usb power problem: solved!

susman · 04-26-2004, 02:11 AM

hello there, for those who have an epia motherboard (or any other motherboard with an additional external usb bracket), the opus 150W and have to deal with usb devices draining out their battery because of the impossibility of powering them down, here is HOW I DID IT

without damaging or permanently modifing a thing)

things you need:

1)external usb hub.
2) two pieces of wire 10-20 cm long.
3)tape
4) a free molex 5 wire plug on your opus (or you can plug in one of those multipliers)

open your case.
follow the wire from the external usb ports to the motherboard. unplug it from the motherboard. usually there are two sets of identical wirings because the usb ports are in couples. from one of these which should be in a row, extract(do not cut or you wont be able to put things back in their original state...):
the black wire and the red wire(should be at the opposites, the otyher two wires are the data rx/tx). now the red wire has to go to the red wire on the free 4 wire molex connector and the black wire to the black wire on the same. since the wires from the usb connector are very small i taped them with a thicker piece of spare wire 5cm long the other end plugged into the molex connector and taped them Very well!

close the case, connect your usb hub to the "modified" usb port and every time your opus shuts down it will shut down the power to the usb gps, bluetooth, irda etc etc. peripherals without draining your battery.

gork · 04-26-2004, 09:19 PM

That is essentially what I did; however, there are a couple small potential problems I discovered with your kind of setup...

USB ports provide a current limiter on the 5V line that has a maximum output of 500mA. They also provide a capacitor to prevent spikes on the 5V rail as a device powers up. If you connect your usb ports directly to the 5V switched power rail (as from a moles drive power cable) you could potentially have an unregulated device (such as one of those USB "cell phone chargers" or the like) damage your system if you use such a thing. You can also have the problem of a USB device causing your computer to reboot as it powers itself off and on and that power cycling causes a large enough voltage dropout on the 5V rail.

I experienced this second problem quite often and at unexpeted times. The final problem is that the USB controller is allowed to dynamically manage current to the attached devices. In theory, this makes a difference to some devices that go to sleep based on the amount of supplied current. Running them off of a constant source shouldn't pose a problem, though it wouldn't be unthinkable to assume that there are problematic devices or drivers out there..

I solved the problem by switching the 5V USB lines with cheap reed relays. They are rated at 500mA (the maximum you could draw from a USB port anyway) and they only cost about a dollar. The USB ports 5v power from the motherboard is run through the relay and the relay is switched by one of the normal switched power rails. This ensures that the USB device power is always managed properly by the USB controller, but the power is always switched off with the PSU.

Also note that the power is still supplied to the Serial/Parallel ports on the EPIA no matter what, though most serial and parallel devices do not attempt to draw power over the ports...

gospeed.racer · 04-26-2004, 09:28 PM

Edit...heres what I did..

I took a molex extension (short) and it happened to have a pair of small guage wires piggybacked onto the 5v positive and ground. I simply ran those to a small RatShack relay (275-240A) with 1 amp rated contacts. My usb hub is wired to a header connector so I can use it internally...mainly for my 802.11 and receiver for my kb/mouse. I split the red wire and soldered it to the N/O relay contacts. Now, I have power to the USB only when the hard drive is spun up. When I standby, it shuts down rapidly and off goes the hub. Now if I can remember to bring home my meter one night, I will check its amp draw in standby.

So do I really need a diode on the relay coil? More specifically, what kind? I have some schottky's laying around from my R/C stuff, but I am sure they are overkill!

gork · 04-26-2004, 11:29 PM

Well, the problem with the relay powering a hub approach is that most hubs now will work with or without external power... so it doesn't really solve the problem if your hub will work without an external power adaptor...

If you still want to go this route --

You don't need a relay.. the 5V lines turn themselves on and off with the power supply.. You can wire them directly to the DC jack of the usb hub, but see below for why this might be a problem.

Even if a usb hub does work non-powered, you can modify the hub to power itself only off of the external power port, but in any case where you power the USB hub off of your PSU, you will be back to problem #2 above where the usb devices powering themselves on and off can cause a voltage drop on your 5V PSU rail causing the computer to reset. This would be the purpose of the capacitor. If you place a larger polarized capacitor and a diode across the + and - 5V leads going into the USB hub you could create a very small tank circuit that would keep the computer from rebooting as USB devices power up. I have not tried this approach, though, so I don't know exactly the best way to go about building it...

gospeed.racer · 04-26-2004, 11:32 PM

Sorry, I think I forgot to mention, mine is converted to internal use with a header connector, and just powers up the inside stuff, so my route was easier than it should have been. I edited my post above to reflect what I did.

gork · 04-27-2004, 12:17 AM

I'd say if your system is stable and you can verify that your USB ports power off completely when the power is off, whatever you have going is probably fine... Your hub may have some of this dampening circuitry internally anyway.

UncleGeek · 04-27-2004, 05:00 PM

Will a POWERED USB hub powered by a 5 v. drive connector provide the regulation needed to the USB devices plugged into it without the 5 v supplied by the mobo?

Ricky327 · 04-27-2004, 05:18 PM

So do I really need a diode on the relay coil? More specifically, what kind? I have some schottky's laying around from my R/C stuff, but I am sure they are overkill!

Yes...its safer to put one in anyway. The spike can damage your HDD, FDD, CDROM or PSU even. 1N400x series should be fine.

Ricky327 · 04-27-2004, 05:20 PM

But why just power off the USB? why not kill the power to the PSU all together?

bosstone74 · 08-17-2004, 11:09 PM

does the M10000 have these issues?

gork · 08-18-2004, 12:20 AM

Yes all the via Mini ITX boards have this USB power problem. As I said in my above post, I used some reed relays to switch the 5V line on and off on the USB ports. It works great!

ashzero · 08-30-2004, 05:39 PM

back from the dead. will this prob go away if you use the ITPS?

archimense · 08-30-2004, 05:42 PM

There is another solution that shuts off all standby power from the OPUS to the motherboard.

Here is the thread: http://www.mp3car.com/vbulletin/show...7&page=6&pp=15

Look at post #78.

Rando and I have both done this mod. Works great, and is really easy to do.

bosstone74 · 08-30-2004, 11:41 PM

This is only a problem if you use standby, right?

It is not a problem if you use hibernation or shutdown...correct?

savante · 08-30-2004, 11:48 PM

Quote: Originally Posted by bosstone74

This is only a problem if you use standby, right?

It is not a problem if you use hibernation or shutdown...correct?

I think for most newer motherboards, its a problem(at least ATX ones). I know my USB GPS light is lit up when my computer is 'off'. I think if you follow that link to that other thread, you will find more information.
Also this may be of interest http://www.mp3car.com/vbulletin/show...0&postcount=83