• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Can I use multiple stereo DACs instead of one multichannel DAC?

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
MOTUs don't have per-stereo-pair digital inputs (4x SPDIF or 4x AES), instead they use ADAT. as far as I know Okto is the only DAC that takes such weird inputs.
This means I cannot use a nanoDIGI 2x8 as input.
This means the DSP will have to be performed in software (deal breaker for me, I don't trust windows with 3000W of amplification), or in the MOTU itself, which would be great.
However in order to determine that I'll need to read the entire manual for the MOTU and its software, every word, and again for every model I'm considering. so unless I can find someone who runs a LX521 from a MOTU, it's too much research work.

My post was answer to a direct question by @Dichotome, not a recommendation for yourself as you've clearly indicated you are not interested in software EQ options. Although if you can live with the volume control knob on the MOTU I think the MOTU paired with a RPi4 running CamillaDSP would be a great option for you but I also understand the software EQ hesitancy.

Michael
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
Ended up running a few measurements to confirm the theory that DACs downstream of the nanoDIGI will not have synchronization issues. I used 4 DACs, 2x Schiit Modius (AK4493 based so no ASRC) and 2x Audiophonics ES9023 (has on chip ASRC). Results were exactly as expected, identical DACs had identical delays resulting in identical phase response. There were differences in delays between the two sets of DACs but these could be compensated for with delay in the nanoDIGI to within 20 deg deviation at 20 kHz. I ran measurements days apart and experienced no shifting of any kind. Even so it probably makes sense to use identical DACs to avoid fussing around with delays

See plot below for details. Top mostly flat trace is phase response of Modius compared to another Modius and ES9023 compared to another ES9023. Bottom trace shows large phase deviation when comparing the Modius to ES9023 without correcting delay. And finally middle trace shows corrected phase response between the Modius and ES9023 after adding a 0.07 ms delay to Modius.

Screen Shot 2021-07-28 at 4.24.46 PM.png


Michael
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
I would guess there would be a slight latency difference (if only for latency differences from the second DAC) but this would be easy enough to measure. I have such a DAC (MOTU Ultralite Mk5), I can make some measurements with an external DAC and post the results.

When attempting a multichannel active DSP system like this it is a really good idea to have the ability to make some basic measurements. Heck, even the Okto dac8 pro has slight latency differences between analog output channels when using AES inputs.

Michael

Ran some tests on the Mk5 to see about this delay, I had about a 15 sample delay at 96 kHz (0.156 ms) on SPDIF output compared to the analog output. To me this is definitely another data point that says if you are implementing a multiway active system it is helpful to have the ability to measure delays and phase response, although if you are only integrating a sub this issue is way less critical.

Screen Shot 2021-07-28 at 4.41.43 PM.png


Michael
 

TimW

Major Contributor
Joined
Aug 15, 2018
Messages
1,065
Likes
1,406
Location
Seattle, WA
SHD Studio ($900) + MOTU ($600) + RPi4 w/ PSU ($70) + Bobwire DAT1 ($70)
I am unfamiliar with Camilla DSP and software equalization in general. I understand you are using the SHD as a source selector and volume control for the MOTU. Then the MOTU is performing crossover and equalization duties right? And this is done using Camilla on the RPi4 correct? So your signal is coming from the SHD, traveling to the MOTU via SPDIF or AES or something, and then the MOTU is sending it to the RPi4 via usb which applies processing and sends it back to the MOTU for DA conversion. Do I understand this correctly?

Also where did you get your miniSHARC with AES output?
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
I am unfamiliar with Camilla DSP and software equalization in general. I understand you are using the SHD as a source selector and volume control for the MOTU. Then the MOTU is performing crossover and equalization duties right? And this is done using Camilla on the RPi4 correct? So your signal is coming from the SHD, traveling to the MOTU via SPDIF or AES or something, and then the MOTU is sending it to the RPi4 via usb which applies processing and sends it back to the MOTU for DA conversion. Do I understand this correctly?

Also where did you get your miniSHARC with AES output?

Yes to all of your questions, seems like you understand exactly what is going on. :)

I am also rather new to software eq and had avoided it because I had no desire for a computer to be my single source and wanted more input functionality. I started playing it with after I saw the thread below.

https://www.audiosciencereview.com/...u-m4-phenomal-dsp-streamer.24493/#post-827339

Once I started playing with it I realized that with the right setup you can basically get software eq to operate exactly like a USB output miniDSP and have decent input functionality. Once you setup it up it is a complete set and forget device. The other nice thing is that the RPi4 has wifi and it is very easy to remote login to the RPi via SSH and update the configuration.

A device like the Ultralite Mk5 is super nice because you can route "normal" sources to it via TOSLINK or SPDIF, apply DSP in the RPi4 and then use the MOTU DAC as output. If you do not need a remote control the volume knob on the MOTU works great. I also really like that you can also use other input devices with software eq, for example you can use a DDRC-24 as a TOSLINK and analog input, have it apply Dirac processing and then route that signal to another DAC, software eq is very flexible.

The setup in the MOTU software itself is very basic, in the MOTU mixer you route computer USB 1-2 to output 1-2, USB 3-4 to output 3-4 and so on. All of the routing of the TOSLINK or SPDIF input signal is defined in CamillaDSP on the RPi4.

I built the miniSHARC with AES outputs, although to be honest it is pretty much all off the shelf parts with a bit of soldering. The build thread is linked below.

https://www.minidsp.com/forum/hardware-support/18550-minisharc-8-channel-digital-output

Michael
 
Last edited:

TimW

Major Contributor
Joined
Aug 15, 2018
Messages
1,065
Likes
1,406
Location
Seattle, WA
Yes to all of your questions, seems like you understand exactly what is going on. :)

I am also rather new to software eq and had avoided it because I had no desire for a computer to be my single source and wanted more input functionality. I started playing it with after I saw the thread below.

https://www.audiosciencereview.com/...u-m4-phenomal-dsp-streamer.24493/#post-827339

Once I started playing with it I realized that with the right setup you can basically get software eq to operate exactly like a USB output miniDSP and have decent input functionality. Once you setup it up it is a complete set and forget device. The other nice thing is that the RPi4 has wifi and it is very easy to remote login to the RPi via SSH and update the configuration.

A device like the Ultralite Mk5 is super nice because you can route "normal" sources to it via TOSLINK or SPDIF, apply DSP in the RPi4 and then use the MOTU DAC as output. If you do not need a remote control the volume knob on the MOTU works great. I also really like that you can also use other input devices with software eq, for example you can use a DDRC-24 as a TOSLINK and analog input, have it apply Dirac processing and then route that signal to another DAC, software eq is very flexible.

The setup in the MOTU software itself is very basic, in the MOTU mixer you route computer USB 1-2 to output 1-2, USB 3-4 to output 3-4 and so on. All of the routing of the TOSLINK or SPDIF input signal is defined in CamillaDSP on the RPi4.

I built the miniSHARC with AES outputs, although to be honest it is pretty much all off the shelf parts with a bit of soldering. The build thread is linked below.

https://www.minidsp.com/forum/hardware-support/18550-minisharc-8-channel-digital-output

Michael
Thanks for the descriptions. Currently I'm using an SHD as my preamp but I would like more outputs. It is nice to know that an interface can be used for output and input at the same time with Camilla DSP. I have a Focusrite Scarlett 18i20 I could use for this and an RPi3 B+ that might have enough processing power. Hopefully the ASR Streamer project continues in the direction I last saw which was making moode with Camilla easier to install and get running with maybe some added functionality.

I also have a miniSHARC with a waveflex screen. Currently it is setup as a 2x2 with the digi board but originally it had the DA8 dac board installed. Would be nice to set it up with 8 digital outputs.
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
Thanks for the descriptions. Currently I'm using an SHD as my preamp but I would like more outputs. It is nice to know that an interface can be used for output and input at the same time with Camilla DSP. I have a Focusrite Scarlett 18i20 I could use for this and an RPi3 B+ that might have enough processing power. Hopefully the ASR Streamer project continues in the direction I last saw which was making moode with Camilla easier to install and get running with maybe some added functionality.

I also have a miniSHARC with a waveflex screen. Currently it is setup as a 2x2 with the digi board but originally it had the DA8 dac board installed. Would be nice to set it up with 8 digital outputs.

The Moode comment is interesting. A few months back I tried to get Moode to work with CamillaDSP and was unable to do so. In general it seems like the audio focused OSes do not play very nice with multichannel DACs.

I also first tried Raspberry Pi OS (Raspian) and had issues with intermittent audio cut outs. Some of these were very audible (sound drops out for a second) but some were so short that I almost thought I was losing my mind and they weren't real. Turns out when I ran some measurements I could see the drop outs on the RTA when looking at the FFT. Once I moved to Ubuntu Server the drop outs went away and measurements were rock solid. To be honest I like the fact that the RPi is only running CamillaDSP and nothing else, makes it very simple.

I understand that most folks want a much more plug and play solution but to me the tinkering/learning is half of the fun.

Michael
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
Thanks for the descriptions. Currently I'm using an SHD as my preamp but I would like more outputs. It is nice to know that an interface can be used for output and input at the same time with Camilla DSP. I have a Focusrite Scarlett 18i20 I could use for this and an RPi3 B+ that might have enough processing power. Hopefully the ASR Streamer project continues in the direction I last saw which was making moode with Camilla easier to install and get running with maybe some added functionality.

I also have a miniSHARC with a waveflex screen. Currently it is setup as a 2x2 with the digi board but originally it had the DA8 dac board installed. Would be nice to set it up with 8 digital outputs.

I thought about this some more and realized I can simplify my setup with software DSP while retaining the same input functionality and hopefully eliminating the channel timing differences on the Okto (need to run some tests to confirm the timing differences are solved).

I put the Okto in USB / AES mode which is designed to take AES inputs and route them to a USB host for processing (in my case a RPi running CamillaDSP) and then do the D/A conversion on the processed signal from the USB host. I am still using the miniSHARC + DIGI-FP as source selection and for ASRC to 96 kHz (which is what I am running CamillaDSP at) but now only use the channel 1-2 AES output from the DIGI-FP to the Okto, all processing is handled by the RPi.

This completely eliminates the need for all of the hoops I had to jump through to get the 8 channels of AES output (McFIFO/McDualXO, WM8805 I2S to AES output boards) and means I can use the much smaller OpenDRC-DI box instead of the larger custom enclosure. This means that I now have more processing power than before and the overall cost is significantly less (obviously sunk cost in my case but hopefully others can learn from my mistakes!). As I mentioned the miniSHARC with 8 channels of AES output cost upwards of $800 to build, the used OpenDRC-DI + RPi4 is less than $400 total.

Michael
 

Dacapalooza

Active Member
Joined
Feb 19, 2020
Messages
121
Likes
43
Awesome this thread exists. Was wondering the same thing. Still have questions.

I actually have Okto8. But since Okto bailed I need new plan. I learned to always have Plan B ready at moment's notice. I always regretted my fire buys when something breaks. And the Okto never shuts off... I want a well researched Plan B.
My system: Oppo 203 -> Vanity -> Trinnov Demon -> Okto8 -> Benchmarks -> Revels

Crazy that even the same DAC can have syncing issues. We don't test syncing amongst the channels here? I was so happy with my dac8....

Funny no one mentioned a clock as a possible resolution to syncing. Which prompted my research. Vanity Pro was just introduced and it has clock out.
Trinnov has clock. So I can clock my entire DAW if I needed to replace my DAC.
If 3 DAC's & entire DAW were using same clock would all syncing issues be resolved?

...as long as the latency difference was a fixed...
Are the channel to channel latencies fixed? Every run? To the end of time? It seems not.
compensate in DSP
Would Room Correction fix the latency/syncing issues automatically? Would not the RC think a delay is a distance thing? And if there is a delay the calibration app's speaker distances would be incorrect? Does RC work in ms? Or better: fractions of ms? Bottom line: Are aforementioned delay amongst my Okto8 channels being compensated for in Trinnov or other RC apps? And if latencies keep changing perhaps monthly calibration is better. I currently calibrate every 6 months.

If I understand the other posts. Some DAC's reclock. Avoid those. Might be BIG limitation. Now I need a DAC that has AES, clock input (if that helps as asked above) and no reclocking. Do most DAC's reclock? Is reclocking guaranteed a latency? So what can 3 reclockings sum to?

Would I not want to avoid reclocking DAC's anyway? I am already reclocked. Twice I think? The Vanity reclocks. Trinnov has FireWire I/O in all their devices. When converting FireWire back to AES, I get the impression that reclocking would be required. Or can somehow Trinnov Convert FireWire to AES using the input signal's clock even though my Trinnov Latency is a whopping 50ms? (Don't worry - My PJ is 40ms. But acoustic delay is 10ms. So I am 20ms off from video. So in 24fps, I am not even 1/2 frame off.)

In the big picture our speakers are not all equal distances so I would think RC, or even mere distance configs, would fall out of sync way more in Speakers than a sync issues within the same DAC or even other DAC's? Since the starting point is way more out of sync in comparison.

Lastly, Is latency the ONLY issue between DAC channels or amongst other DAC's? For example I always see SINBAD is different amongst the channels...
 
Last edited:

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
If 3 DAC's & entire DAW were using same clock would all syncing issues be resolved?

I have not used a clock to sync but it seems like this would work, although I am not sure if you would still have issues when using different DACs if they had different inherent latencies. I imagine your Trinnov acts exactly like the nanodigi mentioned above in that all your digital outputs will be synced with each other.

Are the channel to channel latencies fixed? Every run? To the end of time? It seems not.

I believe you are referring to the Okto here, at least on time scales in the week(s) range the delay differences between channels does not change as long as you do not do the things I mentioned in post #18. The Okto is the only gear I have experienced these non-deterministic delays with. For example he MOTU Ultralite Mk5 has a one sample delay between channels 2, 4, 6, 8, 10 and channels 1, 3, 5, 7, 9 but this is constant and never changes.

Would Room Correction fix the latency/syncing issues automatically? Would not the RC think a delay is a distance thing? And if there is a delay the calibration app's speaker distances would be incorrect? Does RC work in ms? Or better: fractions of ms? Bottom line: Are aforementioned delay amongst my Okto8 channels being compensated for in Trinnov or other RC apps?

The delay on the Okto only matters for DIY active speakers where you are trying to phase match between drivers. Even then an offset on the order of 0.01 ms is probably not important. If you do not have DIY active speakers it won't matter and such a delay is so small that mic positioning will have more variation.

if I understand the other posts. Some DAC's reclock. Avoid those. Might be BIG limitation. Now I need a DAC that has AES, clock input (if that helps as asked above) and no reclocking. Do most DAC's reclock? Is reclocking guaranteed a latency? So what can 3 reclockings sum to?

If by re-clocking you are referring to a ASRC in a DAC I would not worry about it, as demonstrated above there is no issue syncing two DACs with ASRCs as long as they are fed sync'd signals to start with. If you are talking about something like a FIFO designed for jitter suppression then yes you can have issues with syncing but these are not common.

Michael
 
Top Bottom