About power supply and peripherics

[typhoon]

Reading that last post on the main page, I started wondering about the possibility of my power source/setup.

I plan to have the RPi powered by a 2A power source, joined to a spare usb female cable, then joined to a microusb adaptor.

This should be perfectly fine, but since I want to plug a 2,5 '' portable hdd to the RPi, and have it work like that, remotely controlled, I started asking myself if the board would "support" the power drained by the hdd, in terms of mA.

Can someone help me? Thanks in advance.

[RaTTuS]

use a powered HUb for the HD - you should be able to power the RPI from the same hub using usb-> micro USB cable

[typhoon]

I know I can use a powered USB hub that way, but I'm trying to be "compact-minimal" here, I'm going for the smallest possible setup.

A typical USB 2,5'' HDD spin-up should be around 1,2A, so even is the power usage while in operative is less than 500mA (which is ok for the board), I was wondering how the board will react to the spin-up current peak.

If it can't, I'll just solder to the power supply the micro-USB cable directly, along with a power cable for the HDD; still I'd prefer not to.

It would be nice if the RPi could handle it, so I asked since there is no / I could not find info in that way.

[rurwin]

See the Q&A with Pete Lomas currently adorning the front page.


Can the fuses at USB be safely removed to provide >140 mA to USB or is there power concern that requires the limitation? [[Liz: I believe this is a question of interest to almost nobody but Abishur, but asked Pete to answer it because Abishur is a very helpful fellow around these parts!]]

The fuses kick in hard around 280mA and fold back and limit to 140mA. If you remove them then all you have for protection is the 700mA inbound fuse. The tracking on the board is good for 500mA+ so you could if you really wanted too. What about a powered hub – to power the Pi and bigger USB devices.


So the PCB tracks would be overheating at 1.2A, and the polyfuses would open-circuit even on the steady-state current.

[typhoon]

I was referring to that, but I couldn't elaborate it well. In short, RPi can power on USB just small devices, like mouse, kb, usb key, wifi dongle, and not external 2,5 ''HDD since they draw too much current.

Thanks rurwin for dumbing it down for me.

I guess I'm back to choosing between finding a small hub and soldering old pieces a bit...

Well, I have the time before my RPi arrives...

[rpt]

So does this mean that the RasPi's USB ports are limited to 100mA (one unit load) and a request to move to high-power mode and draw 500mA will be denied?

[RaTTuS]

rpt said:


So does this mean that the RasPi's USB ports are limited to 100mA (one unit load) and a request to move to high-power mode and draw 500mA will be denied?


from my understanding yes - it would be impossible

[rurwin]

RaTTuS said:


rpt said:


So does this mean that the RasPi"s USB ports are limited to 100mA (one unit load) and a request to move to high-power mode and draw 500mA will be denied?


from my understanding yes – it would be impossible


Not quite impossible. You could super-power one of the two USB slots. That would subject the PCB tracks to 600mA, which they could almost certainly cope with. You would have to change the (140mA) poly-fuse for the USB slot and the main input (700mA) poly-fuse.

However the power allocation is a software function. Devices are only permitted to draw 100mA unless they are given permission to draw more. Since software cannot know what you have done with the fuses, it will deny permission. So you may have to hack the Linux driver or, if you are unlucky, the setting may be located in the GPU firmware.

[curiousengr]

Most laptops/desktops would be able to supply the full rated 500mA as would a powered hub (for at least 1 port).

Any External USB Hard disk that has only 1 USB connection would be expected to power up within that limit- so no problem if you use a powered hub.

No need to hack the RasPi board and possibly break it!.

If the Disk needed more power then the manufacturer makes 2 USB connectors 1 Power+Data and 1 Power only- you get 2 USB leads.I have an External DVD Writer that is like that.

So you should be fine- as long as you haven't opened the case and replaced the Hard disk with a different one.

I am looking to make myself a distribution board to feed +5v with seperate leads to my 4 port powered Hub and RasPi everything else WiFi dongle,USB Keyboard,mouse should be ok to take power from the RasPi or the Hub.

[forumisto]

there are power supplies with several outputs of 5V and 1A.

If you will have the RPi and the USB hab near, you can use that power supply and join the cables.

[john_wage]

We need a DC booster in order to integrate LED lit LCD-TFT displays on internal power. Because anything above 6" will need between around 9V to 12V for the background lighting.

[Gert van Loo]

...or, if you are unlucky, the setting may be located in the GPU firmware.


I think the USB drivers are completely on the ARM. I will ask about that.

[jojopi]

Gert said:



...or, if you are unlucky, the setting may be located in the GPU firmware.


I think the USB drivers are completely on the ARM. I will ask about that.


Normally in Linux you can force a USB device into a specific configuration, regardless of believed power availability, via /sys/bus/usb/devices/*/bConfigurationValue.

I would be interested to know what the software knows about power by default, however.

[arm2]

Two out the three hubs I've tested have their 5V supply directly connected to 5V on the USB input socket, this would mean that if the external Hub was turned on before the Raspberry Pi's power then the Pi would try and draw its power through the Data USB port but I hear that there is a 140mA fuse on each USB port so it would blow (at 280mA according to Pete's Q&A session). If there was a diode on the USB port power line that would solve the problem but I can see only 4 diodes on the board and they aren't near the USB ports!

Can anyone confirm my deductions authoratively?

[Newbie]

I"m familiar with the 500 ma limit per USB port, but not the (software-negotiated?) 100 ma limit – is that in the USB spec?

Are the polyfuses one-time actual fuses that burn out or open the circuit permanently, or are they really current-limiters that shut down (break the circuit, aka a circuit breaker) for some period of time? If the latter, what is the shutdown period? Milliseconds? Seconds?

Do powered/active USB hubs prevent current flow back from their independent supply to the USB ports on the R-Pi board, or can the board ports also drive up to 500ma per port supplied by the hubs" power supply (assuming the hubs" supply has enough capacity)?

This would all be so much easier if I had an R-Pi sitting in front of me … or even behind me

[rurwin]

The 500mA is negotiated. Theoretically. No device should take more than 100mA until the host (or powered-hub) has enumerated all the devices, calculated the power budget and told the device that it can take more power.

In practise many hubs just supply the power, and I imagine many devices just take the power. However I know for a fact that my phone for example does not take full power unless it is told it can.

Search the RS website and wikipedia for "polyswitch". It's a self-resetting fuse. Its reset period is measured in minutes up to hours.

Powered hubs should not supply power to their upstream port. However we have seen one cheap device on here that does. Supplying power upstream is bad, and could cause problems like blown fuses, damaged equipment and (just possibly) fires.

[Burngate]

Looking at the data sheet for the 9512, it would appear there is nothing to stop a powered hub from supplying power back up the USB output. However, if it did then the fuse would promptly trip, and the Pi would lose that power. Then, when the proper power supply came up, the Pi would come up properly. At that point the Pi would have 5v on both sides of the tripped fuse, but that would be ok.
What might not be ok is when the tripped fuse decided to un-trip itself. At that point there are two PSUs, maybe at different voltages, fighting each other. A maximum of 280mA would flow one way or the other, but it could still cause problems.

[rurwin]

What would probably happen would be that the fuse would blow every time you used that hub, but you would not notice. Those fuses probably degrade each time they trip, so you may find that over time the 5V supply from that USB slot becomes under-powered.

On the other hand, if you had taken out the fuse in order to supply more than 100mA, then lots more current could flow and that could damage something.

[abishur]

I believe diodes 10 and 11 on the back side of the board will prevent power from coming back down the usb lines.

[jojopi]

Abishur said:


I believe diodes 10 and 11 on the back side of the board will prevent power from coming back down the usb lines.


D{10,11} are directly traceable on the Gerber screenshot to the 9512's PRTCTL{3,2} pins.  I believe they are the diodes shown in Figure 2.4 of the LAN9512 datasheet and are only to prevent the peripheral attempting to draw power from the 9512 after the fuse has opened.

It is not impossible that there is a second forward Schottky diode in the same package to prevent reverse power, but it seems unlikely and would be a very big assumption.  Edit: there do not seem to be either tracks or vias connected to the extra pins of D{10,11}, so I believe the extra integrated diodes can be ruled out.

I think rurwin is right that the fuses themselves are the reverse power protection.

[abishur]

jojopi said:


Abishur said:


I believe diodes 10 and 11 on the back side of the board will prevent power from coming back down the usb lines.


D{10,11} are directly traceable on the Gerber screenshot to the 9512's PRTCTL{3,2} pins.  I believe they are the diodes shown in Figure 2.4 of the LAN9512 datasheet and are only to prevent the peripheral attempting to draw power from the 9512 after the fuse has opened.

It is not impossible that there is a second forward Schottky diode in the same package to prevent reverse power, but it seems unlikely and would be a very big assumption.  Edit: there do not seem to be either tracks or vias connected to the extra pins of D{10,11}, so I believe the extra integrated diodes can be ruled out.

I think rurwin is right that the fuses themselves are the reverse power protection.



Good catch!  That's definitely why I preface comments I'm not sure about with belief   I know (see the difference from belief? ) that we've been told by the hardware guys at the RPF that the USB ports (and HDMI for that matter) do provide... what's the word?  voltage back-flow protection? (can I get a better word for that? it's slipping my mind) I would have thought that would be done with a diode, but the figure you referenced doesn't leave room for uncertainty there!

[Jim Manley]

Abishur said:


what"s the word?  voltage back-flow protection? (can I get a better word for that? it"s slipping my mind)


We generally call it "reverse current protection" on this side of the pond, which also covers Murphy"s Law situations like customers inserting batteries backwards (despite preventive mechanical alignment and labeling precautions), connecting cables incorrectly, etc. One way to remember is that current is a measure of electron flow quantity, and voltage is a measure of potential difference (analogous to fluid flow and pressure in fluid mechanics terms).

[abishur]

Jim Manley said:


Abishur said:


what"s the word?  voltage back-flow protection? (can I get a better word for that? it"s slipping my mind)


We generally call it "reverse current protection" on this side of the pond,


There we go! I'm working on a waste water plant HMI right now so all I could think of was black-flow prevention