• 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!

Review and Measurements of Okto DAC8 8Ch DAC & Amp

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
Oh that looks really concerning

Does anyone know whether it's been fixed?

What is your application? It really isn't that concerning for many applications.

If you aren't using multiple AES inputs -> not an issue
If you aren't implementing DIY active speakers -> not an issue

If your application does fall in to criteria mentioned above (say using a DDRC-88D to implement DIY active speakers) you can use the Okto to measure the offset between channels and adjust the timing in the upstream DSP. The only issue with this is you need to understand that there are certain activities (mentioned in my original post about the timing issues) that can cause the delays to change and you will then need to re-measure / re-correct.

EDIT: And I should add that using a DDRC-88D as a DSP for DIY active speakers with the Okto is not the best idea. Much better to use a RPi4 running CamillaDSP with the Okto in USB / AES mode. It is much more powerful, significantly less expensive, will have no timing issue and multichannel Dirac correction is not helpful for DIY active speakers.

Michael
 
Last edited:

Verausci

Member
Joined
Dec 20, 2019
Messages
34
Likes
6
What is your application? It really isn't that concerning for many applications.

If you aren't using multiple AES inputs -> not an issue
If you aren't implementing DIY active speakers -> not an issue

If your application does fall in to criteria mentioned above (say using a DDRC-88D to implement DIY active speakers) you can use the Okto to measure the offset between channels and adjust the timing in the upstream DSP. The only issue with this is you need to understand that there are certain activities (mentioned in my original post about the timing issues) that can cause the delays to change and you will then need to re-measure / re-correct.

EDIT: And I should add that using a DDRC-88D as a DSP for DIY active speakers with the Okto is not the best idea. Much better to use a RPi4 running CamillaDSP with the Okto in USB / AES mode. It is much more powerful, significantly less expensive, will have no timing issue and multichannel Dirac correction is not helpful for DIY active speakers.

Michael
I mean one thing is that I don't think I'd like to spend that much on something for it to have faults

Interesting about the DDRC-88D though as I was considering that for a DSP multi-channel setup from my PC.
What makes you say Dirac is not helpful for DIY active speakers? (My speakers aren't DIY but are active)
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
I mean one thing is that I don't think I'd like to spend that much on something for it to have faults

Interesting about the DDRC-88D though as I was considering that for a DSP multi-channel setup from my PC.
What makes you say Dirac is not helpful for DIY active speakers? (My speakers aren't DIY but are active)

If you are using a PC as source I would just use the PC for DSP. Looking at the specs of the DDRC-88D I am not even sure it can accept a USB audio input.

A multichannel Dirac device like the DDRC-88D is meant to correct complete speakers, not individual drivers in a DIY active speaker where you also implement x-overs / EQ in DSP. I guess you could use it to individually correct drivers but that seems like a bad idea as you still need them to sum appropriately at the x-over and I think manual correction is much better suited for that.

The reason why the channel delay only matters in DIY active speakers is again because you are implementing an x-over in DSP and the delay will impact phase matching at the x-over. Looking at the phase responses such a small delay will really only matter for the mid / tweeter x-over. To give you a feel a 0.01 ms relatively delay is equivalent to moving two speakers about 0.125" apart and 0.02 ms is equivalent to moving two speakers about 0.250" apart. This will matter for a DIY mid / tweeter crossover but certainly not matter if your completed speakers (either active or passive) are on their own distinct channel.

Michael
 

Verausci

Member
Joined
Dec 20, 2019
Messages
34
Likes
6
The reason why the channel delay only matters in DIY active speakers is again because you are implementing an x-over in DSP and the delay will impact phase matching at the x-over. Looking at the phase responses such a small delay will really only matter for the mid / tweeter x-over. To give you a feel a 0.01 ms relatively delay is equivalent to moving two speakers about 0.125" apart and 0.02 ms is equivalent to moving two speakers about 0.250" apart. This will matter for a DIY mid / tweeter crossover but certainly not matter if your completed speakers (either active or passive) are on their own distinct channel.
Oh do you mean in terms of actually sending a pre-crossed-over signal directly to the individual drivers instead of the crossover being done on signal arrival to the speaker unit as a whole?
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
Oh do you mean in terms of actually sending a pre-crossed-over signal directly to the individual drivers instead of the crossover being done on signal arrival to the speaker unit as a whole?

Yes, crossover done upstream of the Okto which outputs filtered signals to individual amplifiers for each driver.

Michael
 

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,823
Likes
2,951
Location
Ichihara City, Chiba Prefecture, Japan

phoenixdogfan

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
3,297
Likes
5,079
Location
Nashville
Oh that looks really concerning

Does anyone know whether it's been fixed?
He's talking about a 1-2 sample delay between (among ) channels when receiving signal via AES/EBU inputs. How significant can this be when at a minimum it's running 44,100 samples per second? That means the channels are out of sync by at most 1/44,100 of a second! That's 1/44th of a millisecond. Wouldn't having each speaker even one millimeter varying from being equidistant to the listener have more effect?

Maybe I'm missing something here, but I fail to see how this temporal discontinuity would ever be audible.
 

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,823
Likes
2,951
Location
Ichihara City, Chiba Prefecture, Japan
He's talking about a 1-2 sample delay between (among ) channels when receiving signal via AES/EBU inputs. How significant can this be when at a minimum it's running 44,100 samples per second? That means the channels are out of sync by at most 1/44,100 of a second! That's 1/44th of a millisecond. Wouldn't having each speaker even one millimeter varying from being equidistant to the listener have more effect?

Maybe I'm missing something here, but I fail to see how this temporal discontinuity would ever be audible.

OK, understood. For the AES/EBU multichannel input, I am sorry that I have no info nor experience/measurement as I have been always sticking to USB ASIO multichannel input into DAC8PRO in my project.

I basically agree with your thought that the possible "1/44th of a millisecond delay" between the AES/EBU inputs could be ignored. Otherwise, digital signal processing running 44.1 kHz would be useless...
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
He's talking about a 1-2 sample delay between (among ) channels when receiving signal via AES/EBU inputs. How significant can this be when at a minimum it's running 44,100 samples per second? That means the channels are out of sync by at most 1/44,100 of a second! That's 1/44th of a millisecond. Wouldn't having each speaker even one millimeter varying from being equidistant to the listener have more effect?

Maybe I'm missing something here, but I fail to see how this temporal discontinuity would ever be audible.

I’m not sure the point of this post. When I first found the issue I quantified it. You can see the result on phase response for 48 kHz and 96 kHz in the plot below.

miniSHARC Okto Phase Overlays.png


I don’t know about you but if I was crossing over a mid to a tweeter at 3 kHz I wouldn’t appreciate a 20 deg variable phase mismatch. This will without a doubt impact combined frequency response at the x-over.

I also quantified what the impact was converting to an equivalent distance.

To give you a feel a 0.01 ms relatively delay is equivalent to moving two speakers about 0.125" apart and 0.02 ms is equivalent to moving two speakers about 0.250" apart. This will matter for a DIY mid / tweeter crossover but certainly not matter if your completed speakers (either active or passive) are on their own distinct channel.

Again, for a mid / tweeter an extra 0.250” spacing between the two is significant and will impact frequency response. For full range completed speakers such a distance is very insignificant and just moving your head will have more impact.

Michael
 
Last edited:

phoenixdogfan

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
3,297
Likes
5,079
Location
Nashville
I’m not sure the point of this post. When I first found the issue I quantified it. You can see the result on phase response for 48 kHz and 96 kHz in the plot below.

View attachment 209356

I don’t know about you but if I was crossing over a mid to a tweeter at 3 kHz I wouldn’t appreciate a 20 deg variable phase mismatch. This will without a doubt impact combined frequency response at the x-over.

I also quantified what the impact was converting to an equivalent distance.



Again, for a mid / tweeter an extra 0.250” spacing between the two is significant and will impact frequency response. For full range completed speakers such a distance is very insignificant and just moving your head will have more impact.

Michael
Ok, so we're saying if you're using the Octo to provide a DAC for each channel of a 7.1 setup via AES/EBU the effect will be negligible, but if we're using a digital crossover and somehow using AES/EBU input to the OCTO for each driver, we could be introducing audible phase shift. But, would it not be more likely that if you're doing a digital crossover, you'd be doing it in your computer (JRiver, Ekio, DePhonica) and passing the LPCM digital via usb to the Octo. I know that's how I do my sub crossover? I suppose there might be a use case with maybe one of the miniDSP products where it takes a USB input, does PEQ and Crossover, and passes AES/EBU to the OCTO, but that's possibly the only case I can think of. And if that miniDSP component has Dirac Live, would not that allow it to compensate for that effect?
 
Last edited:

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
Ok, so we're saying if you're using the Octo to provide a DAC for each channel of a 7.1 setup via AES/EBU the effect will be negligible, but if we're using a digital crossover and somehow using AES/EBU input to the OCTO for each driver, we could be introducing audible phase shift. But, would it not be more likely that if you're doing a digital crossover, you'd be doing it in your computer (JRiver, Ekio, DePhonica) and passing the LPCM digital via usb to the Octo. I know that's how I do my sub crossover? I suppose there might be a use case with maybe one of the miniDSP products where it takes a USB input, does PEQ and Crossover, and passes AES/EBU to the OCTO, but that's possibly the only case I can think of. And if that miniDSP component has Dirac Live, would that allow it to compensate for that effect?

Yes, for a 7.1 system it will have negligible effect. If using an AES output DSP (like the DDRC-88D) to implement active speaker crossovers you can have issues with higher frequency crossovers due to phase mismatch. To give a feel here is a 3000 Hz LR2 x-over with and without a 0.02 ms delay on one of the drivers. To be honest I hadn't modeled this before and it was not as bad as I expected. About a 0.5 dB dip although it is over a rather large region.

1653704582977.png


I agree that software DSP is the way to go with the Okto and I mentioned this in my original response to @Verausci. He of course was talking about using the AES outputs of DDRC-88D to feed the Okto which is why he was asking about the delay (it was originally unclear whether he was implementing active speakers or a 7.1 system).

As I mentioned in my original response to @Verausci I don't think using a multichannel Dirac device to correct individual drivers is a particularly good idea, much better to have manual control of the corrections when tuning an active crossover. You can exactly compensate for the delay in DSP but the problem is that the delay is not consistent. If you power cycle the DSP or switch the Okto mode or power cycle Okto the delay will change and your correction will be wrong.

Michael
 

DWPress

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2018
Messages
966
Likes
1,398
Location
MI
@mdsimon2 - just to be clear: Does this issue only occur when using 1 or more of the AES inputs or is there an issue with multi channel USB input as well?

I used a miniDSP 4x10 for over a decade to manage my active XO 8ch system but now do everything it did (and more) on the computer. All the miniDSP are good devices for a stand alone solutions but just lack horsepower to be truly great. I agree that MC Dirac would not be appropriate for an active XO in the use case sited above. Software Dirac Live can work on a traditional active XO speaker (2 way, 3 way design + sub(s)) but strictly as a 2 channel filter for L&R.
 

phoenixdogfan

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
3,297
Likes
5,079
Location
Nashville
@mdsimon2 - just to be clear: Does this issue only occur when using 1 or more of the AES inputs or is there an issue with multi channel USB input as well?

I used a miniDSP 4x10 for over a decade to manage my active XO 8ch system but now do everything it did (and more) on the computer. All the miniDSP are good devices for a stand alone solutions but just lack horsepower to be truly great. I agree that MC Dirac would not be appropriate for an active XO in the use case sited above. Software Dirac Live can work on a traditional active XO speaker (2 way, 3 way design + sub(s)) but strictly as a 2 channel filter for L&R.
When Dirac introduces DLBC it will be able to treat each unit in a 2.2 system as a separate speaker, but right now, I run my DL 3 on the PC so as to treat my LS 50's + SB 2000 as just a 2.0 speaker system.
 
Last edited:

DWPress

Addicted to Fun and Learning
Forum Donor
Joined
May 30, 2018
Messages
966
Likes
1,398
Location
MI
I don't use Dirac but host filters the same way via a 2ch reverb plugin which contains correction for all speakers + subs, all my hosting and processing done on Mac. Yes, it would be nice if DIBC were to be released in software for us to play with, not likely to happen anytime soon though.

I've thought about using FIR filters exclusively for individual driver correction instead of IIR but haven't taken the time to follow through with the experiment and it's convenient to be able to quickly adjust a PEQ.

Curious that the source is still (for me) 2 channels and that in the end, no matter how we hi/low/band-pass signals, adjust the frequency/phase/magnitude or otherwise manipulate or distribute the channels a final 2 ch convolution filter for 20-20kHz is often what it boils down to for most of us for simplicities sake.

How's that multi channel convolver app coming along @mitchco? ;)
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
@mdsimon2 - just to be clear: Does this issue only occur when using 1 or more of the AES inputs or is there an issue with multi channel USB input as well?

I used a miniDSP 4x10 for over a decade to manage my active XO 8ch system but now do everything it did (and more) on the computer. All the miniDSP are good devices for a stand alone solutions but just lack horsepower to be truly great. I agree that MC Dirac would not be appropriate for an active XO in the use case sited above. Software Dirac Live can work on a traditional active XO speaker (2 way, 3 way design + sub(s)) but strictly as a 2 channel filter for L&R.

It only happens with more than 2 channels of AES input. It does NOT happen with multichannel USB input (Pure USB mode) or 2 channel AES input with software DSP to 8 channel output (AES / USB mode).

For the record such timing issues seem relatively common in multichannel interfaces. For example the MOTU Ultralite Mk5 has a 1 sample delay on half the output channels when using TOSLINK input, however the delay is consistent and does not change. The issue with the Okto is the delay is inconsistent and does change which makes correcting it more difficult, especially if you change operating modes on the Okto.

Michael
 

Ozymandias

Member
Joined
Jun 27, 2019
Messages
23
Likes
11
It only happens with more than 2 channels of AES input. It does NOT happen with multichannel USB input (Pure USB mode) or 2 channel AES input with software DSP to 8 channel output (AES / USB mode).

For the record such timing issues seem relatively common in multichannel interfaces. For example the MOTU Ultralite Mk5 has a 1 sample delay on half the output channels when using TOSLINK input, however the delay is consistent and does not change. The issue with the Okto is the delay is inconsistent and does change which makes correcting it more difficult, especially if you change operating modes on the Okto.

Michael
Thanks for the clear and detailed explanation of the problem @mdsimon2 ! Is @PavelV aware of this issue? So far I’ve only been using the pure usb mode of my DAC8 but I’m planning to get the VanityPro and use that to connect my Blu-ray player to my DAC8 via the AES/EBU inputs so I’m interested in understanding this better. I agree with your previous assertion that this 1-2 sample delay will be inaudible in regular use cases like mine but I just want to understand what part of the DAC8 implementation is causing this. Is there any chance the delay was introduced by the source device feeding your DAC8?
 
Last edited:

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,316
Location
Detroit, MI
Thanks for the clear and detailed explanation of the problem @mdsimon2 ! Is @PavelV aware of this issue? So far I’ve only been using the pure usb mode of my DAC8 but I’m planning to get the VanityPro and use that to connect my Blu-ray player to my DAC8 via the AES/EBU inputs so I’m interested in understanding this better. I agree with your previous assertion that this 1-2 sample delay will be inaudible in regular use cases like mine but I just want to understand what part of the DAC8 implementation is causing this. Is there any chance the delay was introduced by the source device feeding your DAC8?

As described in my original post -> https://www.audiosciencereview.com/forum/index.php?threads/okto-8-owner’s-thread.13899/post-659994 I tried multiple DSPs and all had the same issue with the Okto.

In addition sometimes the delays would occur between channels on the same AES input (i.e. channel 3 would be delayed 1-2 samples from channel 4) and I never observed this behavior when measuring the digital output of the DSPs directly, only when they were used with the Okto.

I notified Pavel when I first discovered the issue in early 2021 but received no response.

Michael
 

Snafu

Active Member
Joined
Dec 27, 2018
Messages
223
Likes
140
What's the availability of stereo dac (sold out at the moment, i see) in near future ?
 

Mario Sanchez

Active Member
Joined
Jan 11, 2021
Messages
161
Likes
276
Rumor has it that topping is dropping a 8 channel D/A Converter come July. No further info yet other than it's going to be USB input only, and 6.35mm TRS outputs a la E50 and D10B, in a 90-series enclosure.
A challenger rises? Okto's supply/production seems to struggle in face of their product's popularity, it would be intriguing to see if topping could build something to rival this beast with better availability.
Source: A friend ;)
 
Last edited:
Top Bottom