[leoduran]

Is not clear to me if the DspNexus is capable of implementing linear phase X-overs (FIR filters), or if it has some other mechanism to achieve the same goal: time/phase alignment.
Thank you!

[HAL]

Audio Weaver has two types of FIR filters to use with user input filter coefficients, standard and fast.

Depends on how many taps the filter requires.  At 192KHz SR and fast memory the standard FIR filter is 3000 taps per channel.  This can be increased to 6000 taps with decimation and interpolation to 96KHz and back to 192KHz for processing.  The fast FIR filter uses 2x the memory of the standard FIR, but can help with throughput.

That is why I help a customer with their DSP crossover implementation to see what is required to assess their processing needs.

[leoduran]

Thanks for your feedback.
At this point I do not know how many taps would be required for my simple 2-way system.
I have a pair of Qual ESL 2905's (which do not have a crossover) and a pair of REL subs, so I'd like to setup an HPF for the panels and a LPF for the subs.

[HAL]

Audio Weaver does not calculate the FIR filter tap coefficient files for crossovers.  That is done with external filter programs that the customer would produce as files of coefficients to use in Audio Weaver. 

In my trials on a few DSP systems, Linear Phase crossovers do not sound as good to me as standard IIR filters that Audio Weaver does produce. 

Using the standard high rate Linkwitz-Riley DSP crossovers has worked very well for integrating speakers and subs with the correct time delay.  Then there is extra processing power to use REW to measure and do at listening seat frequency response correction. 

[leoduran]

Before correcting the room, can/do you (first) measure each channel (driver)? If so, it’s REW suitable for that purpose?…. And can the speakers stay in the listening room?

Basically, can/do you apply per-channel corrections before applying room corrections?

Thanks!

[HAL]

Yes, you use REW to measure each speakers frequency response with the DSP crossover to match levels and then measure the full response at the listening position for frequency response correction.   Once the customer makes the REW in room measurements, the saved files would be emailed to me for processing and Audio Weaver DSP crossover design updates.

The customer will need to purchase a calibrated mic like a Beyerdynamic MM-1 and get the calibration files from the company after purchase for the measurements.

[leoduran]

Thanks, that makes sense!… I presume the MIC would simply be placed near each driver, with the speakers still in the listening room (i.e., I would not have to isolate the speakers).

But it sounds like the applied correction is for level/sensitivity matching, most likely corrected by analog attenuation at the output of the DAC (-3dB steps on DspNexus), is that correct?… If so, what about the frequency response of each driver, is there per-channel correction done to flatten the response of each driver?

[HAL]

It is not necessary to put the mic next to each driver. 

The measurements need to see the entire speaker for correct response due to driver distance offsets.  That can be made on axis quasi-anechoic measurements for the speaker at listening height.  Since the Quad has an internal high voltage passive crossover, there is no way to make separate driver measurements. 

Audio Weaver has high precision digital attenuation to adjust the Quad to subs level matching.  Also, there is analog 3dB step attenuation for the DAC channels.   

[leoduran]

Thanks!… In my case, I meant each whole Quad panel as a driver/channel, as well as each Sub being a channel (4 channels, total).