Over Voltage USB PSUs


21 posts
by theHetman » Tue May 15, 2012 11:28 pm
I got my email for RS inviting me to order today, which of course I did, but just recently I have been reading stories to the effect that the Pi, or rather USB peripherals might be sensitive to over voltage on the PSU.

The video that was embedded into the "Australian" article said that "some newer iPad PSUs" were over voltage and that you didn't want to "fry your Pi" and last week there was the report from the scout troupe that were using the Pi to control a robot but the WiFi/Bluetooth USB devices weren't working partly because of over voltage PSUs. Their were using an expensive lab-bench PSU to provide power and said particularly that an Apple PSU that they had tried put out 5.3V

Now, I was planning on using a spare iPad PSU to power my Pi and this caused me some concern. I decided to test the PSUs that I had to see if there was a problem. I took a foot long USB A plug to USB A socket "extender" and cut the cable in two, stripping back the ends of the lines so that I could get my multimeter on them.

I tested two iPad PSUs and both output 5.15v. I also tested a PSU "brick" that came with a SATA/IDE to USB adapter which has a molex conector on the output and is rated at 2A for the 5V line and 2A for the 12V line. It also was slightly over voltage at 5.16V but all these tests were done with no load and I would expect some drop in voltage on reasonably cheap PSUs when some current was being drawn. For comparison my PC's USB ports also output 5.15V and that PSU is certainly under load.

So what I'm really asking is does a 3% over voltage cause problems for the Pi and any peripherals that might be connected? I can use a really thin USB A to micro USB cable to connect the Pi that would probably drop the voltage a little ;) but seriously, is it really such a problem?

What I plan on doing with my cut extender cable is to put a male molex onto the socket half of the cable so that I can pull USB power from the molex "brick" (although I need to research how to tie the data lines) and for the plug end I plan to solder it to some veroboard with an A socket for output so that I have an easy way to test USB PSUs both with and without a load.
Posts: 88
Joined: Tue Jan 10, 2012 5:42 pm
by fodi » Tue May 15, 2012 11:47 pm
usb standard is 5volts +/- 5% (0.25volt)
plus you measured it without any load (it'd drop with load), so i'd say it will be perfect if it supplies enough amps

plus if i recall correctly, (but take a look at the schematics) the power input does not have any data lines connected, it is only +5V and GND that matter
User avatar
Posts: 112
Joined: Wed Mar 14, 2012 9:03 pm
Location: Hungary
by bredman » Wed May 16, 2012 10:44 am
By exceeding 5v on the input, you risk overheating the polyfuse on the 5v line. This is theoretical, nobody has done any stringent tests on this.
If the polyfuse overheats, you can expect random crashes. The polyfuse may need at least an hour to cool down properly afterwards.
Posts: 1415
Joined: Tue Jan 17, 2012 2:38 pm
by theHetman » Wed May 16, 2012 12:56 pm
Thanks guys. I was aware that the SoC itself was protected by a polyfuse so that it was hard to do damage to it but also that there was no protection for the +5V on the USB ports. I've seem various reports that poor PSUs cause problems for the Pi and also for USB devices but I'm still not sure if this slight over voltage is a really problem or not. 0.15V is within the 5% for the USB spec after all.

As to the question of how to tie the data lines for a USB PSU, I know that those pins are not connected on the Pi's micro USB port but many devices test for voltage on those lines to see if they can draw more that 500mA. An iPad won't charge from a standard USB port for example and the iPad PSU has 2.75V on data+ and 2.05V on data-. I'm guessing that these are signal voltages that tell the iPad it can draw 2A of power. What I'm wondering is what other manufacturers do and if is there any sort of standard as to how these lines should be tied. (I'm very aware that Apple does their own thing).)
Posts: 88
Joined: Tue Jan 10, 2012 5:42 pm
by alexeames » Wed May 16, 2012 5:42 pm
As far as I'm aware (which is not very far, so please check) for most other applications, shorting the data lines does the trick.

This is what I did when I connected up a LM2596S switching reg to a USB socket to run my Pi (not here yet). Before I shorted the data pins (at the socket end) connecting my Samsung Galaxy S2 put it into USB mode (using normal USB cable). After the data pins were shorted, it goes into charge mode.

HTH but check, verify and check again :)
Alex Eames RasPi.TV, RasP.iO
User avatar
Forum Moderator
Forum Moderator
Posts: 2807
Joined: Sat Mar 03, 2012 11:57 am
Location: UK
by mahjongg » Wed May 16, 2012 5:51 pm
theHetman wrote:So what I'm really asking is does a 3% over voltage cause problems for the Pi and any peripherals that might be connected? I can use a really thin USB A to micro USB cable to connect the Pi that would probably drop the voltage a little ;) but seriously, is it really such a problem?


No because all 5V (USB) devices are rated 5V + of - 5%, so any voltage between 4.75 and 5.25V are within normal tolerances.
In fact, having a slightly elevated output voltage wight even help mitigate the voltage drop over the currently used polyfuses for USB protection, which are slightly under dimensioned, and can cause a slight voltage drop (in the order of a few percentage points).
User avatar
Forum Moderator
Forum Moderator
Posts: 9210
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands
by theHetman » Wed May 16, 2012 11:58 pm
Yes, it looks that tying the data lines is supposed to be the standard for USB power and a device can draw up to 1.5A

https://secure.wikimedia.org/wikipedia/en/wiki/USB#Power

That article also explains what Apple does. I also suspect that with the EU drive to standardize phone chargers the PSU wiring will became more standard as well. Apple will still do its own thing of course. Their only concesion has been to sell a 30 pin to micro USB adaptor in the EU.

As to the slight over voltage I'm not too worried about it.
Posts: 88
Joined: Tue Jan 10, 2012 5:42 pm
by bredman » Thu May 17, 2012 8:05 am
For once I must defend Apple. The USB charging protocols were developed by Texas Instruments and are part of the official USB specification.

Apple were just the first to build a mass-market device (the iPod) which depended on this facility.
Posts: 1415
Joined: Tue Jan 17, 2012 2:38 pm
by Jim Manley » Thu May 17, 2012 10:01 am
bredman wrote:For once I must defend Apple. The USB charging protocols were developed by Texas Instruments and are part of the official USB specification.

Apple were just the first to build a mass-market device (the iPod) which depended on this facility.


The original iPod had a six-wire Firewire interface that also provided charging power. mp3 players and other devices made by other manufacturers had USB interfaces for charging and data transfer long before the later iPods shifted to having USB interfaces.
The best things in life aren't things ... but, a Pi comes pretty darned close! :D
"Education is not the filling of a pail, but the lighting of a fire." -- W.B. Yeats
In theory, theory & practice are the same - in practice, they aren't!!!
User avatar
Posts: 1593
Joined: Thu Feb 23, 2012 8:41 pm
Location: SillyCon Valley, California, and Powell, Wyoming, USA, plus The Universe
by plugwash » Thu May 17, 2012 10:44 am
bredman wrote:For once I must defend Apple. The USB charging protocols were developed by Texas Instruments and are part of the official USB specification.

Apple were just the first to build a mass-market device (the iPod) which depended on this facility.

No apple at least initially did their own thing on USB chargers (multiple resistors to indicate current output). Other vendors did too.

Standardising on the approach of shorting the data lines came later.
Forum Moderator
Forum Moderator
Posts: 3217
Joined: Wed Dec 28, 2011 11:45 pm
by theHetman » Sun May 20, 2012 2:29 pm
So, I've now been able to run some tests under load for my various PSUs. I made up a short adaptor cable (USB A male to USB A female) with easy to get at test points in it so that I could test the voltage.

The iPad PSUs when charging the iPad (which I'm guessing is pretty much full load) dropped from around 5.15V to 5.01V and when powering the Pi which take about 700mA dropped to 5.04V.

Using the PSU which has a molex on it and came with an IDE/SATA - USB adaptor the volatage dropped from 5.16V to 4.85V when running the Pi. The iPad wouldn't charge from it but the voltage when it was plugged in was still around 4.85V.

To me that just shows how good Apple's PSUs are. They may be pricy but they are probably worth it in the long run.
Posts: 88
Joined: Tue Jan 10, 2012 5:42 pm
by lewmur » Sun May 20, 2012 10:05 pm
theHetman wrote:So, I've now been able to run some tests under load for my various PSUs. I made up a short adaptor cable (USB A male to USB A female) with easy to get at test points in it so that I could test the voltage.

The iPad PSUs when charging the iPad (which I'm guessing is pretty much full load) dropped from around 5.15V to 5.01V and when powering the Pi which take about 700mA dropped to 5.04V.

Using the PSU which has a molex on it and came with an IDE/SATA - USB adaptor the volatage dropped from 5.16V to 4.85V when running the Pi. The iPad wouldn't charge from it but the voltage when it was plugged in was still around 4.85V.

To me that just shows how good Apple's PSUs are. They may be pricy but they are probably worth it in the long run.
Undoubtedly there are poorly built PSUs out there that don't work right but that is no reason to pay Apple's prices. You can buy a PSU rated a 2 amps for under $5 that will work just fine. A PSU for an IDE/SATA is probably rated at less than .5 amp and that is why its voltage is dropping. Remember, a PSU with a high amp rating won't damage your Pi. Only a high voltage will do that.
Posts: 316
Joined: Sun Dec 25, 2011 3:20 pm
by theHetman » Sun May 20, 2012 10:22 pm
Oh I agree that there are good cheap PSUs out there. All I was saying is that the Apple PSUs are good (and I had a spare one). We just need to be able to source some good ones so that people know what they are buying. As for the IDE/SATA PSU it's a "brick" style PSU (like a laptop PSU) and is rated at 2A on the 5V line (and 2A on the 12V line).
Posts: 88
Joined: Tue Jan 10, 2012 5:42 pm
by lewmur » Mon May 21, 2012 11:10 am
theHetman wrote:Oh I agree that there are good cheap PSUs out there. All I was saying is that the Apple PSUs are good (and I had a spare one). We just need to be able to source some good ones so that people know what they are buying. As for the IDE/SATA PSU it's a "brick" style PSU (like a laptop PSU) and is rated at 2A on the 5V line (and 2A on the 12V line).
As I said, there are poor quality PSUs out there but one that is supposed to have a 2A 5V supply shouldn't have that much of a voltage drop when powering a Pi.
Yes, it would be nice to be able to point people to specific sources for ones known to work. And as more and more people get their Pi's and are able test PSUs, I'm sure that will happen. I was merely trying to point out that, as a general rule, people should have a lot more luck purchasing ones with a higher amp rating. When buying one (ambitiously) rated a 1A or even less, I would expect crashes under heavy CPU load. But the maker would have to be out and out lying for that to happen with one rated at 1.5A or higher.
Posts: 316
Joined: Sun Dec 25, 2011 3:20 pm
by mahjongg » Mon May 21, 2012 5:28 pm
mahjongg wrote:In fact, having a slightly elevated output voltage might even help mitigate the voltage drop over the currently used polyfuses for USB protection, which are slightly under dimensioned, and can cause a slight voltage drop (in the order of a few percentage points).


One remark to that: If the PSU overdoes it, and tries to put (say) 6 volt on the PI, perhaps when unloaded the PSU outputs 6 Volt,. If you then plug it in, with an internal elco in the PSU charged op to six volt, then a protection diode on the R-PI discharges the over-voltage and a high current might flow through the polyfuse which might trigger it (depending on how much charge the elco in the PSU holds), after that the Polyfuse might stay triggered, and/or it may have a higher than normal internal resistance for some time.

I'm not saying this will often happen, but its definitely possible.

I hope this was clear enough to understand, the point is that too much over-voltage might trigger the over voltage protection diode, which may "blow the fuse".
User avatar
Forum Moderator
Forum Moderator
Posts: 9210
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands
by lewmur » Mon May 21, 2012 5:57 pm
mahjongg wrote:
mahjongg wrote:In fact, having a slightly elevated output voltage might even help mitigate the voltage drop over the currently used polyfuses for USB protection, which are slightly under dimensioned, and can cause a slight voltage drop (in the order of a few percentage points).


One remark to that: If the PSU overdoes it, and tries to put (say) 6 volt on the PI, perhaps when unloaded the PSU outputs 6 Volt,. If you then plug it in, with an internal elco in the PSU charged op to six volt, then a protection diode on the R-PI discharges the over-voltage and a high current might flow through the polyfuse which might trigger it (depending on how much charge the elco in the PSU holds), after that the Polyfuse might stay triggered, and/or it may have a higher than normal internal resistance for some time.

I'm not saying this will often happen, but its definitely possible.

I hope this was clear enough to understand, the point is that too much over-voltage might trigger the over voltage protection diode, which may "blow the fuse".

I doubt it. Any decent PSU is going to exceed the Pi's voltage spec when totally unloaded. But it will drop to within specs almost instantly when plugged in. The over-voltage protection circuits are designed to handle that. IMHO, measuring the unloaded voltage of the PSU is meaningless. And I've been dealing with electronic circuit for over 50 years. In fact, the voltage meters we used way back when would have placed enough load on the PSU to get a realistic reading. Today's meters have such a high input impedance that this doesn't happen.
edit: If you want to get a reading on the voltage output of a PSU, at least load it with a 1k resistor. This will only draw about 5ma but will give you a much more realistic reading.
Posts: 316
Joined: Sun Dec 25, 2011 3:20 pm
by xtramural » Mon May 21, 2012 8:26 pm
theHetman wrote:Oh I agree that there are good cheap PSUs out there. All I was saying is that the Apple PSUs are good (and I had a spare one). We just need to be able to source some good ones so that people know what they are buying. As for the IDE/SATA PSU it's a "brick" style PSU (like a laptop PSU) and is rated at 2A on the 5V line (and 2A on the 12V line).

Although I would agree that Apple charges (no pun intended!) a premium for their products, all chargers are not created equal. I've just stumbled across the following. I leant a lot from the article and have a much better appreciation of what a quality charger/PSU entails.

Ken Shirriff's blog: Apple iPhone charger teardown: quality in a tiny expensive package
Disassembling Apple's diminutive inch-cube iPhone charger reveals a technologically advanced flyback switching power supply that goes beyond the typical charger. It simply takes AC input (anything between 100 and 240 volts) and produce 5 watts of smooth 5 volt power, but the circuit to do this is surprisingly complex and innovative.
Posts: 108
Joined: Thu Dec 29, 2011 11:16 pm
Location: Scotland
by lewmur » Tue May 22, 2012 1:45 pm
xtramural wrote:
theHetman wrote:Oh I agree that there are good cheap PSUs out there. All I was saying is that the Apple PSUs are good (and I had a spare one). We just need to be able to source some good ones so that people know what they are buying. As for the IDE/SATA PSU it's a "brick" style PSU (like a laptop PSU) and is rated at 2A on the 5V line (and 2A on the 12V line).

Although I would agree that Apple charges (no pun intended!) a premium for their products, all chargers are not created equal. I've just stumbled across the following. I leant a lot from the article and have a much better appreciation of what a quality charger/PSU entails.

Ken Shirriff's blog: Apple iPhone charger teardown: quality in a tiny expensive package
Disassembling Apple's diminutive inch-cube iPhone charger reveals a technologically advanced flyback switching power supply that goes beyond the typical charger. It simply takes AC input (anything between 100 and 240 volts) and produce 5 watts of smooth 5 volt power, but the circuit to do this is surprisingly complex and innovative.

Porsche builds a good automobile. But I wouldn't buy one to tow around a 10 foot dingy.
Posts: 316
Joined: Sun Dec 25, 2011 3:20 pm
by theHetman » Tue May 22, 2012 3:58 pm
As I said I happen to have a spare iPad PSU so I might as well use it. :)



He also said that he has done a teardown of a Samsung phone charge which he says is almost as good.

http://www.amazon.com/gp/product/B005OSURT2/ref=as_li_ss_tl?ie=UTF8&tag=rightocom&linkCode=as2&camp=1789&creative=390957&creativeASIN=B005OSURT2

but it's only rated at 1.0A.
Posts: 88
Joined: Tue Jan 10, 2012 5:42 pm
by lewmur » Tue May 22, 2012 4:24 pm
theHetman wrote:As I said I happen to have a spare iPad PSU so I might as well use it. :)



He also said that he has done a teardown of a Samsung phone charge which he says is almost as good.

http://www.amazon.com/gp/product/B005OSURT2/ref=as_li_ss_tl?ie=UTF8&tag=rightocom&linkCode=as2&camp=1789&creative=390957&creativeASIN=B005OSURT2

but it's only rated at 1.0A.
A previous post said the Apple charger was 5 watts. That means it would be 1A also.
Posts: 316
Joined: Sun Dec 25, 2011 3:20 pm
by theHetman » Tue May 22, 2012 4:29 pm
lewmur wrote:A previous post said the Apple charger was 5 watts. That means it would be 1A also.

Yes, the teardown was of an iPhone charger which is 1.0A. The iPad charger is 2A.
Posts: 88
Joined: Tue Jan 10, 2012 5:42 pm