This was a fun little project i have been working on that may be of interest to some. I recently built an Objective2 headphone amp with an ODAC in separate enclosures which had the unforeseen side effect of sucking me into the DIY electronics hobby. I then picked up an old Tektronix 2225 analogue scope on ebay and was soon waving the probes around like i knew what i was doing. I literally spent hours looking at AC/DC waveforms, opamp inputs/outputs and generally learning how to use the scope and interpret what i was looking at until i connected the scope probes to the +5v and - wires of a USB cable connected to my PC. The following screen shows is the +5v (top) and - rail (bottom) measured at the end of the USB cable i had my ODAC plugged into. For those who dont know AC power alternates (see avatar), and DC power shouldnt. That should have been two perfectly flat lines, those squiggly bits are noise that should not be there.
I manged to reduce a lot of the noise by unplugging various devices on the PC until i found the main culprit being a cheap power supply for an external hard drive caddy but the more i looked into the traces on the scope the more noise i was finding at various frequencys. The icing on the cake for me was the actual voltage level, in theory USB is 5V DC but i was seeing the voltage level jump up and down constantly. With a USB memory key plugged into the computer and reading/writing data to it the voltage level started fluctuating all over the place. Remember in the case of the ODAC the unit is entirely powered by the USB bus.
The first experiment was to build a simple 5V DC power supply using a 7805 Linear regulator, then cut up a USB cable to splice the power lines in and plug it into the ODAC to test. I started with a digital silence audio track playing, switched to high gain on the O2 amp and set the volume up to full and listened, ...nothing. I then switched between the original USB cable and the new modified one with the power supply until i could confidently say yes, the modified one has less background noise, however slight it may be, there is a difference.
Next test was to see if there is any difference playing high frequency or low frequency sounds so i opened up foobar and the last track i had played was "Where do you think your going" by Dire Straits, a track i know very well so i just hit play and listened. Its always hard to tell when your listening for changes to know if there is a change, or if you just perceive it because you are looking for it but the cymbals at the start of the track, vocals, and guitar all sounded clearer and more precise, but again i'm probably imagining it. Then ~1:15 the drums come in and roll around your head and my jaw hit the floor, the sound was so smooth and full bodied it was like i just received a free headphone upgrade. There really is something to this beyond theory, speculation and subjectivity.
Now that i knew i was onto something i went a step into the overkill territory and ordered a toroidal transformer, rectifier board, and a few regulator boards to build and test from ebay, but more on that later. In the process of researching this i stumbled across the Galvanic USB Isolator chip from Analog devices, and subsequently the DIY PCB kit from Circuits@Home for $25 (http://www.circuitsathome.com/products-page/usb-interfaces)
The beauty of this chip/kit is it removes any direct electrical connection from the USB input to the USB output. When you break down a USB cable you have two data wires, two power wires (+&-), and the outer shielding/metal casing of the plug which is tied to the chassis earth of your pc. (the metal case basically).
You could just cut up a USB cable as i had previously and connect a 5V power source to the output and call it a day however using this board i can also isolate the two data lines via the chip, and the outer shielding via the PCB eliminating any path for noise to find its way to the USB device. For the USB sockets i opted for Neutrik items because the outer shield connection to the enclosure is optional which i have disconnect on the output side. So basically any noise coming from the PC through the cable shielding to the isolator enclosure has no connection to the output socket. The result is perfectly clean data and power lines on the output USB cable.
(Assembled board + a bit of bodgery + Neutrik sockets)
The Neutrik socket to PCB connections were done by cutting up a Belkin USB2 cable. Personally i think the quality of these cables is fantastic as you have x2 28AWG data lines, x2 20AWG copper power lines, a foil shield with drain wire, and a copper outer braiding. This is also the cable i am using from the PC to Isolator box, and from the isolator to ODAC a small 6" cable with integrated ferrite beads.
The enclosure was originally a weird serial to usb device i bought on ebay as i liked the look of it, but the seller has just the enclosures listed now for $10 http://www.ebay.co.uk/itm/251327967350?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649
I have the Power LED wired to the 5v input jack, and the Active LED wired to the USB input jack so they are both functional.
In Part II i will document some of my findings from testing various power supply boards available on ebay. For comparison here is a simple ebay sourced 7805 regulator board on top (~$10), and the USB supply from the PC on the bottom.
So far i have probably spent ~$100 on this project however a basic DC wall adapter and a hacked up USB cable would cost you more like $10 and still have a considerable benefit to your USB powered DAC. I'm just a perfectionist... and far from done yet