Taiko Audio SGM Extreme : the Crème de la Crème
- Thread starter CKKeung
- Start date
The vibration coming from spinning fans and how these vibrations are carried over to the many vibration sensitive parts in a music server make for a more complicated RF environment and are negative for good SQ
Last edited:
What Amanero Firmware is on Howie's Pacific?
Maybe it is a Linux OS optimised one instead of Windows optimised?
It's the right firmware.
Hi dminches,
Indeed, like @EuroDriver posted above vibrations influence electrical performance of parts, especially so in digital circuits. Liquid cooling is worse then regular air cooling as it involves a pump for circulating fluid, usually built into the CPU cooler, and fans mounted on a remote radiator. Let me explain why and start addressing @microstrip 's comments a few pages back.
There are 2 main effects at play, piezo and pyro electric. Pyro is derivates from ancient Greek translating as fire, heat or high temperature, Piezo means to squeeze or to apply pressure.
Pyro electricity causes electrical component characteristics to change by varying temperature.
Piezo electricity causes electrical component characteristics to change by varying pressure force, aka vibration.
2 of the most sensitive, and often discussed, components to these effects are crystal oscillators and capacitors.
Discussing Pyro electric sensitivity for oscillators and capacitors :
For high precision crystal oscillators (referred to as clocks) there are so called OCXO versions available, there are 2 basic ways to cut a crystal, AT and SC cut. AT cut crystals are less sensitive to temperature variation but are less precise then SC cut crystals, however SC cut crystals are only more precise then AT cut inside a specific temperature range, an OCXO utilises an SC cut crystal with an "oven" included in the package to keep the temperature at a specific value.
For capacitors the dielectric value changes with varying temperature, generally speaking the higher the dielectric value the more sensitive it is to these effects. Higher dielectric values are used when capacitors need to be smaller, as a higher dielectric value requires less material for the same capacitance. In digital circuits we need small physical size capacitors for proper high frequency bypassing. Often these need to be placed as close as possible to silicon chip pins, as longer circuit traces and a larger physical size increases ESL (inductance) which provides impedance to the frequencies we are trying to bypass, or filter, decreasing effectiveness. As operating frequencies involved in digital circuits an be magnitudes higher then in analogue circuits, we need smaller physical size capacitors, and these are more sensitive to both pyro and piezo electric effects. The most effective capacitors for bypassing in digital circuits are ceramic capacitors which are abundantly used in virtually every digital device, DACs, clocks, CD transports, servers, switches etc.
Pyro electric sensitivity is positively affected by efficient cooling, so fans can have a positive effect here as fans are magnitudes of orders more effective then passive cooling. You can have a very small heatsink, with closely spaced fins for a large surface area (that will not work for natural convection air cooling as airflow is minimal in those tight areas) and force large volumes of air through it, creating a very effective cooling solution. A passive equivalent solution needs to be dozens of times larger to achieve the same performance. We are even using solid copper heatsinks to boost efficiency as that transports heat twice as effective as aluminium.
Pyro electric sensitivity of components used also defines the equipment temperature operating range, the higher the sensitivity, the lower the range in which it can operate to within specification, or at all. There are so called "wide temperature range" parts / circuit boards / equipment, these are often preferred by audiophiles as they simply provide a more stable performance (same sound) with varying temperature. Of course these are more expensive, just compare the prices of wide temperature spec drives, memory modules, network switches, SFP modules etc. Wide temperature is also often labelled industrial quality, and it generally also has a longer useable lifespan.
Discussing Piezo electric sensitivity for oscillators and capacitors :
Piezo electric effects can be magnitudes more influential then Pyro electric effects, hence fan based cooling positives are generally outweighed by their associated effects on Piezo sensitivity.
Clock oscillators can have mayor increases in noise (phase noise) when exposed to vibration due to Piezo electric sensitivity, this can easily be a 30-40dB increase in phase noise.
Capacitors can exert similar effects, especially the smaller more efficient types (remember there is no way around those in high speed digital circuitry).
As an example, lets compare the Pyro versus Piezo effects on small capacitors, taken from a Panasonic datasheet:
Pyro electric effect, Ceramic X7R are widely used, one of the most effective capacitors for high frequency noise suppression, but note the variance in capacitance outside of general room temperature ranges, ECHU is a much larger type of capacitor:
Now let's take a look at the effect of vibrations to capacitance values, as you can see capacitance drops more with increasing frequency of vibration applied:
There is a more harmful effect of piezo electric activity in capacitors, much akin to crystal oscillator increases in phasenoise, which is actual noise generation from capacitors when subjected to vibrations, this can easily create a 1% or even higher THD (total harmonic distortion) in an environment where you are hunting for sub zero THD:
Great post. Another source of noise that is usually ignored is triboelectric noise - curiously the best dielectrics that are, in general, considered for audiophile purposes, such as teflon, are the worst elements considering triboelectricity.
Its noise must be considered in applications needing very low leakage current and small signals, such as medical applications and high impedance instrumentation.
A good page on it https://experience.molex.com/triboelectric-noise-in-medical-cables-and-wires/
Emile, this is an amazing post. Thank you for taking the time to explain this.
Yes I'm not particularly fond of Teflon, makes you wonder if their "hyper detailed" type of sound is actually detail which is on the source to begin with
View attachment 62684[/QUOTE]
Curiously my experience - mainly with the conrad johnson (ART's and GAT's) and Audio Research REF40 and the Sonus Faber Aida crossover - does not show any trend towards hyper detailed sound at all. For me the big drawback of Teflon is the long burn-in time - they take a lot of time to loose their initial "bright" sound. Please do not ask me why!
Yes I'm not particularly fond of Teflon, makes you wonder if their "hyper detailed" type of sound is actually detail which is on the source to begin with
Curiously my experience - mainly with the conrad johnson (ART's and GAT's) and Audio Research REF40 and the Sonus Faber Aida crossover - does not show any trend towards hyper detailed sound at all. For me the big drawback of Teflon is the long burn-in time - they take a lot of time to loose their initial "bright" sound. Please do not ask me why!
Curiously my experience - mainly with the conrad johnson (ART's and GAT's) and Audio Research REF40 and the Sonus Faber Aida crossover - does not show any trend towards hyper detailed sound at all. For me the big drawback of Teflon is the long burn-in time - they take a lot of time to loose their initial "bright" sound. Please do not ask me why!
Surely there are means towards voicing towards "undetailed"
I did audition a REF150SE and a GS150, and indeed can't say those sounded "overdetailed".
I have the Taiko Extreme now a bit more than 3 months in my system and i really love it. It is my best component i bought the last 5 years in terms of "set and forget", which is quiet uncommon with top level audio equipment.
Built quality is "extreme" and so is the support.
But the sound... that is really unbelievable good. When i got the Extreme brand new with not to many test hours it was allready very good sounding. I had my Aurender W20 next to it so i could switch easy the first 3 days and compare while optimizing the output driver by Emile. At that time the Extreme was clear better sounding, to give it a number say 15-20% better, and i liked the W20 very much.
But it needed a lot of hours to break in. I couldn't believe that a component without analog output stage would need much break in to improve the sound, but it really needs 2 months powered. The sound gets better and better during that time. Last week i tried again the Aurender W20 because i had an issue with my network. Once that was fixed i listened to the W20, it sounded like the system was broken... Than the Extreme, wow great music. I think now it is 80% better.
So much more analog sounding so much more flow and decay is so much better. No harshness at all.
The Extreme in combination with my Pacific Dac is so extremely good and musical that i use it 85% of the time at the moment. This combi is better than a lot of analog imo, sorry..
Steve williams
Site Founder, Site Owner, Administrator
I totally agree. The Pacific DAC with the Extreme has changed completely the way I listen to music as the combination indeed brings it as close to analog as the real thing
Nonsense, nothing is perfect especially in such a capricious hobby as high end audio. In another year or so an Ultimate or Statement Grandioso series X Mk II will be out. Such is the nature of technology and achieving "perfect" will never be reached. And that's half the motivation and fun of it all.
Several of my friends and forum members — Kennyb123, Ray-Dude, Always Learning — and I have been carrying on a back channel discussion about FMCs, SFPs, the EtherRegen (ER) from Uptone Audio and the Optical Module (OM) from Sonore/Small Green Computer. We’ve each heard some of these devices in our own systems but none of us except Kennyb123 have heard them all head to head. Through the generosity of these folks I was lent the pieces I was missing so I could do a comparison in my own system as an Extreme owner and hopefully share some insights from my perspective.
My system is listed in my current signature.
My LAN is fairly simple compared to many peoples’ setups:
DSL modem (WAN port) > copper jumper > Synology RT2600ac WiFi router > copper jumper > TLS OXCO switch > copper > AV system, and copper > Thecus network attached storage (NAS) device. All of these devices are in one room across the entry foyer from the living room where my stereo is located.
I have WiFi turned off in the modem, firewall enabled, domain name server (DNS) enabled. WiFi is turned on in the Synology RT2600ac and it is linked wirelessly with a Synology MR2200 mesh router upstairs. The MR2200 provides strong WiFi coverage upstairs and is linked via copper to an AV system in the master bedroom.
The DSL modem is powered from one rail of an Uptone Audio JS-2 linear power supply, a Startech FMC is powered from the other rail. The Synology RT2600ac is powered from an inexpensive variable linear power supply. The NAS uses its stock onboard SMPS. All of these power supplies are plugged into a Shunyata Research PS-8 power strip using a Shunyata Alpha Digital power cord to the wall. Additionally, a Shunyata Venom Defender is plugged into one of the PS-8 receptacles.
No particular vibration control is used other than original Stillpoints cones under the JS-2, Vibrapods under the cable modem and the FMC or device under test (DUT). I haven’t noted any particular improvement in sound quality with these vibration control devices, but I have them and so I use them.
My baseline connection from LAN to Taiko Audio Extreme server is the one recommended by Emile early on — Startech FMCs and Startech SFPs (1550nm 80km), LC LC fiber and copper jumpers. I’ve been using these even before my Extreme arrived to connect my previous Innuos Zenith server to the LAN, with the addition of another Startech FMC and jumper at the Zenith.
With the arrival of the Extreme, the feed from the LAN became:
DSL modem > copper jumper (Sablon ethernet cable) > StarTech FMC/Startech SFP with -10dB attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
My initial plan for testing involved comparing the baseline feed from LAN to Extreme to the following:
DSL modem > copper jumper > ER “B” side/ER “A” side/Startech SFP w/-10db attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
DSL modem > copper jumper > ER “A” side/Startech SFP with -10dB attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
DSL modem > copper jumper > OM/Startech SFP with -10dB attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
Then repeat without attenuators on the SFP ports.
Finally, repeat again, this time substituting Planet Tech SFPs (1310nm 20km) for the Startech SFPs. Again, with and without attenuation.
I kept all other devices, including power supplies the same and only varied the devices used in place of the Startech FMC. Obviously, the SFPs stayed the same on each end for each trial.
Additionally, I used a ground lead from each device to an AC earth ground; both the Startech FMC and ER have provisions for this, I backed a case screw out a little on the OM to use as its ground point. Like Rajiv in his experiments with the ER, I didn't hear any particular advantage or improvement using the ground lead with any device -- perhaps because I have also removed all SMPS' from my equipment in the computer room/office (my NAS with its internal SMPS has been off since copying all my music to the Extreme). Since the ground wire wasn't hurting anything, I left it attached for each trial.
As indicated above, in the case of the ER, I tried it as designed ("across the moat”), "B" to “A", as well as copper in/fiber out on the "A" side only.
I gave each device under test a minimum of 24 hours powered up with Roon Radio playing and my amplifier off before listening, with the exception of the ER tests which I performed back-to-back without additional warm up time after the first 24 hours. Any other device I thought I might want to quickly A/B against the DUT I kept warm by using an inexpensive variable linear power supply from Radio Shack.
So, that sets the stage with the test conditions and protocol. Next post — what I heard.
Steve Z
My system is listed in my current signature.
My LAN is fairly simple compared to many peoples’ setups:
DSL modem (WAN port) > copper jumper > Synology RT2600ac WiFi router > copper jumper > TLS OXCO switch > copper > AV system, and copper > Thecus network attached storage (NAS) device. All of these devices are in one room across the entry foyer from the living room where my stereo is located.
I have WiFi turned off in the modem, firewall enabled, domain name server (DNS) enabled. WiFi is turned on in the Synology RT2600ac and it is linked wirelessly with a Synology MR2200 mesh router upstairs. The MR2200 provides strong WiFi coverage upstairs and is linked via copper to an AV system in the master bedroom.
The DSL modem is powered from one rail of an Uptone Audio JS-2 linear power supply, a Startech FMC is powered from the other rail. The Synology RT2600ac is powered from an inexpensive variable linear power supply. The NAS uses its stock onboard SMPS. All of these power supplies are plugged into a Shunyata Research PS-8 power strip using a Shunyata Alpha Digital power cord to the wall. Additionally, a Shunyata Venom Defender is plugged into one of the PS-8 receptacles.
No particular vibration control is used other than original Stillpoints cones under the JS-2, Vibrapods under the cable modem and the FMC or device under test (DUT). I haven’t noted any particular improvement in sound quality with these vibration control devices, but I have them and so I use them.
My baseline connection from LAN to Taiko Audio Extreme server is the one recommended by Emile early on — Startech FMCs and Startech SFPs (1550nm 80km), LC LC fiber and copper jumpers. I’ve been using these even before my Extreme arrived to connect my previous Innuos Zenith server to the LAN, with the addition of another Startech FMC and jumper at the Zenith.
With the arrival of the Extreme, the feed from the LAN became:
DSL modem > copper jumper (Sablon ethernet cable) > StarTech FMC/Startech SFP with -10dB attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
My initial plan for testing involved comparing the baseline feed from LAN to Extreme to the following:
DSL modem > copper jumper > ER “B” side/ER “A” side/Startech SFP w/-10db attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
DSL modem > copper jumper > ER “A” side/Startech SFP with -10dB attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
DSL modem > copper jumper > OM/Startech SFP with -10dB attenuator > fiber > Startech SFP with -10dB attenuator/Extreme
Then repeat without attenuators on the SFP ports.
Finally, repeat again, this time substituting Planet Tech SFPs (1310nm 20km) for the Startech SFPs. Again, with and without attenuation.
I kept all other devices, including power supplies the same and only varied the devices used in place of the Startech FMC. Obviously, the SFPs stayed the same on each end for each trial.
Additionally, I used a ground lead from each device to an AC earth ground; both the Startech FMC and ER have provisions for this, I backed a case screw out a little on the OM to use as its ground point. Like Rajiv in his experiments with the ER, I didn't hear any particular advantage or improvement using the ground lead with any device -- perhaps because I have also removed all SMPS' from my equipment in the computer room/office (my NAS with its internal SMPS has been off since copying all my music to the Extreme). Since the ground wire wasn't hurting anything, I left it attached for each trial.
As indicated above, in the case of the ER, I tried it as designed ("across the moat”), "B" to “A", as well as copper in/fiber out on the "A" side only.
I gave each device under test a minimum of 24 hours powered up with Roon Radio playing and my amplifier off before listening, with the exception of the ER tests which I performed back-to-back without additional warm up time after the first 24 hours. Any other device I thought I might want to quickly A/B against the DUT I kept warm by using an inexpensive variable linear power supply from Radio Shack.
So, that sets the stage with the test conditions and protocol. Next post — what I heard.
Steve Z
Last edited:
Because of the number of permutations involved I kept the streaming tracks used for the comparisons to a minimum in an attempt to keep the overall project manageable and importantly, to keep my sanity. Like any audiophile, I unavoidably end up listening to sound as much as music when doing comparisons — what improved, what didn’t improve? But my tolerance for this kind of listening is pretty low and I would much rather relax and just listen to music once a certain minimum acceptable sound quality is reached. Too much concentrated A/B/C testing and I start to lose the will to go on living.
The three streaming selections I used are:
1. Brenda Fricker, Anuna, Never Give All The Heart from The Chieftains "Tears Of Stone" 16/44.1 Qobuz stream
2. Brian Bromberg, The Saga Of Harrison Crabfeathers, "Wood" 16/44.1 Qobuz stream
3. Clifford Curzon, Benjamin Britten/LSO, Mozart Piano Concerto No. 20, Allegro 16/44.1 Tidal stream
With 2. and 3. I used two different sections in each track rather than playing the whole long track through each time. Each of these recordings have places where the digital transfer can sound anywhere from hot to irritatingly bright or hard if everything isn’t up to snuff in the digital path.
A word about what I listen for when doing this kind of test. I'm very sensitive to harshness, hardness and unnatural brightness. I listen for tonal balance and saturation, low-level detail, depth, width and height (particularly depth), and dynamic swing. And generally, I also try to listen for an undefined overall *enjoyability* -- would I be happy listening to my system sounding like this for hours, or would I be unsatisfied and restless?
I've found that for whatever reason, a recording on the edge of being hard or harsh sounding will be tipped over that edge and become hard to listen to by less than excellent digital performance. Sibilant sounds, cymbals, and other high frequency instruments can quickly go from brilliant to bright to hot. As the top end hardens and higher frequencies become more prominent everything seems to flatten out. My test recordings tend to emphasize this in places and so represent a worst case scenario. There is still lots of music that will sound fantastic with any of these configurations though.
So, without further ado --
First off, my feeling is these LAN optimizations are audible and are worth pursuing, particularly since other than the power supplies and one of the devices under test (ER), they are fairly inexpensive in the audiophile scheme of things. However, I am also in general agreement with other Extreme owners that the largest benefits are gained from premium power cords, USB cables, and vibration control. Network optimization is an order of magnitude lower in effect than power, USB and vibration.
As a general conclusion, in all cases I preferred using the Planet Tech 1310nm 20km SFPs over the Startech 1550nm 80km SFPs. This was regardless of device (ER, FMC or OM) and regardless of attenuation used or no attenuation at all. In fact, in some circumstances the choice of SFP made a bigger delta than choice of device (ER, FMC or OM). Likewise, I found I always thought using -10dB attenuators on the receiver ports of the Startech SFPs to be a necessity, or they sounded edgy with any of the devices.
The most cost effective and must-do item from all my experimenting is replacing any Startech SFPs with Planet Tech SFPs. The Planet Tech SFPs made every device (OM, ER, FMC) sound better in my system. At $23.90 USD the Planet Tech SFPs are a bargain.
Specific device results --
Any of the combinations, even the nominally last ranked, are still much better than the 50 feet of copper ethernet I originally used between LAN and server. Also, moving my feed to the Extreme all the way upstream to an ethernet port on my DSL modem and having everything else in my LAN downstream on the other side of a WiFi router fed from the modem's WAN port, was as large a positive improvement as choice of device.
1. Last place, but still very honorable mention:
The combination of Startech FMC/Startech SFP is one I lived with even before I took delivery of the Extreme. It was a large improvement over a long run of copper ethernet from LAN to server. I did find -10dB attenuators on the receiver ports tamed some brightness which might have been caused by overdriving the SFP receivers with the powerful 1550nm lasers over a very short fiber link.
The Startech FMC can sound very good, but there is better out there and it came in last of the three devices I tested. A Startech FMC with Planet Tech SFP and no attenuators sounds better to me than the all-Startech combination of FMC and SFP.
However, the Startech FMC configurations sound, as someone else on one of the Internet forums observed, somewhat mechanical in nature compared to better combinations.
2. Good, and Better:
EtherRegen (ER) "B" to "A"/Startech SFP/-10dB attenuators, and Planet Tech SFP/no attenuation --
The ER "B" to "A" across-the-moat configuration is less mechanical or electronic sounding than the Startech FMC combos. It also nicely ameliorates any tendency of edgy digital transfers to tip over into brightness or harshness. However, with the Extreme it sounds too polite and sounds smaller and less energetic than any other combination. The Planet Tech SFP helped with the clarity but didn’t cure the small stage size and lack of dynamics.
3. Better, and tied for Best:
ER "A" to "A"/Startech SFP/-10dB attenuator, and Planet Tech SFP/no attenuation --
ER in the "A" to "A" configuration brought back a good part of the dynamics and life and scale of music. The Planet Tech SFP sounded a little softer than the Startech SFP, but in either case the "A" to "A" ER brought back some of the edginess too. This was slight however, and for a person who needs a switch rather than an FMC, the ER is a cost and sound-quality effective choice. And, from the preponderance of reports, if a non-Extreme owner can use it in an "A" to "B" configuration, it might be the best sounding option of all.
4. Best in my situation:
Optical Module/Startech SFP/-10dB attenuator, and OM/Planet Tech SFP/no attenuation --
The OM is perhaps the most transparent to, or maybe I should say the most dependent on the SFP it is paired with. With the Startech SFP even using attenuation, the OM can still be a little edgy. But it sounds big and bold and full of life, if just that little bit hard on top with less favored digital transfers. As a result, it sometimes sounds a little flat and a little forward when challenged with brighter recordings.
Paired with the Planet Tech SFP and without any attenuation, the OM retains all of the life and dynamics of music without any objectionable hardness on top, and with a subjective sense of more depth, dimension and space. It doesn't hide what recordings are, but it doesn't exacerbate their imperfections either. To my way of hearing, it strikes just the right balance in the context of my system and my tastes. I had one of those, “Ahhhhh, this is good!” moments when I started listening to the OM/Planet Tech SFP combination.
I wound up my tests by comparing streamed tracks to their twins residing in local on-board storage, and also tested the claim that with the Extreme and an optimized network, high-resolution (24/96, 24/192) streamed music can beat standard Redbook resolution (16/44.1) stored on-board the Extreme.
Listening to the same resolution streamed, compared to local playback, I thought local playback to be a bit better -- a little cleaner, a little darker and a little more depth. But it is very close and streaming certainly doesn't suffer for the comparison. I can listen to music via the Extreme and Roon Radio streaming all day long and never get tired or restless. Every so often a poorly recorded or digitally transferred track will come up, but that can happen regardless of medium.
Streaming higher resolution tracks and comparing them to their standard resolution, locally stored counterparts proved a little trickier. It's difficult to determine in many cases whether the mix was also tweaked when the source recording was remastered to higher resolution; in the case of MQA high resolution, MQA processing may be affecting the recording in ways that are sometimes enhancing and sometimes detracting (in my experience). However, over a dozen or so non-MQA high resolution Qobuz selections I found that in the majority of cases I did indeed prefer the high-res streamed version over standard resolution locally stored content.
Once again I'll just say that all of this is very much in the context of my own system, LAN and taste in music and musical presentation. And it appears that for whatever the reason, with an Extreme in the mix some of the unique attributes of the EtherRegen don't come out in the same way they do with other servers. The EtherRegen would be my first choice if I needed a switch in the feed to my Extreme because it is still better sounding than a Startech FMC, or a TLS OXCO switch. However for my needs and for what I value in sonic presentation the Optical Module with Planet Tech SFP has just the right balance.
Hopefully this was helpful, or at least interesting and worth the time spent reading.
I hope everyone is well and taking care of their loved ones in this challenging time.
Steve Z
The three streaming selections I used are:
1. Brenda Fricker, Anuna, Never Give All The Heart from The Chieftains "Tears Of Stone" 16/44.1 Qobuz stream
2. Brian Bromberg, The Saga Of Harrison Crabfeathers, "Wood" 16/44.1 Qobuz stream
3. Clifford Curzon, Benjamin Britten/LSO, Mozart Piano Concerto No. 20, Allegro 16/44.1 Tidal stream
With 2. and 3. I used two different sections in each track rather than playing the whole long track through each time. Each of these recordings have places where the digital transfer can sound anywhere from hot to irritatingly bright or hard if everything isn’t up to snuff in the digital path.
A word about what I listen for when doing this kind of test. I'm very sensitive to harshness, hardness and unnatural brightness. I listen for tonal balance and saturation, low-level detail, depth, width and height (particularly depth), and dynamic swing. And generally, I also try to listen for an undefined overall *enjoyability* -- would I be happy listening to my system sounding like this for hours, or would I be unsatisfied and restless?
I've found that for whatever reason, a recording on the edge of being hard or harsh sounding will be tipped over that edge and become hard to listen to by less than excellent digital performance. Sibilant sounds, cymbals, and other high frequency instruments can quickly go from brilliant to bright to hot. As the top end hardens and higher frequencies become more prominent everything seems to flatten out. My test recordings tend to emphasize this in places and so represent a worst case scenario. There is still lots of music that will sound fantastic with any of these configurations though.
So, without further ado --
First off, my feeling is these LAN optimizations are audible and are worth pursuing, particularly since other than the power supplies and one of the devices under test (ER), they are fairly inexpensive in the audiophile scheme of things. However, I am also in general agreement with other Extreme owners that the largest benefits are gained from premium power cords, USB cables, and vibration control. Network optimization is an order of magnitude lower in effect than power, USB and vibration.
As a general conclusion, in all cases I preferred using the Planet Tech 1310nm 20km SFPs over the Startech 1550nm 80km SFPs. This was regardless of device (ER, FMC or OM) and regardless of attenuation used or no attenuation at all. In fact, in some circumstances the choice of SFP made a bigger delta than choice of device (ER, FMC or OM). Likewise, I found I always thought using -10dB attenuators on the receiver ports of the Startech SFPs to be a necessity, or they sounded edgy with any of the devices.
The most cost effective and must-do item from all my experimenting is replacing any Startech SFPs with Planet Tech SFPs. The Planet Tech SFPs made every device (OM, ER, FMC) sound better in my system. At $23.90 USD the Planet Tech SFPs are a bargain.
Specific device results --
Any of the combinations, even the nominally last ranked, are still much better than the 50 feet of copper ethernet I originally used between LAN and server. Also, moving my feed to the Extreme all the way upstream to an ethernet port on my DSL modem and having everything else in my LAN downstream on the other side of a WiFi router fed from the modem's WAN port, was as large a positive improvement as choice of device.
1. Last place, but still very honorable mention:
The combination of Startech FMC/Startech SFP is one I lived with even before I took delivery of the Extreme. It was a large improvement over a long run of copper ethernet from LAN to server. I did find -10dB attenuators on the receiver ports tamed some brightness which might have been caused by overdriving the SFP receivers with the powerful 1550nm lasers over a very short fiber link.
The Startech FMC can sound very good, but there is better out there and it came in last of the three devices I tested. A Startech FMC with Planet Tech SFP and no attenuators sounds better to me than the all-Startech combination of FMC and SFP.
However, the Startech FMC configurations sound, as someone else on one of the Internet forums observed, somewhat mechanical in nature compared to better combinations.
2. Good, and Better:
EtherRegen (ER) "B" to "A"/Startech SFP/-10dB attenuators, and Planet Tech SFP/no attenuation --
The ER "B" to "A" across-the-moat configuration is less mechanical or electronic sounding than the Startech FMC combos. It also nicely ameliorates any tendency of edgy digital transfers to tip over into brightness or harshness. However, with the Extreme it sounds too polite and sounds smaller and less energetic than any other combination. The Planet Tech SFP helped with the clarity but didn’t cure the small stage size and lack of dynamics.
3. Better, and tied for Best:
ER "A" to "A"/Startech SFP/-10dB attenuator, and Planet Tech SFP/no attenuation --
ER in the "A" to "A" configuration brought back a good part of the dynamics and life and scale of music. The Planet Tech SFP sounded a little softer than the Startech SFP, but in either case the "A" to "A" ER brought back some of the edginess too. This was slight however, and for a person who needs a switch rather than an FMC, the ER is a cost and sound-quality effective choice. And, from the preponderance of reports, if a non-Extreme owner can use it in an "A" to "B" configuration, it might be the best sounding option of all.
4. Best in my situation:
Optical Module/Startech SFP/-10dB attenuator, and OM/Planet Tech SFP/no attenuation --
The OM is perhaps the most transparent to, or maybe I should say the most dependent on the SFP it is paired with. With the Startech SFP even using attenuation, the OM can still be a little edgy. But it sounds big and bold and full of life, if just that little bit hard on top with less favored digital transfers. As a result, it sometimes sounds a little flat and a little forward when challenged with brighter recordings.
Paired with the Planet Tech SFP and without any attenuation, the OM retains all of the life and dynamics of music without any objectionable hardness on top, and with a subjective sense of more depth, dimension and space. It doesn't hide what recordings are, but it doesn't exacerbate their imperfections either. To my way of hearing, it strikes just the right balance in the context of my system and my tastes. I had one of those, “Ahhhhh, this is good!” moments when I started listening to the OM/Planet Tech SFP combination.
I wound up my tests by comparing streamed tracks to their twins residing in local on-board storage, and also tested the claim that with the Extreme and an optimized network, high-resolution (24/96, 24/192) streamed music can beat standard Redbook resolution (16/44.1) stored on-board the Extreme.
Listening to the same resolution streamed, compared to local playback, I thought local playback to be a bit better -- a little cleaner, a little darker and a little more depth. But it is very close and streaming certainly doesn't suffer for the comparison. I can listen to music via the Extreme and Roon Radio streaming all day long and never get tired or restless. Every so often a poorly recorded or digitally transferred track will come up, but that can happen regardless of medium.
Streaming higher resolution tracks and comparing them to their standard resolution, locally stored counterparts proved a little trickier. It's difficult to determine in many cases whether the mix was also tweaked when the source recording was remastered to higher resolution; in the case of MQA high resolution, MQA processing may be affecting the recording in ways that are sometimes enhancing and sometimes detracting (in my experience). However, over a dozen or so non-MQA high resolution Qobuz selections I found that in the majority of cases I did indeed prefer the high-res streamed version over standard resolution locally stored content.
Once again I'll just say that all of this is very much in the context of my own system, LAN and taste in music and musical presentation. And it appears that for whatever the reason, with an Extreme in the mix some of the unique attributes of the EtherRegen don't come out in the same way they do with other servers. The EtherRegen would be my first choice if I needed a switch in the feed to my Extreme because it is still better sounding than a Startech FMC, or a TLS OXCO switch. However for my needs and for what I value in sonic presentation the Optical Module with Planet Tech SFP has just the right balance.
Hopefully this was helpful, or at least interesting and worth the time spent reading.
I hope everyone is well and taking care of their loved ones in this challenging time.
Steve Z
Last edited:
maybe.......and maybe not.
my guess is that the more a competitor investigates the Extreme, the less likely they would commit the resources to attempt to beat it at it's own game. meaning to try and go head to head on the build and hardware side. it would be a large risk. i could see maybe attempts to end run the Extreme with software in some way. note i use the word attempts. and, of course, the Extreme can move forward with software too.
i do agree that perfect is not the right word. a better way to put it might be 'pre-optimized'. you can't blame computer audio tweakers for assuming they can improve anything they see. it's inherent in the species. and computer audio has reinvented itself every 6 months for a decade, so it's assumed that is still the case. only time will tell.
Last edited:
I think the likely scenario is that Emile ups himself. I am sure he is not "done."
