Design flaw? Polyfuses for USB current limiting


 
246 posts   Page 5 of 10   1, 2, 3, 4, 5, 6, 7, 8 ... 10
by hippy » Thu May 17, 2012 12:01 pm
A theoretical thought ... If two low-current but high-resistance polyfuses are paralleled that would double the current trip point but halve the resistance; so an existing 140mA trip with 7R could be implemented as two parallel 70mA with 7R giving the same 140mA but at 3.5R.

If polyfuse resistance doubles with halved current rating you're on a hiding to nowhere but it might be something someone more familiar with polyfuses than I am might like to cast their eye over as a possible solution.

If the idea is sound, it might be possible to connect both USB port +5V together so the draw rating of both ports combined rises to 280mA trip value with half the existing resistance. That may help some people get out of a jam but isn't a perfect solution.
Posts: 774
Joined: Fri Sep 09, 2011 10:34 pm
by mahjongg » Thu May 17, 2012 2:58 pm
The theoretical basis behind paralleling two polyfuses is sound in the sense that paralleling two resistors of the same value equals one resistor of halve that value.

But not much gain can be had from such a solution when it comes to polyfuses. Unfortunately If you parallel two 140mA polyfuses, by mounting one on top of the other, they would behave largely the same as one 140mA polyfuse! At least, they wouldn't behave the same as one 280mA polyfuse.

The reason behind that has to do with how polyfuses operate, they have some resistance even when their "active material" is crystalline (that is when the polyfuse is cold) and when current rushes through this resistance they heat up, but it also emit heats so normally an equilibrium exists at a certain current that is below the tripping current. But if a certain threshold of the current is reached (the "tripping point"), the heat "melts" the crystalline material which through its material properties sharply increases its resistance, which makes the material even hotter, at that point the process enters a phase of escalation, and the heat and resistance increase quickly and dramatically until the polyfuse has such a high resistance (hundreds of times its normal resistance) that current decreases to a level (much) lower to that of the normal current to the USB device. But now almost all the voltage falls over the polyfuse, so the polyfuse stays hot, this remaining current is called the "holding current".

So that is why a certain rated polyfuse needs to have a certain resistance, because at a certain current enough energy must be developed to "trip" the polyfuse. Unfortunately at the sizes of the polyfuses used (which is about as small as they get), and at the tripping current chosen, the polyfuse therefore "must" have the resistance it has, otherwise it would not get hot enough to trip.

If you mount two polyfuses on top, they each get halve the current, so produce halve the heat, but as they are so closely coupled they exchange heat, and thus heat up each other! In the end they behave like a single resistor when used this way (soldered on top of each other) Luckily they also represent twice the mass, and so in principle should need twice the amount of energy to get hot, but in fact a polyfuse is soldered to a board, so a large portion of its heat is drawn away to the PCB. With two polyfuses on top of each other this isn't the case, so at least the top one will heat up quicker, negating for a large part the benefit of having two.

However the idea to mount a new polyfuse on top of the other is a good idea, (for patching an existing board) just not one with the same rating, but one considerably higher, lets say 500mA.

But, a small warning, after soldering wait some while for the, (having been hot from soldering) polyfuse to recover, before using it.
User avatar
Forum Moderator
Forum Moderator
Posts: 5348
Joined: Sun Mar 11, 2012 12:19 am
by hippy » Thu May 17, 2012 6:00 pm
@ mahjongg : Thanks for the detailed reply. I learned a lot there.
Posts: 774
Joined: Fri Sep 09, 2011 10:34 pm
by simonbr » Thu May 17, 2012 6:33 pm
I have spent some more thought on this subject. It looks like quite a number of users are running into problems where (combinations of) USB devices, that should on first sight be acceptable, refuse to work with the Raspberry Pi. I suspect that the Raspberry Pi design does not comply to USB specifications, even if a restriction of 100 mA maximum current per USB port is observed. Let me try to back this up by calculations. By all means, correct me if I make mistakes or false assumptions.

The main fuse F3 appears to be a 0.7A type. For calculations, I will assume the Raychem miniSMDC075F which has a maximum R1 of 0.45 ohms. By the way, R1 is the resistance of the device one hour after recovery. I think it's sensible to base calculations on this parameter, and not a lower initial (from-factory) resistance; otherwise, what would be the point of having a *resettable* fuse...

I measured the power consumption of my unit with an SD card installed, connected via HDMI to a 1080P TV, and with no USB devices connected, from powering on to the login prompt. I observed it going just a bit higher than 0.4A. I will assume the maximum power consumption of the Pi, excluding USB and expansion boards, to be 0.45A.

If 2 USB devices drawing 100 mA each are connected, the total power consumption gets to 0.65A, and the voltage drop across F3 could be up to 0.29V.

Assuming an R1 value of 6 ohms for F1 and F2, the maximum voltage drop across them would be 0.6V. Assume the 5V power input to be 3% accurate (I'm being kind here) then in the worst case the voltage on the USB devices would drop to 3.96V.

If I'm correct the USB specification requires a minimum voltage of 4.40 V on the downstream port from a bus powered hub. We can regard the Pi as such, but even then it is still far from compliance.

In a previous post I mentioned a possible improvement which would be using a single polyfuse of double the current rating serving both USB ports. As due to the nature of polyfuses a doubling of the rated current translates to roughly a fourth the resistance, the voltage drop across this fuse would be reduced to 0.3V and the worst case USB voltage would become 4.26V. This is however still below the required 4.40V.

I considered more exotic solutions such as combining a polyfuse with a step-up converter, built around e.g. the TS1935CX5, to compensate for the voltage loss, but while this is a relatively low cost component the solution would probably be too expensive. I also thought about a low-resistance electronic protection circuit using a RRIO opamp and a MOSFET, but found it nontrivial to design one that is both low-cost and robust. It was interesting to hear about a previous design of the Pi that used DC/DC converters and would run on a voltage of 6..20V, but going back to that is probably to be ruled out.

The best solution IMO is to do away with F1 and F2 altogether. Because what are they there to protect? Not the power supply; it is supposed to be handled by F3. Not the Pi itself; as long as the PCB traces from F3 to the USB ports can handle about 1.5A (which is easily achieved) it is sufficiently protected by F3 as well. Not the USB devices, because any that could draw 1.5A and be damaged by it would not survive getting connected to any average PC.

The drawback of ditching F1 and F2 would be that overloading/shortcircuiting the USB ports would trip F3 and take down the whole Pi. However, no permanent damage would be caused. The SD card could become corrupted, but it is easy enough to restore from a backed-up image. In addition, I doubt that F1 and F2 would always be able to prevent this happening anyway. Two USB devices may draw up to 280 mA each before F1 and F2 trip, and although a current of 1A will not trip F3 it may cause 700 mA rated USB chargers to shut down. And in case of a sudden short-circuit I would not be surprised to see them brought to their knees before F1 or F2 is able to trip (polyfuses are not that fast).

So all in all I don't see serious disadvantages of leaving out F1 and F2, while without them, the voltage on the USB ports would remain at least 4.56V, which is well within USB specs for the downstream ports of a bus powered hub. It may resolve many of the USB related problems that users of the Raspberry Pi are having.

Best regards
Simon Brouwer
Posts: 23
Joined: Sun May 13, 2012 5:02 pm
by mahjongg » Fri May 18, 2012 12:37 am
Simon I agree with your assumption that the USB polyfuses serve no real purpose in the R-PI.
They were probably used because it is common practice to have them, but the common practice supposes they are fed by a PC power supply that could easily source 50Amps!
Polyfuses by nature do not react instantly so even with the current 140mA ones shorting the output of a USB port -will- short the 5V sully, and thus crash the processor (and this indeed might damage the SD-card content if you are unlucky).

If you look at a datasheet of a polyfuse you will see that a certain size of polyfuse, say an 1812 size, will simply need a certain amount of power to trip, and this power is regardless of the actual current that flows through it, as long as the sum of resistance of current is met it will trip, as the resistance will mean a certain voltage, and voltage times current equals power (p=U * I) euqals heat dissipated in the polyfuse.
Lets say we are using a littlefuse 1812L014 (littlefuses 140mA 1812 size device) the resistance of this polyfuse after it has been soldered to the board is described as 6 Ohm max. It tripping current is stated as 340mA. The power required to make it trip is U * I, and we know I, its 0.34A, but we also know U (max) its 0.34A x 6 Ohm = 2.04 Volt (lets ignore the practical implication this has for the moment) this means it needs 0.34 x 2.04 = almost 0,7 Watt

So a typical 1812 polyfuse needs 0,7 Watt of dissipated heat to trip!

But as we have seen such a polyfuse has 6 Ohms of resistance.
Lets say we are using a USB device and it draws a 100mA (the typical rating of most low speed USB devices), that means that over a 6 Ohm polyfuse it will drop 0.1 * 6 = 0.6 Volt!
This is already way too much, as we normally have only 0.25V of tolerance in total!
User avatar
Forum Moderator
Forum Moderator
Posts: 5348
Joined: Sun Mar 11, 2012 12:19 am
by Morgaine » Fri May 18, 2012 7:10 am
I agree with Simon's analysis. And obviously omitting F1 and F2 would provide the cheapest solution too.

However, Simon explains that his solution does have one disadvantage:
simonbr wrote:The drawback of ditching F1 and F2 would be that overloading/shortcircuiting the USB ports would trip F3 and take down the whole Pi.

In contrast, the solution I gave in post p78716 does not have that disadvantage.

To recap, my suggestion was to make F1 and F2 take power not from the board's +5V rail but directly from the charger input (same as F3). This would make all three polyfuses independent, so that F1 and F2 could be sized to allow a full 500mA from a suitable power supply. It would also allow F3 to be sized more appropriately to protect the board's electronics without having to factor in downstream USB usage.

That would make a lot more design sense, although it's only applicable if the Foundation is contemplating an incremental change in PCB layout.

Morgaine.
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am
by simonbr » Fri May 18, 2012 7:54 am
Morgaine wrote:...to make F1 and F2 take power not from the board's +5V rail but directly from the charger input (same as F3).
Morgaine.


Then, using the current type of fuses for F1 and F2, and assuming a 3% accurate input voltage, the minimum USB voltage would be 4.25V, still considerably lower than the USB specs requirement.

With a single fuse of double the currrent rating, shared between the USB ports, the minimum voltage could get to 4.55V. It might be a problem to find the suitable component.

When the fuses are connected in parallel rather than in series, the power supply would not be protected against the same maximum current as with the current design.

Best regards,
Simon Brouwer
Posts: 23
Joined: Sun May 13, 2012 5:02 pm
by Morgaine » Fri May 18, 2012 8:51 am
simonbr wrote:Then, using the current type of fuses for F1 and F2, and assuming a 3% accurate input voltage, the minimum USB voltage would be 4.25V, still considerably lower than the USB specs requirement.

You misread my proposal. I wrote:
Morgaine wrote:F1 and F2 could be sized to allow a full 500mA from a suitable power supply.

In other words, they would not keep their current values, as that would be silly. They would be sized to have no significant effect on USB supply voltage within the rated USB current range, and only trip far above it on shorts.

Their current values are fallout from the current bad design that put them in series with F3, a polyfuse whose role should be to protect the board, not the downstream USBs. The current design couples these two entirely independent requirements together, wholly inappropriately, and hence introduces the problem by design. And the current design also has the secondary problem of not even giving us the option of using a more powerful supply to cater for more demanding USB loads. That's bad.

I do however agree with you that with the current board layout, your suggestion is good. Mine requires new board layout, so it's only appropriate for Pi v1.1 .

Morgaine.
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am
by mahjongg » Fri May 18, 2012 12:11 pm
Morgaine wrote:I agree with Simon's analysis. And obviously omitting F1 and F2 would provide the cheapest solution too.

However, Simon explains that his solution does have one disadvantage:
simonbr wrote:The drawback of ditching F1 and F2 would be that overloading/shortcircuiting the USB ports would trip F3 and take down the whole Pi.

In contrast, the solution I gave in post p78716 does not have that disadvantage.

To recap, my suggestion was to make F1 and F2 take power not from the board's +5V rail but directly from the charger input (same as F3). This would make all three polyfuses independent, so that F1 and F2 could be sized to allow a full 500mA from a suitable power supply. It would also allow F3 to be sized more appropriately to protect the board's electronics without having to factor in downstream USB usage.

That would make a lot more design sense, although it's only applicable if the Foundation is contemplating an incremental change in PCB layout.

Morgaine.


You have to remember that when the USB port is shorted the R-PI will -always- reset!
A polyfuse will never react fast enough to prevent this, nor does a typical PSU have the ability to keep the power up when the current consumption suddenly increases tenfold.
So it doesn't make sense to try to find a configuration where the reset of the PI doesn't happen when a bad USB device is connected.
User avatar
Forum Moderator
Forum Moderator
Posts: 5348
Joined: Sun Mar 11, 2012 12:19 am
by Morgaine » Fri May 18, 2012 11:39 pm
mahjongg wrote:So it doesn't make sense to try to find a configuration where the reset of the PI doesn't happen when a bad USB device is connected.

The main motivation for polyfuses on this board isn't USB problems at all, because USB devices are by and large stable commodity items. The Pi board has itself created the USB trouble that people have witnessed, by making board behavior erratic when USB devices draw what it considers to be too much power. That's the Pi's fault, not the fault of the USB devices. That's why I suggest divorcing the two current paths.

Your point about the Pi resetting if the USB devices short is true, but that occurrence would be extremely rare. Most of the time the USB devices will be working properly, as commodity devices normally do, so the main result of my proposal will be to let them work normally, taking as much current as they need without affecting the Pi at all.

If you know that you have power-hungry USB devices and so you attach a beefy power supply, everything should work fine. The current design doesn't allow that.

The USB isn't the culprit here, it's the Pi that has made properly working devices affect the board adversely as a result of its power design strategy.

Morgaine.
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am
by mahjongg » Sat May 19, 2012 1:22 pm
never said anything about anything being the "culprit", but obviously any fuse would have a reason to be used. Logically a fuse in a power-line is used to protect against what would happen if the power-line after the fuse would be shorted. Thats not "blaming the short" in any logical sense.

Using such fuses in series with the USB power lines is standard practice, and for what other reason would that be to prevent that a short on any USB device (or cable) would short the 5V supply, which in a normal PC could generate enough current to burn up said USB cables, and maybe burn up PCB traces on the PC's motherboard too. So its standard industry practice to have them, but the standard industry practice does not translate to a valid reason to have them on the PI, as a normal R-PI power supply doesn't have enough power to do much damage, unless a 5A laboratory power supply was used.

You could argue that the fuses could be useful if something from the outside, connected to the PI's USB ports would supply power to the R-PI's USB port, and the R-PI itself developed a short, but this is a very unlikely scenario, and most USB hubs (because these are the candidates for a device that might put power back into the USB port) are themselves reasonably lightly powered.


My remark was only made because some comments indicated that the USB fuses somehow could protect the RPI from crashing when such a short occurred, which obviously isn't true.

Routing the power from the Power input to the USB fuses around the main fuse is simply a good idea because it prevents having two fuses in series, which serves no real purpose, and adds the resistance of the to fuses together, increasing the voltage drop, and thus increases the potential for failure.

And yes, if the R-PI designer(s) would have thought of the resistance a polyfuse has, and therefore had chosen a more sensible value of perhaps 750mA, then, providing the external PSU used could cope with it would have been able to power even devices wanting more than the standard 100mA.

Perhaps their reasoning was that these ports would always be used up by a mouse and a keyboard, or if the user wanted to add more devices, a powered hub would be used, thus the ports would never have to supply more than 100mA. So they chose a fuse with a tripping point roughly 50% above that number, not realizing the implications.

I have to say I cannot blame them, as I in the far past I have made the exact same error when trying to protect the 5V power input of a device (made from 5V powered logic consuming roughly 1.5 Amp) with a 3A fuse, (glass tube kind) only to find out the device would not work with a 5V power supply because of the resistance of the fuse. I had to bridge the fuse with a wire to solve the problem, as I calculated I needed a 15A fuse to get a reasonable low resistance between the PSU and the 5V logic, and a 15A fuse wasn't a practical proposal, so I was better of having none. The original thought was that the fuse should help protect the device (together with a reverse diode, and a six volt overvoltage protector) against using the wrong power supply (one with a higher voltage than 5.0 Volt, or with a reversed polarity),

Luckily the R-PI doesn't have that problem, because the used the micro USB connector, (always used with 5V and a knows polarity) but they still use a overvoltage protection device, (D17) and a fuse, (F3) which luckily for them has a sufficiently low resistance, (because of the R-PI's much lower input current, they could use an 1.1A polyfuse with a sufficiently low cold resistance for the currents that normally run through it) although I still have a feeling it could sometimes be a factor, if for some reason the input polyfuse develops a higher than normal resistance, for example when a unusual high current has run through it recently.
User avatar
Forum Moderator
Forum Moderator
Posts: 5348
Joined: Sun Mar 11, 2012 12:19 am
by Morgaine » Sat May 19, 2012 1:47 pm
A large part of the problem stems from the choice of cellphone chargers as if they were proper "USB power power supplies", with performance to USB specs expected. From experience, we now know that those expectations were flawed, unfortunately. More often than not they're utter **** manufactured to the lowest possible cost and only adequate for charging the particular cellphone with which they are bundled.

This sad reality should encourage the Foundation to think again about their choice of power for the next generation of Pi board. Even if the old micro-USB is retained, an alternative power input should be provided as well for those who demand voltage regulation and more power for the USBs.

These cellphone chargers are effectively unregulated, and the series resistance of polyfuses just exacerbates it. Unregulated supplies should not be used in 2012 for hi-tech modern electronics. We left those bad times behind us 2-3 decades ago.

Morgaine.
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am
by prof » Sat May 19, 2012 3:50 pm
Having spent a few hours a couple of weekends back trying to work out why a USB device I was using kept repeatedly initialiasing itself I have to say that some clear and easily located documentation on issues like this is definitely needed.

I would also add my vote to the "lose the 140mA" fuse camp. Many people faced with the perfectly resonable task of connecting, for example, a WiFi dongle won't want the extra bulk of a powered hub and will probably opt for the simplest options which are either to solder a wire directly from the power socket to the USB power lines or to build a bypass cable which achieves the same thing.

Whilst this is not the most sensible fix since it eliminates all USB protection, it is far easier than messing about with surface mount devices and hence far more likely to be done.

Whilst I accept that Gert had good intentions in including this protection I fear it will only lead to confusion, frustration and possibly even damaged boards for no real advantage.

Please change the design in Rev 2 boards and get the polyfuses swapped for 0 ohm resistors as soon as possible. If nothing else, there's still time to get it into the Model A design.

In a school environment if something does not work in a completely predicatable manner it simply won't get used. The RP project is a good idea which doesn't deserve to fail because of something as trivial as a fuse.
Posts: 2
Joined: Sat May 19, 2012 3:28 pm
by Morgaine » Sat May 19, 2012 5:14 pm
STOP PRESS. In a related thread, rurwin has just pointed out that the USB spec for "USB chargers" allows the voltage to drop to 3.6V up to the rated current, as long as it supplies 5V at up to 500mA. Forum link.

3.6v ???? :shock:

That's way out of spec for the USB +5V rail, not to mention the other devices on the Pi board. Even before the polyfuse voltage drop.

This goes a long way towards explaining the poor performance we've been seeing in chargers, and in all likelihood it also explains many cases of Pi system instability.

Morgaine.
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am
by rurwin » Sat May 19, 2012 5:19 pm
I've edited that post. The charger is required to supply 4.75V minimum up to the current given on the case label, and drop to 3.6V at a current no lower than that and no higher than 150% of that. So long as no more that the "rated" current is drawn, the charger should supply 4.75V to 5.25V.
User avatar
Forum Moderator
Forum Moderator
Posts: 2904
Joined: Mon Jan 09, 2012 3:16 pm
by Morgaine » Sat May 19, 2012 5:33 pm
So we can rely on the current rating marked on the charger without worrying that the voltage has gone out of spec, in theory. That's a relief, it was looking really bad there for a while. :P
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am
by mahjongg » Sat May 19, 2012 10:03 pm
As I have always maintained, a "charger" isn't electrically the same inside as a power supply.
The difference is that power supplies measure their output voltage and take corrective action when the output voltage isn't what it should be, (they have a regulator circuit) chargers typically lack such a "feedback" mechanism, and just output a voltage without any correction for loaded or unloaded conditions. At most the chargers have an output resistance that prevent dropping too much over the normal current range, but without a regulator they do drop their voltage linearly with the output current, just like a battery.

In practice, if the circuit inside ( a switching power supply/charger) lacks an opto coupler over the isolation barrier, it acts like a charger, if it has the feedback opto coupler its probably a power supply. Obviously I'm talking in generic terms here, its possible the switcher outputs a higher (but variable) output voltage, and has a standard low-drop regulator to convert it to a regulated +5 Volt, but such solutions are quite rare.

I've seen a photo shoot of a disassembly of a cheap chines charger here in the forum somewhere which was quite revealing, it revealed perfectly what utter garbage is on the market.

Having replaced the standard input jack for the mini-USB port might have been a perfect solution for preventing the wrong kind of power supply (12V instead of 5V, or rare ones with reversed polarity, that is the pin-hole not being the plus, but the sleeve being the plus) but unfortunately it brings this new problem with it.

I have pledged before that a warning should be issued not to use an USB charger, but only an USB power supply, but I agree that for many the difference will be unclear, and even an expert cannot always identify a charger from the outside.

Still as practice has proven, some chargers do indeed work, if they are well dimensioned, and do not drop too much with varying loads.

But morgaine, I wouldn't trust all chargers not to drop to precarious levels when loaded with say 600mA, even if they state 5V 1000mA.
User avatar
Forum Moderator
Forum Moderator
Posts: 5348
Joined: Sun Mar 11, 2012 12:19 am
by Morgaine » Sat May 19, 2012 10:22 pm
Oh I wouldn't trust them at all. But for a short while we thought the very poor quality we've witnessed in USB chargers was a product of a wonky spec. It turned out to be only a wonky Wikipedia. :P

But yeah, that certainly doesn't mean that in practice chargers are trustworthy. Experience in the Pi community has definitely shown otherwise.
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am
by Timo » Sun May 20, 2012 12:09 pm
I think bridging F1 and F2s outputs helps. So one USB device could draw up to 200mA or two USB devices can share the current as they need (e.g. 150mA + 50mA). This would be at least an easy home-made way.

fuses12bridgedsmall.jpeg
Outputs of fuses 1 and 2 bridged. (Picturepart taken from http://eames.us/wp-content/uploads/2012/05/DSC_3127.jpg by alexeames.)
fuses12bridgedsmall.jpeg (46.3 KiB) Viewed 2381 times
Posts: 8
Joined: Sat Aug 27, 2011 9:04 am
by mahjongg » Sun May 20, 2012 12:27 pm
Good idea, unlike soldering a second same value fuse on top of them, each one would be able to dissipate heat away just as well. So In practice this really would halve the internal resistance to (both) USB connectors, without side effects.

This could probably fix many of the "repeating character" keyboard problems.
User avatar
Forum Moderator
Forum Moderator
Posts: 5348
Joined: Sun Mar 11, 2012 12:19 am
by hippy » Sun May 20, 2012 1:23 pm
Timo wrote:I think bridging F1 and F2s outputs helps.


Agreed. I'd suggest doing the bridging on the underside of the board at the USB socket pins rather than soldering to the fuse tags. That should be easier and less risky for people who aren't professionally skilled at soldering.

On the USB power supply front ( ignoring the fuse issues ); when it was announced that a USB supply was to be used, the potential problems with that were raised, including the suitability of supplies and the inability to tell what the suitability of any supply was.

I stated that it was essential that the Foundation ( and now RS and Farnell as suppliers ) provide a power supply rated as suitable for R-Pi use, and recommended people use that. The foundation indicated it would have such a supply available but I don't know what actual progress there has been in that direction.

It's been repeatedly stated that not all USB power supplies or chargers will work and that even if one appears to work initially it may not longer term ( and may even be dangerous in some cases ), but that doesn't seem to have been taken on board.

With an ''official power supply"; that some others don't work, is less of a problem, a RTFM issue and a user's choice if they don't choose to use an official supply. At present though R-Pi users have little choice but to choose something and hope it is is suitable.

This 'hit and miss' nature of the R-Pi is what worries me most in selling it at this stage to the general public; you need the right SD Card, the right power supply and need to pick the right USB devices or it might not work. Hopefully things will get resolved shortly though problems like this seem all too easily waved away and dismissed with "it's a developer's board" when sales aren't to developers.
Posts: 774
Joined: Fri Sep 09, 2011 10:34 pm
by mahjongg » Sun May 20, 2012 1:44 pm
Yes soldering the +5V pins of the USB ports together would be safer and easier, we need a picture of that.

The polyfuse issue would be easy to fix for the "educational release", by increasing the rating of the input fuse slightly, (to say 2A instead of 1.1A) and increasing the rating for the USB polyfuses to 1.1A.

I'm more worried about other factors that cause power problems, such as using chargers instead of real power supplies, and of using too cheap USB cables with very small AWG for the power and ground leads. But that can be solved more or less with beter education of the public.

The SD-card issues should also resolve themselves in the end, its not that the R-PI is doing something wrong, its interface is very similar to all other devices that are using SD-Cards, it only that we are using it more intensely than most other devices would. having a switchable power supply for the card would be nice, (switching from 3V3 to 1V8 under software control) but really not many devices have that, so its not essential to have.

In the end, this is why there is this trial run, to find problems like this.
User avatar
Forum Moderator
Forum Moderator
Posts: 5348
Joined: Sun Mar 11, 2012 12:19 am
by abishur » Sun May 20, 2012 2:13 pm
A while back they had a Q&A with Pete Lomas (who is both a trustee and hardware designer for the Pi) and I asked the following

Can the fuses at USB be safely removed to provide >140 mA to USB or is there power concern that requires the limitation?

His response was this

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.

Of course the pi has a 1A limiting fuse on the input but as long as you're up for modding your pi you could replace the two fuses (I wouldn't want to just remove them) at the USB ports, and then either replace the incoming limiting fuse or use the 5V source/sink pins tucked into the gpio strip
Dear forum: Play nice ;-)
User avatar
Forum Moderator
Forum Moderator
Posts: 4263
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
by hippy » Sun May 20, 2012 2:22 pm
abishur wrote:Of course the pi has a 1A limiting fuse on the input


Apparently 750mA ( 700mA ? ) fitted to the boards being shipped.
Posts: 774
Joined: Fri Sep 09, 2011 10:34 pm
by Morgaine » Sun May 20, 2012 2:49 pm
Timo wrote:I think bridging F1 and F2s outputs helps. So one USB device could draw up to 200mA or two USB devices can share the current as they need (e.g. 150mA + 50mA). This would be at least an easy home-made way. (Image embedded.)

Great hardware hack, Timo, thanks!

I think we need a site for comprehensive articles on useful hacks for the community. The ideal place would have been this forum, but it isn't any longer because last night the admins disabled post editing after 30 minutes, so it's no longer possible to work on a well-considered post and hone it into perfection. Pity, but the forum won't be usable for extended descriptive articles improved incrementally.

Morgaine.
Intolerance is a failure of education. Education is predicated on tolerance of the uneducated.
User avatar
Posts: 141
Joined: Mon Mar 12, 2012 1:13 am