Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Art Welter]

Quote from: Art Welter on Today at 01:06:15 pm

    A bandpass box can exhibit as much as 720 degrees of phase shift, could that be flattened with FIR?
    If it could, what would the overall latency of a FIR filter capable of reducing 100 to 720 degrees of phase shift down to 20 Hz be?

I would estimate at least 50ms, and probably more like 100ms.

Slap echo between stage and mains,  not good.

Appears the FIRmaker is working down to below 80 Hz, wonder what the latency is ?

[Peter Morris]

Since you guys have kept this thread alive, I would like to make a 'clarification' to what I have said.  This is what happens when I jump into a discussion based on memory for something I haven't worked with for years.

For continuous time (analog) filter functions, the IPR of a linear phase filter will be strictly symmetric about a point in time.  That time will be dependent upon the choice of t=0, and the amount of phase slope (group delay) of the filter.  Anyway, for a sampled filter, the impulse response values will only be strictly symmetric for delay times which are integer multiples of the sample period.  For arbitrary delay times, the numeric values will not be symmetric, but if you ran them through a reconstruction filter (D/A convertor), the resulting analog IPR would be symmetric, within error margins.  For these fractional sample time delay values, the FIR IPR table values would correspond to resampling the reconstructed analog function at a slightly time shifted set of points that no longer were positioned symmetrically to the IPR.  We had used just such filters thirty some years ago for synthetic aperture radar DSP.

For beamsteering as FIRmaker does, to deal properly with arbitary box heights, hang geometries, and steering angles, you would want to have finely variable delay times, and not just time shift the signal by an integer number of sample periods.  Therefore I would bet the FIR tables it generates are not sampled symmetically.  Whether FIRmaker generates linear phase filters is an interesting question.  Linear phase filters would delay all the frequencies equally, to accomplish beam steering.  Whether there is anything to be gained in the quality of the steering by delaying different frequencies by different delay times is doubtful to me.

If anyone is interested, phased array techniques in radio and radar go far  back in time:  http://en.wikipedia.org/wiki/Phased_array

Hi Jon,

You might like this plot – same speaker, same phase but  totally different amplitude response.

[Kenneth Berger]

Just a quick note about how may channels of dsp/amps you need to accomplish real benefits from FIRmaker style room FIR's.
Having looked at some examples of single and double box FIRmaker examples you have to think about bandwidth and acoustic centers and overall array height to determine how useful it will be over what bandwidth.  You could conceivably do all your HF drivers individually (FIR>DSP>AMP) mids every other one (FIR>DSP>AMP) and LF every 4 drivers / enclosures (FIR>DSP>AMP) and get most or all the benefits.

[ProSoundWeb Community]