DIY USB A - B cable
May 27, 2013 at 6:49 PM Thread Starter Post #1 of 16

WALL-E

100+ Head-Fier
Joined
Dec 12, 2011
Posts
202
Likes
20
I didn't see any similar thread to this it would be very useful to have ours projects well documented step by step, why pay so much for "know-how"?
This thread is dedicated for those interested in building their own cables.
 
This will start with my latest project - USB A - B cable.
I needed USB cable which meet the USB 2.0 spec over longer runs, as we know the maximum cable length is 5m I'll be back to this later.
I always wondered why not use Ethernet cable as they are special designed to transmit digital data such as a USB cable through the current loop and it is also adapted for transporting the information at a much greater distance than the USB. So I chose the cable category (Fa) cat7 actually Cat7+is improved version of 7a, made by Swedish Supra Cables.
 
Supra Cat7+ spec.
 

 
The USB 2.0 standard defines a High Speed transfer rate of 480Mb/s.The transmission is done from a transmitter to a receiver through differential data line, to limit data reflection, the following must match:
- Transmitter output impedance
- Line characteristic impedance
- Receiver input impedance
The USB 2.0 standard defines a cable with 90Ω ± 10-15% differential impedance.The cable may have no more than 30 ns signal propagation delay on entire length.Cable is 4 wire.The data wires are 28 AWG twisted pair, the power wires are 20 to 28 AWG.The cable shielding must cover at least 65% overall.
 
The Supra cat7+ have everything what I need to build my own USB cable. Excellent screening characteristics 100% coverage(individual and overall screen) for minimal interference, low delay skew, bandwidth 1300Mhz, low capacitance and efficient noise rejection maintain signal integrity in the long run interconnects, PE- insulated wires with the exact 100 Ohm impedance design in wide range.
 
Note : This is a simple tutorial.
 

 
As you see, the Ethernet cable consists of four pairs of conductors, each of the conductor pairs has its own shielding, in addition to the outer shielding, we will need only three pairs, one twisted pair for the (data -), (data+), another pair will be combined together to reducing the resistance of wire and connected to the pin 4 GND and the same we will do with the Pin 1 +5vdc, the remaining pair needs to be cut off.
 


 
USB Connectors Pinout
1. +5 VCC
2. Data-
3. Data+
4. GND
 





 
The braid is to be terminated (crimped) to the plug metal shells at both ends, as close to 360° as possible, to prevents EMI/RFI interference.
 
                                    
  ^ TEST BEFORE YOU PLUG ANYTHING IN ^
 


 
Done! Here you go 2.5m high quality USB A - B cable at low cost.
Hope this was helpful.
 
reference - some useful links:
http://www.usbmadesimple.co.uk/ums_6.htm
http://www.interfacebus.com/USB-Interface-Description.html
http://www.gbs-elektronik.de/fileadmin/download/manuals/TN_Choosing_USB_Cable.pdf
http://sdpha2.ucsd.edu/Lab_Equip_Manuals/usb_20.pdf
http://www.cypherinstruments.co.uk/applications/AN004.pdf
 
 
P.S. About over longer cable runs: I tested so far 12 meters cable! (before I split it into shorter lengths) connected to my Burson Conductor DAC with MacBook pro, iMac and my very old Toshiba laptop streaming music up to 24/192kHz without any glitches or even single dropouts and finally tested the cable with the HP printer and it worked without hassle.
It shouldn't work! It gave me a lot to think about.The cable quality does matter and parameters such as characteristic impedance is the goal, rather than esoteric materials.
 
May 27, 2013 at 7:10 PM Post #2 of 16
Interesting.  Just an FYI, but the limiting factor on USB length is voltage drop.  USB is also designed to carry power.  A typical copper conductor at 28ga will drop one full volt at 500ma and 15 feet - a half volt at 250ma. 
 
May 28, 2013 at 7:06 PM Post #3 of 16
No doubt especially when the USB receiver is bus powered It should be taken into attention and use ticker wire for power lines. My one have enough "copper" in power lines each pair contains of 22awg connected in parallel, effectively reduces the voltage drop to min IMO.
 
 
May 29, 2013 at 12:19 AM Post #4 of 16
Quote:
No doubt especially when the USB receiver is bus powered It should be taken into attention and use ticker wire for power lines. My one have enough "copper" in power lines each pair contains of 22awg connected in parallel, effectively reduces the voltage drop to min IMO.
 

Of course - 22ga - especially run in parallel - is exponentially better at limiting voltage drop over single 28ga.  Since the power wires are the limiting factor in USB cable distance, it would've been appropriate to mention that you upsized them to 22ga and ran them in parallel when making your claims about improved length over the USB standard. No offense, but I guess all I'm trying to say is that "cable quality" really had nothing to do with it.
wink.gif

 
May 30, 2013 at 8:00 PM Post #5 of 16
USB over long distances Cont'd.
 
I did not mention this in the main thread the main reason why I started to look for best suited for this purpose cable was the distance between iMac and my DAC, 6m in a straight line I haven't had much luck with USB over longer cable runs I've tested generic USB2 cables bundled with printer, camera etc. I have lying around in my apartment, simply it doesn't work for me, much longer than that (1.5m + 3m) and I've had issues with devices not being recognized. To be honest Burson Conductor DAC USB does work without power from USB, like most modern DAC nowadays have the USB receivers an independent from 5V DC power from computer USB, what he need to work is just D+/D- line and GND.
 
Power wires are the limiting factor in USB cable distance doesn't change the fact that cable quality has an impact too. The issue with the cable length is related to signal quality, right cable will (do) probably push it over the 5m limit. After I had success making a USB cable longer more than twice the max limit I cut it off to the length of 8.75m, this is my primary set-up now. iMac(audirvana)>>Supra Cat7+ 8.75m USB A to B>>Burson Conductor>>navigate with Remote App on iPad.
 
I did a cable test which allowed me to extract the another limiting factor on USB length. I've used the fourth pair of conductors (normally not connected) and soldered to the existing one D+/D- data line, ran them in parallel. This cause the impedance (mismatch) drop by half. The USB 2.0 specification defines a 90-ohm nominal characteristic impedance. If every component in the chain (transmitter - line - receiver) is designed with a 90-Ohm impedance, there won’t be any reflection and there won’t be signal energy loss due to reflection in theory.
so what happened?......by doing this, the exact characteristic of cable impedance has been disrupted increasing signal reflection as a result the devices not being recognized on Toshiba laptop and with the Mac not working at all. The same way I've tested my 2.5m USB cable shown in this tutorial, this time was better the device has been recognized, but not working properly intermittent dropouts and disconnecting.

Summarizing characteristic impedance about 90 ohms as uniform as possible, right ga of power wires the quality of the connector and cable assembly shielding, quality of the cable shielding both EMI/RFI should be kept as small as possible all of does matter in terms of usb length.
 
Jun 4, 2013 at 10:28 PM Post #6 of 16
Thanks for the tutorial. I'll try this soon.
 
Aug 6, 2013 at 3:43 PM Post #8 of 16
Quote:
Awesome thread, but did I miss somewhere where you got your USB connectors from?! Links to parts and supplies would be useful for those looking to do this themselves.

Thanks, I ordered them from DHC link below,
http://www.doublehelixcables.com/store/index.php?main_page=product_info&cPath=4&products_id=68
 
Apr 28, 2014 at 5:35 AM Post #11 of 16
  Hi, where did you source the Supra Cat7+ cable?

I bought it off ebay http://www.ebay.ie/itm/Supra-CAT7-Network-Cable-Unterminated-per-Metre-/301154615865?pt=UK_Computing_NetworkingCables_RL&hash=item461e36c239
if you looking for a USB connectors too, you can get them from Parts Connexion,10$ for a set.
 
Feb 19, 2015 at 5:19 AM Post #13 of 16
  Of course - 22ga - especially run in parallel - is exponentially better at limiting voltage drop over single 28ga.  Since the power wires are the limiting factor in USB cable distance, it would've been appropriate to mention that you upsized them to 22ga and ran them in parallel when making your claims about improved length over the USB standard. No offense, but I guess all I'm trying to say is that "cable quality" really had nothing to do with it.
wink.gif

 
Hi guys, just wanted to bump this thread.
 
I have a desktop mic that has a terrible buzzing sound. When i put the diy usb cable i got from a dac i purchased some time ago the buzzing goes away. Hence, I wanna make me a good shielded USB A to B cable but I have the following questions.
 
What cable should I get if I do not want to use a Cat7 ethernet?
What to do you mean run 22ga in parallel? Are you talking about a coaxial cable? Then what do you use for D+/- ? How do run it in parallel?
The custom I have has actually two cables going into both connectors and a third slim silver one. From what I remember these were coaxials but I have no idea how they are soldered.
 
What I really want to know is what cable should I purchase exactly :)
 
Thank a bunch :p
 
Feb 19, 2015 at 4:25 PM Post #14 of 16
   
Hi guys, just wanted to bump this thread.
 
I have a desktop mic that has a terrible buzzing sound. When i put the diy usb cable i got from a dac i purchased some time ago the buzzing goes away. Hence, I wanna make me a good shielded USB A to B cable but I have the following questions.

I'm not sure if I understood you correctly but for me it looks like ground loops problem rather than faulty usb cable but I can be wrong.
 
What cable should I get if I do not want to use a Cat7 ethernet?
What to do you mean run 22ga in parallel? Are you talking about a coaxial cable? Then what do you use for D+/- ? How do run it in parallel?
The custom I have has actually two cables going into both connectors and a third slim silver one. From what I remember these were coaxials but I have no idea how they are soldered.
 
What I really want to know is what cable should I purchase exactly :)
 

I choosed Ethernet cable for one reason, length, length....., much over 5 meters and low cost of course. It has all parameters to make it possible on distance above 5 meters, like constant impedance over a wide range, wide bandwidth and very low attenuation.
I don't know how long cable u will need if shorter then 1.5 meter I would not recommended Ethernet cables, not they are bad but they are stiff.
Go for flexible 90 ohm data cables like this one http://wire-cable-tubing.wireandcable.com/item/coaxial-wire-cables/90-ohm-usb-aerospace-cables/w-3627-26
 
No I talked about cat7+ and spare pair of conductors which I combined together to reduce the resistance in power line.
 
Feb 20, 2015 at 8:35 AM Post #15 of 16
OK, this was what I thought. Now it makes sense to me. Thanks a lot!
 
Btw, I see you have combined orange,white and green, white for the ground and power. So coloring doesn't matter? I have some ethernet cable lying around so I will give it a shot. Around 2 meters is what I need but I can go with less as well.
 
Ground loops you say? On the cable? I have all my equipment into a power strip which is connected to a surge protector which then goes into a grounded socket. When I actually discovered the cable was fixing the issue, I was on a laptop not connected in a different room.....I made all sorts of experiments :D
 

Users who are viewing this thread

Back
Top