Saturday, August 23, 2008

Converting your old USB 1.1 cables into USB 2.0 cables

The three biggest problems facing humanity today are: greenhouse gass emissions, the price of fossil fuels and all the USB 1.1 cables that we need to toss out now that we've moved on to USB 2.0 devices. As it just so happens, problem number 2 will take care of problem number 1, so the only thing left for humanity to fix is problem number 3. And here's how!

According to the specs, USB 2.0 cables are shielded and the two data lines are a twisted pair to prevent data loss from high speed transfer. (What property of physics makes a twisted pair more successful I don't know, but maybe Bill and Hillary know.) And just in case the original cables were a little thin (more on that later), we might want thicker wires for less resistance and more power. So I decided to combine two old USB 1.1 cables to make one good USB 2.0 cable, thus doubling up the wire thickness.

The spec says that the shielding is a copper braid and an aluminum shield grounded at the host. I assume this shield is like the tube in a coaxial cable. This shield is simply the coaxial's ground wire, so I decided to try using the USB's power supply lines as a copper braid, since one of them is a ground wire.

Starting out with a pair of cables from the dollar store my total investment was $1.98 to make a USB 2.0 cable that costs $30 or more at local electronics stores, and anywhere from $2.50 to $5 from online stores, plus shipping. So, we can save a few bucks with this project. In this day and age many people have much more time on their hands than money. That's a good thing, because this project took about 8 hours to complete.

Step one is to strip the two layers of insulation off the cable. Inside are four wires, each with their own insulation. With a thin blade you can cut halfway into the cable and do minimal damage to the insulation on these four wires. In the end you'll see that randomly exposed copper on 8 different wires rarely touch each other. But, we'll test for shorts just to be sure. Be careful to leave the outer layer of insulation as one long piece, since we can use it later for our finished product.

After exposing the four wires we need to slice open the plugs. Make one slice along the side of the plug, then pry it open and remove it entirely to expose the innards. There you'll see a body of glue that encases the four wires and their soldiered connections.

Slice the glue and wires clean off the plug behind the metal shroud, far enough back not to slice the four pins to which the wires are soldiered. In this case, 1/8 inch was plenty. Then you need to pick, prod and poke the remaining glue to remove all the bits. Slicing a 3 dimensional checker-board pattern works well. Repeat for the other 3 plugs.

Now it's time to double up the wires. Might as well twist them together since twisting is on the mind. Sure beats stripping them, then twisting them, then applying five feet of heat shrink tubing. It's good practice because when twisting them you'll see that you need to apply the same tension to each wire, and hold them at the same angle. This makes for a proper twisted pair. If you apply more tension to one wire, that wire will tend to remain straight, while the second wire wraps around it like a red stripe on a barber pole. This seems to make for ok copper shielding, so both techniques can be used later.

In the photos you'll see that I made my red and black pairs nearly perfect, while my green and blue pairs were more like the barber pole. There were almost an extra three inches of wire sticking out. I turned ghost white! This means that the data signal, once split, travels through two wires of different length, then arrive at different times at the other end. But won't the first data signal be arriving in the long wire just when the second data signal arrives in the shorter wire? I HAD to do the math.

For a worst case scenario, assume electricity travels at the speed of light (through these superconducting, 99 cent wires), and data transfer is always at the max 480 MHz, then how many feet of wire does it take to carry a bit of information? The speed of light is just under 300 mega-meters per second. Dividing by 480, it seems a bit of information needs 0.62 meters of wire, just over 2 feet. So, for any instant of time, there's over 1 foot of wire with a high voltage and 1 foot with a low voltage. Three inches shouldn't be a problem. I chilled.

The next step is to cut the ends off all four pairs so they appear to be the same legnth, strip an inch of insulation off, and twist the copper together.

Now take the two data cable pairs, green and white. As you can see in the photo, my 99 cent cables had a blue wire instead of the required white wire. So cheap! We all know white insulation is better for high speed transfer than blue insualtion.

Anyway, use your best talent now to make a proper twisted pair. Once complete, hold one end of that pair under your foot to apply tension, and wrap the red pair around it barber pole style. Repeat with the black pair, making sure the red and black don't overlap. At this point use a bit of tape to keep the ends from unraveling.

Now it's time to soldier the four wires to the four pins in each plug. In the drawing you'll see that in a USB A plug, if you hold the plug so the contacts are facing down, the contacts are numbered 1 through 4, starting from the left. Quite simple. For a mini-B plug, the contacts are facing up, also numbered from left to right.

For the regular B plug, the contacts are numbered clockwise and counterclockwise, contacts facing up and down, from left to right and from right to left, starting at the upper left and the upper right, female and male plugs respectively. So easy to remember, even a cave man could do it.

In all three types of plugs, pin 1 is red for positive voltage, pin 4 is black for negative voltage (a ground), pin 2 is white for data negative, and pin 3 is green for data positive. Simpy soldier them this way and use a multimeter in the resistance (ohms) mode to check for shorts between any of the four wires, fixing any with a bit of electrical tape.

If you have an open, you need to replace a section of wire where the break is. Find the break by poking the multimeter's super-sharp, finger-pricking, blood-drawing tips through the insulation at different points along the wire. With the big mess you'll make, this is a good time to check your blood sugar.

I measured the resistance of the thin wires in the original 99 cent cable, which was about 4.5 feet long. They were all between 4 and 4.5 ohms. After doubling up the wires, they were now down to 2.5 ohms each. The original cable was labled 28 AWG, giving a copper diameter of 12.6 mills (thousandths of an inch). By USB standards it is good for 0.81 meters, or about 2.66 feet.

Why these cables were 4.5 feet long I don't know, but they worked fine. Doubling the thickness effectively makes each twisted pair 17.8 mills, almost equal to one AWG 25 wire of 17.9 mills. That is good for a length of USB cable about halfway between 1.31 meters (AWG 26) and 2.08 meters (AWG 24), about 5.6 feet. Now our 4.5 foot cable is legitimate in length.

Store bought USB 2.0 cables advertise AWG24 for power and AWG 28 for data, neither up to USB spec for their length of 10 feet. Seeing that nobody seems to follow the specs, I don't feel too bad for doing this crazy project.

All that's left to do is squeeze the wires into the original outer layer of insulation, squeeze on the plugs, and tape it up.

Now it's time to test this clunky beast. Interestingly, my old 900 MHz motherboard's built-in USB 1.1 port is faster than the USB 2.0 ports on the PCI card I just bought, as tested with a real USB 2.0 cable. So I used a brand new laptop with built-in USB 2.0 ports.

When writing to a 1 gig thumb drive, data transfer was actually faster with my cable, about 34 vs. 27 Mbps. Both are a far cry from the maximum 480 Mbps, but at least the cable is not to blame. More importantly 34 Mbps is more than the 12 Mbps maximum for a USB 1.1 port, so the cable is at least a partial success.

Dowloads from the thumb drive were much faster, at 105 and 45 Mbps for the new and old computers, respectively. Well into USB 2.0 territory. One bad result was when using a 2 gig thumb drive. For all intents and purposes, it didn't work. To be honest, transfers worked fast, but there was a 2 minute delay every time I clicked on a file or file folder, or opened up the Windows Explorer program in the first place.

But back to good omens: When using the home-made cable with a USB 2.0 camera and a USB 2.0 wireless adaptor, Windows XP decided I had indeed built a USB 2.0 cable, and gave me no pop-up window nagging me that "this device can perform faster". I could transfer photos and try to connect to all the neighbor's routers just as fast as I could with a real USB 2.0 cable. In fact, the photos in this article were uploaded to my pc through the actual cable seen in the photos. Eight hours well spent!

Tools required: thin knife (such as razor blade or Exacto), wire stripper, multi-meter, electrical tape, cellophane tape, soldiering iron, soldier
Parts Required: two USB 1.1 cables from a "dollar" store


Anonymous said...

for better results use cable # 18.

WoodHugger said...

I assume that's AWG 18.

Anonymous said...

So how long did it take you to save $3.00? Seems like a lot of effort for not much reward--well, other than proving it can be done. :D

WoodHugger said...

The reason I did this project was the $25 retail price for USB 2.0 cables back in 2008. Little did I know I could get them mail order for $1.80 to $5, depending on length.

Total time is probably 8 hours, with the TV on, and using a low power soldering iron with a fat tip. This makes the soldering job a few hours.

I recently bought a 45 watt iron, and it's much better than the 30 watt I used for years. But, I think I will look for a 60 watt.