Auralic Aries & Sonore microRendu listening impressions

[scan80269]

Hiya...

I'm not trying to criticize anyone, (even Auralic), but in what world does a NIC "perform better" at a slower speed? It just doesn't make sense from the perspective of traditional high-end audio, - where speed is a such an important element of "sounds good"

Yeah, - for sure, - the complications of component interaction in a high performance system get muddled a bit... but something tells me that there's something wrong with either the hardware, or hardware isolation in the Aries. When you couple that with a company that has a pretty substantial "hardware problem:" i am not too sure about the viability of buying expensive band-aids.

 

When you release a product in these times with only 512MB of RAM, - that's a huge indication to me that they don't know what they're doing, or at the very least, over-emphasizing the importance of software.

There is no hard and fast rule, so I cannot and will not make any blanket claim that streaming at 100Mbps can always deliver better SQ than at gigabit speed. The NIC does not perform "better" at 100Mbps, but gigabit speed is arguably overkill for audio streaming purposes. 24/384K audio has a ~18.4Mbps bit rate, and there is no currently known high-resolution audio format that gets bottlenecked by 100Mbps networking and resulting in dropouts, disruptions, etc.

 

As to why 100Mbps can deliver better SQ than gigabit (at least in certain setups), my theory is that compared to gigabit, 100Mbps results in: (1) lower electrical noise generation within the streamer (receiving audio stream from Ethernet) and (2) lower level of high-frequency noise conducted into the streamer via the Ethernet cable.

 

(1) lower electrical noise - is accomplished by the Ethernet PHY working with only two active pairs of differential signals over the cable, and also at 1 bit per symbol (rate = 125 Mega symbols per second). Gigabit uses all 4 differential signal pairs and at 2 bits per symbol, so the PHY has to work much harder in real time to recover a data stream at 1000Mbps (gigabit), and it is easy to picture more electrical noise generated by the Ethernet PHY (within the NIC chip) at the gigabit speed, perhaps not at 10X the noise level vs. 100Mbps, but a substantial difference nevertheless.

 

(2) lower level of high-frequency noise conducted by Ethernet cable - this is mainly due to 100Mbps (100Base-TX) working with only two differential signal pairs, compared to gigabit using all 4 pairs. Whatever the amount of high-frequency noise coupling onto these differential pairs, the Ethernet PHY on the receiving end should see less noise coming in over 2 pairs compared to 4 pairs.

 

Theories aside, my adoption of 100Mbps streaming for Aries Femto was the result of careful listening tests based on control experiments. Prior to this tweak I did not know what to expect, and whether the SQ would improve or degrade, but I have separately heard that the Sonore microRendu may not perform at its best when talking at gigabit, so I started becoming curious as to how the Ethernet link rate may affect the Aries Femto SQ, thus the experimentation at 100Mbps.

 

Within this past year or two, my friend and I have encountered so many counterintuitive tweaks for our digital audio systems we almost cannot count them with our bare hands. Adding more equipment into the audio chain is supposed to cause sonic degradation, but in a number of cases the SQ actually got better in specific aspects. My friend and I regularly attend orchestra concerts in a good hall to calibrate our ears, so good concert hall sound is how we gauge the listening experience, and is essentially our yardstick as opposed to the "Hi Fi" sound or some other measure. We would call a tweak beneficial only if it brings the sound closer to the live concert hall experience. I fully believe and acknowledge that what I would call better SQ other people may not. As usual, YMMV.

[scan80269]

Very interesting scan!

Do you know exactly which RJ45 signal pairs that is reuired for 100Mbps streaming? I see more and more proof of similarities between my finding in USB cables with your and others findings in Ethernet cables.

I do wonder if not a DIY unshielded signal pair cable will be the next step for me to try out instead of the FMCs which actually add additional power supplies to the audio chain? With two of these it should be fairly easy to make that experiment:

 

[ATTACH=CONFIG]31555[/ATTACH]

I am just not sure which ones I´ll need to connect for 100Mbps streaming....so I would truly appreciate a pointer from you or someone else (gladly with a description what the specific pins do as well)!

 

Thanks for a great thread!

 

Ps. I also want to try to separate the two twisted signal pairs on a distance from each other to see if that will make a difference as well...since it will most likely reduce the capasitive coupling along the lenght of the wires. Ds.

 

Based on the TIA/EIA-568-B.1-2001 T568B wiring standard, the pin definitions of the 4 twisted pairs are as follows:

 

* Pair 1: pins 5 & 4 (blue)

* Pair 2: pins 1 & 2 (orange)

* Pair 3: pins 3 & 6 (green)

* Pair 4: pins 7 & 8 (brown)

 

For 100Base-TX, pins 1 & 2 and pins 3 & 6 are active. For gigabit (1000Base-T), all 4 pairs are active.

 

I converted a standard CAT5e cable to "100Mbps only" by insulating pins 4,5,7 & 8 at the RJ45 plug with superglue gel. My friend will make me a cable with pins 4,5,7 & 8 not connected at either RJ45 plug to achieve the same result.

 

Increasing separation between the 1/2 and 3/6 pairs should reduce alien crosstalk, which is critical for 10Gbps operation (CAT6a), but should be less critical for gigabit or 100Mbps operation. It would be interesting to determine the effects on SQ at 100Mbps, though.

 

I have on order from BlueJeansCable several CAT5e & CAT6 cables (I already have some CAT6a from them) of 1-foot and 7-foot lengths, and will be comparing them directly for SQ merit. My friend has discovered that 7-foot/2-meter cables can in some cases sound better than 1-foot cables. His theory is that at 2-meters, any signal reflections arriving at the receiver end are delayed sufficiently to not have an impact on how the initial signal waveform is seen by the receiver chip. The same reasoning may apply to other cables such as coax, AES/EBU and possibly even USB cables. Normally, shorter is better for cables, but this assumes the cable has ideal impedance (single-ended or differential) with no part of the signal being reflected due to impedance discontinuities, which is never 100% true in practice. "Return Loss" is a measure of the loss caused by signals in the cable reflecting as a result of changes of impedance of the line.

 

The pin 3/6 pair likely has the worst impedance discontinuity due to it being terminated into non-adjacent pins. Crosstalk between pairs is likely worst between the 3/6 and 4/5 pairs. The 4 twisted pairs are supposed to have different numbers of twists per length to reduce alien crosstalk. The device in your picture looks interesting, but I don't know how much of an impedance bump it will have on each twisted pair. Differential impedance will change whenever the spacing between the two wires of a pair changes.

 

I look forward to hearing your findings on this experiment!

[scan80269]

Hi Larry,

 

The post that I read that made me ensure I bought 100MB FMCs was with respect to the microRendu by Jesus of Sonore. The mR NIC is 100MB and 1GB input speeds can overload its stack and cause dropouts. This was then further backed up by another poster who posited that 1GB NICs are noisier.

 

I think you have done precisely the right thing, tried for yourself and decided what sounds best in your system.

 

M

I found these statements in Chris' microRendu review, part 1:

 

"The microRendu contains a 10/100/1000 Gbps Ethernet interface. This interface is limited to 470 Mbps due to the internal i.MX6 bus. Audiophile needn't worry about this "limitation" because 470 Mbps is still hundreds of Mbps more than is required for even the highest resolution audio files."

 

I was able to confirm the microRendu establishing gigabit link speed with my GS108 switch. The data throughput is apparently below line speed, but even 470 Mbps is overkill for high-resolution streaming.

 

I would agree that 100Mbps FMCs are the way to go, and I'll be buying a pair of these for myself.

[scan80269]

Hi Scan,

 

It just occurred to me to ask: have you ever done a head-to-head between the Aries Mini and the Aries Femto? If you have the ability and interest to do that, your results would be fascinating, particularly with the isolation tweaks in place.

 

My Aries Mini has scaled so far in SQ from when I first got it a year ago, especially with all the isolation tweaks, that I now wonder if there is anything to be gained by "upgrading" to a microRendu or big Aries.

If you are referring to comparing Aries Femto with Aries Mini without using its internal DAC, I haven't done an apples-to-apples comparison, so I'll need to schedule another listening session at my friend's place.

 

The Aries' FemtoClocks for USB and digital outputs is possibly one factor accounting for SQ difference between the Femto and the Mini. When the Mini first showed up last year, my friends and I did some casual listening and agreed it delivers roughly 80% of the SQ of the Femto, and this was with the internal DAC, so we all thought the Mini is one heck of a bargain, even though it didn't quite measure up to the Femto, which was expected given the huge price difference. If the Mini is used as a renderer with the isolation tweaks & LPS, etc., the difference with Femto may be even smaller, and the audibility of this difference may depend on the rest of the system.

 

Stay tuned!

[scan80269]

Sorry if I wasn't clear. The configurations I'm referring to are with an external DAC.

 

I suppose the microRendu will need to use DLNA output mode for an apples-to-apples comparison with the Aries Mini?

[scan80269]

Hi Scan,

 

This is on my list of things to do, strip out all the USB bits and see what really counts. The question in my mind is what is the Line Isolator solving? I spoke to a EE I know who wasn't convinced that it was either a grounding issue or a leakage loop, as he didn't feel that either would have the sound signature that we both have described.

 

M

This is only a (educated?) guess: an Ethernet isolator acts as a low-pass filter to attenuate high-frequency (RF) noise conducted into Aries by the Ethernet cable.

 

Long cable runs can become antennas, which can transmit (radiate the transported signal) or receive (pick up noise from the environment). An Ethernet isolator can help reduce the junk picked up by the Ethernet cable and carried into the device. The transformers inside the EMO EN-70HD isolator are of higher quality than typical Ethernet pulse transformers, so RF noise is less able to (capacitively) jump through the isolator.

 

Another possibility is the Ethernet isolator somehow helping to deliver a better signal waveform into Aries. This is admittedly difficult to quantify, at least without expensive network analyzer equipment that can measure eye patterns for the Ethernet twisted pairs.

 

I don't even know the input and output differential impedances of the EMO EN-70HD isolator (they are not listed in the datasheet), but judging by the return loss graph, I would guess that those differential impedances should not be far off the 100 ohm spec for Ethernet cables. It is unlikely the EN-70HD aggravates Ethernet signal integrity by inducing reflections down the cable. I only wish I have the means to do the measurements to support or debunk these hypotheses.

[scan80269]

If you have a managed switch, such as the GS108T, can't you just set the max port speed on in the switch? Does this sound differently to the cable tweak?

 

I almost forgot I have a GS108T along with GS108v4. I'll need to configure one GS108T port to 100Mbps speed to test with Aries. Any way to avoid having to add a FS105 Fast Ethernet switch (just to force Aries down to 100Mbps) is worth trying.

[scan80269]

When I first tried my FS105 switch (V2 7.5v) in my setup I struggled to hear any difference between gigabit and 100mbps being fed to the Aries, the EN-70HD then arrived and I installed it together with the switch between my wireless bridge and the Aries. It all sounded great to me so I left it that way.

 

Last night I returned from a week away from home and did a comparison with and without the switch. With very "fresh" ears there was a noticeable difference, particularly on rock music, where there was a bit more fullness without the switch. With the switch in place the presentation tended to be on the leaner side, like some mid-bass presence was missing. Either way sounds good but I prefer the sound without the switch.

 

As I will be regularly traveling for a week every month in the future, I think I will leave any critical listening comparisons to the time when I return home with refreshed ears

 

I found that even the FS105 switch revision and power supply type affect the sound. Between FS105 v1 (7.5V) and FS105 v3 (12V) my friend and I preferred the latter. Between the Netgear 12V wall wart supplied with FS105 v3 and my modified Teradak X1/X2 PSU (set to 12V) we also preferred the latter. Even for a simple switch there are variations that can yield sonic differences, so people may settle into different setups.

 

It looks like the FS105 v3 switch has lower power consumption than its predecessors. There is a switching regulator inside for energy efficiency, but it may be a noise source. I'm working on replacing it with a low-noise linear regulator.

 

I also believe that the best switch is no switch at all, so I've been separately exploring other ways to force my Aries to stream over 100Mbps without using a FS105 switch, such as the "100Mbps only" Ethernet cable, or setting one port of a GS108T switch to 100Mbps speed to be dedicated to Aries.

[scan80269]

I have several tweaks queued up for listening evaluations at my friend's place this coming Friday:

 

1. Vinnie Rossi MINI-PURE-DC-4EVR power supply for Aries Femto

 

2. Modified Netgear FS105v3 Fast Ethernet switch (internal switching regulator replaced with ultra low noise LDO regulator) - powered by UpTone LPS-1 supply (no SMPS contribution)

 

3. Custom "100Mbps only" Cat5e Ethernet cable

 

4. Blue Jeans Cable Cat5e, Cat6 & Cat6a cables of 1' and 7' lengths (6' for Cat6a)

 

I just took a listen to my Aries Femto powered by the Vinnie Rossi MINI after only a 12-hour burn-in, streaming over 100Mbps with the custom "100Mbps only" Cat5e cable & EMO EN-70HD isolator, and my usual Vega DAC over USB (with WireWorld Starlight Platinum cable & VBus2 isolator), with the Vega balanced XLR outputs run through a Jensen PI-2XX transformer to my Taurus MkII headphone amp, then finally to my HD800 headphones with a Stefan AudioArt balanced cable (non-hardwired). Without giving too much away, I will only say "WOW!!!!"

 

I can't wait till Friday to listen to the MINI/Aries Femto/Vega setup at my friend's place. It should be great!

[scan80269]

Sorry austinpop! Not possible! I have checked this carefully when l was looking for unregulated battery supply for the Mini. 14-18v...not more and sadly not less!

Sent from my iPhone using Computer Audiophile

 

I also confirmed that a 12V power supply is good for Aries Femto but NOT for Aries Mini. 14V is the absolute minimum for the Mini.

 

This past weekend, I auditioned 3 different PSUs for the Aries Mini at another friend's place, with QUAD ESL 57 electrostatic speakers and Cary Audio 300B monoblocks. There were six of us and we were unanimous in rating the PSUs in order of improving SQ:

 

* Aries Mini included SMPS wall wart

* Modified Teradak 30W 15V linear power supply

* SBooster BOTW P&P ECO 15-16V linear power supply + SBooster Ultra

 

The Aries Mini appears to have fewer aftermarket power supply options than Aries Femto. These won't work with Mini: UpTone LPS-1, UpTone JS2, Vinnie Rossi MINI PURE-DC-4EVR, Sonore Signature Series Power Supply. I suppose Vinnie's MEGA power supply when available next year would be an option, as is Teddy Pardo, SBooster, etc.

[scan80269]

Yup, two LPS-1s user set at 7V and wired in series makes a quite nice PS for the Aries Mini.

Sent from my iPad using Computer Audiophile

 

Yes, but with other power supplies there is no guarantee it can work...

[scan80269]

I'm not an electrical engineer, but the Aries ships with a 16V LPS. Not sure if using the Rossi 12V could damage it. Personally, I have Ken Lau's LPS and really like it.

Before the arrival of the Vinnie Rossi MINI PSU, I have been powering my Aries Femto with an UpTone JS2 linear power supply set to 12V for well over a year without issues. I believe member jtwrace has used this same configuration. Empirically, the Aries Femto appears to have a DC input voltage latitude of several volts. I would not go below 12V, though.

 

Kenneth Lau's PSUs are beauties, and should be a great option for Aries Femto.

[scan80269]

That depends on what plug is used on the Aries Mini.

 

If it is another 2.1mm plug, then easiest is to splice two 2.1mm cables together as the inputs to another as the output.

 

Cut off one end of each of the three 2.1mm cables

 

Wire the negative of one of the cut ends of the input cables to the positive of the other.

 

Connect the output cable negative to the unattached negative at the input cable junction and the output cable positive to the unattached positive.

 

Before using, test with a meter set to test continuity to confirm:

 

1. The negative of one of the input cables connects to the output negative.

 

2. The positive of that cable connects to the other input cable's negative.

 

3. The positive of the other input cable connects to the output positive.

 

Next plug them into the two LPS-1s set to 7v, power them up, then use the meter set to DC voltage to confirm 14v output AND the correct polarity. COMPARE polarity to the existing PS TWICE just to be sure!!!

 

If all good, plug it in and enjoy!

 

Greg in Mississippi

Aries Mini uses a 5.5mm/2.5mm plug. A 2.1mm plug will not fit (I already tried).

[scan80269]

I have several tweaks queued up for listening evaluations at my friend's place this coming Friday:

 

1. Vinnie Rossi MINI-PURE-DC-4EVR power supply for Aries Femto

 

2. Modified Netgear FS105v3 Fast Ethernet switch (internal switching regulator replaced with ultra low noise LDO regulator) - powered by UpTone LPS-1 supply (no SMPS contribution)

 

3. Custom "100Mbps only" Cat5e Ethernet cable

 

4. Blue Jeans Cable Cat5e, Cat6 & Cat6a cables of 1' and 7' lengths (6' for Cat6a)

 

I just took a listen to my Aries Femto powered by the Vinnie Rossi MINI after only a 12-hour burn-in, streaming over 100Mbps with the custom "100Mbps only" Cat5e cable & EMO EN-70HD isolator, and my usual Vega DAC over USB (with WireWorld Starlight Platinum cable & VBus2 isolator), with the Vega balanced XLR outputs run through a Jensen PI-2XX transformer to my Taurus MkII headphone amp, then finally to my HD800 headphones with a Stefan AudioArt balanced cable (non-hardwired). Without giving too much away, I will only say "WOW!!!!"

 

I can't wait till Friday to listen to the MINI/Aries Femto/Vega setup at my friend's place. It should be great!

 

Here are the listening results for some of these tweaks, at my friend's venue:

 

1. Vinnie Rossi MINI-PURE-DC-4EVR power supply for Aries Femto

 

Compared to Auralic LPS: Excellent clarity & transparency, but sound became a bit flatter, bass impact diminished, and music became slightly less engaging. An "A" to "A-" drop.

 

 

2. Modified Netgear FS105v3 Fast Ethernet switch (internal switching regulator replaced with ultra low noise LDO regulator) - powered by UpTone LPS-1 supply (no SMPS contribution)

 

Compared to stock FS105v3 powered by modified Teradak X1/X2 9VA PSU set to 12V: slight enlarging of instrument sizing; 1st violins sounded larger (wider).

 

 

3. Custom "100Mbps only" Cat5e Ethernet cable

 

Not tested. My custom cable was not long enough to take over from the stock Cat5e cable. FS105v3 switch was used to force Aries to network at 100Mbps.

 

 

4. Blue Jeans Cable Cat5e, Cat6 & Cat6a cables of 1' and 7' lengths (6' for Cat6a)

 

Not tested; ran out of time. My friend previously evaluated 1' Cat6a and 2m/6' Cat5e between Aries and EN-70HD isolator and said the 2m/6' Cat5e cable sounded slightly better.

 

Admittedly these results today are somewhat unanticipated, but not totally surprising. My friend and I have some hypotheses:

 

A. Auralic LPS feeds 16V into Aries Femto. Vinnie Rossi MINI PSU outputs 12V. Aries Femto SQ may vary by input DC voltage. The switching regulators inside Aries Femto can work with 12V or 16V or anything in between, but the current draw and thus switcher operating point will differ depending on what the input voltage is. I plan to put this to the test by using a lab grade adjustable power supply to compare Aries Femto SQ between 16V and 12V inputs.

 

B. Modified FS105v3 switch is fed by lower DC input (5V) but higher current than the stock switch (12V). Similar case as A. The linear regulator in the modified switch is also less efficient than the switching regulator in the stock switch, so the modified switch consumes a bit more power overall. The stock switch does not work with 5V or 7V input, so instead I ran 12V briefly into the modified switch, but the stock switch still yielded superior SQ. The sonic difference here is much smaller vs. the PSU change for Aries Femto.

 

So far, none of the linear PSUs tried have been able to beat the Auralic LPS in getting the Aries Femto to excel at this location.

 

Please note these observations are very specific to my friend's setup and venue, so results can easily vary with different setups elsewhere. For example, I heard the exact same Vinnie Rossi MINI PSU deliver an audible SQ step-up with my headphones setup, though in this case I went from UpTone JS2 PSU @12V to Vinnie Rossi MINI PSU @12V so there was no input voltage change to Aries Femto. The lesson I learned here is that equipment synergy has to be achieved on a case-by-case basis, and often by trial-and-error.

 

The SQ at my friend's place is currently at a very high and enjoyable level, especially prior to today's evaluations. The soundstage is deeper than ever before, instruments very focused, imaging is rock steady, abundant micro-detail with stunning clarity, excellent dynamics with strong bass impact, easy-to-hear instrumental articulation and vocal expressiveness. Overall a very holographic and engaging sound with very strong concert hall ambiance. Reversing the orientation of the EN-70HD isolator was the most recent tweak with positive SQ results. Everything tried today took the SQ in the wrong direction. Hope to have better results next time!

[scan80269]

Here's an update to power consumption measurements for Aries Femto & Aries Mini, as DC current figures (as opposed to AC power last time) which should be more accurate, as the power dissipation of the LPS is not lumped into the numbers. Voltage fed into both streamers was 16.0V.

 

 

This time the numbers are closer among the three networking options. Both streamers were located much closer to the wireless router, with direct line of sight between the two. For both Aries Femto & Mini, the 100Mbps networking option continues to represent the lowest current (& power) consumption.

[scan80269]

Hi scan!

Have you ever tried a 100base-T ethernet cable in your setup?

Hi Cornan, I'm not sure what you mean by a 100base-T Ethernet cable, but all my Ethernet cables are Cat7, Cat6a, Cat6 & Cat5e, but nothing lower. These support both 100Mbps (100Base-TX) and gigabit (1000Base-T). My friend made me a custom "100Mbps only" Cat5e cable with connectivity in just two twisted pairs (pins 1/2, 3/6) and it has delivered very good SQ results.

[scan80269]

It seems that pin 4&5 is the best choise if I want to try 100Base-T1 when searching the web. Anyway, I will let you know if it works and sounds good or not when I have got all the parts in house! 

 

Sent from my Nexus 7 using Computer Audiophile mobile app

Hi Cornan, I'm not 100% sure, but I suspect in order for your DIY Ethernet cable to work, it will need to have two twisted pairs on pins 1/2 and 3/6 respectively as a minimum, in order to be compliant with 100Base-TX. A cable with only one twisted pair on pins 4/5 will not deliver connectivity, as virtually no audio or standard networking gear with an Ethernet port is compliant to 100Base-T1. I guess you will find out for yourself when building the DIY cable.

 

When I opened up my Netgear FS105v3 Fast Ethernet switch, I noticed that each of the five RJ-45 ports has electrical connectivity to the Ethernet IC on pins 1/2 and 3/6 only. Pins 4/5, 7/8 of each port are not connected to anything. This lead me to ask my friend to crimp me a "100Mbps only" Cat5e cable supporting only these two pin pairs.

 

One thing to note about using such a custom "100Mbps only" Cat5e cable to interconnect two gigabit capable devices (e.g. gigabit Ethernet router/switch to gigabit-capable streamer): the two devices will initially handshake to gigabit link speed, but quickly fail due to only 2 of 4 twisted pairs having connectivity. In many cases, the link speed will fall back to 100Mbps (talking over 1/2 & 3/6 twisted pairs) and stay there, though this is not always guaranteed to work, as it is dependent on how the devices handle a gigabit data transfer failure in their firmware/driver.

[scan80269]

Thanks scan!

I have already suspected that it might not work. I will probably start with 100Base-TX with each pair on a distance from eath other. Thanks for letting me know what awaits me! [emoji4] Happy New Year! [emoji898][emoji95][emoji483][emoji92]⭐️[emoji1323]

 

 

Sent from my iPhone using Computer Audiophile

 

For your DIY Ethernet cable, I recommend making the two lengths of wire for each twisted pair to be close as possible, to minimize intra-pair skew. Also, the twisting of the wire pair should be done as evenly as possible to maintain consistent cable impedance. CAT5e cables typically have different numbers of twists per length for each of the four twisted pairs, to reduce alien crosstalk, though there is apparently no standard on numbers of twists per length. If you can maintain large physical separation between the 1/2 and 3/6 pairs then it may not be beneficial to do different numbers of twists per length for the two pairs.

 

Happy 2017!

[scan80269]

There is obviously no audio timing information transported over Ethernet. The digital audio stream is sent over Ethernet TCP packets at gigabit or 100Mbps, at a symbol rate that has nothing to do with any known digital audio sampling rate. So while the Ethernet is not part of the audio signal path, it is part of the audio transport path. I don't think Cornan is arguing that Ethernet is part of the audio signal path, but I'll let him speak for himself.

 

Strictly speaking, even the USB link between streamer and DAC does not carry audio timing information. The USB isochronous/asynchronous transfer mode used here allows the receiver (DAC) to vary the data pace of the send (receiver) to some degree, mainly to prevent buffer overruns or underruns, but overall the USB link still operates at nominal 480Mbps (for USB 2.0 HighSpeed) which is also distinct from any digital audio sampling rate. It is the DAC that generates the audio-based clock feeding into the digital-to-analog converter chip.

 

Other DAC inputs such as S/PDIF coax, AES and Toslink do involve digital audio timing information, but I wouldn't lump USB into this group.

 

What's been puzzling to me until fairly recently is that since the Ethernet transport merely serves to get the digital audio data into the streamer for converting into USB stream, and is twice removed from the digital stream entering the D-to-A converter, that Ethernet tweaks can impact the SQ of the overall system. Assuming such tweaks (Ethernet isolator, cables, etc.) do not impact Ethernet data transport integrity (i.e. no bit corruption of audio data from music source to DAC), then the SQ changes need to be explained by something other than the bits having gone bad.

 

I currently subscribe strongly to the effects of leakage loops, which can exist in just about any section of a digital audio playback system, so the benefits of effective isolation are real. And with John Swenson having just confirmed my belief that Ethernet PHYs vary significantly in power consumption (& thus noise contribution) by incoming signal integrity, I can see how any tweaks that can help uphold better signal integrity in transport links such as Ethernet and USB can improve SQ.

[scan80269]

Thanks @scan80269!

Correct, to me the similarity between USB and Ethernet is in how various noises and possibly DC currents is effecting the signal. I am not saying it is a part of the audio signal path...but the various ethernet noises will indirectly affect the streamer and the audio signal path.

 

 

Sent from my Nexus 7 using Computer Audiophile mobile app

 

With Ethernet, since there are pulse transformers behind each RJ-45 jack in every piece of Ethernet gear, it should be impossible for DC currents to be carried by the Ethernet cables, assuming the cables are unshielded (U/UTP).

 

The audibility of an Ethernet isolator such as EMO EN-70HD (used in conjunction with unshielded Ethernet cables) suggests some possibilities:

 

* Isolator alters the Ethernet signal integrity (e.g. waveform) at the streamer/renderer input

 

* Isolator attenuates some form of high-frequency noise carried by the cable into the streamer/renderer

 

What's even more mind-boggling to me is how reversing the EN-70HD orientation can cause an audible SQ change. Per EMO, the EN-70HD is fully symmetrical for gigabit networking operation, and I have no reason to dispute this. However, there are "energy absorbing" components attached to only one side of this isolator. These components should not be active during normal operation, but perhaps they contribute capacitance (intra-pair and inter-pair) that affects differential impedance and thus signal integrity of the Ethernet signals running through the device. It would be interesting to compare the EN-70HD with a Baaske MI-1005 isolator which AFAIK does not have similar over-voltage protection components as the EMO. I'm almost willing to bet that the Baaske isolator orientation has much less of an effect on SQ compared to EN-70HD. Another experiment to be queued up!

[scan80269]

Very Interesting read scan!

 

There is already several trustable people who have posted great SQ improvements going from Baaske MI-1005 to EMO EN-70HD so I think there is definately something happening there. There is ofcourse a possibillity be that all these people are using shielded Ethernet cables at the input of the Ethernet isolator. I do not know for sure.

 

Capasitive coupling created inside router & FMCs or along the lenght of the Ethernet cable is always an issue to consider as well as ground noises IMO.

 

One of my personal theory is however that the DC current and leakage loops could be injected and shared through the input of the router (ie. from service provider) plus inside the router by switched and linear mains connected PSU and also affect the signal output of the router somehow. "Luckily" my router is using F-plug input, so I have ordered a F-plug DC blocker to test this theory of mine. It is just a long shot. I will also try powering the upstream FMC with battery vs SMPS as well...but the downstream FMC always powered by battery.

These test should atleast give me a pointer if there is some truth in those theories.

 

I am very much looking forward to know what you find out with your experiments. Keep it on!

I currently have a Jensen ISO-Max VRD-1FF isolator on the RF input of my cable modem, and a Baaske MI-1005 isolator between the modem and wireless router. I have not done listening tests with and without these devices, but consider them preventive measures to be left in place permanently.

 

The Baaske was bought used and cost much less than the EMO EN-70HD.

[scan80269]

I've got a combined router/wifi-modem. No switch. NAS (dedicated to music) and microRendu are connected to this modem. To improve SQ I'm considering two options.

1) Add EMO EN-70HD between modem and microRendu.

2) Add a switch and connect NAS and microRendu to this switch. This way NAS and microRendu are a bit isolated from the modem. Later I can upgrade the switch psu.

 

What would have the best influence on SQ?

 

Synology NAS, LPS-1, microRendu, Job INTegrated, Penaudio Cenya.

I'll speculate that both options can affect SQ. Why not try both? Another option that may be worth considering:

 

3) Leave NAS connected to modem. Add a Fast Ethernet switch (e.g. Netgear FS105v3) between modem and microRendu to force 100Mbps networking speed for microRendu audio streaming. As an alternative to this switch, a custom "100Mbps only" Ethernet cable with connectivity in just 2 twisted pairs (pins 1/2, 3/6) can be used to force 100Mbps.

[scan80269]

@Cornan and @austinpop,

 

I just measured my Jensen ISO-MAX VRD-1FF CATV isolator, and there is no DC continuity between the two ends. I beeped both the center conductor and outer thread and got nothing. Jensen website mentioned this device is no good for satellite setups due to the DC needed for the satellite dish being completely blocked. Not sure if the Supra is built the same way?

 

The Jensen isn't exactly cheap, but at least there's no need to insert a DC blocker in line.

 

BTW, we have drifted rather far off-topic...

[scan80269]

@austinpop, you are certainly welcome to post your mRendu findings on this thread!

 

One key operational difference between Aries Mini and mRendu is that the former supports memory playback, and therefore allows the Ethernet link to go mostly quiescent in between tracks, whereas the mRendu stream buffering is much shallower, so the Ethernet link is active on almost a constant basis during playback. A hypothesis here is that the "often-active" Ethernet activity of mRendu may put it in a SQ disadvantage against the Aries Mini/Femto. As far as I can remember, the mRendu was sonically competitive with Aries Femto running 3.x firmware (without memory playback feature), but got leap-frogged by Aries when memory playback got enabled by 4.x firmware. This was a quantum jump in SQ of Aries Femto that so far has not been matched by mRendu powered by LPS-1 and SBooster VBus2 added to the USB link to DAC.

 

Of course, your results may well be different, and I look forward to hearing from you.

[scan80269]

Hi Scan,

 

In my headphone setup, after attaching the Vinnie Rossi MINI PSU @12V to my Auralic Aries Femto

(previously had the Auralic LPS attached), with my Audeze LCD-X I heard increased resolution,

increase in dynamics and pleasantly surprised with better imaging.

 

Curious if you got a chance to do that test "using a lab grade adjustable power supply to compare Aries Femto SQ between 16V and 12V inputs."

 

Unfortunately for my wallet, I noticed a subtle, but noticeable improvement on my wifi router which accepts 12V when I attached the 12V Vinnie Rossi MINI PSU. So am thinking moving my current 12V MINI PSU to the router permanently ... and considering ordering another MINI PSU for the Auralic Aries Femto , though unsure whether to order 16V (when the Vinnie Rossi MINI high voltage becomes available) or just another 12V Vinnie Rossi MINI PSU. Hence my curiosity about your tests with a lab grade adjustable power supply, if you had a chance to do them. Thanks.

 

John

Hi John,

 

I also heard a sonic improvement with headphones after connecting the Vinnie Rossi MINI to my Aries Femto, so a clean, isolated power supply can work its magic on a streamer/renderer.

 

I just did a quick test with my Aries Femto being powered by an HP E3631A adjustable DC power supply, and alternated the voltage between 12V and 16V, and did not hear any difference using my usual headphone setup including Vega DAC, Taurus headphone amp, HD800 cans and usual tweaks like Jensen transformer, EN-70HD Ethernet isolator and VBus2 isolator. I'm going to propose to repeat this experiment at my friend's place with speakers instead of headphones. Unfortunately that HP power supply has a rather noisy fan which can interfere with critical listening, but we'll try our best to discern any difference.

 

I did discover that the Aries Femto can handle a DC input voltage as low as 10V, and as expected, the current consumption goes up with lower voltage, so the power consumption (voltage x current, in wattage) is roughly the same. Changing the DC input voltage (within the supported range) alters the operating point of the switching regulator inside Aries, and it would be interesting to hear how that affects SQ, if at all.

 

For your router, even though the Vinnie Rossi is the ultimate supply, you may be able to achieve similar results with a lower cost linear power supply, such as Teddy Pardo, Kenneth Lau, KECES, or even Teradak. Is your HDPLEX linear PSU the 100W model? If so, the 12V output may be worth a try.