USB optical cable : stunning IF powered & implemented correctly

[cat6man]

11 minutes ago, Superdad said:

The SlimRun cable uses a switching DC-DC boost converter at the upstream end to convert 5V (either just from the computer VBUS or from an external 5V supply you might attach) to 12V to drive the long pair of thin copper wires that run alongside the optical cable.  In the downstream end there is another DC-DC switching converter (buck) that turns that 12V back into 5V—for both the VBUS and for the USB3>USB2 translator chip (which has its own internal 5V to 3.3/1.1V regulators).

Nothing “optimized for audio” here...

 

yuck!

thanks for the explanation.

 

so an alternative to diy-ing surface mount scale components (not a forte of mine as i've learned to limit myself to tube-sized diy projects) could be that an iso-regen at the DAC end with a good PS might be a useful thing to try.

fortunately, i already have one to try. :)

 

anyone here already tried an iso-regen terminating the slimrun cable?  links?

 

[matthias]

2 hours ago, cat6man said:

anyone here already tried an iso-regen terminating the slimrun cable?  links?

 

@CJH 

Two posts in this thread

 

Matt

[tgb]

On 11/16/2020 at 1:43 PM, hopkins said:

Hi,

 

When you say "USB optical cable enables a new hifi paradigm", what exactly have you compared this to in order to reach that conclusion ?

 

thanks

Hi Hopkins,

I pointed out "new paradigm" because you can have a nice SQ with a do-everything PC as a source. A dedicated audio server becomes useless.

My background :

- tested NAS as music server (LPS on NAS > optical network to renderer) : worse

- I moved from NAS to audio server (NUC / fanless / LPS ; Win10 with optimized processes or Daphile as OS) : worse.

 

Cost of optical cable + LT3045 based PS = 350Eur

Standard PC as a source = no extra-cost / versatility / multimédia source

Then, see signature, I put some cash in the USB-spdif interface.

Optical cable + optimized interface => here is the good SQ.

This way : you "hear" the cash you spent.

 

I'm not saying that the current trend to build an audio server is useless.

I'm saying :

- it's expensive

- technically it is not "smart" in a sense that it's way more efficient & cheaper to clean the noise of a basic PC than building an audio server that generate a very low noise. Technically not smart because the noise is not "encrypted" in the data stream. Thus, optical isolation & reclocking (fine reclocking) after any basic server is fine enough to create a clean spdif stream.

- optical usb cable is an exemple , Grimm MU1 & Innuos Phoenix are other examples that show that "good cleaning" after a server can be dead efficient, much more cost efficient than building a dead silent audio server.

Rgds

[tgb]

On 11/16/2020 at 4:38 PM, juanitox said:

 

 

Christmas time is here ...  😀

You're right juanitox !

 

I posted that photo on purpose, to show that a "audio" server can be a basic laptop, parked on a basic shelf as workstation when not out, in the room next door (or can be in the ceiling).

It looks strange because is not a "nice audio gear", but it can streams audio thru USB... so why bother ? what else !?...

Rgds

[tgb]

On 11/16/2020 at 5:33 PM, cat6man said:

 

merci for the link to the other forum.

i'm confused by the 12v reference.  my slimrun cable inserts 5v at the upstream connector, so i don't understand the 12v to 5v buck converter issue?  is this not the slimrun cable you are working with on the other forum?

Hi cat6man,

these cables are USB cable, thus they use the 5VDC given by the computer.

In the pc-side connector, the 5V is used to convert the data stream from copper to fiber.

Then, in the upstream connector, you need to do the opposite : fiber to copper. To do that, you need power. Then, you have 2 options :

1. the simpliest : like the Fibbr Alpha USB optical cable : the upstream connector has a plug to supply power (5VDC)

2. the "most complexe" but "most convenient" for any end-user : you use the power given on PC-side. You can't use 5VDC because the cable is too long, thus if you supply 5VDC from the pc-side plug, you'll end up with 2-3?VDC at the upstream plug : not enough to run the fiber>copper conversion. So, the trick is to send from the pc-side plug a "high" voltage current to be able to have at the end connector "still" this high voltage, and then be able to convert fiber>copper.

 

I'll take time to post the mod I failed on the Fibbr, that may be interesting.

[tgb]

On 11/16/2020 at 6:02 PM, Superdad said:

Nothing “optimized for audio” here...

Of course :-)

But SQ is there to show that this technology is fine for audio.

 

You have 2 things there :

1. the optical isolation.

2. the conversion process fiber>copper

 

On paper you may think : the benefit of 1. may be crashed by 2. (due to noisy regs inside)

In practice : no.

=> 

The benefit of 1. is huge (copper based stuff like galvanic isolation can't compete)

The 2. process is not so noisy... much less than we can expect.

Quite surprising but finally SQ is very good.

 

I should receive some Corning by end of this week. After my fail in tweaking the Fibbr, I'll have a try on a Corning the cut the boost & buck converters & power the upstream connector with external 5 & 3.3V. We'll see if this mod is relevant or not.

Rgds

[cat6man]

i have an hdplex 300 providing the 5v at the pc side.

 

this weekend, i'll add an iso-regen powered by a paul hynes LPS.

[Qhwoeprktiyns]

12 hours ago, tgb said:

Hi Hopkins,

I pointed out "new paradigm" because you can have a nice SQ with a do-everything PC as a source. A dedicated audio server becomes useless.

My background :

- tested NAS as music server (LPS on NAS > optical network to renderer) : worse

- I moved from NAS to audio server (NUC / fanless / LPS ; Win10 with optimized processes or Daphile as OS) : worse.

 

Cost of optical cable + LT3045 based PS = 350Eur

Standard PC as a source = no extra-cost / versatility / multimédia source

Then, see signature, I put some cash in the USB-spdif interface.

Optical cable + optimized interface => here is the good SQ.

This way : you "hear" the cash you spent.

 

I'm not saying that the current trend to build an audio server is useless.

I'm saying :

- it's expensive

- technically it is not "smart" in a sense that it's way more efficient & cheaper to clean the noise of a basic PC than building an audio server that generate a very low noise. Technically not smart because the noise is not "encrypted" in the data stream. Thus, optical isolation & reclocking (fine reclocking) after any basic server is fine enough to create a clean spdif stream.

- optical usb cable is an exemple , Grimm MU1 & Innuos Phoenix are other examples that show that "good cleaning" after a server can be dead efficient, much more cost efficient than building a dead silent audio server.

Rgds

 

To call this a new "hifi paradigm" seems to me exaggerated, and unsubstantiated. 

 

Unfortunately, your tests will never let you realize this because you are only focusing on one aspect of the digital chain: the quality of the USB signal. There is a lot more going on after: conversion of the USB to spdif (in your case), then handling of the spdif signal in the DAC. 

 

Simply "tweaking" a Singxer to put in better power and clocks is unfortunately not sufficient. An XMOS chip will itself generate a lot of noise, irrespective of how clean the power you supply it with. This noise will spread, disrupt the clocks, etc... 

 

[tgb]

hi hopkins,

of course "new paradigm" is too much & closer to a joke than a complete break thru in hifi world 😀

 

I focus only "on one aspect of the digital chain: the quality of the USB signal" :

you talk about it like minor stuff. Why not, it is partial stuff.

But I can use a basic laptop as a server : is it minor stuff ? minor stuff to be able to put money on the usb>spdif conversion because putting $$ on server side is not relavant anymore ?

just a question...

 

There is a lot more going on after: conversion of the USB to spdif (in your case), then handling of the spdif signal in the DAC. 

yes of course. Each step can be improved, as always.

For instance in my case : the NewClassD NeutronStar2 clock on the spdif part. To see if this clock brings something, that's easy : transcoding at server-side at 24/192 to use it, or transcoding at 24/176 to use the Crystek still in place.

- before : corning powered by power bank : diff between the NeutronStar2 & Crytek was obvious but... not huge or night/day.

- after : corning powerd with good PS : diff between the 2 clocks is dead clear.

As always, a modd somewhere in the chain improve "what is downstream this modd". And the usb optical cable is in that mods category.

Same effect when I change the PS on the USB clock : improvement, good point. But without Corning upstream, the improvment would have been much much lower. Same as NeutronStar2.

 

Simply "tweaking" a Singxer to put in better power and clocks is unfortunately not sufficient. An XMOS chip will itself generate a lot of noise, irrespective of how clean the power you supply it with. This noise will spread, disrupt the clocks, etc... 

It's on my todo list. 1VDC required for the xmos chip.

SolidCore gave me the idea => his posts is there => https://audiophilestyle.com/forums/topic/29553-my-very-mini-review-of-the-singxer-su-1-ddc/?do=findComment&comment=744476

 

Please note the SolidCore mentions "improvements" (good point), but not night/day change as you think because this bad 1VDC would pollute everything at miles around it.

I exaggerate a bit, but you too ;-)

Rgds

[Qhwoeprktiyns]

What you don't seem to understand is that irrespective of its power supply, computer components (processors, like the XMOS) generate noise! In other words, the noise does not only come from power supplies. 

 

Imagine your refrigirator is making a lot of noise, and you call a repairman, who suggest installing a power filter to supply "cleaner" power... That makes just as little sense as trying to fix the problems of digital audio (which are quite complex) just by adressing it by tweaking power supplies! 

What you are doing may provide some very marginal benefits, but the effort involved is probably not worth it. 

 

Instead of wasting your time with this type of pointless tweaks, you should look into what serious/professional engineers are doing to address all these issues. 

 

However, good luck :) 

 

 

 

 

 

 

 

 

 

 

[fds]

If the computer is a laptop running on battery, I do not see how it will pollute the mains. Also I am not sure how much of the noise will travel down the USB cord, in particular, if this is made of fiber. What will remain though is some computer-caused jitter that will travel down the USB line towards the DAC. However, now with a good well-powered reclocker such as an IsoRegen on a good LPSU or a Phoenix, that jitter will be reduced significantly. Thus, I think @tgb has a very promising approach there ... In fact, I do find the term "new paradigm" quite suitable. 

[Qhwoeprktiyns]

52 minutes ago, fds said:

If the computer is a laptop running on battery, I do not see how it will pollute the mains. Also I am not sure how much of the noise will travel down the USB cord, in particular, if this is made of fiber. What will remain though is some computer-caused jitter that will travel down the USB line towards the DAC. However, now with a good well-powered reclocker such as an IsoRegen on a good LPSU or a Phoenix, that jitter will be reduced significantly. Thus, I think @tgb has a very promising approach there ... In fact, I do find the term "new paradigm" quite suitable. 

 

Heads-up: its all been tried before...

[tgb]

Hi hopkins,

you're right... I should ask an engineer in electronics. But none around, thus I took the 1st engineer around, myself ;-) Not in electronics but in chemical engineering. That engineer knows a bit more in chemistry / ind. processes / fluids mechanics than electronics... but he was available right away thus I said to him  : go for it ! :-)

 

2 hours ago, hopkins said:

What you don't seem to understand is that irrespective of its power supply, computer components (processors, like the XMOS) generate noise! In other words, the noise does not only come from power supplies.

What else ? Anybody knows that. Nothing new there.

so, what next ? if I follow this remark, any DAC with an xmos chip is bad stuff (because this chip kills the SQ). Why not, but there are quite numerous DAC with xmos chips on the market , for years, and quite numerous people aere happy with the SQ

2 hours ago, hopkins said:

What you are doing may provide some very marginal benefits, but the effort involved is probably not worth it. 

If there was a marginal benefit, I would not take time to post about a loosy tweak. No sense.

 

2 hours ago, hopkins said:

you should look into what serious/professional engineers are doing to address all these issues.

Thanks for the joke :-)

Honestly, are you sure we can call "serious engineer" (in electronics) that put a basic regulator to get 3.3V to power a clock ?

A clock !!!!???

That's (maybe) the most important component in any digital device !

Rgds

[tgb]

Hi,

FYI, I've posted in the French forum the mod on the Fibbr that I failed a fwe days ago => http://forum-hifi.fr/thread-18535-post-428069.html#pid428069

(no slang / idoms : Google Translate should work ;-) )

 

@fds thanks for you reply. Sorry, no time, I'll reply to your post tonit.

Rgds

[Qhwoeprktiyns]

33 minutes ago, tgb said:

Hi hopkins,

you're right... I should ask an engineer in electronics. But none around, thus I took the 1st engineer around, myself ;-) Not in electronics but in chemical engineering. That engineer knows a bit more in chemistry / ind. processes / fluids mechanics than electronics... but he was available right away thus I said to him  : go for it ! :-)

 

What else ? Anybody knows that. Nothing new there.

so, what next ? if I follow this remark, any DAC with an xmos chip is bad stuff (because this chip kills the SQ). Why not, but there are quite numerous DAC with xmos chips on the market , for years, and quite numerous people aere happy with the SQ

If there was a marginal benefit, I would not take time to post about a loosy tweak. No sense.

 

Thanks for the joke :-)

Honestly, are you sure we can call "serious engineer" (in electronics) that put a basic regulator to get 3.3V to power a clock ?

A clock !!!!???

That's (maybe) the most important component in any digital device !

Rgds

 

Yes, unfortunately it seems that the XMOS chip does kill SQ if used as is. The noise it generates is measureable, and you may find it surprising, but some engineers (not chemical engineers) have actually measured it.  Moreover, outputting I2S signal from XMOS, with high bandwidth, just carries the noise downstream. You can put the best clock behind an XMOS, with the best power supply, it will not prevent noise from spreading from the xmos to the clock. 

 

Unfortunately digital audio cannot be solved by the type of amateurish endeavors you are pursuing... 

 

What is the remedy? You'll soon find out (though not by reading the "french forum") :) 

 

Until then, have fun with your tweaks! 

[tgb]

Hi hopkins,

Thanks for calling me an "amateurish" DIYer. I see it as a compliment ;-)

The  "amateurish" DIYer goes thru trial&error, he's learning, he's making assumptions that don't last for ever, he knows that he knows very few things, no brainer for him to go one way & turn around if it's finaly a wrong way...

I like that sticker on my head :-)

 

Regarding the noise generated by the xmos chip :

- I posted info above : I know, it's noisy...

- I'll power it nicely when I have time

 

In the meantime, I have a modded SU-1 that give a SQ that is miles away from the stock version

When I modd the PS on the xmos, SQ will improve. 

Fine with that.

 

Honestly, that's a bit strong to focus on xmos noise as the "SQ killer", given that no product on the market uses 1 or 2 steps linear regulation to power perfectly that chip... (I don't know about boxes >10k$... )

Rgds

[Qhwoeprktiyns]

1 hour ago, tgb said:

In the meantime, I have a modded SU-1 that give a SQ that is miles away from the stock version

When I modd the PS on the xmos, SQ will improve. 

Fine with that.

 

Honestly, that's a bit strong to focus on xmos noise as the "SQ killer", given that no product on the market uses 1 or 2 steps linear regulation to power perfectly that chip... (I don't know about boxes >10k$... )

Rgds

 

I have really big doubts on the fact that SQ can significantly improve by simply changing the power supply on a chip. 

 

As I mentioned earlier, a chip, any chip, will produce noise, regardless of how clean the power supply is. It is exactly the same thing as believing that you can get high fidelity from a computer by simply changing the power supply. 

 

If you think you are getting phenomenal results by doing these things, fine, there is no harm in trying :) 

[ASRMichael]

1 hour ago, hopkins said:

 

I have really big doubts on the fact that SQ can significantly improve by simply changing the power supply on a chip. 

 

As I mentioned earlier, a chip, any chip, will produce noise, regardless of how clean the power supply is. It is exactly the same thing as believing that you can get high fidelity from a computer by simply changing the power supply. 

 

If you think you are getting phenomenal results by doing these things, fine, there is no harm in trying :) 

is exactly the same thing as believing that you can get high fidelity from a computer by simply changing the power supply.
 

YOU CAN! Add something along the lines of Sean Jacobs LPS! 

[Zauurx]

Just try, especially if the audio programs are a little greedy.

[Qhwoeprktiyns]

Lol, its a French invasion 😁