Martin Poulin
November 2018,
​
I'm still in Alberta… for who knows how long. Tinkering about the next setup (multi-way and multi-speaker)
I do plan on a "5.1" system, but with every five channels active, DSP compensated and the ".1" to become eight mono subwoofers .
If the three front speakers are 4-way, that brings me to 12 channels required. (4x3=12) The rear will be 2-way for sure and maybe 3-ways, here goes four more channels (16 channels required already). Add eight subwoofers (may be mono but DSP will likely be somewhat independent), four channels will be likely required for subs.
This bring me to a total of 20 channels.
​
Given the rear speaker could be changed to 3-way and maybe even 5-way for the front main (L+R+C), I’d like at least 25 channels available.
If every subwoofer has an independent DSP instead of being used in pair, it would require as many as 29 channels.
Options for such setup are VERY limited! Standard product configuration call for 32 channels.
I may use up to 90% of such product (29/32=90%), leaving room for limited expansion.
​
I know I have to “think outside of the box” as no commercial product offers what I’m really looking for, especially if I want to implement FIR style of filters.
Also, when searching for something you don’t know exists, it's quite time consuming and challenging to come up with a solution that bridges the gap. I've dedicated countless hours on this matter and here’s my solution. It's not ideal but it does work, doesn't break the bank and can be improved with time.
​
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
​
Introduction:
​
1. When using passive stereo speakers, life is very compromised but improvements are easy, given you can use digital tools and don’t believe in snake oil theory.
At the risk of getting the analog purists infuriated, digital room correction and DSP can significantly help the average set up. Convolution applied to a given signal can correct for room irregularity and compensate quite a bit for possible speaker shortfalls. Left and right independent equalization can compensate for room acoustics and re-center the phantom center image, right in the middle, where it should be. If loudspeaker positioning is compromised, "stereo enhancer" processing could be loaded to great positive effect in a single playback system. I've been guilty of this more than once. Many 3rd party products do just that, and should be seriously investigated. Your average CD is "engineered and processed" by many people, so doing it once more by your ear isn’t a big deal. If you want pure "electronic-free" sound, go to a classical music concert for proof.
-One just have to accept one extra stage in the analog signal, insert a DSP unit and correct your sound.
​
2. When using 2-way active speakers (4 channel crossovers required and 4 amplifiers), life becomes very easy. Two stereo amplifiers and a basic mini DSP 2x4hd will do the job for very little money. These are easy to use, and provide decent sound. A single home theater amplifier would do as well… (they boast 5 channels) and voila!
This is exactly where I started, 2-way passive bookshelf speaker (6" woofer +1" tweeter) + two stereo subwoofers on a mini DSP 2x4.
-Eventually, I grew to an active 3-way set-up using the same 2 ways speaker with an active crossover + my subs, and upped the game further thereafter.
​
3. Moving to a 3-way set-up proved a tad more expensive and opened the door for more serious results. Plenty of brands offer solutions, with 3-way DSP + crossover preamp at various price points. From affordable miniDSP models to more expensive DEQX units or even analog tubed Marchand crossovers (no DSP here). The 3-way set-up was just a start for me, but good results can be realized… if implemented correctly.
​
4. The 4-way set-up, (or 3 way + subs) is where I just started to become excited.
Here you start to see the various shortfalls of hardware limitation. Most high-end DSP units offer six channels (ie: stereo 3 ways is the maximum). For 12 channels, DEQX recommends piggy backing two of the six channel units! Yes this works, but it's pricey at $5000USD for each of their HDP-5 models. And no matter how good the DEQX may be, it’s not my dream to own one, as every channel processing is limited to 10 bands, this is a serious limitation for bass management. The DEQX also uses an undocumented "hybrid IIR/FIR filter". No clue what they really do, so it’s suspicious to me, like most marketing BS.
By comparison, my midi DSP 4x10HD has just five bands per channel, half the DEQX capacity and only has a maximum of 48 dB IIR crossover slope ability, but they do offer eight analog outputs, plus two more in digital domain (SPDIF) for 10 processed outputs. More outputs but not enough, and it costs 10 times less than DEQX. Those eight analog outputs are the reason I bought two of the miniDSP 4 x10HD. But their analog section isn’t sophisticated and they’re fairly noisy units. I'm not aware of a high-end consumer product offering more than eight analog outputs. At the moment, this is a niche market. I’m now foreseeing about 20 channels required on my next system...all with independent DSP. Buying, four DEQX for $20KUSD however? Not gonna happen...
​
5. At 5-way set-up (10 channels) is where the issues are apparent. MiniDSP and DEQX can't do it unless you piggy back 2 units.
Serious and impossible, unless going with full professional gear (IE: studio grade quality). No more Stereophile recommending you their components of the year LOL.
Besides…many professional interfaces are limited to eight analog outputs, with few offering 16 outputs, and only a rare selection offer 32 channels.
​
6. At 6-way set-up and more. (above 10 channels) My need is for roughly 28 fully DSP, active channels: "my solution"
​
My Motu 1248 isn’t exactly the top end interface, but it provides me with the ability to free myself from the compromised interface of the mini DSP, with using the "unlimited" processing power of my computer instead. (currently Windows 10, 64 bits, Intel I7 with 16Gb of Ram)
​
I can also try implementation that I could scale up later as the 1248 only has 8 analog outputs.
This way, I can process all eight channels independently inside the computer and send the result directly to all of the eight analog outputs of the Motu1248 (ie: totally bypassing the internal Motu DSP as per below screenshot).
​
Audio Path to Follow:
​
-
JRiver 23 media player re-samples every track to 88.2 kHz and sends the digital stereo output to VB Cable (free audio re-routing software)
-
VB Cable operate in "WMD mode" it is also clocked at 88.2 kHz, it receive the signal and send it to asio4ALL software (free software as well)
-
ASIO4ALL acts as a converter to link the VB cable WMD signal to the Motu "Asio" mode, all at 88.2 kHz.
-
PetalBoard 2 software (free) receive the Asio signal and allow to split incoming signal to infinity.
-
Every resulting signals can be processed at will. Just add VST plug in and route the signals to any output as required. (8 outputs shown as per my motu1248)
-
I use 4 instance of Fabfilter Pro-Q (Asio compatible VST Plug in, not free) for crossover duty and DSP processing. ​
-
I then add required time delays per stereo output to time align my speaker. I use a free VST plug in called Sound Delay. (unlimited delay possible).​
-
All eight resulting channels (4 ways stereo) are routed to the Motu1248 per Petal Board output map.
-
Latency is low, Motu is just used as a DAC, no processing done by the hardware, the computer does everything.
​
The above screenshot that show all 5 software's running simultaneously took me about 10 hours to get working!
​​
Update: April 2020, I do not use that solution anymore but it does work.
​
I used the wrong software. I tried lots of different software options and combination to get something that finally worked. Driver compatibility issues, rebooting computer multiple times, sampling frequency errors, no sound, crash on DSD files playback, up-sampling… you name it! Seemed like it would never work. Discussion forums were of limited help, as nobody I know uses this kind of setup. Eventually however, it worked… and worked like a charm. So powerful!
​
Of course, I'm committed to a single sample frequency to have it work properly. Arbitrarily, I used 88.2 kHz, but I may also try 96 kHz. Using 192 kHz would also work, but it won't be able to carry enough channels later, on a single USB 2.0 cable, you can only transfer so much data, 192 kHz doesn't allow 8 simultaneous channels. So, it must be: 44.1, 48, 88.2 or 96 kHz.
​
The processing power required by the CPU is much less than initially feared. A I7 computer is already leaps and bounds above the internal capacity of both the miniDSP and the Motu 1248. I'm using 96 dB/octave IIR slope with 12 EQ bands per channel, and 1000 mS delays on every channel (knowingly using more CPU power than will be required in real life, worst case scenario) Yet… with eight channels operating simultaneously, I peak @ ~25% CPU utilization rate when playing harder to process DSD files.
My computer draw about 5% CPU usage at idle, so 20% net CPU is used for audio processing in my worst-case test scenario, plenty of room left!
​​
This put a big smile on my face, because I know that on normal programming, DSP processing would represent maybe 12% net CPU for eight simultaneous channels. Speculating to ~20 channels, it may hit 35% net CPU usage while playing DSD files, leaving lots of room to implement FIR crossover slope (they are CPU hungry). On ~20 channels FIR+DSP set-up, I anticipate 60-75% CPU will be required. Probably quite doable even if I have to use a bigger buffer to cover other Windows 10 functions (system scan or antivirus check for example).
​
​