Making a usb cable _ data only

[easp]

Buying any USB cables is willy-nilly, and none that I know of that are easily available give any performance figures whatsoever.

Sure they do. They are USB cables. The USB standard specifies cable characteristics.

 

[dmills]

Without the ground connection the USB will probably be very unreliable or not work at all. Some ground connection is needed to keep the common mode levels within the allowed range.

It is worse then that, during the initial phase of the enumeration the USB lines are use in a single ended manner, at the every start of the action, voltages relative to ground code for the speed of the client device, and IIRC there are a few other places it uses single ended signalling.

Cutting the ground will not work at all well.

The specs, including the physical layer details are readily available on USB.org - Welcome

Regards, Dan.

 

[Max Headroom]

Sure they do. They are USB cables. The USB standard specifies cable characteristics.

I said 'easily available'...that means Ebay etc.
Don't expect tight conformance from these sources...and btw most of the world uses these cheap type cables supplied as oem accessory.

One of my questions was does anybody know of a source of HQ USB cable....so far no sensible answers.

Dan.

 

[SY]

Amazon, Frye's...

 

[dmills]

90 Ohm twisted pair is not that uncommon a thing, this Item # W-3627-26, 90 Ohm USB Aerospace Cables On Whitmor/Wirenetics for example looks like it would be entirely suitable, but I would expect that the usual suspects (TE/Belden and the like) would have something suitable.

For short runs at lowish speeds, a tightly made AES cable (~110 ohms nominal) would probably be fine, as would using a few pairs from cat 5 (~100 ohms) both are close enough to 90 ohms to keep the return loss reasonable, that said USB cables are so cheap and so readily available, why bother?.

If the wire carrying a digital protocol that is inherently bursty is making a significant difference to your audio then something is very wrong, time to have a look at the eye pattern.

Regards, Dan.

 

[multisync]

if you need or want electrical isolation on your usb devices, there are several affordable galvanic usb isolators available on ebay. Your external usb device must have its own power source. I tried an inexpensive one and it does do what it says it does. The downside is a loss speed on the budget ones. I have used the one I bought up to 24 96 without any issues.

 

[vladimirb0b]

To the OP: You can find a "USB Injector" that will cut the 5V out, but probably not the GND. You would have to ask the manufacturer about the GND.

 

[DF96]

One of my questions was does anybody know of a source of HQ USB cable....so far no sensible answers.

You have already been given a very sensible answer: if the cable really is USB-compliant then it is good enough and can perfectly transfer data. There is nothing better than perfection so there is no such thing as a 'high quality' USB cable. The only two options are compliant and non-compliant - and USB is probably sufficiently robust that some non-compliant cables will work OK for short distances.

 

[vladimirb0b]

Devil's advocate; I saw THD+N increase with USB DAC withADuM isolator in the circuit... So there can be noise injected onto the USB that affects the signal.

Based on that, I would say that a shielded cable, and maybe even a ferrite, could help sound quality.

 

[rif]

A few observations

1) in the OP's first post, he says his usb soundcard can work with an external PS. So the soundcard should already take into account that there will be 5V on the usb connector in addition to the external PS. No special cable needed.

2) Buy a premade usb cable. In order to say usb compliant, or somesuch, it probably has to pass the standards detailed at usb.org (someone said it before, the best usb cable is a usb cable)

3) if you really want to make your own cable, dont try to reinvent the wheel with microphone or other cable. Buy some bulk Belden or other reputable usb cable. Trust that they have done the research and testing that you can't. Buy some connectors from digikey and spend some time making up a few. While i dont think it will make an audible difference, it may be enjoyable and fulfilling- this is diyaudio after all.

I may be confusing #2 with hdmi standards whose beanding is more rigorous?

 

[Max Headroom]

I couldn't find reference to USB cable on the Belden site.
And yes I know there are plenty of cable assemblies on eBay, Amazon etc.

Dan.

 

[ginetto61]

The best USB cable is... a USB cable.

Hi ! sorry !!!
I thought I had subscribed to the thread but actually I did not
However, Thanks a lot for the kind reply.
Yes of course I will be using the right connectors and golden ones.
But it is the cable in between that I should select.

If you willy-nilly substitute in something else, the USB receiver MAY be robust enough to still function, but I wouldn't count on it. Are you deliberately trying to screw up the performance?

I am looking for electrical isolation. I have seen a noise floor spectrum of a dac with and without isolation from the pc. The one with isolation was much cleaner and lower. So my guess is that electrical isolation in the worst scenario is not damaging and in the best is beneficial.
So actually I am looking for isolation from the pc.
But I read also the usb isolators can introduce issues.
My dac has an external power supply so it does not need usb bus power.
Thanks a lot again, gino

 

[ginetto61]

Hi to All !
Sorry for the belated reply but initially I saw no replies and thought of having subscribed to the thread but I had not.
Thanks to all your very kind and valuable reply. This is a first batch of some of them.

But surely mic cables are designed for carrying audio signals? That must be better than a USB cable meant for carrying just bits of data?

Hi ! yes this is my idea. A good quality mic cable with an impedance similar to the usb specifications, that I do not know by the way. In the end the two data wires look to me like a twisted pair.
Usually the quality of wires used is very very low. I have opened one.

I agree with SY: The best USB cable is... a USB cable. You need a balanced cable with a very specific impedance and good shielding. A good USB cable will give you that. Even a good microphone cable may not give you that.

I see. The real problem is selection of a good usb cable. But my other goal is to provide electrical isolation between dac and pc. So I would need only wire for data. Plus shield.

Buying any USB cables is willy-nilly, and none that I know of that are easily available give any performance figures whatsoever.
In my experimentation different generic USB cables can indeed cause subjective differences.
So just what is good USB performance and by what measures ?.
Dan.

I am interested as well.

Examples please, including bulk supply of HQ cable.
Dan.

Interested again. Thanks for any recommendation of suitable bulk cable. At least which Z is optimal.

Correct characteristic impedance would be a start (a mic cable is unlikely to provide this). Sufficiently low series resistance (a mic cable will do this, but this is probably less important than correct impedance). If a cable says it is USB-compliant and the maker is not telling lies then it is good enough for audio.

Thanks a lot. Problem is … which is the correct impedance for a usb cable ? what usb-compliant means in electrical terms ?

Yes mic cables are designed to carry audio, not data. Why do you think audio is easier to carry than data with ten times the bandwidth?

Interesting point. I am sure there are tests to check the quality of signal transfer. I have opened a usb cable and the quality looked very poor indeed. A decent mic cable is so much better than it is amazing.
Correct impedance could be indeed an issue.

He was making a joke. There are very few people on this forum who know more about signal transmission than DF96

Thanks for the valuable advice.

If you're really motivated, you could check out the official specs at usb.org

I am checking now. Thanks for the link.

"In my experimentation different generic USB cables can indeed cause subjective differences." As usual, you slip in the "subjective" word there! Thus you conveniently exclude yourself from having to measure, justify or explain anything. So all further discourse is pointless. Just your opinion. "So just what is good USB performance and by what measures ?"
It works! I can pull any cable from the spaghetti scramble in my cable box and it works: Audio, streamed TV, Gigs of data. In the past I have had a few USB cables that gave trouble. I decided they were faulty and binned them. Not worth the trouble to disect them.

Thanks a lot for the valuable advice. Maybe I should focus more electrical isolation. This is a must for me. I want complete electrical separation from the pc. It can be beneficial. And for sure it is not a bad thing to have.

Why not take an existing usb cable, cut it in two close to the end, strip back the conductors and only reattach the data ones? Perhaps you will need the ground since the usb data is differential.
Let's say you did this, would the computer even recognise the device? Would it be reliable?

Hi ! I indeed opened one and cut the power and ground wire and it worked.
But the data wires looked so poor. And so similar to a normal twisted pair like the ones found in mic cables. I will try to find the recommended impedance.

Conductors which are normally used for power should not be disconnected (or removed) as this may modify the characteristic impedance of the data conductors. Instead, the power conductors should be grounded or connected to whatever impedance best mimics the output of the power supply. I guess a power supply will be the best mimic of a power supply?

This is difficult to understand. I would start from the usb specs … what they say about the data wires impedance. That is the reference. And then try to find a mic cable with Z close to that value.

Without the ground connection the USB will probably be very unreliable or not work at all. Some ground connection is needed to keep the common mode levels within the allowed range. If noise is an issue it should be solved in a different way, e.g. with digital isolators on an I2S bus if this is available. Not by removing the ground (shield) of the USB cable.

Hi and thanks a lot. But I would really like to stay away from usb isolators as much as possible.
I am not comfortable with them. They can introduce issues. Maybe I am completely wrong on this point.

this seems interesting
https://www.adafruit.com/product/2107

Thanks for the link. But if possible I would prefer not to use them.

I have a similar'ish USB isolator powered from a car battery.
IME it does provide useful improvement in sound with tighter/cleaner positioning of sound sources (L/R and depth imaging) and cleaner sound overall..ie removal of film of 'haze' over the whole sound.
For best results use a USB type/gender adaptor directly between the isolator and the DAC.
Dan.

Hi and very interesting. Clearly the car battery has less noise than the pc usb power.
But I just need to transfer the usb signal, not the power. The interface has already its own psu.

90 Ohm twisted pair is not that uncommon a thing, this Item # W-3627-26, 90 Ohm USB Aerospace Cables On Whitmor/Wirenetics for example looks like it would be entirely suitable, but I would expect that the usual suspects (TE/Belden and the like) would have something suitable.

Thanks a lot indeed. So 90 Ohm is the recommended impedance. Good to know.
At this point yes … also a 90 ohm coaxial cable should be just perfect. Maybe they are good also for RF signals transfer. The Hz should be similar to usb frequency ?
For me it is important to have established that the main parameter is impedance. Good !

For short runs at lowish speeds, a tightly made AES cable (~110 ohms nominal) would probably be fine, as would using a few pairs from cat 5 (~100 ohms) both are close enough to 90 ohms to keep the return loss reasonable, .

I guess that as long as the Z is respected any good quality cable should be just great.
I see the cat 5 a little complex cable … I would prefer a twisted pair or a coaxial one.

that said USB cables are so cheap and so readily available, why bother?.
If the wire carrying a digital protocol that is inherently bursty is making a significant difference to your audio then something is very wrong, time to have a look at the eye pattern.
Regards, Dan

I see your point. A good cable from a renowned manufacturer should be more than enough.
Then I should open it and cut the power&ground wires to get the isolation.
Thanks a lot. I am still thinking.

Regards, gino

 

[ginetto61]

However, I am reading around that even usb certified cables could not be real 90 Ohm cables
I think that a mini microphone cable should be around 100-110 Ohm ?
Maybe a little Z mismatch will not be that detrimental for signal integrity ?
They are quite easy to work with as well.
What do you think ?
Thanks a lot again, gino

 

[DF96]

If USB needs 90 ohms then a real USB cable will be somewhere near 90 ohms. As it happens, if you twist together almost any pair of plastic insulated wires you will end up not too far from 90 ohms; perhaps a bit below 90.

I suspect the problem (the usual problem!) is that you are working beyond your understanding - and so seeking simple 'recipes'. The key thing about a USB cable is that it is a twisted pair with a particular characteristic impedance; you want to do a 'good' one yet you didn't know that impedance is the key issue, and my guess is that you are rather hazy on what is meant by characteristic impedance and how to design for it etc.

 

[SY]

DF, you're missing the ingenuity. He doesn't have a problem here and he desperately wants to have one- so the solution is to do something to actually cause the problem that he's worried about. See?

 

[AndrewT]

.......... so the solution is to do something to actually cause the problem that he's worried about. See?

No see !

 

[AndrewT]

I have to ask, because I have seen questions before:
If one uses a separate isolated supply to feed 5V power into a USB device, then do we also need to maintain a zero volts connection between the 2 usb devices?

Or will the data transfer across the 2 data wires with the power from the isolated PSU?

 

[ginetto61]

If USB needs 90 ohms then a real USB cable will be somewhere near 90 ohms.

Hi ! I said 90 ohm because I understand that this is a standard requirement.

As it happens, if you twist together almost any pair of plastic insulated wires you will end up not too far from 90 ohms; perhaps a bit below 90

Impedance between conductors, cables for digital aside, is not specified in the datasheets.
AES cables are 110 ohms and coax usually 75 and 50 ohms.
I cannot find impedance for mic cables. Maybe it is not an interesting parameter in their applications.
If it is close to 90 ohm I will buy some and try. The cost is small.

I suspect the problem (the usual problem!) is that you are working beyond your understanding - and so seeking simple 'recipes'.

actually I am trying to understand before working.
Honestly, I have seen many many boutique cables, very expensive ... but I also read extremely positive reviews. So I am wondering if a DIY project could be reasonable.

The key thing about a USB cable is that it is a twisted pair with a particular characteristic impedance; you want to do a 'good' one yet you didn't know that impedance is the key issue, and my guess is that you are rather hazy on what is meant by characteristic impedance and how to design for it etc.

it is exactly for this reason that I have asked ... because I do no know the requirements for a usb cable.
I understand that many people make their own rca and xlr cables.
So I wonder if also building usb cables could be a good way to get a very nice cable with limited amount of money.
The fact that usb connection is good only for 6 meters means to me that is not a very robust connection.
Ethernet is good up to 100 meters this must mean something.
But if I will not able to find a suitable bulk cable the project stops of course.
I will buy cables from a good brand and live with that.
Thanks a lot again, gino

 

[DF96]

I cannot find impedance for mic cables. Maybe it is not an interesting parameter in their applications.

Audio characteristic impedance is frequency dependent for a mic cable, and so will never be specified. Nevertheless a mic cable is likely to have a fairly well-defined RF characteristic impedance, but it is unlikely that this will be specified. It is possible that the anti-handling-noise layer will make a mic cable quite lossy for RF.

The fact that usb connection is good only for 6 meters means to me that is not a very robust connection.
Ethernet is good up to 100 meters this must mean something.

Not at all. USB was designed for short distances. Ethernet was designed for longer distances. Most of this is a matter of drivers, receivers and protocol timing windows; very little to do with the cable quality.

Buy a USB-compliant cable. Use it. If this is not good enough then don't use USB.