of the bass driver to provide the other pole. You have to consider the system, not just one part of the system to find the frequency and phase response. My advice: don't bother trying to correct the xover response ... plot the actual system response and get that right.

Edited by fatfreddiescat - 04 May 2015 at 12:05am

The filter really needs to be 'tuned' to each particular system you´re using, so it´s a bit more complicated than just chucking in an all pass. You need to analyse the whole system with something like SMAART and tweak the phase then. I think you´re overcomplicating the whole thing, really.