Adequate Power Supply Critical to Pi Stability


79 posts   Page 1 of 4   1, 2, 3, 4
by Jim Manley » Thu May 03, 2012 6:12 am
This won't be news to anyone who has been using an adequately-rated power supply for their Pi, but, a typical cheaply-made cell phone charger allegedly rated at a nominal 5 volts at 700 ma may not be enough to ensure stable operation of the Pi.  This is especially true if the micro-USB cable is not constructed with conductors of adequate gauge (wire thickness) and/or the cable connectors don't make good enough contact under full load.  We had been experiencing at least once-daily mysterious crashes on a Pi in the UK I've been accessing remotely from California.

After the owner upgraded to a 2.1 amp supply with a more substantial cable, we've been running for several days with no more crashes.  What is particularly curious is that the Pi would crash even with nothing else running besides an xterm window on LXDE on the Debian 13 April release, with nothing executing in the xterm window when the crashes occurred.  In theory, the Pi should have been drawing well under the 700 ma maximum stated in the Pi hardware power requirements spec.

I had previously turned off the screen-saver, thinking that some of the more graphically-intense savers might be pushing the Pi to the limit of current capacity of the cell phone power supply, but, that was not the case.  The GPU wasn't even being used during any of this period, which is even more surprising.  The owner hasn't had the chance to measure voltage and current output of the original supply under various Pi operating loads, but, it's pretty clear now that it was the culprit.

The moral of the story is that, even if you have a power supply that is allegedly rated at an adequate voltage and current for the Pi's requirements, it may not actually be delivering the required power in terms of voltage and/or current.  You cannot have too much current capacity as long as the power supply is providing a regulated 5 volts (+/- 0.25 volts, i.e., 4.75 to 5.25 volts under load).  Any current in excess of the Pi's requirements can be routed to an active USB hub, so, upward of several amps of current capacity would be just fine.

If I ever receive a Pi board, I will be taking it to all of the cell service providers in my area and trying to power it with as many different micro-USB equipped device chargers as are available, running the Pi up to full operating load, including the GPU.  I'll publish a report by both phone and charger manufacturer, where it's specified on the charger, along with the current ratings, which should be printed on the chargers.
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: 1358
Joined: Thu Feb 23, 2012 8:41 pm
Location: SillyCon Valley, California, USA
by shirro » Thu May 03, 2012 6:43 am
A lot of cheap no-name stuff is mislabelled.

Check out this video: 

Most electronic goods are sold with a power supply with the correct rating. While the usb power option is convenient for people with a stack of quality USB chargers on hand RS and Element 14 probably should put a Pi+PSU bundle up front as the main option to make things less error prone for people who aren't too sure.
Posts: 248
Joined: Tue Jan 24, 2012 4:54 am
by rurwin » Thu May 03, 2012 6:51 am
I'll be testing PSUs soon as well, although maybe not as thoroughly. It certainly seems that the Foundation may have been wrong in assuming that USB chargers produce the current they say they do and that all USB cables are reasonably constructed.

For those people not geared up with volt meters etc., the best advice at the moment is probably to buy a package from RS or Farnell, even if it is more expensive than your local phone shop.
User avatar
Forum Moderator
Forum Moderator
Posts: 2949
Joined: Mon Jan 09, 2012 3:16 pm
by Jongoleur » Thu May 03, 2012 7:02 am
I've got a number of USB chargers of at least the approximatly correct output and variously sourced cables.  WHEN I get a Pi, I'll give them all a go and provide a suitable pass/fail list...

:-)
I'm just a bouncer, splatterers do it with more force.....
User avatar
Posts: 1179
Joined: Thu Aug 11, 2011 12:47 pm
Location: O'erlooking the sea, and all those effin windfarms...
by shirro » Thu May 03, 2012 7:15 am
So far everything I have tried with an Apple logo works but they probably cost more than the Pi.  I have a couple of other USB power supplies that claim to be 1A that I can test.

What is the suggested testing method. Check the voltage across the testpoints – but with the Pi idle or running something CPU intensive – what about a load on the USB ports? Will it make much difference?

BTW my guess is if your charger was bundled with something like a phone or a tablet from a major manufacturer it will probably be to spec. I would guess that it is hard to fool a major electronics distributor as well. Wouldn't be too sure about the cheap stuff at the discount store.
Posts: 248
Joined: Tue Jan 24, 2012 4:54 am
by fulburncams » Thu May 03, 2012 7:24 am
Good point re the USB loads,  I am having a nightmare getting USB-Wifi adapters going ( as are others).  I did a quick analysis of currents for wifi adapters and found these can be in range 130-150mA ( right on the USB polyfuse limit).  Subsequently adding a powered USB hub only made matters worse as there is apparently a USB hub issue currently being investigated ,  see here :-

http://www.raspberrypi.org/for.....mouse-wifi

and ..

http://www.raspberrypi.org/for.....r-hub-wifi

Cheers

Steve
User avatar
Posts: 70
Joined: Wed Nov 30, 2011 8:29 pm
by rurwin » Thu May 03, 2012 8:24 am
shirro said:


What is the suggested testing method. Check the voltage across the testpoints – but with the Pi idle or running something CPU intensive – what about a load on the USB ports? Will it make much difference?


Probably a bit of everything. Jim reports that it fails when it's idle, but only once a day. Maybe there is a current spike every now and then. The current taken by the RaspPi is the big unknown. It is possible that that is where the fault lies; maybe it demands 1.5A for a millisecond now and then.

Those of us without RaspPis are limited to testing the power supplies with dummy loads such as the Useful and cheap power-supply tester I posted.


BTW my guess is if your charger was bundled with something like a phone or a tablet from a major manufacturer it will probably be to spec. I would guess that it is hard to fool a major electronics distributor as well. Wouldn't be too sure about the cheap stuff at the discount store


I had a USB charger fail in service with sparks and smoke, although it gave several months of good service before that. I bought it from an independent mobile phone shop on the high-street. So I would agree that only major distributors are set up to vet their products to the extent required. It is obviously not enough to go down the cash-and-carry and buy a load of chargers trusting what it says on the label.

Most people put their phones on charge for hours on end, and would not notice if it only charged at the 100mA rate instead of the 500mA rate, let alone the 1A rate. So wrongly marked products are not easy to detect. Neither HMRC nor anyone else has the responsibility of testing such goods, so it falls on the distributor to do so. (Theoretically it should be the importer.) Only a distributor that is capable of doing so will even know it is necessary. That does not include wholesalers, discount stores or even independent mobile phone shops. It should include electronics distributors or any company that puts their brand on the goods.
User avatar
Forum Moderator
Forum Moderator
Posts: 2949
Joined: Mon Jan 09, 2012 3:16 pm
by johnbeetem » Thu May 03, 2012 3:09 pm
Sometimes you need a 'silly scope to see what's really going on with an iffy supply.  For example, your multi-meter may be reporting 5.0 volts but if you look at the voltage on a 'scope you'll see 5.3V when the 110/220VAC input is high and 4.7V when the 110/220 VAC input is low.

For my BeagleBoard, I salvaged a Linear Technology switching regulator demo board which spits out 5V +/- 2% at 1.5A max from a 6VDC to 25VDC input, e.g., a spare 9V or 12V non-regulated wall wart.  The only part I'm worried about re: RasPi is the micro USB cable.  Maybe I'll get a micro USB connector from Digi-Key and make my own cable with real wire.
User avatar
Posts: 942
Joined: Mon Oct 17, 2011 11:18 pm
Location: The Coast
by rurwin » Thu May 03, 2012 3:43 pm
I've just got a micro USB socket for my dummy load — so about the same as a micro USB plug to make up a cable.

The pins are a bit small and close together ;-)

Yes, those are millimetres. Fortunately we only need to connect the outer pins.

I've got three 'silly scopes. Now that is silly.
User avatar
Forum Moderator
Forum Moderator
Posts: 2949
Joined: Mon Jan 09, 2012 3:16 pm
by Raspberry wino » Thu May 03, 2012 4:03 pm
I've seen so many people have problems due to dodgy power supplies, using power supplies from some other equipment which don't match the markings and poor connections that I'd like to take a different approach.

I have a beefy 5V regulated PSU which can supply at least 3A. I'm dubious about the tiny micro USB connector so I want to know this: can I supply power to a Pi via the GPIO pins? Since the 5V is available to measure or supply other circuitry, I'd assume yes but can someone give a definitive answer please?
Posts: 49
Joined: Mon Mar 05, 2012 11:48 am
by rurwin » Thu May 03, 2012 4:53 pm
Yes you can, the sole proviso being that since it avoids the polyswitch fuse, you should  take steps to prevent over-current externally. ie you should insert a 1A fuse between the power-supply and the Raspberry Pi.
User avatar
Forum Moderator
Forum Moderator
Posts: 2949
Joined: Mon Jan 09, 2012 3:16 pm
by Raspberry wino » Thu May 03, 2012 5:48 pm
Good, thanks that makes sense. I can easily add a 1A fuse. Most likely though, my R-Pi will be part of a post 10k batch where RS/Farnell save a few pence per unit by leaving off the GPIO pins making connecting much harder for anyone not equipped or not confident with soldering.
Posts: 49
Joined: Mon Mar 05, 2012 11:48 am
by mahjongg » Thu May 03, 2012 5:52 pm
A (classical) 1 amp fuse will probably have too high an internal resistance, so you again introducing problems, use at least a 5Amp fuse, or a 1 Amp polyfuse with a low initial resistance. Low being a tiny fraction of an ohm, like 0.05 Ohm.

P.S. I am convinced all future R-PI will come with pinheaders
User avatar
Forum Moderator
Forum Moderator
Posts: 6016
Joined: Sun Mar 11, 2012 12:19 am
by jbeale » Thu May 03, 2012 6:00 pm
I had a look at a few micro-USB connectors, it seems the actual connector pins can support up to 1.8 amps. Somewhat surprising considering how small they are.
User avatar
Posts: 2121
Joined: Tue Nov 22, 2011 11:51 pm
by Raspberry wino » Thu May 03, 2012 7:26 pm
How much voltage can drop across a tiny length of admittedly very thin fusewire? If it"s more than 0.1 or 0.2 V I'd be very surprised. I can measure it under load and if it"s significant drop I can adjust the PSU output as there"s a trimmer. Alternatively, a polyfuse would be good if I can get one but I know nothing about them.

Replacing it with a 5A fuse would make it pointless!

I"m amazed such tiny connectors are rated for 1.8A and I suspect the high quality ones  meet the spec., probably not run-of-the-mill cheapo ones.

I really hope you're right and they don"t lose the very useful GPIO pins to save a few pence.
Posts: 49
Joined: Mon Mar 05, 2012 11:48 am
by Pencoys » Thu May 10, 2012 9:48 am
Raspberry wino said:


I've seen so many people have problems due to dodgy power supplies, using power supplies from some other equipment which don't match the markings and poor connections that I'd like to take a different approach.

I have a beefy 5V regulated PSU which can supply at least 3A. I'm dubious about the tiny micro USB connector so I want to know this: can I supply power to a Pi via the GPIO pins? Since the 5V is available to measure or supply other circuitry, I'd assume yes but can someone give a definitive answer please?



Not sure if thats a good idea. What is the PCB track capable of carrying from the GPIO ?.

I doubt it was ever intended as a power out from the GPIO so probably not a good idea to power in at that point. My take is that the +5v on the GPIO may be intended as a 'reference' rather than a supply pin. I have taken 'proper wire' to C5 -ve and the input to the poly fuse connecting with a standard 2.5mm barrel connector to a 5v 2A wart. works fine, but still no wireless networking. :-(
User avatar
Posts: 14
Joined: Fri Jan 13, 2012 8:48 am
by timinator13 » Sat May 26, 2012 9:34 pm
My Pi appears stable from a 2A tablet power supply, but the network won't stay connected. Could this be a supply issue?
Posts: 8
Joined: Mon May 21, 2012 10:09 pm
by timinator13 » Sun May 27, 2012 6:53 am
I replaced the wireless mouse with a wired one and it all works now.
Posts: 8
Joined: Mon May 21, 2012 10:09 pm
by lewmur » Sun May 27, 2012 2:03 pm
For those in the USA, the HP Touchpad store http://www.shopping.hp.com/en_US/home-office/-/search-SimpleOfferSearch?PageSize=15&search=&SearchTerm=touchpad&aoid=35252 has their wall wart on sale for $5 including shipping. This is a high quality unit well regulated to 2A.
Posts: 284
Joined: Sun Dec 25, 2011 3:20 pm
by jkirk » Thu Jun 28, 2012 4:00 am
I have an HP TouchPad, so I tried that power adapter. It's better than most, but even though it says it's 5.3v 2.0 amp output, I'm only getting about 4.73-4.77 during load. Without load it goes up to about 4.88 but that's about it. It seems to be doing OK...I had a USB mouse and keyboard, Ethernet and and HDMI cable all plugged in and I was able to get into X and had no problems.
So far, I'm just disappointed in the speed... load times are very slow and video is pretty slow. I assume getting a faster SD card would probably help the loading times a bit... but the overall sluggishness of the GUI is disappointing. Hopefully better drivers will come out soon.
Posts: 2
Joined: Thu Jun 28, 2012 3:55 am
by jbeale » Thu Jun 28, 2012 6:48 am
My Pi just came today, yay! Before touching anything I measured the notorious polyfuses and found:
F1: 3.45 ohms
F2: 2.67 ohms
F3: 0.28 ohms
(each is net resistance, not including the meter leads)

I started off with just a keyboard (DELL RT7D50, "max 75 mA")and an old portable TV with composite video. Booted up with no problems. I was measuring DC input current, which was 325 mA after boot. Voltage across F3 varied between 93.7 and 94.8 mV (confirming the fuse resistance was about 0.28 ohms) but I found TP1-TP2 voltage was only 4.4 V. If I'd been more ambitious and connected the ethernet, that might have been a problem. Apparently my nice long 6-foot microUSB cable was not helping! Using a shorter cable brought me up to 4.8 V, and the ethernet did work OK when I subsequently tried it.
first-boot.jpg
first boot of the new R-Pi
first-boot.jpg (53.57 KiB) Viewed 10610 times
User avatar
Posts: 2121
Joined: Tue Nov 22, 2011 11:51 pm
by rurwin » Thu Jun 28, 2012 8:03 am
jbeale wrote:My Pi just came today, yay! Before touching anything I measured the notorious polyfuses and found:
F1: 3.45 ohms
F2: 2.67 ohms
F3: 0.28 ohms
Those are fantastic numbers, less than half of what I have and what others have posted (and like you, I tested them before anything else.)

I wonder if new fuses are being shipped or if it's just that you have some good components.
User avatar
Forum Moderator
Forum Moderator
Posts: 2949
Joined: Mon Jan 09, 2012 3:16 pm
by Jim Manley » Thu Jun 28, 2012 9:03 am
jbeale wrote:My Pi just came today, yay! Before touching anything I measured the notorious polyfuses and found:
F1: 3.45 ohms
F2: 2.67 ohms
F3: 0.28 ohms

Welcome to The Pi Club for Men, where we're not just the owners and executives, we're customers and users! :D

Those polyfuse numbers are nice. I should measure mine when I rescue my Fluke meter from another more pressing task (read income-generating, vs. the Pi, which so far is nothing but income-consuming :) ).
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: 1358
Joined: Thu Feb 23, 2012 8:41 pm
Location: SillyCon Valley, California, USA
by jbeale » Thu Jun 28, 2012 1:20 pm
rurwin wrote:
jbeale wrote:My Pi just came today, yay! Before touching anything I measured the notorious polyfuses and found:
F1: 3.45 ohms
F2: 2.67 ohms
F3: 0.28 ohms
Those are fantastic numbers, less than half of what I have and what others have posted (and like you, I tested them before anything else.)

I wonder if new fuses are being shipped or if it's just that you have some good components.

Hmm- and even with that, I started at 4.4 V on the board, due to my long USB cable! I had even gone to the trouble of pre-qualifying my 5V USB wall-wart supply with a hacked USB cable wired to a 5 ohm resistor load, to confirm my supply could deliver 0.95 A at 4.9 V. No wonder people have trouble with power.
User avatar
Posts: 2121
Joined: Tue Nov 22, 2011 11:51 pm
by Salamander » Tue Jul 03, 2012 11:46 am
rurwin wrote:I wonder if new fuses are being shipped or if it's just that you have some good components.

Good question. Maybe different components for different batches (first 10K batch, RS production batch and Element14 production batch). Another variable to take into account :twisted:
User avatar
Posts: 49
Joined: Fri Mar 30, 2012 1:41 pm