• 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

For some reason I too decided to update from 1.5 to 1.6. I found that with 1.5 and DoP in Roon, there was a pop when switching formats between DSD and flac. No pops so far in 1.6.
 
Ok, tried again with AES/EBU output and still no front knob volume control. I have latest firmware, tried usb and AES/USB input. Only using standard XLR cables, not 110ohm, but no digital dropouts etc.
I did control digital volume with GLM - to be honest not convinced I can hear any difference over analogue connection to my Genelecs.
My unit was an early one #31. Wonder if there is a hardware difference?
 
AudioJester said:
Ok, tried again with AES/EBU output and still no front knob volume control. I have latest firmware, tried usb and AES/USB input. Only using standard XLR cables, not 110ohm, but no digital dropouts etc.
I did control digital volume with GLM - to be honest not convinced I can hear any difference over analogue connection to my Genelecs.
My unit was an early one #31. Wonder if there is a hardware difference?
Click to expand...

If you run “xmosusb --dacstatus" what are the results? See here to get the xmosusb tool -> https://github.com/fabriceo/AVDSP_DAC8/tree/main/usbtools.

Michael
 
AudioJester said:
Ok, tried again with AES/EBU output and still no front knob volume control. I have latest firmware, tried usb and AES/USB input. Only using standard XLR cables, not 110ohm, but no digital dropouts etc.
I did control digital volume with GLM - to be honest not convinced I can hear any difference over analogue connection to my Genelecs.
My unit was an early one #31. Wonder if there is a hardware difference?
Click to expand...
I think I remember that the early version was using the DAC0/1 i2s wire and not a separated one as now ... I ll check with Pavel but I m afraid that's the point. There is always a possibility to reroute the AK4118 input so it goes on the new i2s line (the one used for DSD7 in fact) but that requires a very fine rework.
 
amirm said:
This is a review and detailed measurements of the Okto Research DAC8 8-channel DAC and headphone amplifier. I was sent a prototype unit for measurements. From my understanding, the company has been producing the internal DAC module for a while for the DIY market. But now are building complete systems. So what makes this an early production unit is not the DAC module but the integrated version with input and other features. The version I have has the Okto DAC8 module plus an off-the-shelf UBS Input. My understanding is that the final units in production in April will also have their own USB input implementation plus 4xAES/EBU. It will cost 989 EUR in this configuration (about $1,120 as of this writing). This version will be known as DAC8 Pro. I will just be calling it DAC8 from here on.

There will be two other versions, one with DSP and the other in stereo configuration where four channels are folded into one to get better performance. See: https://www.oktoresearch.com/index.htm

Due to large number of outputs, the unit comes in "rack mount" 1U size (at least I think it complies with that spec). But through out any impression of dull, industrial look out of the window. The DAC8 employs two beautiful and highly legible OLED displays flanking its rotary control, giving it an exquisite look:


From what I have read, they have done the mechanical design (and production?) themselves too. The metallic part of the front panel looks a bit cold to me so if there is an option in black would be nice. Any misgivings on that part goes away the moment you interact with the unit. Display changes are instantaneous, giving you a very solid feel that you are making changes. I love all the key information nicely visible including the sample rate, input, volume, etc. Love it!

I had no trouble navigating the unit with the rotary control. The DAC8 though is able to learn the remote control of your choice or you can get it with an Apple Remote.

One minor thing is the top: it has screws on each side but none in the middle. This allows the top or flex a bit. I suggest putting a single additional screw in the back middle to remedy this.

The back panel on my unit is not fully populated per intro:


I am too lazy to usually show you all the menus but I thought I make an exception here. :) Pushing the rotary control changes the left side to:

View attachment 23762

A second click changes the right side to this for volume. The menu will scroll down to let you lower the levels in all 8 channels:
View attachment 23763

I turned on the 18 dB boost and it produced severe clipping at anything above -19 dB. I am not sure of its use other than for digital input that is at very low level. I hope @Okto Research can explain what they think this is useful for.

System menu is such:
View attachment 23764

Display shows the usual stuff:
View attachment 23765

And finally the ESS Sabre DAC settings which I left alone:
View attachment 23766

Okto Research advertises superlative specs for this product -- numbers which we have not seen in our reviews. Is it the usual marketing games or real? Let's find out.

DAC Measurements
As usual, we start with our dashboard view:

View attachment 23767

Wow! What a way to impress. Distortion+noise of just 0.000127%? And with such excellent channel matching? This places the SINAD at 118 which is a new record:

View attachment 23785

Yes, a new SINAD king has been crowned!

In honesty though, there is an unfair advantage here. Can you spot it? The output voltage is 3.3 volt, not the usual 4 volt we like to see. Okto says they can program this to any level one wants. Usually distortion sets in when levels are increased so I expect some performance will be lost if this were boosted to 4 volts. Fortunately in use you are not going to need more voltage since you have a volume control in DAC8 so will be driving an amp below these levels anyway.

As if to deliver a one-two punch, the DAC8 delivers exemplary stable and noise-free linearity results sailing to -120 dB (20 bits of resolution):

View attachment 23769

Dynamic range pushes the limits of my analyzer so likely the actual DNR is better than reported:

View attachment 23770

With threshold of hearing being -116 dB, this is one silent DAC!

Jitter shows tiniest imperfection made more visible by very low noise floor:

View attachment 23771

I had to remove audiosciencereview.com sticker at the bottom of the graph since the noise floor slid under it!

DAC8 is using ESS DAC chip so for sure we will see our famous "ESS THD IMD," yes?

View attachment 23772

There is a tiniest hint of that -- maybe -- around -32 dB or so. Or then again it could just be the variations we see throughout the graph. Whatever, it bests my reference DX3 Pro DAC I use for this test.

Distortion and noise relative to frequency shows flat response which indicates plenty of bandwidth/gain:
View attachment 23773

With some 30 dB lower distortion at 20 kHz, you do get something for paying more than $99 for your DAC (read: Khadas Tone Board).

For multi-tone test, I switched to a new method as the one I was using was too finicky and would generate garbage results if the delay through the DAC was not what it was expecting. So don't quite compare this to previous measurements:

View attachment 23774

The accumulation of distortion products stops around -150 dB. Our tones are around -20 dB meaning there is almost 130 dB of distance between the peak and distortion products!

Folks, we have to stop here. I can't find any holes in this design. Surely we find a fertile in the headphone amp though. It has to be a throw-away feature made to fill the spec sheet....

Headphone Amplifier Measurements
Let's jump right into our power versus distortion+noise:

View attachment 23775

From distortion and noise point of view, the DAC8 easy bests my DX3 Pro reference even in its low-gain mode. Alas, power is highly limited due to maximum of 1.6 volt output. So we don't even make it to 10 milliwatts. Those of you who have the DX3 Pro and operate it in low gain mode would have almost the same amount of power here.

Switching to 33 Ohm load gives us more power as expected but still not as much as we like:

View attachment 23776

Again, power output is near what DX3 Pro produces in low gain mode.

No, the 18 dB boost doesn't help here. Again, it just makes the clipping point -18 dB on the volume control.

Output impedance is excellent at 1 ohm which is close to limit of how low I can measure with my fixture:
View attachment 23777

Headphone Listening Tests
I plugged in my HD-650 into DAC8 and was greeted by what you see in the measurements: there is no hint of distortion to max 0 dB volume level. Due to limited power, there is no sensation of "subwoofer in your ear" as I like to see. :D What was there though is likely sufficient for many though. I was able to listen at levels between -9 to 0 dB.

The situation was a bit better with Hifiman HE-400i. A very pleasant experience but no chance of rattling off the screws on the drivers.

If you are interested in this product for headphone listening, I would start with what is there and see if it is enough. If not, then a pairing with Massdrop THX AAA 789 would be in order.

Conclusions
Let me confess that I had a few down moments prior to testing the Okto DAC8. There is such a large backlog of gear to review and it seems I am jumping from one measurement to another. Nothing cured that better though than putting the Okto Research's DAC8 on the bench. The DAC8 Pro is superbly engineered and delivers on objective measurements like there is no tomorrow. Heck, I got tired of putting exclamation marks in my review above. :) Could we have predicted this? To some extent yes: despite being a small company, they used the major competitor to my Audio Precision analyzer from Rohde and Schwarz. Time and time again we have seen that companies that measure their designs, product superb products. You have to verify what you build folks.

And this is not just some 2-channel DAC. You have 8 channels of output that should satisfy anyone who needs that many channels now or in the future. It is remarkable that Okto achieved such excellent results without combining four or 8 channels together as other companies do.

Then we have highly responsive, well design and clear displays. And excellent connectivity in the final version with more inputs to let you tailor this with your favorite DSP upstream.

The only miss is lack of infinite amount of power in its headphone stage for anyone wanting to use it that way. But what is there is high performance and distortion and noise-free.

As you can surmise, the Okto Research DAC8 is an incredibly well designed and featured product. It is my absolute pleasure to recommend the DAC8.

I am also so happy to find a device designed and manufactured in Europe (Prague). I always feel bad giving high marks to US built devices that wind up being very expensive to import into EU.

------------
As always, questions, comments, recommendations, etc. are welcome.

They say blueberries are good for you. We have a few plants but need to buy some more for the fruit to last us past summer season. So please be generous with your donations using:
Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
Click to expand...
I’m trying to connect mine to the net with direct wiring., are there any things to watch out for?
 
Peterbean said:
I’m trying to connect mine to the net with direct wiring., are there any things to watch out for?
Click to expand...
Given the simple fact that the Okto Research DAC8 Pro does not have any network interface of any kind whatsoever, your question leaves me confused, to say the least.

Also, the full quote of AmirM's review appears to be totally pointless. Please edit you post so that readers can have a rough understanding which products exactly you want to connect to what and what you actually expect to achieve in doing so.
 
Latency and output level?

Sorry if this has been covered in this massive thread, I could not find it.

I did some measurements on my Okto 8ch with the latest FW (1.6). I wanted to check latency and adjust level to the rest of the system since I am using other DAC's in parallell.

The Okto was set to Attenuation "0" (volume on full), Pure AES in, all channels level, SRLP filter (the others ring a lot).

I used a BSS DSP unit with one analogue output card to compare and one digital output card feeding the Okto as source. Input signal was a controlled analogue MLS signal via the a BSS analogue input card and the identical signal was fed both the analogue and digital output cards so level and latency should be identical. When measuring I used the Okto analogue out as reference.

Feeding the Okto with digital AES (48kHz / 24 bit) signal on AES input 1&2 (same signal) the output on the Okto analogue out 1&2 was 5.8 db lower than the input signal, when compared the analogue out signal direct from the BSS analogue card. That sounds like a lot?

Anybody else measured the output of the Okto compared to input?

I also tried to measure the latency of the Okto and used the Okto analogue signal as a reference and looked at how much I needed to delay the direct signal to have a timing match between them and came to that the latency in the Okto is 2.1250 ms . That also sound high.

The latency in both the BSS output cards (analogue and digital) is in the femto-range, so it should be neglectable when comparing timing.

Thoughts??

Kind regards
//Rob
 
Wish I could verify or weigh in on that; I don't own one yet, I've put an order in for one this week however.

I wonder if your finding around the 5.8dB difference would apply to models of the DAC8 PRO whose max output voltage is modified.

Likewise, when you say you observed 2.1250ms difference between analogue outputs 1 & 2 - I'm assuming that's when fed via AES ? Does it behave the same when in pure USB mode? I suppose in USB mode a user might have more freedom to work around this on the software side.

I was exploring an RME HDSPe AES which I'm sure might provide the tools to tweak around this. (That all depends how well that kit would play with pipewire in Linux though, I don't want to be landlocked to Windows or Mac, may open a thread about this).

@mdsimon2 did make some references to the clocking of AES stream earlier in the thread. Would your findings indicate inadequacies in the local oscillator?
mdsimon2 said:
They have already stated they are against a word clock input:

Q: Why doesn't the dac8 PRO have a word clock input?
A: When processing AES/EBU inputs, the clock is encoded with the signal. This means that even if multiple DACs are used, they will always operate synchronously with the source device. Additionally, the clock recovered by the dac8 PRO from the AES/EBU stream has lower jitter compared to the word clock received from a standalone unit, which is then internally multiplied upon reception to match the frequency used by the DAC chip. You can find online measurements comparing these two methods using competitor DACs. These comparisons typically show superior analog performance using the clock recovered from AES/EBU signal.

The other thing that goes unsaid in the FAQ is that although the XMOS uses the AES recovered clock, the DAC chip itself always uses a local oscillator as described here -> https://www.audiosciencereview.com/...nts-of-okto-dac8-8ch-dac-amp.7064/post-333521.

I don't see how SPDIF/AES are "far behind" Dante/Ravenna, other than that they require physical cables. Seems to me like you are searching for a solution to a problem that does not exist.

The biggest unknown in all of this is how connecting two Oktos to a single PC will behave. Can they use the same driver? Do they need to use different drivers? Even if they are clocked the same does the PC data start at the same time or is one DAC offset from the other? Pavel described these issues in a reply to you three years ago -> https://www.audiosciencereview.com/...nts-of-okto-dac8-8ch-dac-amp.7064/post-450024.

Michael
Click to expand...
 
sebackman said:
Feeding the Okto with digital AES (48kHz / 24 bit) signal on AES input 1&2 (same signal) the output on the Okto analogue out 1&2 was 5.8 db lower than the input signal, when compared the analogue out signal direct from the BSS analogue card. That sounds like a lot?
Click to expand...

All I would take from this is the BSS has a max output level that is 5.8 dB higher than the Okto, which means it is around 8 V, certainly not uncommon for pro interfaces. The only way this is an issue is if you think the output voltage of the Okto isn’t meeting the spec’d output voltage (default is 4.2 V).

sebackman said:
I also tried to measure the latency of the Okto and used the Okto analogue signal as a reference and looked at how much I needed to delay the direct signal to have a timing match between them and came to that the latency in the Okto is 2.1250 ms . That also sound high.
Click to expand...

That doesn’t seem off spec and is not objectively very high. Okto quotes as low as 1.5 ms playback latency which I am sure depends on sample rate and filter. I am out of town but I can make some latency measurements when I get home with the different filters.

Just my 2 cents but I always like using a digital loopback as a timing reference as it gives perfectly flat magnitude and phase response, you still will have some additional latency from the ADC on the channel from the Okto so it is definitely a worst case measurement. I know you say you see there is no difference in latency between analog and digital output on the BSS, but that seems highly unusual as the DAC and ADC will always add some latency, the only way I see this being true is if the digital output has some added delay to it.

Michael
 
Hi There
the groupe delay of the SRLP filter at 48khz is 8.75/48000 =182 micro seconds
the treatment of one sample reception goes via the SPDIF decoder, then the xmos serial register, then it is sent to the usb stack and comes back and sent to the DAC via another xmos serial register and then the DAC needs to deserialize it . so let say 6 samples lead-time = 120us. so 302us in total.
if you measure 2.125ms, the delta should be due to the delay in the ESS DPLL which is always activated in AES mode. I think you can change the DPLL length in one menu. try to reduce it but there is a minimum level to get it locked properly.
For the 5.8db difference this is a mater of defining what 0dbFS means (unless your dac is not delivering 4.2vrms of course)

best regards
fabriceo
 
Thanks for input.

Sound like measured latency sounds reasonable with AES in som that should be fine then. Just expected lower. The latency is not really a problem as I can introduce a delay on the other DAC's which are all the same mark/model, so that should be just fine in the DSP. I prefer using the same DAC's in one system but here I use 16-chnnels but only 6 (fronts) or possible 9 (including the center) are critical (DSP based active 3-way 5.2 system)

Regarding level I just found it to be a high number, 5,8db. Maybe that is to avoid digital clip in all scenarios. Its just a relative number so when knowing what it is it can be compensated. The best way would be to send a digital AES signal at 0dbFS, as suggested by member fabriceo, and see what comes out.

I don't have good sound card with AES out, so If somebody could do that it would be great.

Kind regards
//Rob
 
mdsimon2 said:
I am out of town but I can make some latency measurements when I get home with the different filters.
Click to expand...

Here is what I get with default DPLL BW AES of 8, PureAES mode, 48 kHz, using a RME Fireface 800 as source and ADC with a SPDIF loopback as timing reference.

FRLP: 1.77 ms
SRLP: 1.23 ms
FRMP: 1.16 ms
SRMP: 1.12 ms
AFRLP: 1.77 ms
HFRMP: 1.43 ms
BW: 1.77 ms

And then I tried adjusting with the DPLL bandwidth with SRLP, for everything other than 0 which gave no lock I got 1.23 ms. There were some slight changes to the digits after 1.23, but nothing significant.

Obviously these results still have the variable of the ADC latency, so should be worst case.

Michael
 
Last edited:
Great Michael. Thank you.

Just for me to understand as I have never used a RME unit. You mention an ADC, is that a separate box or the ADC in the RME? Is the SPDIF loop-back internal in the RME or did you connect SPDIF out to in, by cable on the RME. Or was the loop-back done in a separate box? If so, what was that?

I understand that SPDIF and AES are in principle the same save for the voltage level and one "control bit". The signal fed to the Okto was that SPDIF or AES and if SPDIF from the RME was is the same signal as the loop-back?

Sorry for all the questions but since I'm mixing two different DAC's in the same system understanding why I got +2ms and you got 1,23ms is important in my system. I need to delay my other DAC's to compensate for the Okto latecy. And gain.

Can you please check levels also, ie what is the level out, at 100% volume, compared to input. Just a compare measure would be great. My finding of +5db sounds high.

Kind regards and thank you
//Rob
 
sebackman said:
......
Sorry for all the questions but since I'm mixing two different DAC's in the same system....
Click to expand...

If you have no exact and consistent synchronization mechanism for the two DAC's, it would be very much difficult or even impossible to achieve your goal, I believe.

I have performed my intensive "experiments" on quasi- or sham- synchronization of two, three, and even four independent USB ASIO DAC's in one audio system by applying "suitable delay" for "kick-up timing" synchronization, but after all I simply gave up using such non-consistent, highly possibly non-reproducible, sync method in routine audio setup.

- Can I (we) temporarily synchronize outputs of multiple DAC units (each of them has own independent ASIO driver) in 10 micro second (0.01 msec) precision in DSP-based multichannel audio setup? #783

- Can I (we) temporarily synchronize outputs of multiple DAC units (each of them has own independent ASIO driver) in 10 micro second (0.01 msec) precision in DSP-based multichannel audio setup? Part-2: Simplified experiments without using audio mixer: #804


And in my post #805 there, I wrote as follows;

Hello again @mdsimon2 and dear ASR friends,

I believe that now I should end-up my experiments and discussion on the very much niche and limited "temporarily compromising synchronization method" for multiple ASIO DAC units in our Windows-PC + DSP-based multichannel audio setups.
(please PM me, if we need further discussion on this topic.)

I (we) would like to seriously implement full synchronization of multiple DAC (or audio interface) units through ASIO routing, we need to utilize (pro-grade) audio interfaces (or DAC units, if any) which actually do support such usage, for example as I could find this clearcut FAQ page for RME Fireface UFX series; I hope your web browser would properly translate the page into English.


My post #842 too should be noticed;
WS00006493.JPG

I wrote there;
In this case, it is very nice that their USB ASIO driver is so designed to recognize multiple Fireface UFX (up to three inter-synced units) as just one multichannel USB DAC/ADC for full utilization of all of the total channels for multichannel DAC processing.

In home audio HiFi stereo DACs (or even many HiFi multichannel DACs like OKTO DAC8PRO which I use now), however, the dedicated ASIO driver does not recognize multiple units USB-connected to one PC, and they usually have no master-slave clocking (hardware and software) features/functionalities.

Now I know well that if I would be really serious in using multiple DAC(-ADC) units in-sync on my single PC, I need to go into professional DAC-ADC world, like RME Fireface, supporting inter-sync clocking and having USB ASIO driver designed for recognition of multiple units by two or three USB cable connection to them.

BTW, at present, up to 16-Ch DAC processing would be enough at least for my multichannel home audio project, and therefore I assume MOTU 16A or ANTELOPE Studio Synergy Core (and other professional DAC-ADC capable of 16-Ch or more) would be preferable as "one-unit" choice/solution for me.
 
Last edited:
Hi dualzmak,

That is way above my pay-grade. I my case it is not that complicated as no USB is involved at all.

My system (aktive 3-way (LCR) & 2-way system (rear) 5.2 (+ 2 subbs) = 15 channels used) uses a buildable DSP system (BSS) with the option of analogue and/or digital input and/or output cards. Each card has four channels and one unit hold 4 cards (16 channels). The system can handle up to 128 channels in and out.

The BSS system comprises of a DSP core and user selectable input and/or output cards. Each physical unit can take 4 cards and can be expanded up to 128 channels with additional units/cards. I have two input cards with together 8 channels digital input, 4 analogue output cards, DAC's, (16 channels) of which I use 2 and 2 digital output cards (8 channels) that feed AES to the Okto (8 channel unit).

I feed my DSP unit 7.1 SPDIF 48kHz/24bit digital signal that goes into a sample-rate converter (in the DSP) and into the DSP system. The DSP does its magic in the digital domain (41 bits floating point) and sends 8-channels to 4x 4-channels DAC cards using the same clock (using only 2 cards = 8 channels analogue) and 2x 4-channels digital output cards using and embedding the same clock.

The analogue output from the DAC output cards (8 channels) feeds part of the system and the digital output cards feed the Okto for the rest of the system. The digital output cards can be set to SPDIF or AES and will carry the clock in the signal. The DSP is "smart" and will make sure that all outputs are in sync, ie add latency so all output regardless of SW or HW are in sync. The Okto receives 8-channel AES signal in and, as far as I can understand, picks up the clock from channel 1, which should give a system in clock sync.

What I need to do, as opposed to your system, is to compensate for the Okto latency (AES in to analogue out) compared to the DSP DAC's which are in timing sync with the digital out feeding the Okto.

-And to compensate for the fact that the Okto seem to have some attenuation on the output compared to the input.

Does that make sense?

Kind regards
//Rob
 
You must log in or register to reply here.

Similar threads

D
Minidsp SHD and Okto Research DAC8 Pro
Replies
4
Views
1K
ZolaIII
ZolaIII
W
Okto DAC8 update with DSP
2 3
Replies
43
Views
9K
montyliam
M
T
Active Crossover (Jubilee) and Amp Input Sensitivity
Replies
2
Views
379
Chris A
Chris A
M
MiniDSP Flex HTX vs Okto DAC 8 Pro (AVR replacement)
Replies
5
Views
2K
phoenixdogfan
phoenixdogfan
amirm
  • Poll
Topping Centaurus R2R DAC Review
23 24 25
Replies
497
Views
90K
Chagall
Chagall
Back
Top Bottom