USB Port Current Boost (solved)

[Ed Raket]

Has anyone bridged F1 & F2 and replaced the standard 47 microFarat for a 220 microFarat yet?
This way both USB ports may stay oparational...

I just tried this exactly.

47uF -> 220uF and bypassed F1 & F2.

The result however was not any better than the non-cap pipass.

Top port could use higher powered devices connected before power-on but could not hot-plug them. Board would reset.

Bottom port could not deal with higher powered devices at all, but still worked with lower powered devices.

Ok thanks, good to know...

[transmition]

I then did a similar mod to the previous poster by adding a wire from the input directly to the USB sockets. Actually I added a thin twisted pair of wires and also bypassed the ground tracking. Having done this I can now hot plug the hub with no problem. I left all the polyfuses in position and not shorted.

Could you please explain how you bypassed the ground tracking? I'm a bit confused.

The elinux wiki stated that if you bridge the fuses with 1ohm resistors, the hot plugging issue would be resolved, does anyone know why?

[g4eml]

Transmition, see my Previous post on 30th June.

All of these 'fixes' are trying to get the Pi to do something it wasn't designed or tested for. The USB circuitry is very minimal to keep cost down. This, coupled with the individual variations in power supplies, cables and USB devices, means that there cannot be any guaranteed solutions. But that is all part of the fun.

[lrvick]

For those having problems after fitting the 'pi-pass' it might be worth trying adding a second wire to bypass the pcb tracks of the ground connection. (When increasing the current capability you need to consider both the supply and the ground Impedances.) Additional capacitance as mentioned above also wouldn't hurt, although I haven't found this necessary in my case.

This can be achieved by running a second wire from the end of D17 nearest the board edge to the two USB ground pins. Actually I used a 30swg twisted pair of wires to bypass both the power and ground.

I did this mod some time ago on two Pis and have so far not seen any problems with hot swapping devices or non responsive ports.


...Colin.

This is one I have not tried. Makes perfect sense. Not positive which pins your talking about though.

Feel like snapping a pic?

[g4eml]

Irvick,

Sorry, I can't take a picture as my Pi is boxed up and not accessible.

I am sure I have seen a picture somewhere on the forum of the same mod done by someone else. I seem to remember It was done with a red and black twisted pair of wires. I don't have time to search for the post though, sorry.

Colin.

[lrvick]

I finally went a little crazy and spent several hours working on this problem with multimeters, digital bench power supplies with current meters, fully isolating usb power, monitoring drops and resets...

I have what I now appears to be a solution.

The bottom USB port issues were resolved for my by simply providing adequate grounding to the port in addition to the added positive as suggested.

I bridged the USB V- pins on the USB to T2, and found both ports to become equally stable as a result with low and high current devices.

With that problem out of the way, I then started trying to figure out the issues with the board resetting under conditions like plugging in a wifi dongle, or when wifi dongles hit high tx/rx.

Even with 220uF cap in line (in either direction) and the added ground, the board resets (and sometimes freezes) still intermittently persisted.

In frustration I removed the F1 and F2 polyfuses entirely to ensure the ports were only getting current from the added wire. Much to my surprise, the resets persisted.

The only logical conclusion was that the initial current demands of higher power devices like the USB-N10 were actually trying to draw backwards through the board even at this very early input point, thus robbing the board of power (and/or tripping the main polyfuse).

So! In addition to removing the F1 and F2 polyfuses to isolate the ports, I added a diode to prevent and any backwards drawing from the board.

The board is now rock solid (provided you are feeding in a 1.5A+ power supply). I can boot up with devices in either port, hot swap them over and over etc. I can no longer find any situation that can cause a board reset, as the USB power is completely isolated from the board.



Notes: The diode I added a PR1007 that I stole from a perfectly working power supply in my rage-hackery. It is rated at 1A forward current and prevents up to 1000v reverse voltage which is certainly overkill. I am sure a diode that is easier to obtain and far more suitable for this use case exists. It would also probably be fine to leave the F1 and F2 polyfuses in place with this solution, but I have no real reason to put them back now.

[mahjongg]

Note that if you use a diode (of any kind) in series with the power to the USB device, you end up with a similar voltage drop that you were trying to remove by bypassing the polyfuse (a few hundred millivolts).

The problem is actually simple to understand;
A good capacitor will , at the moment you connect it to a voltage source, act like a short.

That is at "time zero" any capacitor can be thought of as being a simple piece of wire!

So if the USB device contains a capacitor, what you do if you plug that USB device into a working R-Pi, is that you essentially short the output of the USB port's +5V to GND, if even if its only for a very short time.

If you take no other precautions nothing at all will prevent that the +5V collapses to 0V for a (very) short period! Even a very short breakdown of the +5V will make the SoC crash, and reset.

So what to do?

Its simple, the capacitor shouldn't be allowed to be connected through a very good conductor to the +5V of the R-PI, and the R-PI itself should have another filled capacitor that is able to deliver enough current so the voltage on the R-PI stays "up", even when while plugging the USB device in, through a small resistance the USB device shorts the +5V for a very short time. Note that the "very small resistance" is essential, without it you will still put essentially zero ohm at time zero over the +5V. The resistance doesn't have to be big, just big enough. Something like the resistance of the above mentioned diode will indeed do the job, (but will also fail to provide 5.75V), a very small resistor of half an ohm or so will do the same, but without the permanent voltage drop of several 100mV.

[lrvick]

Note that if you use a diode (of any kind) in series with the power to the USB device, you end up with a similar voltage drop that you were trying to remove by bypassing the polyfuse (a few hundred millivolts).

The problem is actually simple to understand;
A good capacitor will , at the moment you connect it to a voltage source, act like a short.

That is at "time zero" any capacitor can be thought of as being a simple piece of wire!

So if the USB device contains a capacitor, what you do if you plug that USB device into a working R-Pi, is that you essentially short the output of the USB port's +5V to GND, if even if its only for a very short time.

If you take no other precautions nothing at all will prevent that the +5V collapses to 0V for a (very) short period! Even a very short breakdown of the +5V will make the SoC crash, and reset.

So what to do?

Its simple, the capacitor shouldn't be allowed to be connected through a very good conductor to the +5V of the R-PI, and the R-PI itself should have another filled capacitor that is able to deliver enough current so the voltage on the R-PI stays "up", even when while plugging the USB device in, through a small resistance the USB device shorts the +5V for a very short time. Note that the "very small resistance" is essential, without it you will still put essentially zero ohm at time zero over the +5V. The resistance doesn't have to be big, just big enough. Something like the resistance of the above mentioned diode will indeed do the job, (but will also fail to provide 5.75V), a very small resistor of half an ohm or so will do the same, but without the permanent voltage drop of several 100mV.

Currently the USB ports are stable for once with wifi adapters (which is a first from all of these mods). They are however as you said, only getting about 4.5v which is not ideal for long term and more sensitive devices, even if it is "working".

So, if I am following you correctly... the short may be happening in the USB device itself and the diode in my setup simply preventing the short from reaching the board?. Also, that you are recommending replacing the diode in my setup with a 0.5ohm resister AND a 220uF cap to maintain the same stability while also maintaining a full ~5v/1A+?

Thus something like:

(+5v) -- (0.5ohm resistor) -- (+ 220uF cap -) -- (wire) -- (USB V+ pins)

If so, I will certainly test that out tomorrow.

[Lob0426]

We keep having the discussion of the issues and the possible fixes. Now we have one. apparently, viable cure for the USB problems. What we do not have is any schematics of the possible circuits.

@Irvick:
(+5v) -- (0.5ohm resistor) -- (+ 220uF cap -) -- (wire) -- (USB V+ pins)

?...(-5v(TP2?)) -- (USB V- pins)

This is the closest we have so far. It does not show the ground circuit he used which looks like he used TP2. Another point is; that if we are basically hacking around the onboard power for the USB then why not cut the cord completely to board power? Remove the polfuses and don't jumper them. Then you do not need the diode @irvick used. Apparently you need a small value resistor (@mahjongg) in the +5v.

Would putting a polyfuse (like 500ma or so) in the +5v jumper act as the resistor and give us some protection in the bargain?


And by the way good work @Irvick. Now we are getting someplace here.

[Ed Raket]

Note that if you use a diode (of any kind) in series with the power to the USB device, you end up with a similar voltage drop that you were trying to remove by bypassing the polyfuse (a few hundred millivolts).

The problem is actually simple to understand;
A good capacitor will , at the moment you connect it to a voltage source, act like a short.

That is at "time zero" any capacitor can be thought of as being a simple piece of wire!

So if the USB device contains a capacitor, what you do if you plug that USB device into a working R-Pi, is that you essentially short the output of the USB port's +5V to GND, if even if its only for a very short time.

If you take no other precautions nothing at all will prevent that the +5V collapses to 0V for a (very) short period! Even a very short breakdown of the +5V will make the SoC crash, and reset.

So what to do?

Its simple, the capacitor shouldn't be allowed to be connected through a very good conductor to the +5V of the R-PI, and the R-PI itself should have another filled capacitor that is able to deliver enough current so the voltage on the R-PI stays "up", even when while plugging the USB device in, through a small resistance the USB device shorts the +5V for a very short time. Note that the "very small resistance" is essential, without it you will still put essentially zero ohm at time zero over the +5V. The resistance doesn't have to be big, just big enough. Something like the resistance of the above mentioned diode will indeed do the job, (but will also fail to provide 5.75V), a very small resistor of half an ohm or so will do the same, but without the permanent voltage drop of several 100mV.

Currently the USB ports are stable for once with wifi adapters (which is a first from all of these mods). They are however as you said, only getting about 4.5v which is not ideal for long term and more sensitive devices, even if it is "working".

So, if I am following you correctly... the short may be happening in the USB device itself and the diode in my setup simply preventing the short from reaching the board?. Also, that you are recommending replacing the diode in my setup with a 0.5ohm resister AND a 220uF cap to maintain the same stability while also maintaining a full ~5v/1A+?

Thus something like:

(+5v) -- (0.5ohm resistor) -- (+ 220uF cap -) -- (wire) -- (USB V+ pins)

If so, I will certainly test that out tomorrow.

@Irvick: Nice job! please keep posting these changes you make, and post them with a photo so we can "see" what has been done for easy understanding

THX for your effort, this is really getting somewhere now!

[lrvick]

We keep having the discussion of the issues and the possible fixes. Now we have one. apparently, viable cure for the USB problems. What we do not have is any schematics of the possible circuits.

@Irvick:
(+5v) -- (0.5ohm resistor) -- (+ 220uF cap -) -- (wire) -- (USB V+ pins)

?...(-5v(TP2?)) -- (USB V- pins)

This is the closest we have so far. It does not show the ground circuit he used which looks like he used TP2. Another point is; that if we are basically hacking around the onboard power for the USB then why not cut the cord completely to board power? Remove the polfuses and don't jumper them. Then you do not need the diode @irvick used. Apparently you need a small value resistor (@mahjongg) in the +5v.

Would putting a polyfuse (like 500ma or so) in the +5v jumper act as the resistor and give us some protection in the bargain?


And by the way good work @Irvick. Now we are getting someplace here.

My _current_ setup is:

[remove F1 & F2 polyfuses]

(+5v) -- (PT1007 diode) -- (wire) -- (USB V+ pins)

(-5v(TP2)) -- (USB V- pins)

The diode was still required in my case even after the polyfuses are removed as it prevents a short in the board at "time zero" as explained by mahjongg. Without this diode, even with removed polyuses, the board can still reset via a short in the USB device itself. The diode in conjunction with the removed polyfuses ensured complete isolation so this short never reaches the board.

The caveat to this approach, is that this diode consumes ~500ma itself, thus the USB ports are only getting ~4.5v. This means the board is fully protected, and the ports are getting stable power without drops, but the power they are getting is not quite in the acceptable range. Even if it is enough to make most USB devices "work" it could be bad for them long-term.

Tonight I will be testing mahjonggs proposal of a 0.5ohm resistor followed by a 220uF cap in place of the diode to ideally prevent the short while also preserving the full ~5v to the ports.

Will post results either way.

[transmition]

I attempted the pi pass, going off of and I'm not exactly sure what went wrong. I can still boot, but I'm still getting repeated keystrokes. My OK led isn't on, just my power one, and I'm using a 1A iphone charger. The keyboard is one from an old Dell Dimension 3100, with LEDs and USB ports (though nothing is plugged in, and the LEDs do turn on)

Here are some pictures:

[mahjongg]

The diode was still required in my case even after the polyfuses are removed as it prevents a short in the board at "time zero" as explained by mahjongg. Without this diode, even with removed polyuses, the board can still reset via a short in the USB device itself. The diode in conjunction with the removed polyfuses ensured complete isolation so this short never reaches the board.

The caveat to this approach, is that this diode consumes ~500ma itself, thus the USB ports are only getting ~4.5v. This means the board is fully protected, and the ports are getting stable power without drops, but the power they are getting is not quite in the acceptable range. Even if it is enough to make most USB devices "work" it could be bad for them long-term.

The diode is simply acting as a resistor in this case, (thats why it solves the reset problem) but the downfall, (as I explained) of using a diode is that one again you have 500mV (not mA) of drop over it (or 200mV if you use a Schottky diode). This means, that just as before the USB device is only getting something like 4.5 Volt, with a 1 Ohm resistor it would get something like 4.9 Volt (0.1 Volt per Ohm at 100mA).
The diode function of the diode in the meantime is a little bit irrelevant.

[Lob0426]

I attempted the pi pass, going off of and I'm not exactly sure what went wrong. I can still boot, but I'm still getting repeated keystrokes. My OK led isn't on, just my power one, and I'm using a 1A iphone charger. The keyboard is one from an old Dell Dimension 3100, with LEDs and USB ports (though nothing is plugged in, and the LEDs do turn on)

Here are some pictures:

Look for solder bridges, solder that ran over to other solder pads or components. It looks like you may have one over to the USB socket pad. You should also look near the incoming polyfuse you got awfully close to the component below it. It takes very little solder to connect and create a short. Tin your wires first (heat with iron and stick solder to them). You then will need less solder to make a solid connection.

[mahjongg]

Far too complicated with all the wires and stuff, and the diode.
The addition of the diode will negate what you are trying to do, which is not to have a few tenth of a volt loss between the voltage coming in, and the USB device.

No wires needed!

Simply take two 1 Ohm quarter watt or eight watt resistors.
simply solder a resistor over both sides of the green polyfuse, (F1) . That is one wire on either side of the same polyfuse. Repeat for the second polyfuse (F2). That is All!

The tracks on the PCB board should be plenty wide enough for say 500mA to the USB ports.
There is no reason at all to solder a wire between one ground point and another, especially not on a board as small as this.

All you need to do is to solder two 1 ohm resistors over the polyfuses.
I doubt that plugging in an USB device will reset the board with an 1 Ohm resistor, but in case it does there are two ways to mitigate that effect.
First rise the resistance a little, say to 2.2 Ohm.
or
Add more capacitance on the 5V on the R-PI's board side, best place is over the existing cap, the yellow tantalum cap in between the two polyfuses.
Polarity of the cap is indicated on the board, (plus on top) a good value would be 100uF and at least 10Volt.

[Lob0426]

@mahjongg: You may be right about that for the power circuit. But there seems to be a problem with the -5v circuit also, to the USB. Several who have tried the PiPass have lost the lower port. When a jumper is ran to TP2 it seems to cure the lower port and for some the resets also. So some of the problems might be on the other end of the circuit. If the resistors over the polyfuses works, then just increasing the polyfuses rating should work also, though they may not have just enough resistance. It would be nice if @dom, @Gertvanloo or even @Eben could weigh in on this. I know they are busy and might not want to take an official stance on RasPi hardware mods.

[mahjongg]

@mahjongg: You may be right about that for the power circuit. But there seems to be a problem with the -5v circuit also, to the USB. Several who have tried the PiPass have lost the lower port. When a jumper is ran to TP2 it seems to cure the lower port and for some the resets also. So some of the problems might be on the other end of the circuit. If the resistors over the polyfuses works, then just increasing the polyfuses rating should work also, though they may not have just enough resistance. It would be nice if @dom, @Gertvanloo or even @Eben could weigh in on this. I know they are busy and might not want to take an official stance on RasPi hardware mods.

There is no "-5V" (negative voltage) circuit on the PI!
If you are referring to the ground return, then I can say that absolutely no reason at all exist to "reinforce it", both Gnd pins of the USB have an identically hard connection to the ground plane, and to the capacitor C32.
Also, the upper port and lower port are identical!
Please refer to the schematic.
It can be downloaded here:
http://www.raspberrypi.org/wp-content/u ... s-R1.0.pdf

And yes, just increasing the polyfuse ratings would work also, if you can buy 700mA polyfuses that fit the board these would be an ideal solution.

Forget all the nonsense mentioned earlier in this thread, brought on by experimentation without real understanding of the issues involved. No need for wires and diodes! Just replace the polyfuses with a higher rating, or simply with resistors in the neighborhood of 1 Ohm. If the +5V on the R-PI dips when plugging in a USB device, it's due to "inrush current issues", the empty capacitors on said device that gets charged from the PI's own onboard capacitor, (C32 47uF located near the USB ports placed there especially to provide power to charge capacitors on USB devices) if you do not limit the inrush current with which charging an empty capacitor happens the required currents will be enormous, and capacitor C32 on the PI can't "keep up", a 1 Ohm resistor will limit this current enough so that capacitor C32 on the R-PI can deliver enough power.


For more details about "inrush current limiting" on USB devices when hot plugged, read paragraph 7.2.4.1 of the official USB specs, that can be found here:
http://www.poweredusb.org/pdf/usb20.pdf

Hope this will satisfy any doubters.

[transmition]

I cleaned up the soldering a bit, and checked for accidental bridges, and couldn't find any. I discovered that the Dell keyboard is rated at 1.5A, though I also tried with an Apple keyboard (that also has usb ports, which of course had nothing plugged in) that was rated at 200mA. Both keyboards still had repeated and missing keystrokes. The "OK" LED stays on up until arch linux starts the network. It does not matter whether an ethernet cable is plugged in or not, once it tries to start up the network the LED turns off.

More photos:








The shine from the camera makes it look like the connection between the usb solder pad and ground are touching, but they aren't. I feel pretty frustrated with my pi right now

[transmition]

I just noticed that even if I boot up the pi with nothing plugged in, the OK light still turns off.

[itimpi]

I just noticed that even if I boot up the pi with nothing plugged in, the OK light still turns off.

With the latest firmware the OK light indicates tht the SD card is being accessed so one expects it to occasionally flash to indicate activity - not stay on.

On a different issue, I would not expect any keyboard that includes a USB hub to work plugged directly into the Pi as the hub part will almost certainly try and draw more power than the Pi can supply.

[jbeale]

For what it's worth- if anyone in the R-Pi design department reads this thread- a different approach to USB power is taken by the Olimex "OlinuXino Micro" which is a small ARM board (advertised as a "Raspberry Pi alternative" although it is lower-spec in many areas). They use a Silergy SY6280 protection device on their USB host port +5V. This is a programmable current limiter, also with a shutdown input and reverse-current protection (no back-feeding from cheap powered hubs). The FET on-resistance is a low 0.08 ohms. Output current limit programmable from 0.4 A to 2 A. Comes in a small SOT-23-5 package, and is presumably cheap. I'd never heard of this part before, maybe because the documentation (if any?) is apparently only in Chinese, and it seems part is not sold in the US. Board schematic:
https://github.com/OLIMEX/OLINUXINO/blo ... _B.sch.pdf

[mahjongg]

There is an abundance of such parts, the reason the PI doesn't use them is simply cost. These IC's costs dimes, a polyfuse cents.

[lrvick]

Simply take two 1 Ohm quarter watt or eight watt resistors.
simply solder a resistor over both sides of the green polyfuse, (F1) . That is one wire on either side of the same polyfuse. Repeat for the second polyfuse (F2). That is All!

Went ahead and tried this verbatim. Wireless adapters work in either port and everything just... works. No capacitor upgrade was needed.

Still not sure why the bottom usb port would never work with the other solution when it didn't have the added ground, but it hardly matters anymore.

This is clearly the solution we need to move on to guys. Go pick up a couple 1 Ohm quarter watt resistors and enjoy your happy usb ports

[Ed Raket]

Simply take two 1 Ohm quarter watt or eight watt resistors.
simply solder a resistor over both sides of the green polyfuse, (F1) . That is one wire on either side of the same polyfuse. Repeat for the second polyfuse (F2). That is All!

Went ahead and tried this verbatim. Wireless adapters work in either port and everything just... works. No capacitor upgrade was needed.

Still not sure why the bottom usb port would never work with the other solution when it didn't have the added ground, but it hardly matters anymore.

This is clearly the solution we need to move on to guys. Go pick up a couple 1 Ohm quarter watt resistors and enjoy your happy usb ports

That is great news! could you messure the voltage before and after the fuse/resistor combi?

and maybe post a picture...

[Lob0426]

That is good news. Did you plug and unplug to see if your RasPi was resetting?

I just checked my USB polyfuses F1 is 7 ohms , F2 is 5 ohms and F3 is .001.
So 1 ohm is a lot better than that.