USB Port Current Boost (solved)

[Burngate]

n00b question:
Why are you using 15V fuse? Shouldn't it be like 6V?
Is it ever going to trigger when it's operating voltage is 15V but the board runs @5V??

Fuses work on current, not voltage. While they're in their "on" state they don't care what voltages are around - the volts on either end are (almost) the same.
It's only when they're blown that they care about volts - now there's different volts at either end, and they can only stand so much.

For a classic fuse - piece of thin wire - too many volts can give you sparks etc.
For a PTC there's still current flowing even when it's blown, 'cos it's not got infinite resistance, and the more volts the more current, and so more power, and you could end up with it no longer resettable.

So (like most things) they've got a maximum voltage, but less is ok

[Lob0426]

n00b question:
Why are you using 15V fuse? Shouldn't it be like 6V?
Is it ever going to trigger when it's operating voltage is 15V but the board runs @5V??

Fuses work on current, not voltage. While they're in their "on" state they don't care what voltages are around - the volts on either end are (almost) the same.
It's only when they're blown that they care about volts - now there's different volts at either end, and they can only stand so much.

For a classic fuse - piece of thin wire - too many volts can give you sparks etc.
For a PTC there's still current flowing even when it's blown, 'cos it's not got infinite resistance, and the more volts the more current, and so more power, and you could end up with it no longer resettable.

So (like most things) they've got a maximum voltage, but less is ok

Thank you @Burngate for explaining that. I kinda missed the point of his question!

[Stomp]

Thank you @Burngate for explaining that.

+1
I should have read Polyfuses explained before asking

[jbeale]

I just received my 2nd R-Pi, this one from stock (yay MCM Electronics) which took just a week to ship from Ohio to California. I was pleased, but surprised to see that the USB fuses F1, F2 are now just 0-ohm jumpers (photo below). I was able to confirm they measure less than 0.1 ohms. Is this the new standard going forward? I haven't been paying attention, I guess. The input fuse F3 is still a fuse and appears to be the same 750 mA rating type as before (measured on this board around 0.35 ohms cold).

https://picasaweb.google.com/1099282360 ... 0375008498

[Burngate]

You've obviously recieved a Chinese pirate copy - it won't have all the features a proper Pi would have, such as not working with some keyboards, and the like.

Either that, or the factory ran out of poly-fuses, so they put on those links, hoping no-one would notice

[RaTTuS]

what are the serial numbers and build dates

[eben]

I just received my 2nd R-Pi, this one from stock (yay MCM Electronics) which took just a week to ship from Ohio to California. I was pleased, but surprised to see that the USB fuses F1, F2 are now just 0-ohm jumpers (photo below). I was able to confirm they measure less than 0.1 ohms. Is this the new standard going forward?

Yes. After seeing how many people have decided to modify their boards with zero-ohm links, and having done some safety and reliability testing, we've decided that the best course of action is to remove the USB fuses from the design. Right now they're linked out, but on a subsequent board revision the pads will be removed entirely.

[liz]

And I'm impressed someone noticed so quickly!

[ejsolutions]

...we've decided that the best course of action is to remove the USB fuses from the design. Right now they're linked out, but on a subsequent board revision the pads will be removed entirely.

Interesting step to take (cheaper than better specification polyfuses?) and presumably there's no complications from voltage dipping/spikes nor hot-swapping.
Given that this looks to be the new standard, does this now mean that replacement polyfuses or low value resistors should be forgotten about, for existing Model B hardware?

[jbeale]

I just received my 2nd R-Pi, this one from stock (yay MCM Electronics) which took just a week to ship from Ohio to California. I was pleased, but surprised to see that the USB fuses F1, F2 are now just 0-ohm jumpers (photo below). I was able to confirm they measure less than 0.1 ohms. Is this the new standard going forward?

Yes. After seeing how many people have decided to modify their boards with zero-ohm links, and having done some safety and reliability testing, we've decided that the best course of action is to remove the USB fuses from the design. Right now they're linked out, but on a subsequent board revision the pads will be removed entirely.

Wow, a reply from both Eben and Liz, cool! Anyway I've been reading these pages every single day since before launch, so I know the fuses are an item of interest. Speaking of which, I think it's also worth keeping an eye on this issue (SD card pins not contacting well on this board): http://www.raspberrypi.org/phpBB3/viewt ... 28&t=15573

[pluggy]

Good move. The polyfuses have been a pain in the a*** for a lot of people, saves having to bridge the things like I did.

[Lob0426]

This is a smart move. there should be less posts about keyboards not working. The only caveat will be the people that will insist upon trying to use their G15 gaming keyboard and a WiFi dongle directly to the RasPi, which together will draw more than F3. These people will then complain that their RasPi constantly resets on them.

Nope; I am not trying to be negative about this change, it is really needed. And it is good that the Foundation is keeping tabs on what is being found out about the RasPi. But there is always going to be the people that cannot use common sense and will not read the FAQ's or the Wiki.

Any chance that F3 will be replaced with the 1.1A that was in the original power specs for the production board at the revision?

I have one (1.1A), on my modified board. A USB HDD still will not run off of the RasPi with any reliability, but a WiFi dongle does just fine with it. It is not the Magic bullet, but it seems to be more versatile than the other two stock RasPi that I have. Less issues overall. I have not seen a single repeated keystroke, since I modified it, and it used to have them occasionally. This was in fact caused by my F1 being 7.2 ohms, outside of spec. I believe there are many such instances of this, which is why some people just seem to have keyboard issues, no matter what they try.

Also is a future revision going to address the 1.8v rail issue at the same time?

Honestly I have not had the 1.8v issue with any of the RasPii that I have seen. (5 total, my brothers, my three and the one I sent in for RMA)

When this revision happens, I am sure your partners are going to be happy with two less components to source.

Definitely a step that will reduce problems for the majority of the people out there. Again great news, and I hope there are a few more tweaks in the works. Nothing big just some small cures here and there?

[cesarvog]

I'm curious about one thing:

If F3 was supposed to be 1.1A (but somehow is 700mA currently), AND newer boards have F1 and F3 replaced with something that essentially is a piece of wire with 0 ohm resistance, then I guess that means the PI will only have around 100mA if one would connect two devices that require the standard 500mA from both USB ports. Of course, in that scenario, I suppose the PI would not even start.

So, my question is: what is the maximum current the PI's PCB board rails can handle?

I ask this because if someone would provide a verified answer with a value exceeding 1.1A, then I would start thinking of replacing F3 with a polyfuse that can handle such a larger voltage. This is because, I would very much like to use a single PSU the 700mA needed by the PI + the 500mA required by each USB port connected devices (which equate to 1.7A). Capiche?

[cesarvog]

I found this polyfuse capable of 2A: http://br.mouser.com/ProductDetail/Litt ... UK5dWJQ%3d
I wonder if the PCB rails can handle 2A...

[cesarvog]

On the above post, when I say "then I would start thinking of replacing F3 with a polyfuse that can handle such a larger voltage". Of course, I mean amperage, not voltage.

[cfcubed]

<snip> I was pleased, but surprised to see that the USB fuses F1, F2 are now just 0-ohm jumpers (photo below). I was able to confirm they measure less than 0.1 ohms. Is this the new standard going forward?

Yes. After seeing how many people have decided to modify their boards with zero-ohm links, and having done some safety and reliability testing, we've decided that the best course of action is to remove the USB fuses from the design. Right now they're linked out, but on a subsequent board revision the pads will be removed entirely.

I've wondered why my MCM-late-august Pi did not have as many USB issues as I see others have here, it too has 0R resistors in the F1 & F2 spots.
My Pi has even fewer issues now w/the current Wheezy updates/upgrades + Aug 24th firmware (kudos!).
So shorting F1 & F2 seems sanctioned now & those w/earlier boards w/o the skills could seek out a friend w/a soldering iron.

Questions:
1) Are board & BoM revisions like this enumerated publicly anywhere?
Haven't tried to find this nor firmware revision notes anywhere yet myself...

2) Why not sub w/1R resistors per this?:

The best solution would be to bypass the USB polyfuses with two 1 ohm resistors, low enough for a minimal voltage drop, but high enough to limit inrush current to something the 47uF cap can deliver <snip>

BTW having quite a bit of fun with this great little board even in its current form. Seems others are starting to understand these growing pains & see the cup as well over 1/2 full:)

[digital_addict]

I'm curious about one thing:

If F3 was supposed to be 1.1A (but somehow is 700mA currently), AND newer boards have F1 and F3 replaced with something that essentially is a piece of wire with 0 ohm resistance, then I guess that means the PI will only have around 100mA if one would connect two devices that require the standard 500mA from both USB ports. Of course, in that scenario, I suppose the PI would not even start.

So, my question is: what is the maximum current the PI's PCB board rails can handle?

I ask this because if someone would provide a verified answer with a value exceeding 1.1A, then I would start thinking of replacing F3 with a polyfuse that can handle such a larger voltage. This is because, I would very much like to use a single PSU the 700mA needed by the PI + the 500mA required by each USB port connected devices (which equate to 1.7A). Capiche?

I believe the spec for that micro USB connector is 1.8amps for all four pins. But people have put up to 2 amps through the micro USB. The usual is 1amp for it as connected right now.

Gert has stated that the traces were designed for 500ma. There probably is some safety margin there as well.

"Pushing the envelope." could be a usefull phrase.

[mahjongg]

A few comments, that the F3 fuses is labeled 75 doesn't mean that when the pi draws 750 mA it will trip! 75 just means that under no temperature and other conditions the fuse will trip if the current stays under 750 mA, the average trip current will however be at least twice that.

Also, in most cases the real limitation won't be the polyfuse, but the power supply.

I do not know why the RPF chose to use zero-ohm resistors, instead of 0.5 Ohm resistors, (and I asked). I can only guess that they see the voltage drop as the mayor problem, and zero ohm is the most efficient solution. Also note that such a very low resistance also facilitates "hub backfeeding" With zero-ohm resistors you can power a PI reliable through a self-powered hub that is back-feeding the PI, without the need to use a hub port to feed the PI thru the "front entrance".

There is still no real proof that there really is a significant "hot-plug" problem, and then it can be solved by simply not hot plugging, most USB devices on a hub will work well when they are connected prior to turning on the system. Time will tell if using zero-ohm is the most optimal solution. In any case, removing the fuse is a very big step forward!

About the question "will the traces be a limitation", I can be short, I do not believe that trace width will really be any kind of limitation (for the normal copper thickness that is used in the PI's PCB), unless the trace width are hair-breath wide, and unless you want to draw 10 Amps through them. Its much more reasonable to believe that the wires in a normal USB cable will be the limiting factor. Anyhow, in any normal scenario using normal USB devices I predict the traces won't be an issue.

I think the decision to replace these fuses with (essentially) wire bridges will solve in a practical sense the power problems, also the "hub power feedback" problem will be a thing of the past, as with zero-ohm resistors its no problem at all if the hub feedbacks power, provided that the hub itself forms an effective protection against shorts (on for example the GPIO pins) as there is no no fuse to protect the PI against such shorts when feed by the hub this way.
In fact feedback power from a hub will become the most effective way to power a Pi.

[Lob0426]

@ mahjongg:
I have seen problems with hotplug caused reset on my back-powered Raspi. It happens every time I plug in a USB stick. I attribute this to the fact that I was not very smart when I soldered in my jumper inside the hub. I put the jumper to the input side of the hubs 2A polyfuse rather than on the other output side. Probably the same "inrush" current issue that was experienced with The PiPass. I have not moved it to the other side, which may just solve the problem by adding just a little resistance. As you say I just do not hotplug anything for now. I will have to move the jumper wire and see if it solves the problem or not. Just been lazy about it.

@cesarvog:
I would not try a fuse as large as 2A. Bigger is not always better. For one, the USB traces on the board are not designed to handle as much as 1A. The USB traces were designed for 500ma to the USB ports.

For two the Micro B USB connector, that is the power connector will not handle a steady draw at that amperage. On all the newer ones F3 is a 750ma part. Also on the newest units there are 0ohm resistors so no fuses at all. If you put a 2A fuse at F3 then there will be no protection for the RasPi or the USB device attached if there is a problem.

Example: you attach a cell phone to the RasPi with your 2A fuse in place, the phone will try to charge from that port. There is no circuit on the RasPi to deny the request for a 1A charge. In a relatively short time you will find that your USB port does not work any more. You burned out the traces on the RasPi.

For high draw, a powered hub is the best option overall. The only way you might carry that kind of power on the RasPi is to use the Pipass as described earlier in the very post. I just do not know why you would need that kind of amperage through your RasPi's USB. Nothing else on the RasPi will draw that much power.

[mahjongg]

I really doubt that 1A will "burn out any traces", there won't be enough energy loss in them, even if they are rated 500mA. What will happen is that the voltage loss to your USB device will again become significant.

If at 500mA there is no significant voltage loss to the USB ports (that is what "rated for 500mA" means, that means there will also be negligible energy loss in them, and twice that negligible amount will still only be a very small amount, maybe just enough to get the traces at most slightly warm, not burn them out.

That said, i still have to see how thin these traces really are, but if at 1A they burn out they would also be unsuitable to transfer 0.5A.

I also do not see the need to use a 2A fuse, except if a lower amperage polyfuse still introduces a non-negligible voltage drop. Look at the typical resistance of the fuse, not at their rating, their rating is only interesting when limiting a short circuit current.

In any case for normal use these traces shouldn't be too much of a problem.

[Lob0426]

There is the perfect world where that 1A ramps in and remains steady on a perfect circuit that was over designed because the space was available.

1. How much were the traces designed to carry.
2. How much over capacity were they able to work into those traces.
3. Are those traces up to those design specs.

With the new board there is less protection. There is no current protection other than F3. There no protection from spikes. The capacitor at the USB is too small to really help either. Drawing 1A over that circuit, would just plain be dumb to do.

I remember reading that the traces may, may be able to handle 700ma. Is his board "perfect" i.e. there were no manufacturing mistakes on his PCB. Does the device draw 1A, or will it draw whatever it can get. There is no current negotiation on the RasPi. Do you think USB traces would survive having my iPad connected? It probably would, but only because the iPad would decide not to charge. How many boards will be able to handle 1A. There will be a small number, with defects, that will not handle even the 500ma. Drawing 1A over both ports will be safe.

And again why put a 2A fuse at F3. The core RasPi uses at most 500ma. Anything over that is going to the USB. If you stay within the 2A you are putting 750ma per port. But you could draw 3A over that fuse and only see a percentage of voltage drop due to increased resistance. Or 1.25A per port. Or 2.5A for one port. Lower voltage with higher amperage equals more resistance so more heat. Nope, best answer is still a powered hub and protect the RasPi. This would only be an option if you were building a special use device.

The difference in ohms between a 1A and a 2A polymeric resettable fuse is negligible. Less than a tenth of an ohm. My change from a 700ma to a1.1A was from .18 ohms to .1 ohms.

[mahjongg]

And again why put a 2A fuse at F3.

The only real reason is to lower the resistance of the fuse with another fraction of an ohm, each fraction of an ohm, wen drawing 1A trough it (that is 400 Ma drawn by the PI itself, one USB port drawing 100mA, and another one drawing 500mA) would lower what they all see by a fraction of a volt, and that fraction of a volt can be relevant!
For example if you draw one amp through a PSU that is able to deliver exactly 5.0 volt at that amperage, but your USB power cable has a realistic resistance of say 0.1 Ohm, your PI at it mini USB port would receive 0.1 volt less, so 4.9 volt, now if the F3 fuse was an 1A one (a 0.2 Ohm fuse) then it would drop 0.2 Volt, and the PI would receive 4.7 volt, and would fail (minimum is 4.75V) if the fuse would be a 2A one and thus only 0.1 Ohm it would drop only 0.1 volt, and the PI would get to work with 4.8 Volt. Obviously this is just an example, but I just wanted to explain that indeed just 0.1 ohm more or less polyfuse resistance can mean a working or a failing system.

Just a few "rule of the thumb" figures for normal copper thickness of 0.17mm (half ounce per square feet).

A trace of 5 cm long and 0.5mm wide would be about 0.1 Ohm.

See : http://circuitcalculator.com/wordpress/ ... alculator/
So if that is more or less similar to the trace from the PI's input connector to the USB ports, (but I think the trace would be wider than half a mm) then running 1A through them would result in a 0.1 Volt drop. It would also mean that the dissipation (energy turned into heat) in the total length of the trace would be P=U*I = 0.1 x 1 = 0.1 Watt, which is such a small number that it would hardly warm up the trace. If the current would be double that (2A) the energy dissipated in the trace would be four times that, or 0.4 Watt, still not really significant.
By the time the current would reach levels which would be threatening to the trace, the fuse would have been long blown, and unless the "PSU" is a 200 Watt PC PSU the 5V would have collapsed completely. I simply see no valid scenario where the traces would burn out, unless you are able to keep up 5V while sending 10A trough the trace, which is absurd when using a normal PSU.

[Lob0426]

I wish someone would tell this to the people who layout PCB's. They should be making traces about a molecule wide. Then we could shrink the RasPi about 40%.

The traces might be a .5mm wide, but what is their thickness? The calculator will not take into account manufacturing defects either.

I still will not advocate running 1A into a single USB port that was designed for 500ma! I still believe you will damage a RasPi, if you were to try to charge your cell phone from it, if it were drawing 1A. It is not made for it. I might attemp it for my own use but I will not tell anyone else it is a good idea.

Example: I have told the forum that I am back-powering a RasPi through the USB. I have not at this point advocated that this is a safe way to power a RasPi. I qualify the statement that it is running fine with "so far" or I have seen no damage "yet". I am still testing it, and will withhold an opinion until it has proved to me it will not damage a RasPi. I am willing to risk my equipment, but will not state that another person should follow my example. This is really true when you only have ONE RasPi to risk.

Example two; I will advocate a person assemble a RasPi with a Motorola Atrix Lapdock. It has proven to be an easy modification to the cables and is a solid performer when it is completed. It is working for dozens of people with no real proof it is a hazard to a RasPi or it's user.

There are people that are out there, That will take these discussions to heart. Not knowing that there is a whole load of factors, that are not discussed. If they try the concept that is being discussed. It might just destroy the Only RasPi they were able to afford. That would be a real shame if they lose out on the experiences that the RasPi could have given them. I have three, one up in smoke will still leave me with the "experience".

[mahjongg]

The traces might be a .5mm wide, but what is their thickness?

Normally the default is 17 micron, but some PCB's use double that (35 micron) for example for heat spreading. The calculator also defaults to 17 micron.

An experienced PCB designer (lay-out'er) will know all this, for signal traces obviously much smaller tracks are used, sometimes for power supply "traces" whole planes (layers) are used, and obviously GND is always a copper plane, and sometimes even a double thickness on (35 micron).
The PI has been designed to be as small as possible (but no smaller), so probably the engineer has been careful not to oversize tracks and such.

If the PI has zero-ohm resistors mounted it should be quite safe to back power it from a hub, just like any other "peripheral" you plug into the hub. The hub must take care of the safety (be fused and such).

Obviously you should not advocate doing something stupid to your PI, but I think that delivering normal amounts of power to USB ports should not be an issue, obviously USB ports are designed for no more than 500mA each, so you should not want to push 1A through them. But using an USB device using say 400mA should be safe, and should work, the bigger issue is whether your normal 5V 1A power supply can take the extra load.

The zero-ohm patch isn't there so that you can run extreme currents through it, (and I will certainly not advocate doing that) it is there so that a device that uses a little over 100mA (say 250mA) will still work, because it will still get the full 5.0 Volt, not just 4.4 Volt. For that purpose the trace isn't a limiting factor, and "burning out traces" isn't something that could happen during such normal use, only when you have both a very strong power supply and a short circuit there is a chance of trace damage, but that is quite normal.

[Lob0426]

That is the point I was trying to get to! There are readers here who might not be able to seperate the theoretical discussion from a real world use of the RasPi.

If the traces are up to any normal spec then they can, and I am sure were designed to carry at least a percentage more then The 500ma A USB port is supposed to supply. but when manufacturing in large volumes quality is sometimes traded for quantity. Thin traces are a real likely scenario.

That is the point I was trying to get to! There are readers here who might not be able to seperate the theoretical discussion from a real world use of the RasPi.

I am using a 500ma poly fuse with no problems so far. The 0ohm resistors are going to pass more voltage into a back-powered board. Currently I am seeing 4.25v to 4.35v at TP1/TP2 some of this is the drop through the polyfuse. More drop is from components pulling a load, before the test points which are now at the wrong end of the board. I do believe it is safe to power a RasPi this way, but It has not been confirmed by any other RasPi geeks. One (mine) test device, does not make a valid test!

Delivering 500ma to each port, what they were designed at, should be perfectly safe. You of course will have to use a larger PSU. 500ma x 2 + 500ma more for the RasPi means you will need a PSU rated at least 1.5A. The problem is that the polyfuse at F3 is 750ma. This puts that load right at the specified trip point for F3. So what amperage can you really run through a RasPi with an unmodified F3. Most should handle about 1A with little or no problem. So the realistic figures would be 500ma shared between the two USB ports + the 500ma for the rest of the board, a B model of course. If the user is capable of replacing F3 with the original Specification 1.1A then you could possibly use 500ma at each port. I ran at least 1A (USB HDD) into the 2 USB ports with the 1.1A F3 in place. I was using a 2A PSU. I replaced F2 with a 700ma polyfuse and it made no difference to the chirping USB HDD (single port). There was just not enough voltage getting through a single port! That HDD runs fine on a single port of a powered hub. So what was the likely culprit causing the low voltage? F3, Trace resistance, F2 being drawn above its rating? All three draging the voltage down a bit each?

The 0ohms are indeed going to allow people to connect a better variety of devices to their RasPi. But there is still that limit at F3 to contend with. Connect a 250ma keyboard and a WiFi to the RasPi and you may or may not be successful, on the B model. There will be more overhead for the USB on an A model as it will not be supplying a LAN9512. That assumes that F3 remains 750ma on the A model! Hopefully it will as the smaller polyfuses actually cost more to buy than the larger ones.