RME HDSPe AES card/TotalMix Fx and active XOs: subs and LFE

[etc6849]

I got a decent deal on a b-stock RME HDSPe AES card and with the advent of 4k UHD bluray playback on my PC, I want to eliminate my Emotiva XMC-1 AV processor.  I will use my PC for all decoding and room correction (Dirac Live) and supplement it with another Xilica XD4080 to handle additional bass management.  This means no standalone preamp or home theater processor.

 

Going to buy another Xilica XD4080 (#2) and send AES outputs for 7.1 to two Xilica processors where one XD4080 #1's active crossover is used for:

driving mono subs, FL and FR (tri amp outputs) and center amp, and the other XD4080 #2 used for driving my rear speakers and side speakers (eventually I'll fully tri-amp or quasi bi-amp my rear speakers and add side speakers).

 

My question is what is the best option for handling bass?

Option A:  I could use the TotalMix Fx software console to sum all 8 channels and send it out via AES and let the Xilica do a LPF and time delays for this mono signal and send it out to the subs.  The problem with doing this is that if there was LFE content above the cutoff frequency (say 90Hz) it would be lost, although I could send it out FL and FR using XD4080 #1.

 

Option B:  FL, FR and C would get crossed to sub out internally by XD4080 #1, LFE would also be sent to sub's analog out (AES signals from PC: FL & FR, C & LFE so two XLR AES cables).  Additionally, I'd send AES out from XD4080 #2 into XD4080 #1; this signal from XD4080 #2 would contain low frequencies for rear and side speakers and be summed in XD4080 #1.  This wastes half an input pair (due to the way the Xilica assigns AES I/O), and I am also uncertain about latency here, although since I plan to use two of the same speaker processors, latency differences would likely not matter (although the sub signal fed out #2's AES out would be slightly delayed as it went through a Xilica digital input already).

 

http://xilica.com/products/xd/

 

https://www.rme-audio.de/en/products/hdspe_aes.php

 

Small drawback to the above:

1. I would need to find a foobar2000 plugin that would down sample anything above 96kHz for input into the Xilica's.

2. No Atmos/DTS-X decoder right now for consumer software playback (although this could eventually change).

 

Lots of pluses:

1. RME uses 42 bit processing and seems like it would be a superior volume control to what any preamp can offer (including an analog pre driving the XD4080's analog input).

2. PC does all decoding, which makes things future proof.

3. Lower noise floor as the XMC-1 is doing unnecessary A/D and D/A conversions for speakers that are triamped.

4. 40 bit processing with DSP in the Xilicas, 42 bit processing with DSP in RME card and 32 bit room correction done through software using a CPU (not a DSP).  All are superior to what any processor under $15k can do.

5. Much more versatility, FIR filtering for LF/MF and MF/HF if I later triamp more speakers.

6. Much better sub alignment as I can pick the order and filter type for how each speaker is crossed to subs.

7. No worry of upgrading processor when HDMI or HDCP changes in the future.

8. Avoids ADC of the Xilica channel inputs (right now when AV processor feeds these there is unnecessary D/A and A/D conversion taking place).

 

[Chris A]

I found one drawback: DSD files (i.e., SACD music files, especially multichannel).  There are two legal roadblocks that prevent electronics manufacturers from doing two things:

 

1) Convert HDMI data to another digital format directly without first converting to analog, and,

2) Allow multichannel DSD-over-HDMI except on special devices (i.e., there are no PC-based HDMI buses that will allow you to pass DSD files over them, and only on certain disc players). (Multichannel DSD files cannot be sent over S/PDIF or AES3 to my knowledge.)

 

This means that multichannel DSD files--which I find are probably the highest fidelity in terms of their perceived bass and extreme treble response (i.e., no phase shifts due to mastering, which leads to much more natural sound in multichannel recordings).  You're thus stuck with an internally automatic DSD-PCM conversion in the source player/server/PC if trying to use the above PCIe-->AES3.  All of these DSD-to-PCM conversions in my experience remove the fidelity I hear (easily hear) with DSD multichannel files playing directly on my Oppo player-->HDMI-->preamp/processor-->balanced analog to the DSP crossover(s)-->loudspeaker drivers.

 

Additionally, there is evidence that conversion to analog, and then back to digital is below audibility in terms of degrading the sound quality.  This is something that I've experienced with Xilica processors: they are totally transparent. 

 

Chris