AUDIO over IP - REDNET 3 & 16 Review. AES67 Sets A New Standard for Computer Audio

[mourip]

   
 
My mind is kind of fuzzy but I could have sworn a few months ago (well before this update but not sure what version it was) not seeing 176K in my RedNet Control app (and beginning to panic) but instead seeing it and being able to set it in the Dante Controller instead.
 
As far as the Antelope is concerned, not a big deal for me but I can see how it could confuse people because you will get playback whether you manually reset the sample rate or not. As you mentioned, most Pro studios likely set one rate per session and do not switch things  track by track. Perils of using Pro  gear.
 
I know several people set Jriver to 192K and upsample everything. I tried that for a while but switched back and actually prefer using native sample rate with the Yggy.  Does upsampling everything bypass the Schiit Mega-Burrito Filter?

 
You are right. I do see 176.4k listed in the Dante Controller app under Device view. If I choose it I am prompted with a warning that I will not be able to receive or send 192K however. Unfortunately I have a fair amount of 192K downloads now as that is my preference when available.
 
You are also right about the setting on the Antelope being confusing as even with it set to 192K I still get music when SRC Follow is set and using a slower rate than 192K.
 
Oh well. Such is the price of the bleeding edge...

 

[rb2013]

  In researching my latency issue, I came across the following from the AES67 Config section of Audinate's Dante Controller Users Guide:
 
For supported devices (Brooklyn II v3.9.x devices and up), the Device View also includes an AES67 Config tab. The AES67 Config tab allows the selection of AES67 mode for the device.
AES67 is a standard for audio over IP interoperability.
Devices in AES67 mode are able to transmit and receive AES67 multicast flows to/from non-Dante AES67- enabled devices.
Between Dante devices, Dante's native audio transport protocol is used instead (even when AES67 is enabled for both devices).
 
Based on this information, I concluded the following:

• The RedNet 3, being a Brooklyn I type device, does not support AES67.
• Dante's native audio transport protocol (not AES67) is used for AoIP communication between the DVS and the RN3.

 
Can someone correct me, if I've misstated anything?

Yes you have that right.  The AES67 compatibility was later added to DANTE in the BK2 card.  For direct connection - that is PC>REDNET>DAC (or Mutec>DAC) this makes no difference. AoIP refers to AVB, AES67 and other protocols that use Ethernet packet audio (and video) delivery over a LAN.  Many proprietary systems came into existence for transporting high-quality audio over IP based on TCP, UDP or RTP,  So I refer to only AOIP, a term I coined here on this thread, meaning DANTE and RAVENNA - AES67 or not.  This is to avoid confusion when referencing these particular audio over internet protocols.
 
AOIP AES67 is a new standard and as time progresses will allow new devices developed to 'speak' to each other over a LAN.  But at this stage the AES67 std is pretty minimal - such as needing to support only up to 96K SR's.  This is why the REDNET AM2 is included.  Ravenna which is AES67 compliant - far exceeded it's min std supporting SR up to 384K PCM and DSD.  DANTE has a 192K SR limit.
 
Bottomline - no need for AES67 activation for a direct connection.
 

 
​Correct. My understanding is your point #2 is also true for those with Brooklyn II (D16). The default setting is the Dante  transport protocol. However, for those with Brooklyn II they can switch to AES67 if they have a need (environment with mixed Ravenna and Dante gear for example)

Correct.   But only interoperability to 96k SR is required to be AES67 compliant.

 

[rb2013]

Here is a great comparison matrix of the current AoIP protocols.
 
Note only AES67, Dante, Ravenna, Q-LAN and UMAN use an 'IP' based protocol.  Livewire is now part of the AES67 collaboration.
 
https://en.wikipedia.org/wiki/Comparison_of_audio_network_protocols

 

[jazzfan]

 
Correct. My understanding is your point #2 is also true for those with Brooklyn II (D16). The default setting is the Dante  transport protocol. However, for those with Brooklyn II they can switch to AES67 if they have a need (environment with mixed Ravenna and Dante gear for example)

 
 

  Yes you have that right.  The AES67 compatibility was later added to DANTE in the BK2 card.  For direct connection - that is PC>REDNET>DAC (or Mutec>DAC) this makes no difference. AoIP refers to AVB, AES67 and other protocols that use Ethernet packet audio (and video) delivery over a LAN.  Many proprietary systems came into existence for transporting high-quality audio over IP based on TCP, UDP or RTP,  So I refer to only AOIP, a term I coined here on this thread, meaning DANTE and RAVENNA - AES67 or not.  This is to avoid confusion when referencing these particular audio over internet protocols.
 
AOIP AES67 is a new standard and as time progresses will allow new devices developed to 'speak' to each other over a LAN.  But at this stage the AES67 std is pretty minimal - such as needing to support only up to 96K SR's.  This is why the REDNET AM2 is included.  Ravenna which is AES67 compliant - far exceeded it's min std supporting SR up to 384K PCM and DSD.  DANTE has a 192K SR limit.
 
Bottomline - no need for AES67 activation for a direct connection.
 
Correct.   But only interoperability to 96k SR is required to be AES67 compliant.

 
Thanks for the clarification! I appreciate your helpful responses.

 

[johnjen]

So I looked at the wave forms at the input power connector to the FMC (Fibre Media Converter) with both the stock SMPS wall wart and a $46 LPS (Linear Power Supply) from China.
 
And while there are observable differences in ‘noise’ it is all ‘riding on top’ of the added’ noise’ that is being fed back into the power supply itself, by the active circuit under load (the FMC).
 
Describing it would be much more problematic than showing pictures, but even that would be problematic, so I figured I’d measure the ‘static’ noise into a 25KΩ resistor and use that as a baseline set of measurements and then show what the scope displays.
 
So I ignored the 9volts of dc and just looked at the amount and types of noise, of the ac signal riding on top of the 9vdc.
 
The Tek scope I’m using (468) has 3 ways of measuring the peak to peak voltages, which are Norm, Average and Envelope.
 
So here is a brief description of what they measure and why their numerical results are so different.
 
Norm is a short duration semi-frozen version of the waveform that shows the fast and more long term voltage spikes and hi-frequency noise.
Average is a sample of a medium time window view of the waveform that ignores ‘random’ voltages (the fast spikes), since it is looking for the average voltage (repetitive frequencies).
Envelope is a long term sample display of all of the voltages fast and slow, repetitive and random showing the sum total of all of it.
And these measurements were made at low frequency sweeps so I’d see all of the low and the higher freqs as well.
 
IOW the Norm is a fast snapshot of all of the voltages and frequencies including the spikes and fast transients but the displayed waveform is VERY dynamic and is constantly changing, faster than we can keep up with the flow of information.
The Average slows down the pace of change and focuses on where most of the noise resides but tends to ignore the fast spikes and transients.
The Envelope slows everything way down and shows all of everything in a cumulative display, but looses all the detail of the waveforms themselves, but it does show the maximal voltages.
 
So these are the numbers from the ‘static’ (P-P) measurements.
                 SMPS        LPS        ≈ %        multiplier
Norm         20mv        7.3mv      36%        x2.7
Average     3.54mv    1.26mv     36%        x2.8
Envelope   197mv     40mv        20%        x4.9
 
IOW you could say that the LPS is from 3 to 5 times less noisy, at least in this set of tests.
 
But that is only looking at peak to peak voltages and not the frequencies, nor the periodicity of this noise.
Put another way it is giving us the amounts of noise as a voltage but not the ‘quality’ of the noise.
That requires a personal evaluation and assessment, because there are no tests that directly relate a measurement of the noise to the SQ that results.
 
So I listen for audible changes that result from using the LPS vs. the SMPS.
And what I hear are SQ improvements very similar to those that the RN3 has made.
Specifically there is an increase in focus, especially where 2 (or more) ‘voices’ share very similar sonic attributes, such as background vocals, duets, and massed instruments, etc.
This increase in focus allows being able to differentiate between them and follow one or an other much more easily.
 
Now granted the improvement of using the LPS vs the SMPS isn’t anywhere near the same magnitude as when the RN3 was introduced into my system.
But that they exhibit nearly the same type of sonic improvement is telling in and of itself.
 
So this is the power supply I used, it’s a cheapy…
http://www.ebay.com/itm/External-Linear-Power-Supply-DC-for-CM6631-XMOS-DAC-USB-converter-led-display/131034415051?_trksid=p2047675.c100009.m1982&_trkparms=aid%3D888007%26algo%3DDISC.MBE%26ao%3D1%26asc%3D38828%26meid%3D1c6ed440b70b4896ba764695308d87d3%26pid%3D100009%26rk%3D1%26rkt%3D1%26sd%3D162096079417
 
And it has massive over capacity compared to the power draw of the FMC, so it’s loafing along, even though the LPS is only rated at 1.66 amps.
 
JJ

 

[jelt2359]

I see there's an upgraded Talema transformer version, have you tried that?
 
http://www.ebay.com/itm/Talema-transformer-External-Linear-Power-Supply-PSU-for-DAC-with-digital-display-/131737843438?ssPageName=STRK:MESE:IT

 

[rb2013]

  So I looked at the wave forms at the input power connector to the FMC (Fibre Media Converter) with both the stock SMPS wall wart and a $46 LPS (Linear Power Supply) from China.
 
And while there are observable differences in ‘noise’ it is all ‘riding on top’ of the added’ noise’ that is being fed back into the power supply itself, by the active circuit under load (the FMC).
 
Describing it would be much more problematic than showing pictures, but even that would be problematic, so I figured I’d measure the ‘static’ noise into a 25KΩ resistor and use that as a baseline set of measurements and then show what the scope displays.
 
So I ignored the 9volts of dc and just looked at the amount and types of noise, of the ac signal riding on top of the 9vdc.
 
The Tek scope I’m using (468) has 3 ways of measuring the peak to peak voltages, which are Norm, Average and Envelope.
 
So here is a brief description of what they measure and why their numerical results are so different.
 
Norm is a short duration semi-frozen version of the waveform that shows the fast and more long term voltage spikes and hi-frequency noise.
Average is a sample of a medium time window view of the waveform that ignores ‘random’ voltages (the fast spikes), since it is looking for the average voltage (repetitive frequencies).
Envelope is a long term sample display of all of the voltages fast and slow, repetitive and random showing the sum total of all of it.
And these measurements were made at low frequency sweeps so I’d see all of the low and the higher freqs as well.
 
IOW the Norm is a fast snapshot of all of the voltages and frequencies including the spikes and fast transients but the displayed waveform is VERY dynamic and is constantly changing, faster than we can keep up with the flow of information.
The Average slows down the pace of change and focuses on where most of the noise resides but tends to ignore the fast spikes and transients.
The Envelope slows everything way down and shows all of everything in a cumulative display, but looses all the detail of the waveforms themselves, but it does show the maximal voltages.
 
So these are the numbers from the ‘static’ (P-P) measurements.
                 SMPS        LPS        ≈ %        multiplier
Norm         20mv        7.3mv      36%        x2.7
Average     3.54mv    1.26mv     36%        x2.8
Envelope   197mv     40mv        20%        x4.9
 
IOW you could say that the LPS is from 3 to 5 times less noisy, at least in this set of tests.
 
But that is only looking at peak to peak voltages and not the frequencies, nor the periodicity of this noise.
Put another way it is giving us the amounts of noise as a voltage but not the ‘quality’ of the noise.
That requires a personal evaluation and assessment, because there are no tests that directly relate a measurement of the noise to the SQ that results.
 
So I listen for audible changes that result from using the LPS vs. the SMPS.
And what I hear are SQ improvements very similar to those that the RN3 has made.
Specifically there is an increase in focus, especially where 2 (or more) ‘voices’ share very similar sonic attributes, such as background vocals, duets, and massed instruments, etc.
This increase in focus allows being able to differentiate between them and follow one or an other much more easily.
 
Now granted the improvement of using the LPS vs the SMPS isn’t anywhere near the same magnitude as when the RN3 was introduced into my system.
But that they exhibit nearly the same type of sonic improvement is telling in and of itself.
 
So this is the power supply I used, it’s a cheapy…
http://www.ebay.com/itm/External-Linear-Power-Supply-DC-for-CM6631-XMOS-DAC-USB-converter-led-display/131034415051?_trksid=p2047675.c100009.m1982&_trkparms=aid%3D888007%26algo%3DDISC.MBE%26ao%3D1%26asc%3D38828%26meid%3D1c6ed440b70b4896ba764695308d87d3%26pid%3D100009%26rk%3D1%26rkt%3D1%26sd%3D162096079417
 
And it has massive over capacity compared to the power draw of the FMC, so it’s loafing along, even though the LPS is only rated at 1.66 amps.
 
JJ

Great post.  I would say the TeraDak DC-30W is a far superior Chinese LPS - R-Core transformer, better and more capacitors - they should have even lower noise.  I have taken the stock DC-30W and replaced the caps with Nichicon HW's for better performance.  The HW's have a 10,000 hr life, low impedance and high PSRR.  This is what I'm using to power the BURL B2 Bomber DAC/Dante board.


I then added a iFi DC Purifier - which reduces the noise down by 300 to 100,000 times lower .  This filter plugs right on to the DC cable.  So with this combination - extremely low noise both on the DC and feeding back into the AC line.  This is the other major issue with SMPS's they feed a lot of high frequency noise back into the local AC grid.

 
 

Ultra wide-band design, effective from 1Hz to above 5GHz
iFi leaves no stone unturned; the DC iPurifier is a ground-up, ultra-wide band design.
Effective from 1Hz all the way up to 5GHz. Crucially the strongest performance is in the audible range of 20Hz to 20KHz.

 
On my FMC I used the cheaper TeraDak X1/X2 with the caps replaced with Nichicon HW's.  I heard an immediate improvement over the SMPS's.
 

 
When i was using the REDNET 3 and Mutec MC-3+ USB - both with internal SMPS's I had them on a separate AC line isolator and filter from my DAC and a separate one for my PC (using a high PSRR Seasonic fanless PS).
 
All of this attention to power isolation and filtering has yielded great sonic improvements - noted by a blacker noise floor, increased clarity and transparency, and greater detail.

 

[johnjen]

  I see there's an upgraded Talema transformer version, have you tried that?
 
http://www.ebay.com/itm/Talema-transformer-External-Linear-Power-Supply-PSU-for-DAC-with-digital-display-/131737843438?ssPageName=STRK:MESE:IT

I saw that option but elected to go the low cost route instead, mostly to see how a 'bare bones' LPS would fair against a comparably priced SMPS.
And the power rating for the 'enhanced' transformer goes up, which is also unnecessary, due to the low power draw of the FMC.
 
And granted the SMPS wall wart that come stock with the FMC is way cheaper, but this was the cheapest LPS I could find.
It's even cheaper than the ifi ipower wall wart.
 
And it's interesting to note that the ifi ipower rates their 1µv noise rating only in the audio band, not broadband.
 
I see this as a bit disingenuous seeing as how SMPS create noise well outside of the audio band (20-20KHz).
And this LPS is being used to power a digital, not an analog circuit to begin with.
 
All in all I'd say even a cheap LPS provides an audible improvement over a cheap SMPS, and probably not even a cheap SMPS at that.
 
JJ

 

[mourip]

  I saw that option but elected to go the low cost route instead, mostly to see how a 'bare bones' LPS would fair against a comparably priced SMPS.
And the power rating for the 'enhanced' transformer goes up, which is also unnecessary, due to the low power draw of the FMC.
 
And granted the SMPS wall wart that come stock with the FMC is way cheaper, but this was the cheapest LPS I could find.
It's even cheaper than the ifi ipower wall wart.
 
And it's interesting to note that the ifi ipower rates their 1µv noise rating only in the audio band, not broadband.
 
I see this as a bit disingenuous seeing as how SMPS create noise well outside of the audio band (20-20KHz).
And this LPS is being used to power a digital, not an analog circuit to begin with.
 
All in all I'd say even a cheap LPS provides an audible improvement over a cheap SMPS, and probably not even a cheap SMPS at that.
 
JJ

 
I added fiber to my setup a week or so ago but started with it between my switch and my PC. My PC has two ethernet ports with one dedicated to DVS. I cannot say this made a difference.
 
A couple of days ago I moved the two FMC's and the fiber connection to the link between my PC and my D16. This made a difference although it was subtle.
 
This afternoon I replace the supplied SMPS to the FMC that is on the D16 end of the link with 9V from my HDPlex LPS. This made a noticeable positive difference, giving me a cleaner sound and more distinct instrument positioning.
 
Thanks for letting us know about your own experimentation.

 

[johnjen]

My setup is very simple.
My Mac Pro feeds my RN3 directly from it's 2nd ethernet port, and there is nothing else involved (except for the 2 FMC's inline between the Mac and the RN3).
No switches, routers etc.
 
The difference this setup makes, while close to the ToP (Threshold of Perceptibility ≈15%) was easily discernible in my system and the addition of the FMC and LPS bumps that ≈15% up to ≈20-25%.
 
But numbers don't really tell the tale as in it's the sonic and acoustical changes that make for compelling differences.
Being able to more easily differentiate 2 (or more) similar 'voices' from each other tells me that the timing of the reconstructed analog signal is closer to the original with less 'smearing', and more individuation of each 'voice'.
This isn't something where numbers can be assigned, or be used to reveal such differences.
 
Yes it's subtle or can be depending upon the systems overall degree of focus and inner and micro detail, and how familiar we are to these changes in our systems.
 
This is really what these RedNet boxes are revealing, at least for me, is to be able to hear 'into' the music and pick out any voice and focus upon it enough to follow it independently of all the rest of the voices, no matter what else is going on, musically.
 
That these FMC's and associated gear helps in this, is just more evidence of what is going on and what is 'important' in cleaning up the digital pathway from source to dac.
 
JJ

 

[mourip]

  When i was using the REDNET 3 and Mutec MC-3+ USB - both with internal SMPS's I had them on a separate AC line isolator and filter from my DAC and a separate one for my PC (using a high PSRR Seasonic fanless PS).
 
All of this attention to power isolation and filtering has yielded great sonic improvements - noted by a blacker noise floor, increased clarity and transparency, and greater detail.

 
Speaking about "When I was using the Rednet 3" how about an update on your experience with the Burl B2 DAC with Dante.
 
Where would you place the unmodified Burl with Dante on your DDC list?

 

[rb2013]

   
Speaking about "When I was using the Rednet 3" how about an update on your experience with the Burl B2 DAC with Dante.
 
Where would you place the unmodified Burl with Dante on your DDC list?

I really wish I could say more right now. 
 
When I have started many of my threads - speaking about major audio finds - I initially received outright dismissal and often scathing and unfair flame attacks.  Sometimes personal - accusing me of 'hyping' some of the audio stuff I found. 
 
Since generally these were either previously completely undiscovered , like the Russian vintage 6N23P tubes I referred to as my 'Holy Grail' tube http://www.head-fi.org/t/761078/6922-tube-review-17-top-6922-6n23p-e88cc-cca-7308-e188cc-tubes.  Or cutting edge computer audio stuff like the Singxer F-1 http://www.head-fi.org/t/803111/xmos-xu208-usb-bridges-the-latest-gen-has-arrived. 
 
And of course this thread  - much of these battles were deleted and at times I have been banned (still banned from the Lyr Tube thread where I first started posting about the Russian tubes to great early rebuke) and then reinstated.  Many of the attacks were from members of the audio industry - so had strong credibility - funny one of those from my old Gustard U12 thread days (leading to that thread being locked) - is now in the audio business SELLING F-1's and SU-1's and promoting them on my XU208 thread!  How we've come full circle...
 
But time has proven my findings legitimate and they have all gathered a popular following.  This only came after a long period and others eventually trying these audio finds and reporting good results.  So giving me credibility.  But the battles were unnecessary and not good for anyone, least of all me.
 
I'm a bit smarter now...so I'm going to do this BURL thing differently.  I will not post about it (other then to say I HAVE sold my REDNET and Mutec - that you can see from the closed classifieds - so you can make your own assumptions about the SQ of the BURL/Dante post mods). So no possibility of accusations of me 'hyping' anything.
 
Do this guerrilla style.
 
Cheers!

 

[occamsrazor]

  I'm a bit smarter now...so I'm going to do this BURL thing differently.  I will not post about it (other then to say I HAVE sold my REDNET and Mutec - that you can see from the closed classifieds - so you can make your own assumptions about the SQ of the BURL/Dante post mods). So no possibility of accusations of me 'hyping' anything.

 
That's a shame. I for one would really like to hear more about the Burl B2 Bomber DAC with Dante. And as you have a number of reference points to compare it to, you are in a good position and it would be useful to hear more. But up to you of course...

 

[InsanityOne]

   
That's a shame. I for one would really like to hear more about the Burl B2 Bomber DAC with Dante. And as you have a number of reference points to compare it to, you are in a good position and it would be useful to hear more. But up to you of course...

I agree, if just buying the Burl B2 does the trick so that one does not need to buy any other components (RN3, Mutec, Etc.) then that would be a really great cost savings to share with fellow Head-Fi'ers that are interested in this topic of "getting the best audio from your PC".
 
- InsanityOne 

 

[rb2013]

   
That's a shame. I for one would really like to hear more about the Burl B2 Bomber DAC with Dante. And as you have a number of reference points to compare it to, you are in a good position and it would be useful to hear more. But up to you of course...

I want to flesh out the mods first - I will get to post about it at some point.  I just hope that BURL keeps making them. 
 

  I agree, if just buying the Burl B2 does the trick so that one does not need to buy any other components (RN3, Mutec, Etc.) then that would be a really great cost savings to share with fellow Head-Fi'ers that are interested in this topic of "getting the best audio from your PC".
 
- InsanityOne 

Well you are right the BURL DAC plus a DANTE card is around $2700 - the RN3+Mutec is around $2200 - that is without the DAC.  Then factor in a decent SPDIF or AES cable so maybe you're between $2400-$2600 (unless you want to connect these multi $K boxes with a $50 cable - whatever works for you.  My experience is that the SPDIF/AES cable does make a significant difference).  Add in another decent BNC/AES cable between the RN3 and the Mutec...it all adds up.  Upgraded power cables - you need three for the RN/Mutec/DAC - versus one for the BURL,etc...
 
If the RN D16 is chosen then you are well past the BURL cost before adding in the DAC...