USB Cables....

I think we all know by now what this section is for.
Neal
Shed dweller
Posts: 2299
Joined: Fri Aug 03, 2007 10:57 am
Location: From the land of the Bodgers

#1 USB Cables....

Post by Neal »

Something that's bugged me for a while is why should USB cables sound different, after all USB signaling is differential and any competent cable should sound the same as any other. The question seems to divide peoples opinion and starts wars on forums!....I've always been of the opinion that if the cable meets the USB spec then it should have no influence on the sound....but now I'm not so sure. During my own experiments with USB recently I've noticed that no cable IE: using a small gender converter sounds better than a 2m USB 2.0 approved Belkin cable and there are many reports on the internet of people finding the same.

Trying to find out why this might be is not so easy but I came across this document that may give some plausible reasons as to why. I hope I interpreted correctly....The first point that stood out to me was that its incorrect to think of the signaling as purely differential as the signaling lines also have to work in common mode. The characteristic impedance is different for differential mode and common mode. Getting the impedance correct for differential mode and at the same time setting it for common mode is not that easy and is further complicated by the use of a noisy power rail and ground within the same cable sheath.

As the signal pair twist runs down the length of the cable the impedance varies as the rotation angle of the twist changes in relationship to the parallel run power and ground wires, this unbalances the differential signaling. Also the specification of the twist per inch is very loose for USB v2.0 varying from between one twist per 6cm to one per 8cm....although this article says 3cm is found in a typical cable. So seemingly compliant same type cables will have very different characteristic impedances. Impedance is also going to be affected by the end termination and connecter, I wonder how `matched' many connectors really are...the impedance also varies with frequency complicating it further.

The ideal differential impedance is 90R and for common mode 45R but this has a 10% tolerance associated with it so there's some variation. The report goes on to say many measured and modeled cables are actually 85R and 35R impedance so we have a mismatch in the cables and most likely a mismatch in the end node connectors and termination which will have to have been engineered with the ideal parameters in mind....all this is likely to cause signal reflections in the cable IE: the cable will have a poor return signal loss ratio.

There's some other talk about propagation velocity and loss both of these are in the USB cable spec. but these losses are seemingly negligible in reality leaving the impedance mismatch as the most probable cause of differences in cables.

This may not be the main reason but I think its helped and certainly shows (to me anyway) that USB is quite a loose specification and not the seemingly fit and forget interface many would believe.

https://www.ee.washington.edu/techsite/ ... 1-0009.pdf
User avatar
Nick
Site Admin
Posts: 15707
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#2

Post by Nick »

Interesting stuff Neal, but I have to ask how what is basically layer 0 effects can effect the data that comes out of the endpoint in DMA blocks?

The connection between a spdiff signal and the recovered i2s data stream makes it a lot simpler to see how those effects could create effects down the line.

There is so much we dont know :-(
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
Neal
Shed dweller
Posts: 2299
Joined: Fri Aug 03, 2007 10:57 am
Location: From the land of the Bodgers

#3

Post by Neal »

Don't ask awkward question Nick! :) I wish I knew...
Andrew
Eternally single
Posts: 4206
Joined: Thu May 24, 2007 2:18 pm

#4

Post by Andrew »

How about our old friend jitter?
Analogue, the lost world that lies between 0 and 1.
User avatar
Nick
Site Admin
Posts: 15707
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#5

Post by Nick »

Andrew wrote:How about our old friend jitter?
Yep, that makes sense for spdif as the receiver may recover the bclk clock from the input signal, but thats not the case for USB, especially async.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
Clive
Old Hand
Posts: 374
Joined: Tue May 22, 2007 9:17 am
Location: Cheshire, England

#6

Post by Clive »

In addition to very short connections I've found that a cable with no +5V wire sounds better (usually the ground is needed). You need a 2nd USB for the +5V and sometimes a ground, this comes together at the type B plug.
David
User
Posts: 123
Joined: Wed Jan 05, 2011 3:12 pm

#7

Post by David »

Hi Guys,
I am reporting here on a small experiment I did over the weekend and the results. I read Clive's review of the JK Dac which I have, and how it sounded better run through a USB A to B adaptor rather than through a lead.

I can't actually fit the adaptor on my Mac Mini, which dos all the music, so was unable to try the adaptor solution.

So Clive in your review you then compared a Kimber AG silver USB cable and said (I paraphrase) that the Kimber was as good or even better for overall sound.

So I then hashed up a USB lead with solid silver connectors by doing the following. I got a cheap gold plated USB lead and carefully took off the insulation and cut into the two data cables inside as close to the USB plugs at either end as I could. I then tried it in my Mac Mini. Nothing. I then tried it with my other DAC an Arcam R Dac, and it worked very nicely with a very noticeable increase in sound quality.

I then tried the same lead in my Macbook Pro Laptop. The lead worked with the JK dac in one of the two USB ports, and again gave a very noticeable improvement in quality.

So without going into why a silver cable makes digital sound better what's going on here?

How come my hashed up cable works with the Arcam Dac but not the JK?

How come the hashed up cable works with the JK on my Macbook pro, but only using one of the two USB sockets?

Needless to say the electrical connections measure perfectly.
David
Clive
Old Hand
Posts: 374
Joined: Tue May 22, 2007 9:17 am
Location: Cheshire, England

#8

Post by Clive »

Hi David.

Since the JK review I've made very short USB cables using soledrable plugs Maplin sell. I've also tried Elijha Audio cables for the JK. All these cables omit the +5V wire as the JK dac can be charged via a separate cable. Overall I now have better results with these short cables than with the Kimber USB cable. Silver does seem to sound better better but any short cable without the +5V wire is better than the silver Kimber.

As for why your cable doesn't consistently work I can't say; something to do with the ground connection? The JK dac doesn't have a +5V connection in the type B socket but that shouldn't matter.

Here's my cable with the Maplin connectors:
Image
User avatar
Ali Tait
Eternally single
Posts: 4373
Joined: Fri Jun 08, 2007 8:10 pm
Location: Galashiels

#9

Post by Ali Tait »

At a previous get together at my place, we tried four different USB leads, and all agreed there were obvious differences between them. The TQ cable was easily the best, but also by far the most expensive. Leads were used from a laptop to a V Link.
User avatar
Ali Tait
Eternally single
Posts: 4373
Joined: Fri Jun 08, 2007 8:10 pm
Location: Galashiels

#10

Post by Ali Tait »

I also tried the TQ between my Drobo and Droboshare, which is cabled to the router, then cabled to the TFS via Ethernet. Made no difference we could detect.
David
User
Posts: 123
Joined: Wed Jan 05, 2011 3:12 pm

#11

Post by David »

Thanks for the replies.Those tips are very useful, Clive! guess who's off to Maplin tomorrow! So far, I have kept the 5v+ in too so there might be a bit more to come from that direction. Sounds like I shouldn't bother with the Kimber too!
Neal
Shed dweller
Posts: 2299
Joined: Fri Aug 03, 2007 10:57 am
Location: From the land of the Bodgers

#12

Post by Neal »

Device detection, the simple way anyhow, is for the host to look for either D+ or D- being pulled to 3v3 when something is plugged in...if the 3v3 is derived from Vbus and you've cut that line then its not going to be found...

I would also suggest that perhaps silver cable has nothing to do with any change in sound but more to do with the completely different geometry of your DIY cable. :wink:
Clive
Old Hand
Posts: 374
Joined: Tue May 22, 2007 9:17 am
Location: Cheshire, England

#13

Post by Clive »

Neal wrote: I would also suggest that perhaps silver cable has nothing to do with any change in sound but more to do with the completely different geometry of your DIY cable. :wink:
I did try copper and silver in the same config but th differences were small. I don't claim any magic due to silver.
Clive
Old Hand
Posts: 374
Joined: Tue May 22, 2007 9:17 am
Location: Cheshire, England

#14

Post by Clive »

Neal wrote:Device detection, the simple way anyhow, is for the host to look for either D+ or D- being pulled to 3v3 when something is plugged in...if the 3v3 is derived from Vbus and you've cut that line then its not going to be found...
Not only cut that line, the connection doesn't exist in the type b on the jkdac32....
Neal
Shed dweller
Posts: 2299
Joined: Fri Aug 03, 2007 10:57 am
Location: From the land of the Bodgers

#15

Post by Neal »

Clive wrote:
Neal wrote:Device detection, the simple way anyhow, is for the host to look for either D+ or D- being pulled to 3v3 when something is plugged in...if the 3v3 is derived from Vbus and you've cut that line then its not going to be found...
Not only cut that line, the connection doesn't exist in the type b on the jkdac32....
Interesting, theres something called back power that can cause issues with some host systems when a self powered device is on but the host is off. The device is supposed to check Vbus to see if the host is on and if not disconnect the D lines. Back power charges the host via the D lines and reportedly can cause device boot or startup failure. I wonder if the JDAC simply goes to sleep after a while if no data is sent.
Post Reply