Taiko Audio SGM Extreme : the Crème de la Crème

[Taiko Audio]

Dear Emile & WBF Extreme NIC&Switch users:
Ben Lau of Volent Hong Kong told me today that some audiophile ethernet cables didn't work with the NIC & Extreme Switch, such as the ethernet cables of Siltech and Crystal Cable.

I understand that all Siltech and Crystal ethernet cables are 100M, except Da Vinci, the new flagship ethernet cable of Crystal Cable which has a spec of 1G (Ben hasn't tried the Da Vinci ethernet cable yet).
Is this the cause?

What's your experience?
Which brands/models of audiophile ethernet cables work & don't work with Extreme NIC and/or Extreme Switch?
Please share.

Thanks!

Hi CK,

Although reducing the network speed to 100Mbit is a strategy employed by some "Audiophile" switches with a school of thought that lower speed translates to lower noise we simply do not agree with this. Further to this many unexplained Roon problems, like sudden stops during playback are caused by 100Mbit network speeds and solved by moving up to 1Gb, with a vast decrease in error messages in Roon's logs. For XDMS the negative impact on SQ by running at 100Mbit is very obvious. Therefor we do not support 100Mbit networks, but we do support diagnosing and curing the cause of running a 100Mbit network.

 

[Taiko Audio]

12v on switch is minimum voltage for best SQ so will be interesting if the 7v minimal will also be best sounding on router ?

The issue here is the LPS. Both the Switch and Router are very power conversion efficient (~99%). That means that if the power consumption would be 12 watts (just as an easy example), the current draw would be 1A as voltage*current=power. If you use a 7V supply the current would go up to 1.7A (12W divided by 7V = 1.7A). Typically higher current = higher noise in a linear regulator, but there's more to it then that. A LPS is not a very power efficient supply. As a commercial manufacturer you would not typically use a different transformer for each different output voltage. Perhaps you'd use a 15V AC secondary transformer which supplies about 21V DC after rectification (+/- 10% to account for wall voltage variations) for 15V and 12V out after regulation, in a LPS the excess voltage is discarded, transformed into heat, increasing the current draw from your powergrid, and the current spikes for recharging the capacitor bank which is also a type of noise, the more inefficient it is (the more power is transformed to heat) the worse this gets. So it just depends on the LPS design which voltage will perform best. As 12V is a very common voltage you could assume those to be relative efficient designs, but in the above example with a 15V AC transformer secondary the 15V LPS could perform "better". Now if a manufacturer uses the same transformer for a 5V , 7V and 9V supply, the 9V would be most efficient. Therefor it's impossible to give you an exact recommendation without knowing the design specifics of the LPS as the Switch and Router really don't care much which voltage is supplied as long as it's in range.

 

[CKKeung]

Hi CK,

Although reducing the network speed to 100Mbit is a strategy employed by some "Audiophile" switches with a school of thought that lower speed translates to lower noise we simply do not agree with this. Further to this many unexplained Roon problems, like sudden stops during playback are caused by 100Mbit network speeds and solved by moving up to 1Gb, with a vast decrease in error messages in Roon's logs. For XDMS the negative impact on SQ by running at 100Mbit is very obvious. Therefor we do not support 100Mbit networks, but we do support diagnosing and curing the cause of running a 100Mbit network.

Thanks Emile!

I also don't like the approach of making switches & ethernet cables only 100M so as to "tune" the sound or "lower the noise".

Nowadays the CAS components are improving at lightspeed and with much lower background noise when compared to several years ago.

 

[Taiko Audio]

Thanks Emile!

I also don't like the approach of making switches & ethernet cables only 100M so as to "tune" the sound or "lower the noise".

Nowadays the CAS components are improving at lightspeed and with much lower background noise when compared to several years ago.

Yes, in fact if you look at it from a design / component perspective, 1Gb switches are much more modern / advanced / further developed, and can actually end up consuming less power overall then antiquated 100Mbit switch chips (I think I did mention the entire switch only uses 1.3 watts with DAC cables?). I did not have much success with 10Gb though, that just uses way more power, runs much hotter and is very difficult to get to sound right.

 

[CKKeung]

Yes, in fact if you look at it from a design / component perspective, 1Gb switches are much more modern / advanced / further developed, and can actually end up consuming less power overall then antiquated 100Mbit switch chips (I think I did mention the entire switch only uses 1.3 watts with DAC cables?). I did not have much success with 10Gb though, that just uses way more power, runs much hotter and is very difficult to get to sound right.

Agree!
During the recent 3-4 years, among all the CAS components I have auditioned, I always prefer the 1G connection if they have both 100M & 1G sockets.

 

[Taiko Audio]

Network card update:

The network card has successfully completed an overnight test run at 100% load at 75 Celcius. We are considering the problem as being solved. This does require both a firmware fix to "tell" the processor it's a 85C part and an adjustment to the voltage monitor to allow for more variance in the supplied voltages which begins to deviate above 67C. Although this is in fact a "non operating" temperature for the Extreme's motherboard and we could simply ignore it there is a degree of manufacturer pride involved which wants it to be "perfect".

Now I did disclose earlier we made an interesting discovery with regards to sound quality in an earlier stage while we were experimenting with heatsinks. Heatsinks do not actually solve anything beyond a few degrees more headroom on the processor as the ambient temperature dominates the issue, maxing out at about a ~5C gain, but one particular design does actually improve sound quality. Because of the inconvenience caused by this problem and, well because it sounds better, and we know you all love heavy stuff (it's about 1 lbs), we have decided to supply this as a free upgrade. It will take us a few days to CNC machine these and then we will ship out replacements to everyone including the new firmware, which actually turned out to be tricky to perform remotely.

A few photos of the prototype:

 

[John T]

Network card update:

The network card has successfully completed an overnight test run at 100% load at 75 Celcius. We are considering the problem as being solved. This does require both a firmware fix to "tell" the processor it's a 85C part and an adjustment to the voltage monitor to allow for more variance in the supplied voltages which begins to deviate above 67C. Although this is in fact a "non operating" temperature for the Extreme's motherboard and we could simply ignore it there is a degree of manufacturer pride involved which wants it to be "perfect".

Now I did disclose earlier we made an interesting discovery with regards to sound quality in an earlier stage while we were experimenting with heatsinks. Heatsinks do not actually solve anything beyond a few degrees more headroom on the processor as the ambient temperature dominates the issue, maxing out at about a ~5C gain, but one particular design does actually improve sound quality. Because of the inconvenience caused by this problem and, well because it sounds better, and we know you all love heavy stuff (it's about 1 lbs), we have decided to supply this as a free upgrade. It will take us a few days to CNC machine these and then we will ship out replacements to everyone including the new firmware, which actually turned out to be tricky to perform remotely.

A few photos of the prototype:

View attachment 106741

View attachment 106742

View attachment 106743

Very Cool...

 

[CKKeung]

Original :
View attachment 106746

New :

 

[MarkusBarkus]

...and that's a new board design, looks like, so no need for us to send you the old boards back? I am adding "copper futures" to my investment portfolio. Thanks for the helpful updates and *cool* solution Emile!

 

[Taiko Audio]

...and that's a new board design, looks like, so no need for us to send you the old boards back? I am adding "copper futures" to my investment portfolio. Thanks for the helpful updates and *cool* solution Emile!

It’s the same board, we’d like the old ones back. The photo above is from the 1st network card iteration which never made it to production (bus powered in stead of externally amongst others).

 

[steve williams]

Very Cool...

that's the word we are all looking for ......"cool"

Kudos Emile for once again finding the problem and quickly coming up with the cure. Looking forward to the new card and heat sinks

 

[Rhapsody]

As the price is not much different between priority and economy Fedex for this type of parcel it would be GREAT if the new cards can be sent out Priority. With US shipments it's almost a week difference in the shipping time.

 

[Taiko Audio]

As the price is not much different between priority and economy Fedex for this type of parcel it would be GREAT if the new cards can be sent out Priority. With US shipments it's almost a week difference in the shipping time.

We are removing economy entirely. Somehow the Fedex shipping manager plugin sometimes bugs up and ships economy in stead of priority, removing economy is an easy fix. We have to rely on automated systems as the volume is simply to high to enter every shipment manually.

 

[seatrope]

Seems like the bracket is part of the heat sink too! Nice.

 

[Ackcheng]

Hopefully this copper cover is painted so it won’t oxidized in the next couple of months.

 

[rando]

Original :
View attachment 106746

New :
View attachment 106745

Following up on earlier question.

What is your prognosis on a real world siting of Extreme mounted in a lower rack position? Test bench and test equipment ovens allow air movement that is not possible inside the design specific environment. Do you feel there will be any need for operational procedures (minimal) exceeding typical use during extended periods of 40C or greater ambient?

Heat soak does appear to be well handled here.

 

[John T]

Sometimes I seriously think we are insane! A little bit is a good thing, as long as you realize it. I had all inexpensive UTP Cables. Router to Spectrum Modem, Router to Switch, Switch to Extreme. I had a 1m Cat7 (JCAT) expensive shielded cable sitting around doing nothing. I inserted that between the Router and Modem. I think it added a bit of heft to the presentation. I don't even know if heft is the correct description? Fuller, but not over the top...

 

[Rhapsody]

Sometimes I seriously think we are insane! A little bit is a good thing, as long as you realize it. I had all inexpensive UTP Cables. Router to Spectrum Modem, Router to Switch, Switch to Extreme. I had a 1m Cat7 (JCAT) expensive shielded cable sitting around doing nothing. I inserted that between the Router and Modem. I think it added a bit of heft to the presentation. I don't even know if heft is the correct description? Fuller, but not over the top...

I was using a $1500 Sig Gold Cat RJ45 cable from the wall to my Switch. I replaced it with an inexpensive Supra Cat8 cable and I preferred it. ( A little bird sitting on my shoulder gave me the hint to try the normal RJ45 cable)

 

[RUIN*ish]

Network card update:

The network card has successfully completed an overnight test run at 100% load at 75 Celcius. We are considering the problem as being solved. This does require both a firmware fix to "tell" the processor it's a 85C part and an adjustment to the voltage monitor to allow for more variance in the supplied voltages which begins to deviate above 67C. Although this is in fact a "non operating" temperature for the Extreme's motherboard and we could simply ignore it there is a degree of manufacturer pride involved which wants it to be "perfect".

Now I did disclose earlier we made an interesting discovery with regards to sound quality in an earlier stage while we were experimenting with heatsinks. Heatsinks do not actually solve anything beyond a few degrees more headroom on the processor as the ambient temperature dominates the issue, maxing out at about a ~5C gain, but one particular design does actually improve sound quality. Because of the inconvenience caused by this problem and, well because it sounds better, and we know you all love heavy stuff (it's about 1 lbs), we have decided to supply this as a free upgrade. It will take us a few days to CNC machine these and then we will ship out replacements to everyone including the new firmware, which actually turned out to be tricky to perform remotely.

A few photos of the prototype:

View attachment 106741

View attachment 106742

View attachment 106743

Brilliant!!
Will this feature be standard moving forward?
Mn

 

[John T]

I was using a $1500 Sig Gold Cat RJ45 cable from the wall to my Switch. I replaced it with an inexpensive Supra Cat8 cable and I preferred it. ( A little bird sitting on my shoulder gave me the hint to try the normal RJ45 cable)

I'll put some hours on this approach, then go back to the UTP. I only have around 10 hours since yesterday...