USB cable comparisons

[sandyk]

I tried the test on the SOtM dx-USB-HD running on battery and it apparently doesn't need any USB power. Can you wire just Pin 2 (Data -) and Pin 3 (Data +) to a USB shell and call it a day? Maybe use solid core silver (22awg) each sheathed in Teflon tubing, with both bound inside shrink tubing? You could dress out the cable however you wished.

 

You will still need either a shield or a 0 volts return wire. You should also try to ensure that the twisted together D+ ,and D- leads have the correct impedance. Without a shield, the cable should be kept as short as possible.

[Harpy]

As always, thanks for the information. You may have no way to know this, but there is a jumper pin on the PCB that allows you to choose between USB Ground or Chassis Ground. That wouldn't some how negate the need for shielding if set to Chassis Ground? Sorry if I seem lacking in my understanding.

[Harpy]

Disregard my above post it was dumb. If the USB cable has to have 90 ohms in the transmission line impedance it's not very likely you can buy some wire and twist it up and achieve the USB2 standard is it? How is it measured? Simple Ohm reading? Would hate to make a cable that was inferior to a 50 cent one.

[sandyk]

Disregard my above post it was dumb. If the USB cable has to have 90 ohms in the transmission line impedance it's not very likely you can buy some wire and twist it up and achieve the USB2 standard is it? How is it measured? Simple Ohm reading? Would hate to make a cable that was inferior to a 50 cent one.

 

Disregard my above post it was dumb. If the USB cable has to have 90 ohms in the transmission line impedance it's not very likely you can buy some wire and twist it up and achieve the USB2 standard is it? How is it measured? Simple Ohm reading? Would hate to make a cable that was inferior to a 50 cent one.

 

You are unlikely to be able to make a DIY USB cable of the correct impedance other than by sheer guesswork.

No, you can't use Ohms law for this. Twisted pairs from a cat.5E cable may be close enough in impedance ? (100 ohms)

Then you have the problem of adequately screening the twisted pairs, especially if they are more than a few 100 mm in length.

[fmak]

You are unlikely to be able to make a DIY USB cable of the correct impedance other than by sheer guesswork.

No, you can't use Ohms law for this. Twisted pairs from a cat.5E cable may be close enough in impedance ? (100 ohms)

Then you have the problem of adequately screening the twisted pairs, especially if they are more than a few 100 mm in length.

 

is available, there is the problem of being able to dress and solder correctly at the plug end to maintain

continuity in impedance.

 

I have only found one aerospace cable that is specified as 90R and only if I buy 100 meters!

[sandyk]

I have only found one aerospace cable that is specified as 90R and only if I buy 100 meters!

 

That's why I haven't gone there. It would be nice to know who the OEM is for some of the double barrelled type cables.

[Harpy]

Here is the cable I would like to try:

 

http://www.l-com.com/multimedia/eng_drawings/CSUZAB.pdf

 

They say it is aerospace - it has the LSZH outer jacket. They have bulk, but $25 to $108 per 100 feet depending on the wire quality. I think it's all USB certified too.

 

USB - Bulk Cable

 

The one listed is about $14 which is pricey for a short non audiophile USB cable.

[fmak]

Here is the cable I would like to try:

 

http://www.l-com.com/multimedia/eng_drawings/CSUZAB.pdf

 

They say it is aerospace - it has the LSZH outer jacket. They have bulk, but $25 to $108 per 100 feet depending on the wire quality. I think it's all USB certified too.

 

USB - Bulk Cable

 

The one listed is about $14 which is pricey for a short non audiophile USB cable.

 

But currently stuck with not knowing overseas postage cost. $14 is cheap compared to audiophile

cables.

 

I note that the power cables are not shielded separately and so the cable may perform better with

grounded power supply, cut at the input end and grounded at the receiving end.

[Harpy]

I thought this was a neat item too (I get on these little obsessive compulsive streaks, sorry):

http://www.l-com.com/multimedia/datasheets/DS_UH401.PDF

 

http://www.l-com.com/multimedia/manuals/M_UH401.PDF

 

They recommend different grounds from the to Pc to the component you are hooking up. My transport has a selectable ground be between USB or chassis. They max out 12 Mbps and not sure if that would be good enough.

 

I note that the power cables are not shielded separately and so the cable may perform better with

grounded power supply, cut at the input end and grounded at the receiving end.

 

They say they do custom cables, but I don't know what their minimum order would be. I on going to Email them and find out.

[YashN]

If the USB cable has to have 90 ohms in the transmission line impedance it's not very likely you can buy some wire and twist it up and achieve the USB2 standard is it? How is it measured? Simple Ohm reading? Would hate to make a cable that was inferior to a 50 cent one.

 

Ha, but you can easily take an existing compliant but cheap USB cable and hack it while preserving the compliance, which is what I did to great effect.

[YashN]

Then you have the problem of adequately screening the twisted pairs, especially if they are more than a few 100 mm in length.

 

Screening is not difficult at all: use adhesive aluminium tape of the length of the wires, get the screening off a satellite cable and then push it over the aluminium tape.

[YashN]

is available, there is the problem of being able to dress and solder correctly at the plug end to maintain continuity in impedance.

 

Strive to get the wire and contact at the plug end themselves in physical contact and then solder them rather than make contact solely through the solder and you should be good.

[fmak]

Strive to get the wire and contact at the plug end themselves in physical contact and then solder them rather than make contact solely through the solder and you should be good.

 

you make out. It's to do with how wires overlap and how you dress when they are

at opposite ends diagonally over the connector. Nothing to do with solder.

[Harpy]

Ha, but you can easily take an existing compliant but cheap USB cable and hack it while preserving the compliance, which is what I did to great effect.

 

What did you hack on the cable? I've seen some where battery or linear power was added to the power lines for devices that need the USB power. For devices that do not need USB power I've seen people cut the 5 vdc wire on a standard cable, or cables that use just D+ D- with 0 volt return line with no shielding. Just trying to get an idea what's all out there and your thoughts.

[YashN]

What did you hack on the cable?

 

I had done the scotch tape experiment with my previous DAC, so this time I wanted to hack a DIY USB Cable for my new DAC (iFI iDSD Nano). I took design inspiration from the LS and Gemini, i.e. focusing on separating the power leads from the signal leads. My DAC cannot do without the power leads (tested this with a switch on +5V), so I ended up keeping them in, except I removed them from the original sheath.

 

I then separately twisted them, double-shielded them and re-soldered them.

 

The difference from the original iFi iDSD cable is astounding.

 

There's room for improvement on the USB jack shielding though (both sides).

 

It's a surprisingly easy thing to try.

 

Of course, if your system is impervious to the USB-related issues as John described them, you may not hear any difference. In my system, they were obvious to us.

[YashN]

you make out. It's to do with how wires overlap and how you dress when they are

at opposite ends diagonally over the connector. Nothing to do with solder.

 

No such issue if you re-use an existing, compliant cables.

 

You do realise, I hope, that if instead of the wire making contact directly with the jack contact or with another piece of the same wire if you cut them off, contact could be made via the solder material and that wouldn't be too good for impedance matching and hence, continuity.

[fmak]

'existing, compliant cables'

 

who makes them; there are only claims, no measurements I can find?

 

HiFiNews measures ete patterns for there tests. Wide range of finding!

[Harpy]

who makes them; there are only claims, no measurements I can find?

 

http://media.digikey.com/pdf/Data%20Sheets/Phoenix%20Contact%20PDFs/1654853.pdf

 

There you go $61 for 1 meter none audiophile cable.

[fmak]

It says: cable type Firewire! No impedance spec either.

[Harpy]

The Wave impedance on the signal line is 100 ohms (+ - 5%). The USB spec is 90 ohms (+ - 15%). No clue on the on the FireWire cable unless IEEE 1394a 4 pin FireWire cable has the same characteristics of USB 2.0 cable.

[bogi]

I am planning to make my own USB cable, length about 40 cm. It will be Y type cable with separate branches for data and power. I cannot find pure 90 Ohm cable without connectors, so probably I have to buy some USB cable and remove connectors from it. Do you have some tip?

 

How important is exact impedance match? For example this is 110 Ohm AES/EBU cable: DH Labs Silver Sonic D 110 AES EBU Cable Bulk per Foot DIY Digital Cable | eBay

[pooger]

For a cable that short, the impedance will probably be more affected by the terminations than the wire itself. The 110 AES cable would work fine I would think. There are most likely 2 conductors and the shield in the cable; you could experiment with only attaching the shield at one end of each leg of the cable (traditionally it was the sending end, but since you have 2 sending ends, and they may have slightly different ground potentials, it could be more complicated. And, depending on if the USB input in your DAC had the shield hooked up or not.) Bare a very short amount of wire at the terminations, 1/8th inch or so, and solder it so it is right on top of the "land" on the connector, rather than at an angle. (This will help with keeping the impedance correct.)

 

Mouser has the A and B plugs; I would really recommend starting with a clean plug rather than rescuing one from an existing cable. Again, keeping the impedance as correct as possible at the connections is at least as important as in the wire.

 

Report back with your impressions. I've made several ~1 meter cables so far, using mainly mil spec silver plated copper in teflon, and have been pleased with the results. (Have also tried bonded cat6 twisted pairs, which worked well too.) My DAC doesn't need the +5 from my computer, so I just run the two datas and a ground as shield.

[fmak]

I am planning to make my own USB cable, length about 40 cm. It will be Y type cable with separate branches for data and power. I cannot find pure 90 Ohm cable without connectors, so probably I have to buy some USB cable and remove connectors from it. Do you have some tip?

 

How important is exact impedance match? For example this is 110 Ohm AES/EBU cable: DH Labs Silver Sonic D 110 AES EBU Cable Bulk per Foot DIY Digital Cable | eBay

 

from Partsconnexion or from eBay Hong Kong for gold plated copper but expensive at $20 a pair.

 

I have used them and liked them

 

You can also consider buying 100R bnc cable and using these, with the advantage that they are each shieldes and that you can add another on top.

 

Impedance is important

[fmak]

I am planning to make my own USB cable, length about 40 cm. It will be Y type cable with separate branches for data and power. I cannot find pure 90 Ohm cable without connectors, so probably I have to buy some USB cable and remove connectors from it. Do you have some tip?

 

How important is exact impedance match?

 

93 Ohm and 12mm in diameter with both Data and Power Runs, separated.

[bogi]

That RG62 cable has proper impedance, but it is a simple coax with 1 shielded wire only and I need to use two wires data+, data- expect of ground. You suggested "you can add another on top", but which one and how about the impedance match?

 

from Partsconnexion or from eBay Hong Kong for gold plated copper but expensive at $20 a pair.

Can you please post a link?