The longer the wire, the greater the Inductance induced Voltage fluctuations are.
The location for this was a toss up, OSWATT Power BUSS, or Idea013.
Foot notes, before we break something expensive. Some mods, should only be attempted when all other options have been exhausted.
0) I'm still brain storming the SATA plug things.
1) the Molex(tm) power plugs are only rated for 50, thats fifty and no
more, plug/unplug cycles. So the solution needs to be cheep enough to
throw away when worn out. "O" and there only rated for like 5Amps, so
the buss plugs the power supply plugs into, needs to be ganged up. like
one plug for each 5Amp draw. The expensive plugs at Mouser Electronics can handle
8amps, so two of them will allow me 16amps on the buss. Thats 16amps on
the 5V (80 Watts), plus 16amps 12V (192 Watts).
2) The ATX standard calls for 16AWG wire for the "Auxiliary" (HDD,
CDROM, etc) Power connectors, 18AWG for the floppy things. I was going
to use excessive 12AWG (5V&12V), and 10AWG (ground) for the long
runs up and down the tower. With a breakout box like things to the
individual plugs, to reduce power loss. Voltage drop, and RFI was
already on my mind before OCZ came out with there excellent product.
The thought crossed my mind to order a slew of them for
"breakout-cables" to the devices. 18 plugs * unit-price (about $25) =
approximately $450... for a throw away item, Ouch! I'll settle
for just one for my graphics card.
3) There are calculators on-line to calculate the inductance of
straight runs of round wire. Coiling up the excess cable in the
computer only increases the inductance. Making the internal noise
transmitting of the wire worse. It is NOT clad-shielded (Coaxial)
stuff. Plane and simple, all you need to know is the thickness of the
conductor (excluding the insulation) in mm or inches (decimal not
fractions), and the length of the wire stretched out straight from the
PSU to the last plug (in inches or mm), for each run of cable. This txt file is what I came up with for my computer.
4) With DC power... When the device demands an increase of power draw
threw an inductor, the voltage drops. When the device decreases pulling
power, the inductor only cares about Amps flow staying the same, thus
increasing the voltage at the device to keep the Amps flow constant.
Allot of Buck/Current-Boost/Flyback switching regulators take advantage
of the inductors properties in a good way. In DC power distribution to
devices, inductance without addiquit protection is BAD. Voltage will
drop during the increase of power draw, stabilize at constant power
draw, then spike during decrease of power draw. The quicker the power
draw is changed with a given inductance, will increase the severity of
these voltage fluctuations.
5) before we go over kill with "snubbing caps" as some call them. The
ATX standard also states the minimum amount of capacitance the Power
supply needs to be able to charge during the start-up cycle, to have
the voltages stable in "X" amount of time. This implies there is a
maximum amount of capacitance the given power supply can charge up as
well. An excessive amount of capacitance will probably trick the power
supply into thinking there is a short, and thus not turn on in the best
case. The lovely little shot-key diodes in the power supply, or
something else in the computer might just release it's carbon in the
worst case (POOF! plume of smoke rises from expensive computer).
6) And last of the footnotes. The closer we can get the caps to the RFI
source (the Hard drives), the less wire between them is transmitting
that nasty RFI. Additional chokes are tricky items. OCZ did it right
with a toroid. Toroids reduce the surrounding magnetic fields far
better then simple bar inductors. The last thing we want to do is
degauss the hard drive platters. How long are your wires, and what are the effects?
Finding Cable lengths that work in tall computer cases often leads to
multiple feet of power cable runs. Anything less, and the plugs wont
reach. All that cable length comes at a price. Wire Inductance induced,
Voltage fluctuations. Like Any DC power system, combating Voltage
fluctuation is done with capacitance. Now I know there are products out
there that have capacitors in them to reduce noise from the Hard drive
or VGA card. They Also contain additional Inductors. Adding Inductance
in the power cabling without a capacitor before the inductor, would be
counter-productive with such long wire runs.
Plane and simple, three feet of 16AWG wire, has an inductance of over
one micro henry (1uH). In electronics and RF terms, Thats a
significant amount of inductance. More then enough inductance for the
changing power draw of a single hard drive, to throw the voltage at the
hard drive well out of spec. No simple multimeter will see these high
frequency inductance induced spikes. However even with the most
primitive Oscilloscope, they become blatantly obvious. As I sit here
and watch the bar graph on my Fluke periodically peg to the beet of the
hard drive chatter, and the numeric display stays solid at 12.073VDC.
Why don't the capacitors in the hard drive dampen these spikes? The
capacitors in most consumer computer gear is there to reduce EMI/RFI
interference, in accordance to various regional standards. There not
there to compensate for a lacking power system. In engineering terms,
adding more capacitance to the product to correct the voltage in a hand
full of rare situations would not be cost affective. In management
terms, Most people don't have more then three feet of power cable
between there PSU and the hard drive. Most people don't need a better,
more expensive to produce, less profitable product.
This leaves me with a power supply and hard drives, that was not
designed for that extra length of cable. The solution sounded simple at
first, till I took some mechanical measurements, and fired up ELSIE to
run some numbers. In my case, I needed to add a measly 0.057uf (0.01uf
+ 0.047uf) to each plug on the Power extension cable. The larger 1uf
caps can be pushed back further on the cable for a convenient mounting
location. The problem is, there is not enough room around the plug to
attach the small capacitors in the same way on every plug. Either there
is no room under a plug, less then enough room behind the plug, or
there precariously stacked on top of the plug blocking the hard drive
There is the other issue of the cable radiating the hard drive noise
All over the inside of the computer case. Making nasty lines on my
screen, and digital noise in my sound card, when ever I'm browsing my
hard drives. At first I thought the Power at the motherboard was noisy,
till I tested and found the power well with in spec, everywhere except
at the lower hard drive rack. So the caps need to be as close to the
hard drives as possible, without voiding the hard drives warranties.
So now I'm thinking about modifying the plugs on the power extension to
make the capacitors fit in the plugs. Why not, 0.01uf and 0.047uf
Polly-film caps are rather small, and inexpensive. Before I ruined my
expensive computer, I grabbed a dead PSU, and a hacksaw. I cut up the
dead PSU's plugs, to see how close to the wires I could get the caps.
Then I stole the Polly-film caps out of that dead PSU, to try to fit
them in various ways in the plugs. I also drew them up in ExpressPCB as
a crude mechanical drawing while brain storming the solution.
UPDATE: OSWATT Wire harness...
Well the plugs are looking cool, hopefully by next week, I'll have that
harness in OSWATT, and more in the works for the Proxies, and
LASI&SPICE-drones errr SETI-Cluster.
BTW, Protecting the conductors?
I think plugging these things into a dead drive while I'm hot-gluing
the exposed conductors to protect them, might ensure that the plug pins
somewhat lineup when the plug is used. Hot glue has some give to it,
T.B.D. The O-scope I had access to, gave up the ghost before I had a chance to use it... Ahgggggg!
This mod also might help the RFI emissions of acrylic PC cases as well, just a thought.