hippy
Posts: 6107
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Power supply news!

Mon Oct 24, 2011 3:46 pm

[quote]Quote from jacklang on October 24, 2011, 13:40
[quote]
liz on October 21, 2011 at 11:12 am said:

There’s a polarity protection diode, a voltage clamp, and there’s a self-resetting semiconductor fuse onboard, so you shouldn’t have to worry about blowing your Raspi up.
[/quote][/quote]

That to me begs more questions than it answers. I don\'t blame liz as she\'s only the messenger.

What voltage drop is there through this polarity protection diode ? Does it mean that if one wants to use hardware connected to the R-Pi which needs 5V from the 5V line then the input voltage has to be 5.6V ? Perhaps it\'s a Schottky Diode but we still don\'t know the voltage drop. Maybe it\'s a \'crowbar\' diode ?

What does the voltage clamp actually clamp to ? Not a 5V rail as there isn\'t one, so presumably to 3.3V. And what form does it take ? If it\'s a diode to 3.3V then is it not simply a diode feed from 5V into 3.3V regulated, effectively dumping 5V into the 3.3V rail ?

[quote]liz on October 21, 2011 at 11:12 am said:
We’ll also be selling a tested and approved power supply from this website
[/quote]

That\'s good news and welcome, but we still have the issue of people not wanting to buy that particular supply, wanting to use one they may already have. It\'s a question of is that compatible with their R-Pi ?

User avatar
abishur
Posts: 4477
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 3:57 pm

Sorry in advance for how long this will be, but there\'s some confusion that needs to be cleared up. I believe it started by this post Gert made

[quote]
Another problem is that some things are self evident for \"those in the know\" and they don\' even realize it. I am an electronics engineer so for me it was \"self evident\" that an external 5V supply means there is no longer a 5V regulator. In this case you can accept it from this authoritative figure that: there is no 5V regulator on the board. (I have the schematics...) [/quote]

Which (ironically?) was made in response to hippy noting that a lot of people are making very authoritative statements when the truth is they\'re just guessing ;)

Now the issue Lobo and I have been talking about is this lack of a voltage regulator. For those who don\'t know, a voltage regulator let\'s you take a higher voltage and maintain a consistent lower voltage (i.e. take a 6-20V DC range and give a steady 5V DC output). Now what we had forgotten is Liz\'s post above (thanks jack!) which states that even though the PSU is not regulated any more it *is* protected. If we give it too much voltage we\'ll blow a self-resetting fuse. (Here\'s a link to wikipedia\'s page on fuses, search for \"self-resetting\" and there\'s a shot paragraph on what a self-resetting fuse is... though the name is pretty self explanatory :P )

Here\'s another post from Gert warning us that the original 1W estimate isn\'t gonna happen
[quote]
Using linear regulators means the device will use more then the originally promised one watt. (Sorry but the price promise is what we really had to keep). I have written a document to explain why for the non-electronic engineers subscribers.
http://www.scribd.com/doc/69700160/Rasp ... More-Power
[/quote]

All of which is to say the r-pi board is safe from over-voltage, but your PSU is still in danger of dying if it\'s not capable of sustaining a continuous voltage high enough to power the board + any peripherals attached to the USB ports (which are powered according to this comment left by liz in the front page post announcing the power supply

[quote]
That’s basically right. Effectively, you have to negotiate for that power; the board will not repond to a device which requests more power than can be supplied. You wouldn’t, for example, be able to be able to plug in 2 USB hard drives, and have them both work. Instead, you’d have to use a powered hub.
[/quote]
)

@Hippy - I wouldn\'t worry excessively about voltage drop. If a 5V source was going to be insufficient then they wouldn\'t say we could use a 5V PSU. These are veteran professionals here, not some fly by night start up ;) There actually is a 5V rail. I forget the post or I\'d add it to this increasingly long post, but the broadcom chip actually separates the input into the different rails... or does the 5V get sent straight through to USB? Well rather than pretend my speculation is fact on this point, I\'ll let Gert clarify, but there is 5V for the USB ports so it\'s not like there\'s nothing for the voltage clamp to clamp onto :)
Dear forum: Play nice ;-)

User avatar
liz
Raspberry Pi Foundation Employee & Forum Moderator
Raspberry Pi Foundation Employee & Forum Moderator
Posts: 5202
Joined: Thu Jul 28, 2011 7:22 pm
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 4:13 pm

I was going to say something much shorter and terser, but Abishur\'s covered it all.
Director of Communications, Raspberry Pi

Johannes
Posts: 64
Joined: Sun Jul 31, 2011 8:05 pm

Re: Power supply news!

Mon Oct 24, 2011 4:24 pm

[quote]Quote from Lob0426 on October 24, 2011, 13:21
A sub \"C\" NiCad or NiMh cell can deliver up to 70amps of continuos current until the cell has been depleted.[/quote]

If there are cells that can deliver that much current, they\'re top notch with a top notch price tag. Even racing packs are typically not rated for more than 40A. Not that it makes a difference...

[quote]Quote from Lob0426 on October 24, 2011, 13:21
Remember that the regulators that are now being used burn wattage independent of the RasPi itself, load or no load.[/quote]

The amount of power \"wasted\" by a linear regulator is proportional to the load current and proportional to the voltage difference between the input and the output. (For example 700mA at 5V drawn from a 12V supply means the linear regulator will use (12V-5V)*0.7A=4.9W to heat the room.) No load means (almost) no loss in the linear regulator. For low loads, switching regulators are hardly better. With the small footprint of the Raspi and given the assumption that there won\'t be any heatsinks on it, the current drawn from the onboard linear regulators is probably very small. Most of the power consumption will be through the internal switching regulator of the SoC. What happens offboard is a different matter entirely: 0.7A is no small current for a linear regulator. 5W of waste heat (see example above) requires a heat sink. It\'s a shame that people will still use linear regulators in battery powered applications simply because switching regulators are hard to find and cost an order of magnitude more. All to save a buck or two on the Raspi.

Johannes
Posts: 64
Joined: Sun Jul 31, 2011 8:05 pm

Re: Power supply news!

Mon Oct 24, 2011 4:55 pm

I am by no means an electronics expert, but here\'s the way I think the protection circuit may work: A fuse directly after the connector, a diode between the circuit-side of the fuse and the other pole of the connector to \"short-circuit\" wrong polarity and a voltage clamp anti-parallel to the diode to \"short-circuit\" excessive voltage, so that wrong polarity or excessive voltage will \"instantly\" trip the fuse. This way there\'s no voltage drop due to the diode because it\'s not involved if the polarity is correct. Please do correct me if I\'m wrong.

User avatar
Gert van Loo
Posts: 2486
Joined: Tue Aug 02, 2011 7:27 am
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 4:59 pm

Yes the 5V coming in goes straight to the USB. (Apart from the fuse of course.) As things stand now that 5V net is also connected to two pins on the GPIO connector. I have had some questions about that: is it 5V in or 5V out, on which I answered that a 5V copper strip does not have \'ins\' or \'outs\'.
If you really want to you can replace the fuse with a bigger one and you can connect two 1A consuming USB devices. Just make sure you pump enough current into the board. There probably is an upper limit at which the copper tracks are not sufficient for the current but I don\'t know where that is and we are then talking silly currents....

kattle87
Posts: 34
Joined: Mon Oct 10, 2011 7:50 am

Re: Power supply news!

Mon Oct 24, 2011 5:13 pm

One question: almost everyone here is putting a 0.7 amp as a possible draw for the raspi.
Do you mean it will continuously draw the same current with no differences given for different CPU/GPU loads?
I would expect some differences between full load and low load. Not sure about it but I really hope this makes any sense. After all, this chip is similar to the ones used in smartphones, so there should be some power saving tricks. And even without any trick at all, I think a CPU barely running anything still consumes less than a 100% one, right?

jacklang
Posts: 166
Joined: Thu Aug 04, 2011 10:59 am

Re: Power supply news!

Mon Oct 24, 2011 5:19 pm

Aarrgh!
I say again 0.7A is including power drawn from the USB ports when the board acts as a host. RPi board on its own is much less, and it is load dependent.

Panyk
Posts: 2
Joined: Mon Oct 24, 2011 4:39 pm

Re: Power supply news!

Mon Oct 24, 2011 5:52 pm

Just going to ask a few questions, some that I\'m wondering, some just to have a definitive answer and clear the air;
1. \"Unpowered\" USB ports refers to the fact that they won\'t have the connections/juice necessary to charge a battery, but they\'ll still have the needed power to run an external HDD or optical/laser mouse, correct?
2. Is the 0.7A figure the maximum that it will draw? Will it vary between the two initial models? (As there\'s apparently been a few uncertainties, is this an accurate figure?)
3. What would happen when the board becomes undervolted? Undercurrented? Overcurrented? The fuse has been stated to trip when the voltage goes too high; will it trip right at say, 5.01V, or at a tolerance value like 5.25?
Apologies in advance if these have already been asked, this\'s just been nagging at me.

bnolsen
Posts: 76
Joined: Sat Aug 13, 2011 5:37 pm

Re: Power supply news!

Mon Oct 24, 2011 6:34 pm

@panyk

reading wikipedia on usb-2.0 the current is broken up into \"100mA\" blocks. An unpowered usb 2.0 port needs to only supply one \"block\" to each device. A powered usb 2.0 port by spec need only deliver up to 5 blocks (500mA).

This implies that the R-PI need only supply 100mA to each downstream USB device. Unless otherwise stated somewhere by the rpi crew I would only assume 1 block (100mA) per usb device will be available in the model B units. One block is enough to power mouse, keyboard, game controller, flash drive, etc.

A spinning drive? You\'ll probably need an external powered hub.

tufty
Posts: 1456
Joined: Sun Sep 11, 2011 2:32 pm

Re: Power supply news!

Mon Oct 24, 2011 6:37 pm

You will *not* be able to spin up a hard disk without a powered hub. Certainly not on 100mA, and you\'d be lucky to do reliably it with anything much less than 2x500mA.

A quick back-of-a-fag-packet calc shows that you\'re not gonna get much more than 100mA per USB port, especially with ethernet taking a bite out of your power budget.

Simon

User avatar
abishur
Posts: 4477
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 6:50 pm

*edit* I was technically incorrect about max voltage supplied by USB ports, I\'ve changed my post to reflect the right information as provided by the r-pi team in this thread.

1. Yes and No, the USB ports are powered (they receive the power straight from the PSU), but at this time there\'s a fuse that limits the available voltage to the ports to 100 mA. If you replaced the fuse with a higher amp one then since the 5V \"rail\" is connected directly to the USB ports then the simple answer is the USB ports will provide as much power as your PSU can handle though I would be wary of going above the usb 2.0 standard of 500 mA per port).

2. The .7Amp figure I\'ve been mentioning is coming off this post on the front page. At the time of this post it\'s the second post on the front page, but the gist of the post is that .7 Amps is a theoretical max draw based on max load of the board itself and a high powered peripheral attached to it as well.

*EDIT*: I made a mistake here saying a Eben estimated 300 mA for model B, the front page says 300 mA for model A. I\'ve corrected the mistake in this post to avoid confusion. */EDIT*

It\'s been speculated (by Eben) that 300 mA would suffice to run a model A. As you can imagine for the Model B, however, two USB ports supplying 500 mA and an additional 300 for just the board means that you\'d actually need a 1.5 to 2 Amp PSU if you wanted to \"max out\" the USB ports, but in the event that the ports ask for more than the PSU has to offer, Liz has already told us the the r-pi is smart enough to not provide more than the PSU has and the device will not receive power.

This is where the powered vs unpowered really comes into play. I\'ve yet to meet a USB port on a computer that was not powered off the computer\'s PSU, but a USB *hub* can be unpowered (relies on the computer\'s USB port to power any devices attached to it) or powered (has it\'s own PSU to provide power to its USB ports). *IF* you need more power than the max provided via the USB ports then you\'ll have to get a powered USB hub. But then that raises the question, if you\'d have to buy a powered usb hub anyways, why not just get a hard drive with its own PSU or even its own wall wart usb adapter?

To answer the rest of question two (Will it vary between the two initial models) the answer should be only slightly provided that there is nothing attached to the boards ;) The only major differences in power *should* be based on what you have attached to the USB ports and maybe the ethernet port?.

3. Just like any other electronic device, when there\'s an undervoltage, the r-pi will turn off :P I also don\'t know the tolerances of the fuse, but a 1% tolerance on a fuse would be pretty ridiculous IMHO, I think the tolerance will be strikingly similar to the tolerance found with cell phone chargers. (We are dealing with a mobile architecture here ;)
Dear forum: Play nice ;-)

User avatar
Gert van Loo
Posts: 2486
Joined: Tue Aug 02, 2011 7:27 am
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 6:50 pm

0.7A should be sufficient to power the BCM2835 + LAN9512 + HDMI + two USB ports drawing about 100mA each. As with most electronic equipment: under power a bit and it will give you unreliable result . If you under power a lot it will fail spectacularly. I honestly don\'t know what happens if you under power it and write to the SD card at the same time....

User avatar
abishur
Posts: 4477
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 7:03 pm

Well, I stand corrected in my math! Thanks Gert, is 100 mA a max for the USB ports or can they take more if I use a higher amperage PSU?
Dear forum: Play nice ;-)

radix
Posts: 4
Joined: Tue Oct 18, 2011 5:35 pm

Re: Power supply news!

Mon Oct 24, 2011 7:59 pm

what will be the maximum safe current passing through raspb ?? for instance i\'m going to plug two external 2,5\'\' hard drives supplied via USB. will be there som current protection to avoid damage of raspb ?

User avatar
Gert van Loo
Posts: 2486
Joined: Tue Aug 02, 2011 7:27 am
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 8:25 pm

[quote]abishur: Thanks Gert, is 100 mA a max for the USB ports or can they take more if I use a higher amperage PSU?[/quote] see my post in this forum from 17:59. You have to use a bigger fuse.
[quote]radix: what will be the maximum safe current passing through raspb ??[/quote]
See my last remark post in this forum 19:50. But that was not serious. You never supply big currents from a distance unless you\'re desperate. You just get noise and voltage drop. Just don\'t!
That is why big USB drives have their own power socket.

Johannes
Posts: 64
Joined: Sun Jul 31, 2011 8:05 pm

Re: Power supply news!

Mon Oct 24, 2011 8:38 pm

radix, no active component on the Raspberry Pi is involved in supplying 5V power to the USB devices. There\'s a fuse, but other than that it\'s straight from the PSU to the USB peripherals (like it is with most powered hubs too). Except for melting the traces if you really really overdo it, current can not damage the Raspberry Pi. That said, you will need to replace the fuse with one rated for a higher current if you want to pass sufficient power for high-power USB devices through the Raspberry Pi.

Liz mentioned earlier that USB devices negotiate for power and that the Raspberry Pi will deny power that it can\'t supply. This is a software thing. Many devices do not properly negotiate for power. There is no way the Raspberry Pi can actually limit power to a USB device to enforce a power negotiation result. It will just \"turn off\" all power (including to itself) by tripping the (self-resetting) fuse if the total current exceeds the fuse\'s limit.

If the fuse\'s limit is close to the estimated maximum current (0.7A), then there won\'t be enough power to connect even just one external hard disk without using a powered hub or modifying the Raspberry Pi.

User avatar
abishur
Posts: 4477
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 8:44 pm

Ouch, can\'t believe I missed that! Thanks again Gert! I\'ll update my last post with the right information to hopefully avoid confusion.
Dear forum: Play nice ;-)

User avatar
abishur
Posts: 4477
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 8:52 pm

Also one last question. Is this just a plain Jane mini fuse that we can pop up and pop a new one back in, or is there a little more involved in replacing that fuse?
Dear forum: Play nice ;-)

User avatar
liz
Raspberry Pi Foundation Employee & Forum Moderator
Raspberry Pi Foundation Employee & Forum Moderator
Posts: 5202
Joined: Thu Jul 28, 2011 7:22 pm
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 11:06 pm

It\'s self-resetting. You just remove the power, and it resets itself. No popping or replacement of any sort required. 8)
Director of Communications, Raspberry Pi

User avatar
abishur
Posts: 4477
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: Power supply news!

Mon Oct 24, 2011 11:19 pm

Oh we were asking because we wanted to swap them out for a higher amperage one to give more power to the USB ports.
Dear forum: Play nice ;-)

jg
Posts: 12
Joined: Mon Oct 24, 2011 10:20 pm

Re: Power supply news!

Tue Oct 25, 2011 1:03 am

[quote]Quote from liz on October 20, 2011, 22:24
Head to the front page, where Eben has put up a splendidly technical new blog post. (Hint - it\'s 5V micro USB.)[/quote]

That\'s a good move, any info on the final orientation of this ?

I can only find the old overlay
http://www.raspberrypi.org/wp-content/u ... berry1.png

and this block diagram (from abishur:?)

https://lh4.googleusercontent.com/-kS29 ... 0Board.jpg

but I can see one rather glaring mistake on the PCB design :

It has connectors \"going in all directions\", which is not a good idea.

Ideally, these should be spun, so that only two opposite edges are used for connectors.

Doing this makes lead-dress, production testing, case design, and future revisions much easier.

User avatar
abishur
Posts: 4477
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
Contact: Website

Re: Power supply news!

Tue Oct 25, 2011 2:39 am

Well, calling it a \"glaring mistake\" would be a misunderstanding (and kinda off topic :P ). As I\'ve said again and again in these forums, this is not a fly by night start up business here. The guys in charge of design are seasoned veterans who work for some very high names in chip design (broadcom for example). The \"going all directions issue\" was deliberate. The ultimate goal of the project is to meet a $25 price point so it will be readily accessible to children. With this is mind there was two large issues with putting the devices on only two edges.

The first is the extra cost of the PCB itself. In order to do two edges the entire board would have to be elongated and as weird as it might sound, it would end up take a larger area and cost more. Yes this would be a small increase of price, but this project is working hard to bring the fullest option possible while meeting that 25 dollar price point. Even a 10 cent increase adds up quickly.

The second, and larger issue, is that as the items are put along less edges, it creates an interference issue within the layers. The ultimate resolution is to add more layers, but that a costly solution. They\'ve already had to increase the layers from 4 to 6 on the final board and that\'s with keeping things spread out the way it is.

So not a glaring mistake, a deliberate and strategic placement. ;) Hope that clears it up for you, but if you want more info, go ahead and make a new thread where we can discuss it more so we don\'t hijack this one :)
Dear forum: Play nice ;-)

jg
Posts: 12
Joined: Mon Oct 24, 2011 10:20 pm

Re: Power supply news!

Tue Oct 25, 2011 3:01 am

[quote]Quote from abishur on October 25, 2011, 03:39
So not a glaring mistake, a deliberate and strategic placement. ;) Hope that clears it up for you, but if you want more info, go ahead and make a new thread where we can discuss it more so we don\'t hijack this one :)[/quote]

OK, if you suggest a forum location and title, I\'ll elaborate on why I (still) think the connector placement belongs in the glaring mistake basket :)
Are there more recent PCB info\'s available ?

User avatar
Lob0426
Posts: 2198
Joined: Fri Aug 05, 2011 4:30 pm
Location: Susanville CA.
Contact: Website

Re: Power supply news!

Tue Oct 25, 2011 3:22 am

@Johannes those sub c cells can deliver a continuos 40+ amps, but they can deliver a whole lot more in a short time if not used properly.
512MB version 2.0 as WordPress Server
Motorola Lapdock with Pi2B
Modded Rev 1.0 with pin headers at USB

http://rich1.dyndns.tv/
(RS)Allied ships old stock to reward its Customers for long wait!

Return to “General discussion”