USB Direct Power

[appdev007]

I'm considering cutting the DC+ trace to my USB ports and putting a jumper directly from the DC+ pin of the power connector. Wouldn't this just create the same situation as you would have with a powered USB hub and eliminate overloading the Pi's traces with powering the USB ports?

[Burngate]

I assume you're likening it to having a (non-back-powering) hub, and feeding power from one of its outputs to the Pi's power input.
It means the poly-fuse (not the track, which will handle as much as, if not more than, the poly-fuse) is only carrying the Pi's current

So, yes, it seems to be a good thing to try, and similar to what I did on a rev-1 Pi.
In that case it was easier, because I could remove the poly-fuses rather than cut tracks, and rather than a straight piece of wire - for reasons I won't go into here - I put an inductor in series.

It doesn't remove the need for a decent power supply - without the poly-fuse resistance between the input and any USB load, quite high inrush currents can flow if you hot-plug eg. a wi-fi dongle, which could brown-out a poor supply

[appdev007]

pi_usb_power_jumper.jpg (59.88 KiB) Viewed 5481 times

Here is an illustration of what I intend to do. I have a 5VDC @ 3A power supply that I want to run the USB devices and the PI with at the same time. The more I look at this I think it might involve removing the USB ports from the board and reconnecting just the ground and data pins via a cable. Then make a Y cable for the power; female barrel jack on one end, male micro usb on another, and a pair of bare wires on the third for the power pins on the USB ports. I've have a 3D printer so I could print a case to hold the pi and the dangling USB ports.

[Burngate]

Your picture has the yellow jumper connecting to the metal-work of the power input socket.
It might be easiest to connect to one end of the poly-fuse - here in red.

pi_usb_power_jumper.png (52.82 KiB) Viewed 5418 times

As for the other end, not knowing where tracks go, I've no idea how to disconnect the power, without first removing the socket, then modifying it.

Best of luck

[FLYFISH TECHNOLOGIES]

Hi,

Here is an illustration of what I intend to do.

Another option:
http://www.raspberrypi.org/forums/viewt ... 93#p496993


Best wishes, Ivan Zilic.

[appdev007]

LOL Ivan, I was thinking about making cables like that!

[appdev007]

Your picture has the yellow jumper connecting to the metal-work of the power input socket.
It might be easiest to connect to one end of the poly-fuse - here in red.

pi_usb_power_jumper.png

As for the other end, not knowing where tracks go, I've no idea how to disconnect the power, without first removing the socket, then modifying it.

Best of luck

LOL, I didn't actually have my PI on me to see that. I just stole pics off internet. I assumed that was the lead and the ones to the left were the housing mounts. ; )

[kyuzumaki]

Can anyone explain why you need to disconnect power from the pi to the usb output port? Surely its the same +5 voltage from the external supply as is running into the polyfuse anyway so the jumper would just supply extra current as needed?

Also has anyone done this already? I would be really surprised if nobody has it seems like the most obvious fix for the power issue..

[Burngate]

On the original Pi - Rev 1 - there were two polyfuses feeding the USB ports, as well as the main input polyfuse.
I modified mine (removed those two polyfuses & fed the USB ports from before the input polyfuse) to get over a perceived problem with hot-plugging devices (and no, it didn't cure everything)

Later versions had 0ohm jumpers instead of those polyfuses, so different hot-plug issues.
Rev 2.0 could still suffer from those.

There is a difference between raw 5v fed into the board and the 5v reaching the USB port - the polyfuse has non-zero impedance.
If a device with a large capacitance across its power rails is plugged in, the instantaneous voltage on the Pi's 5v can drop enough to brown-out the Pi.
Putting the USB load before the polyfuse means the supply should fill that capacitance before the Pi browns out.

That relies on the supply being able to source enough current, which may not be true. In my case, it wasn't true. The only solution was a powered hub - which browned out but with no adverse consequences - and in fact better than cutting tracks,etc.

[kyuzumaki]

So you have done this and it is safe to connect a wire right from the polyfuse to the usb input without removing the usb connectors +5v from the board?

This is what I had thought of doing, connect +5v to the usb header connect ground to header, add a capacitor across the usb + and - pins to prevent brownout. Would this work? I also have a small inductor from an RC car not sure of its rating but could put that in line with the cap to prevent it causing probs when filling or discharging.

Also not entirely sure which side of R51 to connect to but on schematic it seems to be getting the full ground input on one side.

[drgeoff]

So you have done this and it is safe to connect a wire right from the polyfuse to the usb input without removing the usb connectors +5v from the board?

I modified an original Model B (with USB polyfuses) with a wire exactly as the red one in your diagram. I have not experienced any problem.

[klricks]

So you have done this and it is safe to connect a wire right from the polyfuse to the usb input without removing the usb connectors +5v from the board?

This is what I had thought of doing, connect +5v to the usb header connect ground to header, add a capacitor across the usb + and - pins to prevent brownout. Would this work? I also have a small inductor from an RC car not sure of its rating but could put that in line with the cap to prevent it causing probs when filling or discharging.

Also not entirely sure which side of R51 to connect to but on schematic it seems to be getting the full ground input on one side.

If you don't remove the 5V from the USB then there is no point in doing this. The red wire as you have it is equivalent to just shorting across F3 with a wire.. No point.
There are plenty of places to get a ground from without trying to solder to a tiny SMT resistor pad. You could connect to any of the shield tabs such as the SD slot, RCA jack etc.

If you are really worried about the power issues, then just get a B+ model and a good power supply. Though I am not sure why you or the OP think this is necessary at all on the B?

[kyuzumaki]

I have now bridged with the wire as I suggested except the ground comes from the diode (end closest to the edge of board) instead of R51.

raspberry-pi-power-mod.jpg (60.58 KiB) Viewed 4687 times

I spent some time with my multimeter testing prior to that and my usb ports were getting 4.70-4.80v but the supply coming in was a steady 5.00-5.05v

Having now bridged the gap my usb ports are getting a steady 5v and glad to report wifi dongle has been stable overnight. Previously it dropped out every few hours.

If you don't remove the 5V from the USB then there is no point in doing this. The red wire as you have it is equivalent to just shorting across F3 with a wire.. No point.

My ability to follow the schematic is poor but from what you say could it be my polyfuse thats causing this voltage difference? Have I accidentally back powered the pi with a more stable voltage from my (official usb power supply)

Already have a B+ but I have three model B's too.

[klricks]

I have now bridged with the wire as I suggested except the ground comes from the diode (end closest to the edge of board) instead of R51.

raspberry-pi-power-mod.jpg

I spent some time with my multimeter testing prior to that and my usb ports were getting 4.70-4.80v but the supply coming in was a steady 5.00-5.05v

Having now bridged the gap my usb ports are getting a steady 5v and glad to report wifi dongle has been stable overnight. Previously it dropped out every few hours.

If you don't remove the 5V from the USB then there is no point in doing this. The red wire as you have it is equivalent to just shorting across F3 with a wire.. No point.

My ability to follow the schematic is poor but from what you say could it be my polyfuse thats causing this voltage difference? Have I accidentally back powered the pi with a more stable voltage from my (official usb power supply)

Already have a B+ but I have three model B's too.

If the total power usage is allowed to go over 750mA then F3 will begin to increase in resistance and cause a voltage drop. Depending on what other things are connected to the USB, you will be right on the edge with a WIFI dongle attached. That is why it was frequently suggested to use a powered hub with WIFI and other high power devices.

The top pad of F3 in your photo is F3 pin 2 on the schematic and is connected to all points labeled +5V0 including the USB ports.
The bottom pad of F3 is pin 1 and connects only to the micro USB power input.

When you connect your red wire from F3 pin 1 to the USB port (+5V0) then you are bypassing the fuse F3.
So if you were to solder a wire across F3 then you would accomplish exactly the same thing with less wire.
The black ground wire you have is redundant and has no added function.