User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12125
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Design flaw? Polyfuses for USB current limiting

Fri Jul 20, 2012 10:51 pm

Polyfuse F3 has almost no influence on the voltage USB devices receive, and IMHO it doesn't matter much if you bypass it, F1 and F2 do "all the harm", not F3.

User avatar
bazza14
Posts: 65
Joined: Sun Jan 01, 2012 10:05 am

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 12:09 am

mahjongg wrote:Polyfuse F3 has almost no influence on the voltage USB devices receive, and IMHO it doesn't matter much if you bypass it, F1 and F2 do "all the harm", not F3.
I have to disagree with your statement. The variation of the resistance of these poly fuses varies so much from device to device that some may have negligible effect while others with have noticeable effect particularly when you take into consideration the voltage drop in the supply cable and can result in a supply voltage of 4.5V or less on the board (after F3) although you are correct that the major culprits are F1 and F2.
LINUX convert since 2003

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12125
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 1:41 am

Polyfuses differ only widely only if they have been recently tripped, which won't be the case with F3 normally, while even connecting a memory stick directly to the PI can bring F1 and F2 close to tripping, and if that happens repeatedly their resistance can rise to something like 8 Ohm.
F1 and F2 are normally 4 to 6 ohms, while F3 is typically less than 0.2 Ohm, so F3 typically can be discarded as insignificant.
Yes, if you want it all, you could disconnect F1 and F2 from the PCB, and wire the USB connectors directly to the USB ports, but you would then also lose all short circuit protection, and its quite an involved patch, if you don't do it right you might end up bypassing F1, losing all short circuit protection for the board.

User avatar
bazza14
Posts: 65
Joined: Sun Jan 01, 2012 10:05 am

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 2:21 am

Where are you getting your data from? Most people are reporting F3 as being 0.5 ohm and the only data I can find for polyfuses suggests that the resistance can vary by a factor of 6.
LINUX convert since 2003

Clive_G
Posts: 5
Joined: Fri Jul 20, 2012 8:48 pm

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 7:35 am

I've measured (accurately) F3 on two Raspi - 0.2 and 0.3 ohms, so I agree F3 won't be a problem if current draws are kept within spec.

After popping F1 it returned back down to about 4 ohms when the Raspi had completely cooled down.

So, they are doing the job as intended - just that they can throw unexpected issues at you.

Normally USB memory sticks should be OK for F1/F2, e.g. see: http://darts.cse.nd.edu/papers/OBrien08PAP.pdf - I'll try to rig a meter to measure USB current draws.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12125
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 7:23 pm

@Clive_G
interesting document showing that most often these memory sticks do not consume more than 100mA. Problem is, do they work with the voltages left after a typical 140mA polyfuse?
I think not! The document doesn't
mention working voltage at all which seems to indicate using the perfect voltage, 5.0 Volt. But the voltage drop caused by the 140mA fuses will cause it to drop considerably lower, lower even than normal 5V logic would allow,
The thing is, real low power ("single current unit USB devices)" such as keyboards, and certainly mice will work, are explicitly designed to work, with voltages as low as 4.4Volt, and a memory stick might not!

@bazza14
I have the 0.2 Ohm figure from several sources, including the datasheet of the polyfuse.
Note that its hard to measure sub 1 ohm resistances with a multimeter, as the leads, and contact resistance might also be in the neighbourhood of fractions of an ohm, so you need to have some experience in measuring such low resistances. Most often if jou simply read what the meter says you are of by the lead resistance, which often is a few fractions of an ohm. The trick is to measure a short, then what you get then should be subtracted from your real measurement.
The cold resistance of the (140mA) polyfuses are normally between three and six ohm, a factor two, not a factor of six. After a "burnout" and recovery, the resistance will initially be much higher, but will return to almost normal in say a day or so, perhaps in an extreme case to something like 8 ohm after several cycles.

User avatar
bazza14
Posts: 65
Joined: Sun Jan 01, 2012 10:05 am

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 8:48 pm

Wonderful, you have the data sheet for THIS poly fuse, could you provide a link please. Or do you mean a datasheet for A ployfuse such as the one I found which states that the resistance would be between 1 and 6 ohms which is clearly not this device. It is quite possible that later boards are using a device from a different manufacturer which would account for the early boards measuring in at 0.5 and now it's 0.2 I would imagine the BOM only specifies a 700mA poly fuse, I can't imagine they'd insist on a specific part number, so the manufacturers would be free to select the actual device used with any limits specified by the design.
Thanks for the tips on measuring low resistances but your a bit late, I covered that in my electronics apprenticeship 40+ years ago. When I get the rest of my boards I will measure them all and post the results here.
LINUX convert since 2003

simonbr
Posts: 25
Joined: Sun May 13, 2012 5:02 pm

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 9:17 pm

bazza14 wrote: I can't imagine they'd insist on a specific part number, so the manufacturers would be free to select the actual device used with any limits specified by the design.
In fact it is normal in large scale production that you have to specify the exact parts. A serious production company would never use components that are not explicitly listed (as alternatives) in the bill of materials.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12125
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 9:33 pm

well the actual 140mA polyfuses that were in the first pictures of the model B, were recognizable Tyco devices, but as polyfuses all operate using the same physical principles, you can assume that for a given tripping current and body size (1812) they have to be very similar in resistance rate. The datasheet I was referring to was this one:
http://www.littelfuse.com/data/en/Data_ ... _1812L.pdf
And I was referring to the 1812L014
And yes, you are right in that it has minimum and maximum values of 1.5 to six ohms (after tripping), not three to six ohms, so it is indeed a factor four for this particular device, not a factor two as I wrongly assumed. I'm guessing those people who have 1.5 Ohm polyfuses are lucky enough to not have much problems.
Still the question was whether there was any real gain in bypassing the main polyfuse F3 if you had any problems with voltage drops, and I think I can still say with some confidence that its not worth the trouble.

I'm also quite sure the real resistance of F3 (1.1A and 0.75A hold current) isn't half an ohm, but rather lower, the typical polyfuses I have seen are typically 0.04 to 0.21 Ohm (after tripping). But indeed it will depend on the actual rating of the polyfuse itself.

If you think about it, for a typical current of 100mA the voltage drop would be 0.1 Volt per ohm resistance of the total chain, even if F3 really was half an ohm the extra 0.05V wouldn't matter much. While if you have an average 140mA polyfuse of say four ohm, it would mean a drop of 0.4 volt, which is significant, it would mean a voltage of 5V before it would drop to 4.6 Volt after it, which is below the 4.75 (5V +-5%) minimum rating of most 5V logic.

P.S. in fact the actual datasheet for the X14 (Tyco 140mA fuse) can be found here:
http://www.datasheetarchive.com/datafra ... niSMDC014# and the specs are identical 1.5 to six ohm.

User avatar
rurwin
Forum Moderator
Forum Moderator
Posts: 4258
Joined: Mon Jan 09, 2012 3:16 pm
Contact: Website

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 10:41 pm

mahjongg wrote:If you think about it, for a typical current of 100mA the voltage drop would be 0.1 Volt per ohm resistance of the total chain, even if F3 really was half an ohm the extra 0.05V wouldn't matter much.
The voltage drop across F3 depends on the total current, not that fraction of it that goes to the USB ports. A current of 700mA gives a voltage drop of 7 x 0.05V = 0.35V, which is significant since it can make an in-spec 5V power supply insufficient. Fortunately a more usual current draw seems to be around 400mA including mouse and keyboard, which allows a 5V power supply to stay in spec... just.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12125
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 10:44 pm

yes, you are right in that. I guess its the total of all the problems, note that 0.2 Ohm may also well be in the same ballpark as the fuse F3, and can differ just as much, exactly what we see.

Still if you are using non 4.4 Compliant USB devices the USB polyfuses are the biggest obstacle for their use. And even for 4.4V devices it can get tricky, if all the polyfuses have had a hit, note that the high values mentioned are for polyfuses one hour after tripping and at 20 degrees Celsius (according to most datasheets).

The whole issue with all the polyfuses is a big mess.

Its not so much that I'm against bypassing the 5V to the USB polyfuses, its just that if you naively solder a wire between the micro USB port and the 140mA polyfuses what you are really doing is bypassing F1, and that could have dire consequences (as well as being totally aesthetically displeasing :P ) If there would ever be a complete PCB redesign to fix these issues, then yes, without a moments thought it would be much better to route the input 5V directly to the USB fuses, the only drawback would be that maybe then the USB devices are not so much protected against overvoltage by the TISP D17.

It would be much better when all fuses would be chosen to have more suitable resistances, with tripping values as a secondary significance, its not much use to protect your device if it isn't working properly in the first place. During a real short even the worst chargers would be able to deliver enough current to trip fuses that have a fraction of the current resistances.

For example if the fuses F1 and F2 would have been identical to F3 it would have been much better. We wouldn't have the problems with hubs feeding back (a little bit of) power we have now, and a memory stick would probably have worked fine, as would most WiFi adapters and such.

User avatar
bazza14
Posts: 65
Joined: Sun Jan 01, 2012 10:05 am

Re: Design flaw? Polyfuses for USB current limiting

Sat Jul 21, 2012 11:45 pm

simonbr wrote:
bazza14 wrote: I can't imagine they'd insist on a specific part number, so the manufacturers would be free to select the actual device used with any limits specified by the design.
In fact it is normal in large scale production that you have to specify the exact parts. A serious production company would never use components that are not explicitly listed (as alternatives) in the bill of materials.
Actually I'll have to disagree with this. Unless you are building something of very high quality for use in extreme environments it is common (if you are using a contract manufacturer) to specify simple devices such as resistors, capacitors, inductors, PTCs etc eg 10Kohm +/-1% 1206 package Mil spec and let them select the specific device. There are sound operational, logistical, and financial reasons for doing this.
LINUX convert since 2003

User avatar
bazza14
Posts: 65
Joined: Sun Jan 01, 2012 10:05 am

Re: Design flaw? Polyfuses for USB current limiting

Sun Jul 22, 2012 12:23 am

mahjongg wrote: The whole issue with all the polyfuses is a big mess.
Yes it is and I think every thing that can be said about has been said at least once. I think we are all now well informed that if you try use the USB ports for anything other than a keyboard and mouse you may run into problems.
mahjongg wrote: Its not so much that I'm against bypassing the 5V to the USB polyfuses, its just that if you naively solder a wire between the micro USB port and the 140mA polyfuses what you are really doing is bypassing F1, and that could have dire consequences (as well as being totally aesthetically displeasing :P )
I have with great forethought and deliberation soldered wires directly to the USB ports and I've dropped a short across F3 and I've implemented the over current protection in the power supply. As the board meets the spec for which it was designed I doubt very much that there will be a redesign to accomodate our hacks. Although the hack doesn't look pretty (what hacks do) it solves all the problems associated with the ploy fuses and enables me to carry on testing which at the end of the day is all I'm interested in.
LINUX convert since 2003

simonbr
Posts: 25
Joined: Sun May 13, 2012 5:02 pm

Re: Design flaw? Polyfuses for USB current limiting

Sun Jul 22, 2012 4:46 pm

bazza14 wrote: Actually I'll have to disagree with this.
Is that opinion due to actual experience with contract manufacturing in the Far East?

User avatar
bazza14
Posts: 65
Joined: Sun Jan 01, 2012 10:05 am

Re: Design flaw? Polyfuses for USB current limiting

Sun Jul 22, 2012 11:14 pm

simonbr wrote:
bazza14 wrote: Actually I'll have to disagree with this.
Is that opinion due to actual experience with contract manufacturing in the Far East?
No we're a bit south of there what about you?
LINUX convert since 2003

simonbr
Posts: 25
Joined: Sun May 13, 2012 5:02 pm

Re: Design flaw? Polyfuses for USB current limiting

Mon Jul 23, 2012 8:06 am

bazza14 wrote:
simonbr wrote:
bazza14 wrote: Actually I'll have to disagree with this.
Is that opinion due to actual experience with contract manufacturing in the Far East?
No we're a bit south of there what about you?
I have experience with two contract manufacturers in China, a medium sized and a big one, and they both require us to specify the make and type number of every component in the BOM.

Clive_G
Posts: 5
Joined: Fri Jul 20, 2012 8:48 pm

Re: Design flaw? Polyfuses for USB current limiting

Mon Jul 23, 2012 4:39 pm

OK, some actual measurements breaking into the USB leads (ambient temp 25C, a nice warm day up here in Newcastle)

USB PSU (iPad) off load 5.18v on (Pi) load 4.96v

This drops to 4.82v at the Pi itself, then 4.68v after going through F3.

USB keyboard (Logitech K120) and mouse (IT works MCO 05, quiescent) 4mA, mouse active 8mA

The Kingston USB memory stick draws 80mA, at which the voltage on it drops to 4.34V after going through F1 - it happens to still work at this voltage though.

Moral: cumulative voltage drops (leads, connectors, F1/2/3) could become problematic depending on what you are trying to do - plan accordingly.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12125
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Design flaw? Polyfuses for USB current limiting

Sat Aug 25, 2012 12:40 pm

According to this post http://www.raspberrypi.org/phpBB3/viewt ... &start=150 the RPF decided to address the polyfuse problem, and there are reports of new PI's having the 140mA polyfuses replaced by zero ohm resistors.

User avatar
liz
Raspberry Pi Foundation Employee & Forum Moderator
Raspberry Pi Foundation Employee & Forum Moderator
Posts: 5201
Joined: Thu Jul 28, 2011 7:22 pm
Contact: Website

Re: Design flaw? Polyfuses for USB current limiting

Sat Aug 25, 2012 12:46 pm

That's correct - we've designed out those pads for the next board revision, but until that happens we're linking out the fuses. I'm really impressed how quickly someone noticed the difference!
Director of Communications, Raspberry Pi

W. H. Heydt
Posts: 10772
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Design flaw? Polyfuses for USB current limiting

Sat Aug 25, 2012 4:49 pm

For those of us who don't spend a lot of time doing things to the hardware, is there a visible way to tell whether a given board has the 0 ohm units? Or, alternatively, on what manufacturing date did the change take place?

dom
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 5318
Joined: Wed Aug 17, 2011 7:41 pm
Location: Cambridge

Re: Design flaw? Polyfuses for USB current limiting

Sat Aug 25, 2012 4:58 pm

W. H. Heydt wrote:For those of us who don't spend a lot of time doing things to the hardware, is there a visible way to tell whether a given board has the 0 ohm units? Or, alternatively, on what manufacturing date did the change take place?
There's a picture in the other thread:
https://picasaweb.google.com/1099282360 ... 0375008498

Note the "0" on F1 and F2. That means you've got the updated version.

User avatar
Dweeber
Posts: 606
Joined: Fri Aug 17, 2012 3:35 am
Location: Mesa, AZ
Contact: Website

Re: Design flaw? Polyfuses for USB current limiting

Sat Aug 25, 2012 8:22 pm

Which mine doesn't have. Mine looks like:

ImageTaken with HTC Rezound

I've not had any power related issues so far and my application for this unit won't change much.

Will be getting a couple more RPi's in the future where this might be an issue. I'm sure I will get the revised version when I get them (several months away).
Dweeber A.K.A. Kevin...
My RPI Info Pages including Current Setup - http://rpi.tnet.com

W. H. Heydt
Posts: 10772
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Design flaw? Polyfuses for USB current limiting

Sun Aug 26, 2012 1:55 am

dom wrote:
W. H. Heydt wrote:For those of us who don't spend a lot of time doing things to the hardware, is there a visible way to tell whether a given board has the 0 ohm units? Or, alternatively, on what manufacturing date did the change take place?
There's a picture in the other thread:
https://picasaweb.google.com/1099282360 ... 0375008498

Note the "0" on F1 and F2. That means you've got the updated version.
Thanks much. I'll see if the ones that are in transit (shipped Friday from Allied, due to arrive on Wednesday) have the zero ohm or not. A board I got from Element 14 last month with a "1222" production date has non-zero units.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12125
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Design flaw? Polyfuses for USB current limiting

Sun Aug 26, 2012 2:47 am

W. H. Heydt wrote:For those of us who don't spend a lot of time doing things to the hardware, is there a visible way to tell whether a given board has the 0 ohm units? Or, alternatively, on what manufacturing date did the change take place?
Polyfuses look like this:
Image
They are generally green, and have little indentations on their solder sides. as in the above picture on the left and right side.
Resistors (including zero ohm resistors) are generally black, and have no such notches.

User avatar
reiuyi
Posts: 165
Joined: Sun Oct 09, 2011 4:59 pm
Contact: Website

Re: Design flaw? Polyfuses for USB current limiting

Sun Aug 26, 2012 10:26 am

W. H. Heydt wrote:
Thanks much. I'll see if the ones that are in transit (shipped Friday from Allied, due to arrive on Wednesday) have the zero ohm or not. A board I got from Element 14 last month with a "1222" production date has non-zero units.

Date codes always represent PCB manufacturing, not product manufacturing. If I remember correctly, the PCBs were made way earlier than they were actually populated with parts. Now I'm not sure what sort of engineering etiquettes there are concerning adherence to silk-screen part codes (like replacing a polyfuse with a wire-bridge and being indifferent to what the silk-screen says), so if anyone knows please tell us! Though changing the silk-screen settings is a weekly thing, regardless. I still need to bridge my own polyfuses, though. It's a fun little mod that isn't too hard even for the very beginners. All you need is a paperclip (or another wire source), some pliers, soldering iron, tiny bit of solder and good lights/magnifying glass/ 20-20 eyes!

Return to “General discussion”