TOPIC: PWR-ICE Tweeter/Woofer Delay for FIR Filters

The User Manual of PWR-ICE mentions the need of a delay to be set for the woofer output channel, when using FIR filters, due to asymmetric tap allocation (2300 taps for woofer channel, 300 taps for tweeter channel).

I'm a bit puzzled by this, since the woofer output channel (channel 1) with the longer 2300 tap FIR filter should impose a longer delay relative to the tweeter output channel (channel 2) with the much shorter 300 tap filter. So to get the tweeter & woofer signals back into time alignment, and assuming all FIR taps are fully utilized for both channels, shouldn't a delay (20.83ms) be set for the tweeter channel instead of the woofer channel?

Hi scan80269,

Alignment delay and FIR filter length are very different aspects of a design.

Filtering at lower frequencies generally requires filters with longer impulse responses and so the FIR filter taps available needs to be longer. Filtering at higher frequencies generally requires filters with shorter impulse responses.

In addition to the FIR filters, the plugin has delay available on each channel. So you can set the FIR filters to do the mag and phase manipulation you want AND align the drivers and FIR filters using the delay. The alignment delay will depend on both the acoustic driver offset and the location of the IR peaks within in the FIR filters.

Note that the filters don't need to be linear-phase (with symmetric IR's). The filters can be completely arbitrary in mag and phase with the IR peak anywhere within the available taps. For more details, maybe take a look at the FIR whitepaper and the 2-way tutorial on our website.

Kind regards,
Michael

Michael,

Thanks for your detailed response.

I am actually very familiar with the concepts you mentioned, and have been designing linear-phase crossover filters and implementing driver time alignment for 2-way loudspeakers as a hobby for over a year now.

The main reason for my post is to inquire whether the PWR-ICE user manual is correct in mentioning a delay being needed for the woofer channel, presumably to counteract the time difference imposed by the unequal FIR tap counts (300 for tweeter channel, 2300 for woofer channel). If I create a 2300 tap filter for the woofer and a 300 tap filter for the tweeter, then isn't it the woofer signal that is delayed relative to the tweeter channel by something like 20.8 ms, and not the other way around? In other words, with the woofer signal being significantly delayed (through a longer filter) relative to the tweeter signal, shouldn't a delay be inserted into the tweeter channel to align it with the woofer channel?

So should the following sentence in the PWR-ICE manual:

Note that the delay on the woofer output channel will probably need to be set to allow for the different delays created by the two filters.

be corrected to:

Note that the delay on the tweeter output channel will probably need to be set to allow for the different delays created by the two filters.

Hi scan80269,

I see what you mean w.r.t. to the manual but I wouldn't worry about it.

The delay (on either driver) is totally dependent on the design.

Example 1: If both drivers are physically aligned and both FIR filters are designed for minimum-phase, then little to no delay would be needed. (Sometimes maybe a little delay is used in conjunction with polarity reversal to phase align the drivers through the crossover region.)

Example 2: If both drivers are physically aligned and both FIR filters are designed for linear-phase, then the alignment delay on the tweeter would be (2300/2 - 300/2) / 48000 = 21.9ms. Again, the delay might need tweaking.

Example 3: If the tweeter is deeper in the cabinet (like some large MTM designs) and both FIR filters are design for linear-phase, then the alignment delay on the tweeter would be approximately 21.9ms minus the time difference of the physical distance between the woofer and tweeter.

Example 4: Same setup as example 3 but both FIR filters are design for minimum-phase, then the alignment delay on the woofer would be approximate the time difference of the physical distance between the woofer and tweeter.

I could go on... The above examples only mention minimum-phase and linear-phase filters, but the FIR filters can be arbitrary phase with the IR peak anywhere in the FIR coefficients, and so the alignment delay could be anything. Again, if you take a look at our 2-way tutorial, it discusses how the alignment delay is calculated after the FIR filters are designed. It also discusses how the driver measurements should retain any delay from relative physical distance differences, in order to make the design easy and avoid subsequent delay tuning.

Kind regards,
Michael