• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Dante AVIO Review (streaming audio interfaces)

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
32,923
Likes
110,675
Location
Seattle Area
#1
This is a review and detailed measurements of the Dante AVIO Ethernet streaming input and output modules. It was kindly loaned to me by a member. They each cost $144.

The modules have extremely nice quality and feel of robustness to them:

Dante AVIO Review Input and Output Networked Audio Module.jpg


There is an Ethernet connection at the other end that streams data to/from respectively for each module:

Dante AVIO Review Input and Output Networked PoE Ethernet Audio Module.jpg


Power is provided over Ethernet (PoE) so there are no adapters or dongles to worry about. The owner sent me an HP PoE switch which I used for testing.

There is software you need to download to configure them (Dante Controller). They also have a neat Dante Virtual Soundcard which creates either an ASIO or WDM interface (in Windows) to treat the modules as if they are hardwired to your computer. Alternatively the modules can be configured to talk to each other without a PC in the middle allow you to extend balanced audio over Ethernet to wherever that interface goes. And of course with no digital loss in the interconnect. The ADC/DAC modules are not perfect so let's measure them independently and then together.

Note that there is no dynamic sample rate switching. You set the sample rate and that is the only rate that the module works at. You have to reconfigure the module to play other sample rates which are supported up to 96 kHz.

Dante AVIO Analog Output Measurements
Per above introduction, I configured the output module for 44.1 and ran our dashboard:

Dante AVIO Adapter Measurements.png


Note that there are level configurations and these are set to maximum they support. Harmonic distortion and variable noise floor are disappointing, putting the AVIO output in our least favorite category of performance:
best streaming DAC review.png


You basically have 14 bits of distortion-free range. Sweeping the digital input shows there is not much more performance to be had at lower output levels either:

Dante AVIO Adapter Measurements SINAD vs Level.png


Dynamic range was better than I expected from the dashboard:

Dante AVIO Adapter Measurements dynamic range.png


Linearity was also another bright spot:

Dante AVIO Adapter Measurements Linearity.png


Jitter performance showed the same rather high and uneven noise floor:

Dante AVIO Adapter Measurements Jitter.png


Intermodulation vs level starts reasonably but ends on a sour note:
Dante AVIO Adapter Measurements IMD.png


Reconstruction filter once again shows strangeness with noise floor:
Dante AVIO Adapter Measurements filter.png


This becomes a serious issue as usual in our THD+N vs frequency as higher frequencies:
Dante AVIO Adapter Measurements THD+N vs Frequency Distortion.png


But even without that, the low frequency noise+distortion is rather high at nearly 0.008%.

Can't run the multitone test since 192 kHz sampling required for that test is not supported.

Dante AVIO Analog Input Measurements
Here is our usual dashboard for ADC interfaces:

Dante AVIO Input Measurements.png


Oh, this is much more respectable than the DAC!

best streaming ADC.png


Dynamic range is a bit low though, failing 16 bit threshold:

Dante AVIO Input Measurements Dynamic Range.png


IMD vs level is decent:
Dante AVIO Input Measurements IMD.png


Frequency response at 48 KHz sampling is shy of 24 kHz but this is rather typical:

Dante AVIO Input Measurements Frequency Response.png


Linear is what we would expect from dynamic range test:

Dante AVIO Input Measurements Linearity.png


I was going to stop here but thought I run the SINAD vs Frequency and boy, am I glad I did:

Dante AVIO Input Measurements THD vs Frequency.png


Wow, what is up with that low frequency rise? Usually this is caused by poor power supply (negative rail usually). Fortunately low frequency distortion is not as audible but still, like to see better design than this.

Dante AVIO Input and Output Measurements
Let's cascade the two modules, using them as a black box balanced audio extender. This would be our dashboard then:

Dante AVIO Measurements.png


The DAC is operating a bit more optimally due to lower output level giving us a SINAD of 87 dB. The ADC has much lower distortion so it is transparent in this regard.

Sweeping the analog input level we get:
Dante AVIO Measurements THD+N vs Level.png


So you can only lower the performance, not better it than what I have shown.

Conclusions
I really like the plug-and-play aspects of these modules for streaming audio over ethernet. The build quality is excellent as well. The software can be a bit difficult to use but once you figure it out, you can get things done. In understand these are entry level modules from Dante. Still, I like to see at least clean, 16 bit performance out of them. We are short of that so I am disappointed in that regard. Don't even think about using them on your JBL SDP-55 processor hoping to get better performance than internal DACs. You will only make things worse, not better.

In a pro, sound enforcement situation, the robustness should make them a good fit as long as you realize the performance degradation over just running an analog connection to a good ADC.

Overall, I like to recommend the Dante AVIO modules but based on pure performance, I can't. You have the data to decide otherwise.

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

Appreciate any donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
 

dfuller

Addicted to Fun and Learning
Joined
Apr 26, 2020
Messages
953
Likes
1,137
#2
Yeah... Audinate is not known for their analog engineering prowess. The Dante technology is awesome (see the Focusrite RedNet stuff, which is the only real competitor to Pro Tools HD IO in terms of massive low latency IO) but their own implementations not so much.
 
Last edited:
Joined
Nov 18, 2020
Messages
22
Likes
38
Location
Buffalo, NY
#3
Thanks again for this review!

I believe Audinate may have simply rebadged an existing product from Amphenol. For example: https://www.bhphotovideo.com/c/prod...1112_0050_amphe_dante_rj45_audio_adapter.html

Audinate's main thing is selling chips that take care of Dante protocol stuff for use by other manufacturers, talking I2C to whatever existing ADC/DAC is in place, making it easier to implement or add-on. They do provide reference designs though, and I'm betting most of these cheaper products just follow that exact reference design.

They probably have started pushing these cheaper products knowing their flaws, but wanting to expand the market beyond just very high end live consoles & interfaces.
 
Last edited:

3125b

Addicted to Fun and Learning
Joined
May 18, 2020
Messages
704
Likes
850
Location
Germany
#4
Based on performance I could see this being handy in mobile use applications like live events, but for studio purposes, it's probably not good enough, is it?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
32,923
Likes
110,675
Location
Seattle Area
Thread Starter #5
Not in my view. You want very high quality to start so that all the processing that you do on the samples in post still leaves you with honest 16 bit at least.
 
Joined
Nov 18, 2020
Messages
22
Likes
38
Location
Buffalo, NY
#6
I suppose their main goal with these is something useful to have in your toolbox for the "oh crap" last minute moments, where suddenly the band on stage needs an extra input and this will be "good enough". It's difficult to tell apart from the real professional stuff though because marketing will always shroud performance in mystery. There are plenty of higher end Dante products with data to back them up but they're all $$$$$.
 
Joined
Apr 8, 2020
Messages
63
Likes
56
#7
I suppose their main goal with these is something useful to have in your toolbox for the "oh crap" last minute moments, where suddenly the band on stage needs an extra input and this will be "good enough". It's difficult to tell apart from the real professional stuff though because marketing will always shroud performance in mystery. There are plenty of higher end Dante products with data to back them up but they're all $$$$$.
Exactly, the client has suddenly added a new area that you need to get sound to / from very quickly and easily. Perfect device for that application. For driving the main PA; use a Focusrite Rednet device.
 
Joined
Nov 18, 2020
Messages
22
Likes
38
Location
Buffalo, NY
#8
Exactly, the client has suddenly added a new area that you need to get sound to / from very quickly and easily. Perfect device for that application. For driving the main PA; use a Focusrite Rednet device.
Yeah or a Yamaha digital console or snake. You can also put these in "AES67 mode" which will let them talk to various other Livewire, Q-SYS, Ravenna, Wheatnet and AVB products, though with some ease of use niceties missing and only at 48khz on these models.
 
Joined
Apr 8, 2020
Messages
63
Likes
56
#9
Yeah or a Yamaha digital console or snake. You can also put these in "AES67 mode" which will let them talk to various other Livewire, Q-SYS, Ravenna, Wheatnet and AVB products, though with some ease of use niceties missing and only at 48khz on these models.
It would be very interesting to see an AVB device measured as some manufacturers are touting that AVB has a higher potential sound quality than Dante due to the way the AVB protocol reserves bandwidth and is more sophisticated in stamping data packets in and out of hardware (this is what my Luddite understanding is anyways). I’m sceptical, but I would be happy to be wrong.
 
Joined
Nov 18, 2020
Messages
22
Likes
38
Location
Buffalo, NY
#10
It would be very interesting to see an AVB device measured as some manufacturers are touting that AVB has a higher potential sound quality than Dante due to the way the AVB protocol reserves bandwidth and is more sophisticated in stamping data packets in and out of hardware (this is what my Luddite understanding is anyways). I’m sceptical, but I would be happy to be wrong.
Yeah good question, I’m not sure how much Dante’s clocking architecture would affect performance. I assume similar to what you get from USB or S/PDIF I/O on most products. Once you’re on the Dante network it’s bit-transparent.
 

pos

Addicted to Fun and Learning
Forum Donor
Joined
Feb 13, 2018
Messages
524
Likes
597
#11
Not sure about the Dante AVIO, but the Amphenol version specifies different output levels:+18dBu/+4dBu/0dBu/0dBV/-10dBV
(last page here: http://www.amphenolaudio.com/wp-content/uploads/2018/10/Amphe-dante-SD.pdf )

It is not clear if these values are obtained with simple digital attenuation from +18dBu at 0dBFS, or using some kind of level "presets" in the Dante software that would potentially have an impact on the output stage...
 
Joined
Nov 18, 2020
Messages
22
Likes
38
Location
Buffalo, NY
#13
The AES3 version would be the one to test to verify that.
The AES3 AVIO module runs an asynchronous resampler on the input that can not be disabled, so the source clock is always the Dante system. The output as far as I can tell on mine is bit-transparent, but I don't have a way to measure jitter.

On higher end interfaces (like some RedNets) you can use other forms of digital I/O (like AES3 or MADI) as a source clock that the Dante network then follows.
 

spacevector

Senior Member
Forum Donor
Joined
Dec 3, 2019
Messages
395
Likes
668
Location
Bayrea
#14
An awesome review of an interesting and unique product. Thanks amirm!

Is it possible for you to measure the signal delay in the 'extender' mode?
 
Joined
Nov 18, 2020
Messages
22
Likes
38
Location
Buffalo, NY
#15
An awesome review of an interesting and unique product. Thanks amirm!

Is it possible for you to measure the signal delay in the 'extender' mode?
Forgive me for replying to so many posts on here hah.

The primary purpose of these is to integrate with other Dante products, but you can configure them to send directly to each other. Dante stuff works by "handshaking" at a set latency every time. These models have 10/100 ethernet ports, meaning through a single switch there's probably about 400μs between them. The units can be configured to handshake at 1ms, 2ms, or 5ms. They do this by testing the line with PTP first to establish clock, so in this case the remaining 600μs would be buffer. Most higher end Dante products can handshake at 500μs or even 250μs, using standard gigabit switches.

"Guaranteed latency" is one of the selling points of Dante for big professional workflows.

Given that the DAC/ADC also have some inherent latency, I'm guessing you'd get somewhere around 3ms from analog input to output.

With modern gigabit switches, you could probably stack 10+ switches in series and still not cause them to fail the 1ms handshake.
 
Last edited:

infinitesymphony

Addicted to Fun and Learning
Joined
Nov 21, 2018
Messages
965
Likes
1,497
#16
It looks like with a PoE network switch you could connect multiples of these and have some matrix control with Dante Controller.

 

PeteL

Addicted to Fun and Learning
Joined
Jun 1, 2020
Messages
907
Likes
807
#17
The AES3 AVIO module runs an asynchronous resampler on the input that can not be disabled, so the source clock is always the Dante system. The output as far as I can tell on mine is bit-transparent, but I don't have a way to measure jitter.

On higher end interfaces (like some RedNets) you can use other forms of digital I/O (like AES3 or MADI) as a source clock that the Dante network then follows.
typically there is no gain in doing that tough, dante works great with other dante devices, with qos guarantee and as you said latency guarantee, Dante don’t recommend bringing an external clocking system to the equation, unless you want to achieve something specific. What’s also great is that the Protocol looks for the most accurate dante clock on the network and will use it as the master, it’s the right thing to do to use dante for clocking, the other options are fore edge cases. But sure, for this very specific product, it’s just average (mediocre) adc and dac, the concept of the dante network become limited if all you do is stream 2 channel from an analog end to another
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
32,923
Likes
110,675
Location
Seattle Area
Thread Starter #18
Is it possible for you to measure the signal delay in the 'extender' mode?
Sure. It is 1.54 millisecond at 48 kHz sampling. It drops to 1.14 millisecond at 96 kHz sampling.
 

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,782
#19
The primary purpose of these is to integrate with other Dante products, but you can configure them to send directly to each other. Dante stuff works by "handshaking" at a set latency every time. These models have 10/100 ethernet ports, meaning through a single switch there's probably about 400μs between them. The units can be configured to handshake at 1ms, 2ms, or 5ms. They do this by testing the line with PTP first to establish clock, so in this case the remaining 600μs would be buffer. Most higher end Dante products can handshake at 500μs or even 250μs, using standard gigabit switches.
Not sure about the accuracy of this. May be I misunderstood but last time I looked at Dante specs, the latency "handshake" was designed to manage transceiver processing latency (not network latency). If a transceiver at the other end cannot process packets faster than every X ms (including buffering), there is no reason to send packets faster than that. So that is the minimum latency requirement as far as that transceiver is concerned. This is hardwired into the transceiver (which knows what latency happens inside its processing) and is the same on any network (topology). The sender could decide to send packets at a slower speed than that.

If there is a peer-peer direct connection between them then that is the latency for transmission.

However, if it is on a shared IP network, then I am not sure how the network latency is automatically figured out. PTP is just for clock synchronization as far as I understand it.

Then there is the issue of variable latency in a shared IP network unless the network is purpose-built with QoS guarantees for latency.

I don't see this protocol as a simple plug-and-play for a consumer application except for point to pont (or you have network engineering expertise) in which case an USB will do just as well without the transceiver complexity.

In studio settings and local or wide area distribution, absolutely.

Having said that, I don't think the latency is an issue with most simple topologies on private networks but it could make a difference in the ability to do live recording/playback from multiple sources (unless the latencies are as low as the typical audio interfaces).
 
Joined
Nov 18, 2020
Messages
22
Likes
38
Location
Buffalo, NY
#20
Then there is the issue of variable latency in a shared IP network unless the network is purpose-built with QoS guarantees for latency.
The audio bitstream is typically RTP (or a proprietary flavor of it) over unicast, or optionally multicast. Dante supports IGMPv2 and DSCP flags to assist with reliability on shared infrastructure.
 
Top Bottom