How much current can I put into my RPi?

Using a larger source curent capable supply will not cause any problem, as the PI will draw only the current that it needs.

The usb ports are current limited, so the problem there is that you WILL need a powered hub if you need to use power hungry devices.

The power supply you use MUST be regulated as the voltage must be 5.0V +/- 0.2V.

Millard Watson II wrote:

 

Using a larger source curent capable supply will not cause any problem, as the PI will draw only the current that it needs.

True but a high current capable power supply is very unforgiving of any mistakes like shorts. Then you will get in to the track melting situation rather than the over current trip on the power supply being activated.

I think it would be more inexpensive on you to try a generic powered hub if your app works you are fine if not then try a defferent hub. Of course you have a usb cable with cut power leads, so put a current meter inline and check the current draw. you will have a means to test any hub you try.

Thanks Mike for this suggestion.

I might try this when I get a chance (to go out and buy an ammeter).

Fergus Byrne wrote:

 

Is there anyway that I can make sure that the RPi's power is not being drawn by the hub (ie. a data only USB cable)?

 

I attemped to make a data only cable by cutting the Green and White wires in the USB cable, but then the RPi would not recognise it.

According to Wikipedia's USB article (http://en.wikipedia.org/wiki/USB#Physical_appearance), Green and White are the data pair so you don't want to cut them to make a data-only cable.  You want to cut the Red (VBUS) wire, or whatever color is connected to the +5V connector pins.  Leave the Black (GND) line intact.

The best way to make sure RasPi is not driving a powered hub is to power RasPi directly from the hub using a Standard A to Micro B cable.  This does not always work, because the hub may have current limiting circuitry.  Cheap hubs usually connect a common +5V to all downstream ports, so they'll work.  An expensive hub may limit a port's current to 500 mA to protect the other ports from a bad USB device, and that may not provide enough current for RasPi.  A "Y" cable that combines the current from two downstream hub ports should work in this case.

If RasPi and the hub have independent power supplies, there's no way to ensure that the hub +5V is at least equal to RasPi.  If RasPi is higher, it will drive the hub until F1 or F2 fails.  If the hub is higher, it may drive RasPi if it's a really cheap hub and omits an ORing diode in its upstream port.

You usually cannot tell from a hub's published information what it's going to do -- you have to take it apart and play with an Ohmmeter to see how it's connected up, and whether there's an upstream ORing diode.

John Beetem wrote:

 

Fergus Byrne wrote:

 

Is there anyway that I can make sure that the RPi's power is not being drawn by the hub (ie. a data only USB cable)?

 

I attemped to make a data only cable by cutting the Green and White wires in the USB cable, but then the RPi would not recognise it.

According to Wikipedia's USB article (http://en.wikipedia.org/wiki/USB#Physical_appearance), Green and White are the data pair so you don't want to cut them to make a data-only cable.  You want to cut the Red (VBUS) wire, or whatever color is connected to the +5V connector pins.  Leave the Black (GND) line intact.

 

The best way to make sure RasPi is not driving a powered hub is to power RasPi directly from the hub using a Standard A to Micro B cable.  This does not always work, because the hub may have current limiting circuitry.  Cheap hubs usually connect a common +5V to all downstream ports, so they'll work.  An expensive hub may limit a port's current to 500 mA to protect the other ports from a bad USB device, and that may not provide enough current for RasPi.  A "Y" cable that combines the current from two downstream hub ports should work in this case.

 

If RasPi and the hub have independent power supplies, there's no way to ensure that the hub +5V is at least equal to RasPi.  If RasPi is higher, it will drive the hub until F1 or F2 fails.  If the hub is higher, it may drive RasPi if it's a really cheap hub and omits an ORing diode in its upstream port.

 

You usually cannot tell from a hub's published information what it's going to do -- you have to take it apart and play with an Ohmmeter to see how it's connected up, and whether there's an upstream ORing diode.

Now I know why my data-only cable wasn't working.  I cut the ground and the 5v.  I can check this later now I know what was wrong with my data only cable.

Thank you.

What is the maximum current pull without PolyFuses allowed on a single USB port? Is it 500mA (the maximum allowed due to bus wieth)? or a limit of the power supply?

Fergus Byrne wrote:

 

What is the maximum current pull without PolyFuses allowed on a single USB port? Is it 500mA (the maximum allowed due to bus wieth)? or a limit of the power supply?

I assume you're asking how much current a USB port can draw when you remove polyfuses F1 and F2.  This ends up being a complicated answer.  As in my last answer, if you're providing power to RasPi through Micro USB power jack S1 then polyfuse F3 limits you to 1A for all RasPi chips (approx 500 mA) plus whatever GPIO connector P1 is drawing plus whatever your USB devices need.

If you're powering RasPi through P1, then you're limited by the current-carrying capacity of PCB traces.  For example, if the weakest PCB trace can only carry 1.5A then that trace will become a non-resettable fuse if your current exceeds that limit.  We don't know what the limits are because we don't have Gerber plots of the PCB layers and don't know how thick they are.  It's just as well, because each PCB revision would be different.

With fuses F1 and F2 gone or made Ohmless, you also have the possibility of powering RasPi through a downstream Type A USB port (S7).  While this is a violation of the standard use of a downstream USB port, it is done in practice sometimes by oversight and sometimes deliberately.  Drew Fustini describes a wonderful RasPi laptop using a Motorola Atrix Lapdock.  The Lapdock acts as a USB hub but also acts as a cell phone charger.  So it provides current the wrong way from the hub (inside the Lapdock) to RasPi.  In this case, the current from the Lapdock powers P1 and RasPi's other USB port.  You are limited by the Lapdock's current capacity and by PCB traces.

Note that if you power RasPi through S7 with Ohmless F1+F2, you've lost all F3 protection so you must trust the Lapdock or other non-compliant hub to behave properly.  However, with a rev 1.0 RasPi PCB you can replace F1+F2 with larger capacity fuses (say 1.0A) and get your protection back.

With fuses F1 and F2 gone or made Ohmless, you also have the possibility of powering RasPi through a downstream Type A USB port (S7).

Is this really possible?  I had a USB 4 port Hub (powered) and when I plugged it into the Raspberry Pi USB port, some of the power went back into raspberry Pi and started the boot process but it failed after 1 second.

From what your saying, This would succeed if I remove the the F1 and F3 fuses.

Fergus Byrne wrote:

 

 

With fuses F1 and F2 gone or made Ohmless, you also have the possibility of powering RasPi through a downstream Type A USB port (S7).

 

Is this really possible?  I had a USB 4 port Hub (powered) and when I plugged it into the Raspberry Pi USB port, some of the power went back into raspberry Pi and started the boot process but it failed after 1 second.

 

From what your saying, This would succeed if I remove the the F1 and F3 [I think you mean F2] fuses.

By "fuses F1 and F2 gone", I'm actually referring to RasPi 2.0 which -- as far as we can tell without schematics -- directly connects RasPi 5V0 to the downstream USB VBUS pins.  You can get the same effect on RasPi 1.0 by shorting out F1 and F2.  One way to do this is with 0 Ohm resistors that are the same size as F1 and F2, as RasPi has done with some recent boards.  You can also short them out with pieces of wire, leaving F1 and F2 in place.  You can also short them out using a higher-current fuse in parallel, which has the advantage of leaving in some protection.

If you remove F1 and F2 from a rev 1.0 RasPi, you won't get any connection between RasPi 5.0V and your hub's 5.0V, so you will need to provide separate external power for RasPi and your hub.

Your "reboot in 1 second" behavior is well known.  What happens is that the SoC can get enough current to start the boot process, but when it gets under way and needs more than 140 mA, the voltage drop across F1 or F2 (depending on which USB port your hub is plugged into) gets too high and 5V0 drops too low to keep the SoC alive, so it reboots.

So if you're adventurous and don't mind voiding your RasPi's warranty, shorting out F1 and/or F2 and powering RasPi through your hub's upstream connection will probably work.  After all, that's what's RasPi is allowing officially with the 0 Ohm resistor change.

Also, this will only work with some hubs.  A hub is not supposed to send current upstream.  My hub has a diode to prevent this from happening.  However, there seem to be quite a few hubs out there that do source current upstream, so RasPi decided to fuggeddabouddit and short out F1+F2 rather than de-list a bunch of hubs and listen to complaints.

This is what I mean:

Is this what you are also talking about?

Fergus Byrne wrote:

 

This is what I mean: [figure]

 

Is this what you are also talking about?

Yes.  You can power a RasPi that way if both of these are true:

1.  Your USB hub is able to source current through its upstream port.  This is not standard USB hub operation, but some hubs do it anyway.

2.  Your RasPi's downstream USB port is not protected by a 140 mA polyfuse F1 or F2 because:

a) It's a rev 2.0 RasPi which does not have F1+F2, or

b) It's a newer rev 1.0 RasPi which replaces polyfuses F1+F2 with 0 Ohm resistors, or

c) You have modified an older 1.0 RasPi to short out F1+F2, violating the warranty.

Fergus Byrne wrote:

 

This is what I mean: [figure]

 

Is this what you are also talking about?

Yes.  You can power a RasPi that way if both of these are true:

1.  Your USB hub is able to source current through its upstream port.  This is not standard USB hub operation, but some hubs do it anyway.

2.  Your RasPi's downstream USB port is not protected by a 140 mA polyfuse F1 or F2 because:

a) It's a rev 2.0 RasPi which does not have F1+F2, or

b) It's a newer rev 1.0 RasPi which replaces polyfuses F1+F2 with 0 Ohm resistors, or

c) You have modified an older 1.0 RasPi to short out F1+F2, violating the warranty.

Thank you fo rthe reply.