37 Replies Latest reply on Jul 4, 2012 10:25 AM by jamodio

    It works!

    John Beetem

      I got my RasPi today!  It works pretty well, and I did not have much trouble bringing it up.  Since I've been watching and commenting here at element14 and in the raspberrypi.org Troubleshooting section for some time, let's see how my RasPi compares to her sisters:

       

      1.  Before plugging my RasPi in, I checked the resistance of the polyfuses to see if they will deteriorate over time.  I don't have a proper low resistance meter which nulls out lead resistance, so these are approximate.  The main power fuse F3 was less than 0.2 Ohm.  The two USB fuses F1 and F2 were 2.0 - 2.2 Ohm.  I don't care about them, because I'll be using a powered hub.

       

      2.  I'm using a powered USB hub to provide power to RasPi.  My power supply is based on a Linear Tech demo board and gives very reliable 5V which powers my USB hub.  The resistance from power supply GND to RasPi TP2 is less than 0.1 Ohm.  However, the resistance from power supply +5V to RasPi TP1 seem to be at least 0.5 Ohm.  My micro USB cable obviously has better GND conduction (probably a foil shield) than +5V conduction (1 meter of 28 AWG).  I think the +5V resistance is pretty well shared by the micro USB cable and F3.

       

      3.  The procedure described at raspberrypi.org and the RasPi wiki for downloading Debian and copying the image to an SD card (GNU/Linux command line) worked perfectly.  It took a while to download 443 MB from the mirror and to copy the uncompressed 2 GB to a Class 4 SanDisk microSD card, but there were no problems.

       

      4.  RasPi booted the first time (wow!), in spite of some power supply problems.  My power supply produces +5V from whatever 9V-12V wall wart I have handy.  The first one I used did not provide enough current for RasPi (it works quite well for BeagleBoard thank you very much) and this caused my monitor to be unstable, occasionally blanking out and showing green "static".  I switched to a more powerful wall wart and RasPi booted perfectly with no display problems.

       

      5.  My particular RasPi seems to work with surprisingly low voltage.  When I get the login prompt my TP1-TP2 voltage is just 4.65V.  When I start up X Windows it drops to 4.60V.  I guess my peripherals don't need anything like the USB 4.75V minimum.  I'm using a NEC DVI-D monitor which didn't require any config.txt options, along with Logitech wired keyboard and mouse.  I'm guessing that these are all 3.3V devices and having 4.60V HDMI and USB voltages doesn't faze them.  The monitor is actually connected through a cheap "HDMI Switcher".

       

      6.  I tried Midori and connected to raspberrypi.org to announce success.  Ethernet came up and automatically congfigured to my router just fine.  However, Midori did not run well -- normally consuming all compute time -- so I gave up.  I'm going to see if assigning the Level 2 cache to the ARM CPU helps.  GUI-based terminals and text editor were responsive and I had no problem writing and compiling "Hello, World" except that I need to set my keyboard to USA if I want to type any punctuation.

       

      7.  My RasPi gets quite warm.  I wouldn't say blazing hot -- I can leave my finger tip on the SoC and LAN chip for several seconds.  The back of the board gets very warm.  I'm thinking of adding heat sinks and a thermal pad, but first I'm going to try mounting RasPi vertically so that air can flow past both sides.  I'm concerned that the high temperature will cause F3 resistance to increase, lowering my 4.65V down to something that makes the board fail.  This might be the cause of some of the "RasPi works for a while and then stops" posts we've seen in the Troubleshooting section.  I'm planning to put a low-resistance non-resettable fuse in parallel with F3 to prevent F3 problems.  This should get my 4.65V up to 4.8V, which is plenty.  If 5V0 gets too high, RG2 has to work harder and that makes the board hotter.  So I think keeping my present USB cable may actually help things.

       

      8.  Yay, Debian already includes libXft.so!  That's the only unusual library I need to port my software.

        • Re: It works!
          morgaine

          Gratz John!

           

          I'm eagerly reading your experiences, the more fine details of voltages etc the better.  That has been somewhat missing from startup reviews thus far, so I'm very glad to see it.  And measuring the polyfuse resistances before applying power was an excellent idea.

           

          Still waiting for mine ...

            • Re: It works!

              John,

                There seems to be a surprising amount of variation between what people

              are reporting.   You seem to have relatively low resistance F1 & F2 compared

              to 5 to 8 Ohms reported elsewhere.  Your temperatures are consistent with

              others who have said 5 seconds is all they can touch the SoC & Lan chips,

              but then other people report closer to body temperature.   Your 4.6V between

              TP1 & TP2 seems lower than some people report, and I think will cause

              flakiness with various power supplies (as you have already seen), and SD cards,

              and USB devices.  Some people report that 4.8V is not enough.

               

              I wonder what's causing such large variations?

                • Re: It works!
                  morgaine

                  Someone with a hot Pi must have an infrared imager too ...

                  • Re: It works!
                    Roger Wolff

                    coder,

                    the USB cables. I've ordered a set of micro USB cables from farnell and they claim 30 AWG in the datasheet. John says 28AWG. Already a lot better!

                     

                    Calibration of cheap multimeters is probably a bit off. People reporting 4.7 might have 4.8 in reality.

                    Cheap multimeters will show larger voltages when their battery runs out. I've made the mistake: "It says low battery, but it works fine. I HAVE the battery for the swapout, so I'll change it when the meter really stops".

                     

                    I have once measured the ability of my thumb to transport away about 1W of heat. So that's my rule-of-thumb nowadays (pardon the pun. :-) ).

                     

                    My 'pi uses 440 mA alltogether or 2.2W. The power is split between the SOC, the lan chip and the regulators. This means there is less than one Watt in each of them. I can keep my finger (not even my thumb) on each of them without any problems.

                     

                    Someone with an IR gun has measured his chips to be 100f, and the hottest one 120f. That is basically about 40-50 degrees centigrade, right? Up to 52 degrees centigrade you do not feel "auch! that's hot". (that temperature is surprisingly constant between people).

                     

                    The SD card should work downto about 4.4V as the TP1 TP2 voltage: The SD card does not have any connection to the 5V rail. It works off 3.3V, and the 3.3V regulator will has an advertized dropout of less than 1.0V. So 4.4V should be enough.


                    The BCM2835 calls the 5V input the V_BAT. This means that the BCM2835 is probably designed to work with 3.0-4.2V there. But it will tolerate the 5V of the rpi design there. No problem. Again, there should be no problem with less than 5V on the 5V rail of the 'pi. The 1.8V regulator has plenty of headroom. And the 2.5V regulator has very little current draw. So no problems there.

                     

                    That leaves us with the peripherals on the USB ports with too little power. I have a PL2303 serial adapter that is unreliable if I  power my 'pi through an USB cable. It works fine when I power it through a 12V -> 5V DCDC converter and a 30cm cable connected to a bare USB plug I bought.

                      • Re: It works!

                        100F=38C, 120F=49C

                        It sounds like some people are seeing around 40C, and some around 50C,

                        for reasons yet to be explained.  John's Midori seems unusually slow, so

                        there is possibly some unusual bus traffic or interrupt activity going on that

                        might also explain the high temperature.

                          • Re: It works!
                            GreenYamo

                            Congrats on the new arrival !

                             

                            I don't have an IR sensor, but my mostly used Pi, part of the 'original 8,000' gets pretty hot. I have one from a later order which doesn't seem to get quite as warm, but to be fair I probably haven't put it under as much load yet. Perhaps I will rectify that at the weekend.

                             

                            I'll also dig my multimeter out and see what figures my two report.

                             

                            Midori on the original Pi is usable, just.

                              • Re: It works!
                                morgaine

                                Midori's problem is simply that there is no way to disable Javascript in this browser on a per-source basis.as one can do with the NoScript add-on in Firefox.  Because of that,  Midori's advantage in theory doesn't result in an advantage in practice.

                                 

                                It's much slimmer than Firefox and potentially far less demanding of resources, but today's web is at the mercy of clueless website developers who infest every page with wholly inappropriate scripting that reduces even the slimmest of browsers to a crawl.  Webbies should be locked up in a padded cubicle and not allowed access to anything sharp with which they can hurt themselves, or us.  And Javascript is sharp.

                                 

                                Disable scripts in Edit->Preferences->Behavior and Midori suddenly becomes virtually instantaneous on x86 machinery, regardless of website.  Unfortunately then half the web doesn't work.

                                 

                                Next time you see a webbie, give them a good kicking.  It provides useful stress relief, helps redress the balance of the universe, and while they're nursing their wounds then they aren't writing more Javascript.

                                 

                                  • Re: It works!
                                    John Beetem

                                    Morgaine Dinova wrote:

                                     

                                    Midori's problem is simply that there is no way to disable Javascript in this browser on a per-source basis.as one can do with the NoScript add-on in Firefox.  Because of that,  Midori's advantage in theory doesn't result in an advantage in practice.

                                     

                                    It's much slimmer than Firefox and potentially far less demanding of resources, but today's web is at the mercy of clueless website developers who infest every page with wholly inappropriate scripting that reduces even the slimmest of browsers to a crawl.  Webbies should be locked up in a padded cubicle and not allowed access to anything sharp with which they can hurt themselves, or us.  And Javascript is sharp.

                                     

                                    Disable scripts in Edit->Preferences->Behavior and Midori suddenly becomes virtually instantaneous on x86 machinery, regardless of website.  Unfortunately then half the web doesn't work.

                                     

                                    Next time you see a webbie, give them a good kicking.  It provides useful stress relief, helps redress the balance of the universe, and while they're nursing their wounds then they aren't writing more Javascript.

                                     

                                    Thanks for the advice!  Sounds like a great way to filter out useless cobwebs, like not having Flash available.  I think the biggest problem with web sites is that companies give developers fast computers and fast connections so they can't see what the web sites really do in the wild.  They should give them 80386SX machines with 56K dial-up.  Then then web sites will behave much better :-)

                                      • Re: It works!
                                        morgaine

                                        John Beetem wrote:

                                         

                                        They should give them 80386SX machines with 56K dial-up.

                                         

                                        +1

                                         

                                        Although I turn it into comedy to make light of it, unfortunately it's factual.  Webbies have partially destroyed the Web through a combination of inadequate education and poorly conceived web architecture.

                                         

                                        Education is relevant because writing web content does not require engineering knowledge and engineering sense, whereas programming does.  And the architecture of web scripting is such that poor Javascript programmers have no negative feedback because their programming uses resources on other peoples' machines instead of on their own.  That would invite trouble even if the programmers were experts.

                                         

                                        Put those two things together, uneducated programmers and no negative feedback, and the result is a web that spirals towards unusability through resource misuse and exhaustion.  It's a bad combination.

                                         

                                        Morgaine.

                                          • Re: It works!

                                            It would be nice to figure out if John's subjective sense that Midori

                                            is way too slow, and Steve's sense that Midori is "usable, just", are

                                            different ways of describing the same behavior, or if there's really

                                            a difference, possibly involving the SD card, or firmware version, etc.

                                             

                                            There is an easy to run javascript benchmark at

                                             

                                            http://www.webkit.org/perf/sunspider/sunspider.html

                                             

                                            nothing to download, just click on the link:

                                               

                                            "Start SunSpider 0.9.1 now!" and wait for the results.

                                             

                                            Or maybe by viewing some particular page without javascript,

                                            you can measure how long it takes to load.

                            • Re: It works!
                              John Beetem

                              John Beetem wrote:

                               

                              7.  My RasPi gets quite warm.  I wouldn't say blazing hot -- I can leave my finger tip on the SoC and LAN chip for several seconds.  The back of the board gets very warm.  I'm thinking of adding heat sinks and a thermal pad, but first I'm going to try mounting RasPi vertically so that air can flow past both sides.

                              Vertical mouting works very nicely -- the back side of the board is a good radiator and vertical mounting lets natural convection work effectively on both sides of the board.  I can now leave finger on SoC and LAN9512 (no Ethernet traffic) for at least 10 seconds.  Heat distributes all over the board -- the board has very few vias so the power/ground layers conduct heat very well.

                               

                              Now my worst problem is an iffy 3.5 mm barrel connection to my powered USB hub.  Since my USB hub is providing power to RasPi over a micro USB cable, any extra contact resistance at the barrel connector drops RasPi voltage and creates HDMI "static", or causes HDMI to drop off.  Even when I have a reasonably good connection the "static" occurs when the CPU to doing actual work, e.g., compiling some C programs.  I've ordered some parts and I'm going to make my own micro USB power cable.

                              • Re: It works!
                                John Beetem

                                John Beetem wrote:

                                 

                                8.  Yay, Debian already includes libXft.so!  That's the only unusual library I need to port my software.

                                Well, Debian includes the executable for Xft, but not the development files (i.e., *.h).  "apt-get install libxft-dev" didn't work the first time.  It took a bit to figure out I needed to run "apt-get update".  Once I did, "install libxft-dev" worked fine and my software compiled and came up perfectly (well, except for some nits having to do with different fonts).

                                 

                                All-in-all, I'm quite happy with RasPi's performance when running my software.  I don't need a lot of memory or graphics performance, and I don't have a lot of disk traffic.  So RasPi's limitations don't hurt my software.  RasPi seems slower than my 600 MHz BeagleBoard, but I really need to time it to be sure.

                                 

                                Midori still runs very slow.  I tried turning off scripts, but it didn't seem to make much difference.  It just takes a horrific amount of time to scroll the screen (raspberrypi.org and element14.com).  I couldn't even log in at raspberrypi.org.  element14.com login succeeded, but I couldn't write this message using my RasPi.  OTOH, the Google Code page where my software sleeps and dreams loaded quite quickly.  I suspect the problem with the first two is that they try to write myriad files to disk (i.e., a class 4 SD card) and RasPi balks.

                                 

                                With Ethernet disconnected, my vertically-mounted RasPi runs at a reasonable temperature.  With Ethernet active things get quite warm, especially doing Midori.  (The HDMI shell makes a nice radiator.)  I think I'd be better off with a Model A, with an Ethernet dongle to spread out the heat.  OTOH, it's easy to unplug Ethernet when it's not needed.

                                 

                                So all-in-all, as long as I don't try to use Midori and heavy use of Ethernet, RasPi is quite an endearing little machine with wonderful capabilities for a gadget that small and cheap.

                                  • Re: It works!
                                    morgaine

                                    I suspect that I'll be tripping over the need to install the "-dev" packages repeatedly after so many years of running Gentoo, which has no separate "-dev" packages since it's a source-based distro and so every package is a development package.

                                     

                                    But running Gentoo on a Pi will be oh ... so ... painful during package installation and updates, given the slow CPU.  I'll try it anyway at some point, but it sounds like too much suffering so it's best left for Pi v3.0 which will use Cortex-A15.

                                     

                                    That said, I could of course cross-compile Gentoo on an x86 machine for Pi as target ... very appealing indeed.

                                    • Re: It works!
                                      Roger Wolff

                                      John, I'm wondering about your temperature "problems".

                                       

                                      My 'pi has been on for 7 days now. It's had time to warm up. When I put my finger on the SOC, it is just over the "comfortably warm" temperature. Maybe indeed I'd like to pull it away after a few seconds. But if I then put a different finger on there, the temperature has sufficiently dropped that I can keep it there.

                                       

                                      This is consistent with my measurement/guess that my fingers have a heat-carrying-capacity of about 1W (each) and the fact that my raspberry pi consumes 2.2W and divides that almost equally between the SOC, the LAN and the 3.3V regulator.

                                       

                                      Can you measure the powerconsumption of your 'pi? If there are large differences there is something wrong with the design. (We already know of one place that could lead to excessive power consumption, however I would have expected that in 99.99999 % of the cases this wouldn't matter. (as in not likely one of the few that are in the field today))

                                    • Re: It works!

                                      John,

                                       

                                      you wrote:

                                      "6.  I tried Midori and connected to raspberrypi.org to announce success.  Ethernet came up and automatically congfigured to my router just fine.  However, Midori did not run well -- normally consuming all compute time -- so I gave up."

                                       

                                      You might try this suggestion:

                                         set vm.min_free_kbytes =16384 in /etc/sysctl.conf

                                      http://www.raspberrypi.org/phpBB3/viewtopic.php?f=66&t=8438&start=33

                                      • Re: It works!
                                        morgaine

                                        Yay, I received my Pi at last!

                                         

                                        Following John's good precedent, I measured my polyfuse resistances to be:

                                        F1 = 4.0 ohms

                                        F2 = 3.2 ohms

                                        F3 = 0.3 ohms (but maybe close to 0.2, occasional flicker), LS digit questionable

                                         

                                        These values are a lot higher than John's.  I used two different DVMs, one of them a costly Fluke (although not calibrated), but neither has probe lead resistance nulling (I subtracted the 0.2 and 0.3 ohms indicated when the probes are shorted).

                                         

                                        Allowing for the lead offsets, the two DVMs track each other perfectly across this range, so I have confidence in the results.  I don't regard the F3 reading as reliable, since it's on the least significant digit displayed.  It's too high for comfort though (0.5 and 0.6 ohm readings on the DVMs) --- in effect the board's +5V is unregulated because of it.

                                         

                                        F1 and F2 are outrageously high of course, and do more harm than good by deregulating the supply even further.  This is a plain and simple design fault.  Also, the fact that F1 and F2 have so greatly differing resistances is a worry --- clearly this component does not have a tightly defined resistance parameter, even worse when compared with John's.

                                         

                                        I haven't powered the board up yet, as I must be the only person on the planet without a micro-USB charger.

                                         

                                        I'll pop out and buy one today, or an adapter for a barrel connector or something.

                                         

                                        Morgaine.

                                          • Re: It works!
                                            Roger Wolff

                                            Haha! so much time on the forums and then didn't prepare for the coming of the 'pi.


                                            There are way to many people who say things like: this won't work. just from a theoretical viewpoint.

                                             

                                            0.3 ohms at max 0.7A will give you 0.2V maximum drop.

                                             

                                            With a proper regulated 5.0V powersupply, that will give you 4.8 to 5.0V on the line labeled "+5V" on the raspberry pi.

                                             

                                            Why the F*** do you think that won't work? The 5V is connected to the 3.3V LDO. That will work downto 4.3V AT LEAST (i.e. at maximum load on the regulator, but the 'pi is using less). The 5V is connected to the line labeled "VBAT" on the BCM2835. That leads me to suspect that this is intended to work on a single cell LIPO. So it should work downto about 3.0V, but AT LEAST to 3.3V.

                                             

                                            Next it goes to the USB ports. If you read the USB specs, devices are required to work down to somewhere in the low 4V. Not all do, but they are not USB compliant. They migth work on lots of other USB ports that do deliver a higher voltage, but the remain out of spec.

                                             

                                            So with a worst-case 4.8V (4.9 in practise) at the top of F1 and F2, and a 100mA max powerdraw, we'd loose another 0.4V with the 4 ohms on those fuses. That's indeed close to spec.

                                             

                                            If an USB device doesn't work, you have several options.

                                            * Get a different device that does work.

                                            * Get a powered hub in between.

                                            * short out F1 and/or F2.

                                            * if that doesn't help enough, short out F3 as well.

                                             

                                            When you start shorting out those fuses, you'll be relying on the current limiting on the power supply. No big deal. Connect it to a PC powersupply and start shorting out things on the GPIO port and you'll be able to start a fire. Don't do that then!

                                              • Re: It works!
                                                morgaine

                                                Roger Wolff wrote:

                                                 

                                                With a proper regulated 5.0V powersupply, that will give you 4.8 to 5.0V on the line labeled "+5V" on the raspberry pi.

                                                 

                                                And what exactly does a "proper regulated 5.0V powersupply" have to do with a USB cellphone charger?  Virtually nothing in practice.  It's no use talking theory, or saying there's a proper spec for USB chargers, when massive experience in the Pi community has shown that cellphone chargers are very commonly unadulterated junk, and very frequently you have to try several before you find one that works at all even when they claim adequate amperage.

                                                 

                                                It's wrong to think of it as a "power supply", which conjures up ideas of stability, regulation, protection, and even some reasonable headroom.  It's just a charger, and it's made to the lowest possible spec and with the worst quality materials they can get away with.  The cellphone market is totally cutthroat and cellphones are given away free with phone contracts, so it doesn't surprise me in the slightest that the situation is as bad as it is.

                                                 

                                                The end result is not a supply voltage you can rely on, and making the situation worse with extra drop across a 0.3 ohm polyfuse is not clever when the Pi is specifically targetted at using cellphone chargers as its power source owing to the micro-USB connector.

                                                 

                                                Why the F*** do you think that won't work?

                                                 

                                                FUSE?

                                                 

                                                To answer seriously though, it will work in many cases, but it's marginal ,and it's poor design because it causes substantial loss of regulation.

                                                 

                                                The worst part of this though is the coupling together of distinct power paths, totally unnecessarily.  Nothing you connect to the Pi can expect a stable voltage source even if you provide one at the micro-USB input, because F3 deregulates it for all sinks on the board, including the two USBs and the P1 header.  Under no circumstances can the Pi's power design be considered satisfactory when it turns a regulated supply into a highly unregulated one.

                                                 

                                                Just because the nominal +5V for the two USBs might stay within USB spec (if you're lucky), this doesn't mean that it's good power design for something that is powered by P1-pin2 to experience a sizeable voltage drop whenever USB usage changes.  There's nothing wrong with using cheap polyfuses, but the power paths should not be coupled in this way.  F3 should control the SoC alone, F1 and F2 should tap off the power input connector, and P1-pin2 should have an independent polyfuse also tapping directly off the input connector.  F3 as it stands is a misdesign, and all the trip values are too low.

                                                 

                                                Morgaine.

                                                  • Re: It works!
                                                    Roger Wolff

                                                    Morgaine Dinova wrote:

                                                     

                                                    And what exactly does a "proper regulated 5.0V powersupply" have to do with a USB cellphone charger?  Virtually nothing in practice.  It's no use talking theory, or saying there's a proper spec for USB chargers, when massive experience in the Pi community has shown that cellphone chargers are very commonly unadulterated junk, and very frequently you have to try several before you find one that works at all even when they claim adequate amperage.

                                                    Nobody has convinced me of such a thing. Before the first raspberry pi's were being delivered, I started making noises that even though the power supplies might be up to spec, most cables are not.


                                                    We have measured whole ohms in cables we had lying around.

                                                     

                                                    Of course, people have reported that when they got a better USB-charger the pi suddenly started working. But with the high quality charger, they also got a better cable. So it's quite possible they used the new cable as well. And of course they stopped testing the moment the 'pi started working. So it's the new usb charger that did the job.

                                                     

                                                    Pete Lomas who, as far as I know, worked on both the schematics and the layout of the 'pi. Has his habits. He's overly cautios about overvoltages, decoupling capacitors, and fuses. So there are a bunch of "extra" capacitors on the board that wouldn't have been necessary. There are a bunch of clamping diodes that 99% of the users will never use. And there are a bunch of fuses that some (including me) think are redundant.

                                                     

                                                    And a 0.14V voltage drop in practice is the consequence of the fuse. So what?

                                                • Re: It works!
                                                  morgaine

                                                  Followup to measuring the polyfuses and shopping for a few means of applying power to the Pi ...

                                                   

                                                  Short version:  everything worked first time, but the Pi is piping hot! 

                                                   

                                                  Longer version:  Maplins came up trumps with lots of options for micro-USB power.  This one ---http://www.maplin.co.uk/micro-usb-power-supply-393067  --- even says "Can be used with the Raspberry Pi system" on the product page, lol.  Everyone is getting in on the Pi bandwagon.

                                                   

                                                  That 1A power supply (which does seem to deserve the name, rather than merely "charger") supplies my Pi with 5.01V at the input to polyfuse F3, and I see 4.89V at TP1, when running a basic system load comprised of a Logitech universal receiver which runs totally cold, Ethernet, and HDMI into the HDMI input on a Dell 2408 monitor at 1920x1200.

                                                   

                                                  Pulling out the Ethernet cable raised TP1 by 30mV without any significant change at the input.to F3, so this supply is regulating and its lead seems to be very good.  Pulling out the Logitech receiver raises TP1 by only 10mV.  Pulling out the HDMI lead has no effect on TP1.  My peripherals are taking very little current.

                                                   

                                                  Under these condition, the finger test applied to the SoC/PoP (which BTW is a Samsung) has to be removed in under 10 seconds to avoid pain.  Applied to the LAN chip, the finger has to be removed almost instantly --- this device is running very hot indeed, both with and without the LAN cable.

                                                   

                                                  Nothing else to report really.  I don't know Debian specifically but experience from many other distros made everything easy.  The nearest thing to a hiccup was that "apt-get install gpm" failed to configure the mouse, so I had to modify /etc/gpm.conf slightly before it would work (changed the mouse device to /dev/input/mice).

                                                   

                                                  As an aside, I also bought http://www.maplin.co.uk/high-power-usb-charger-with-travel-adaptors-513509 on the expectation that its 2.1 A rating would quite likely offer better regulation, but that's not what I observed:  it delivers 4.75V at TP1 under the same load described above.  The blame might lie with the lead rather than the charger (haven't checked yet), but that's not what appearances would suggest as the lead is nearly twice as thick as the one on the better supply.  Of course, the extra thickness could be all plastic.

                                                   

                                                  So, a successful start with my Pi, but running the LAN9512 that hot is a bit worrying.

                                                   

                                                  Morgaine.

                                                    • Re: It works!
                                                      John Beetem

                                                      Morgaine Dinova wrote:

                                                       

                                                      Under these condition, the finger test applied to the SoC/PoP (which BTW is a Samsung) has to be removed in under 10 seconds to avoid pain.  Applied to the LAN chip, the finger has to be removed almost instantly --- this device is running very hot indeed, both with and without the LAN cable.

                                                       

                                                      So, a successful start with my Pi, but running the LAN9512 that hot is a bit worrying.

                                                       

                                                      Morgaine.

                                                      Thank you for the write-up.  It's interesting that the RasPi forum doesn't seem to have a place to write up success, only problems :-)

                                                       

                                                      It would be interesting to find out the LAN9512 temperature range seen in the field.  Some people have reported blazing hot as you have, others something more reasonable.  My fear regarding the LAN9512 is that its thermal performance may depend heavily on how well its large ground pad is soldered, which is impossible to see without an X-ray (if then).  How hot is the back side of your board?  When I first picked up my RasPi after running her the first time, the back side (which was sitting on the anti-static bag) was very hot. That's when I decided to get out some tools and make my vertical mount frame.  Radiating from both sides keeps her very comfy.

                                                      • Re: It works!
                                                        Colin Barnard

                                                        My Pi was one of the first released batch and my LAN chip gets warm but not hot, which makes me wonder if there has been any component changes or if some manufacturing defect such as partial shorts or poor soldering have it the later batches.

                                                          • Re: It works!

                                                            There have been some apparently minor differences in the markings on

                                                            the LAN chip, but nothing correlated with temperature. 

                                                             

                                                            The thread "Ethernet chip gets raging hot" started on May 15,

                                                            http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=5898

                                                            so this isn't a particularly recent issue.

                                                             

                                                            As far as I know, the only official comment was from Gert in the

                                                            above thread, who wrote:

                                                             

                                                            "It might be warm, it should never be hot."

                                                             

                                                            and later wrote:

                                                             

                                                            "I can imagine that you do not want to give up your Pi now. But when a few more are available I would really send it back so somebody can find out why it gets so hot."

                                                             

                                                            So maybe you will be the first person to take them up on

                                                            this offer, now that we know replacements are readily

                                                            available.  They really ought to publish what they consider

                                                            to be the normal temperature range, as well as what they

                                                            consider to be the normal range for polyfuse resistance.

                                                              • Re: It works!
                                                                morgaine

                                                                @coder27: "Raging hot" describes the temperature of my LAN9512 perfectly too, so if Gert says that it should not be so, it's faulty.

                                                                 

                                                                @Roger: I did your suggested finger-switching to drain the thermal capacity of the chip surface, and you're right, it has only a small thermal reservoir.  Using up to 4 fingers in sequence, the 1st has to be removed as fast as human reactions allow (it's not charring, but seriously beyond cell survival temperature), the 2nd maybe 1/5th second, the 3rd a whole second, and the 4th up to 5 seconds.  I can't extend "Finger Contact Survival Time" any longer than 5 seconds no matter how many times I switch, I'm not into pain.

                                                                 

                                                                In the absence of a thermometer, I must say I like the FCST concept.

                                                                 

                                                                The chip might survive this operating temperature, but I doubt that the board will last long like this because of the physical stresses of expansion/contraction whenever the board is powered up or down.

                                                                 

                                                                Talking to RPF about this is a dead loss as they don't admit any problems (I can't anyway, having been banned for talking politely to Liz), but I'd like to know what Element 14 / Farnell's position is on this.  Should we try to extend board life by heatsinking the LAN9512 chip, or return it as faulty if Gert says that such high temperature is not expected?

                                                                 

                                                                Morgaine.

                                                                  • Re: It works!
                                                                    Roger Wolff

                                                                    Morgaine, I just did the "finger switching" test again as I suggested. :-)

                                                                     

                                                                    It seems my LAN chip is a bit hotter than I remember. I too can't get much beyond 5sec wtihout hurting my finger. It could also be that the outside temperature is now 25 degrees, while before it may only have been 20 degrees.

                                                                     

                                                                    The "can't keep finger on" temperature is pretty much 52 degrees C. No matter who you are or if you say you have fire resistant fingers.

                                                                     

                                                                    If the outside temperature rises, more heat will need to flow through your finger (less is being diverted by the PCB), and the hotter it gets. Even a 2 degree change might mean two out of two people now report unable to hold beyond 5 sec, while before it was "possible to hold indefinitively"

                                                                     

                                                                    If you look at the datasheet, on page 41, it says you can expect a typical power use of about 763 mW.

                                                                    http://www.smsc.com/media/Downloads_Public/Data_Sheets/9512.pdf

                                                                    I don't pretend to be able to estimate the power-use within  a factor of two, so when I say "about 1W", this is certainly within range.

                                                                     

                                                                    Regretably the thermal resistance isn't specified. The Texas instruments VCA8500 provides a "typical" value (i.e. a wild guess that it might apply to the lan9512), of 22 degrees / W. At 0.7W, that would mean only about 40 degrees C. We're seeing more than that.

                                                                     

                                                                    But if the wild guess is off-by-a-factor-of-two, and it's 40 degrees/C for teh lan9512, then 60 degrees sounds like a possible value.

                                                                     

                                                                    With the case at around sixty degrees, the inside will be only about 80 degrees. This is well below the maximum of 150 degrees.

                                                                     

                                                                    Chips are built to work at these temperatures. Sure, I too have heard about some chips working themselves loose from thermal stresses. Before you complain to Farnell, the RPF or me, please wait until you KNOW it will fail because it HAS failed.

                                                                     

                                                                    Some chips use a bit of power, and therefore they dissipate heat. That heat needs to be carried away and a temperature gradient is required before heat starts moving. This is simply the way things work.

                                                                     

                                                                    Now usually these things are hidden inside a case. So you don't get to poke your fingers at it. The 'pi is open, As a learining experience YOU get to poke your fingers at it. Have you learned something today? That's the stated primary goal for the RPF.

                                                                      • Re: It works!
                                                                        morgaine

                                                                        Roger Wolff wrote:

                                                                         

                                                                        Chips are built to work at these temperatures. Sure, I too have heard about some chips working themselves loose from thermal stresses. Before you complain to Farnell, the RPF or me, please wait until you KNOW it will fail because it HAS failed.

                                                                         

                                                                        This is a technical site.  Questions and discussion are welcomed by Element 14, on both technical and product matters, and your suggested "wait until you KNOW it will fail" is not really helpful advice at all in an engineering community.

                                                                         

                                                                        We can do a lot better than that, understand what is going on, and determine whether the board is being operated outside of its thermal spec (it's not device survival that's the worry, but the board/joints).  According to Gert, such LAN9512 temperatures were not expected at design time.

                                                                         

                                                                        The board isn't failing yet, but if it wasn't designed for these thermal stresses then it won't survive them for long.  Element 14's input really ought to be sought regarding what we should do.  I  don't want my board failing on the day after warranty expires when the solder joints finally decide that they've suffered thermal movement long enough, and it's not in the manufacturer's interests for that to happen either.

                                                                         

                                                                        And once they're in stock and I want to buy several Pi, I need to know that this common operating condition is within board spec before ordering.  (Model A is looking very attractive, the absence of LAN9512 making it more so.)

                                                                         

                                                                        Morgaine.

                                                                            • Re: It works!
                                                                              morgaine

                                                                              Adding a heatsink is certainly an option, but if it's required for board longevity then the heatsink should be part of the board design.

                                                                               

                                                                              I have no shortage of small heatsinks here, but at this stage it's more a matter of establishing the principle with Element 14.  If they accept that it may be a good idea to add your own heatsink pending review of the situation with board designers, then they may want to advise accordingly since it could affect product warranty.

                                                                                • Re: It works!
                                                                                  jamodio

                                                                                  Agreed.

                                                                                   

                                                                                  Also if by any chance, as reported before, due the wrong connection of the LAN9512 VDDCORE pin to the 1V8 rail the GPU is drawing some current from the internal regulator on that chip, it will surelly get hotter than normal.

                                                                                   

                                                                                  Ethernet PHY's tend to run warm to hot, but this one seems to be slightly above of what I've seen before.

                                                                                   

                                                                                  -J

                                                                                  • Re: It works!
                                                                                    Roger Wolff

                                                                                    Really, the chip is using a significant portion of a watt. On a 9mm x 9mm chip that's going to make it feel hot to the touch.

                                                                                    Once I have my "spare" pi, i'll see if I can find that 1.8V line and cut it.....

                                                                                     

                                                                                    All measurements so far have indicated a significant powerdraw on the 3.3V line, and that mostly goes to the lan9512. And from there, it can't go anywhere but turn into heat.

                                                                                     

                                                                                    There is a post on hackaday.com with someone who has added a heatsink to his 'pi.

                                                                                      • Re: It works!
                                                                                        jamodio

                                                                                        That would be something hard to do given that it is two not contiguous pins that probably go to the power plane and you still need to keep the connection to the decoupling capacitors required for the LAN9512.

                                                                                         

                                                                                        Unfortunatelly (once again the veil of secrecy) the Gerbers for the board layout are not publically available so it is difficult to tell by just looking at the two pictures of the boards without components how those traces are routed, that btw as far as I can tell are improperly routed since the trace is supposed to go straight from the pin to the capacitor.

                                                                                         

                                                                                        I've seen the hackaday post, which motivated me to search for tiny heatsinks to try since I don't want to use the saw :-)

                                                                                         

                                                                                        So far in my meter the Rpi shows 380-440mA draw now with the keyboard and tiny mouse connected and a NTSC TV monitor on composite video.

                                                                                         

                                                                                        -J

                                                                        • Re: It works!
                                                                          Roger Wolff

                                                                          Measuring the current into your 'pi is a bit more difficult than, measuring the voltage across TP1,TP2.

                                                                           

                                                                          But I'm guessing you're seeing the same as me, judging from the voltages you measure across F3. We're using the "has too much resistance" fuse as a current sense resistor!

                                                                           

                                                                          My 'pi has 140mV across F3. Yours has only 120mv. Either your polyfuse is a bit better, or your 'pi is using slightly less power. But in the grand scheme of things: Your pi is using the same amount of power as mine.

                                                                           

                                                                          All power used gets converted into heat. Your pi is not producing more heat than mine.

                                                                           

                                                                          The only think I can think of that would make the lan9512 hotter than it should be is if your 3.3V is WAY too high. We measured 3.31V on ours the other day. Measure it, the tab is 3.3V, TP2 is GND.

                                                                           

                                                                          On the other hand, that would increase total power usage....

                                                                           

                                                                          The thermal resistance between where the power is generated and the outside of  the package is much less for the lan than for the soc. That might lead to a percieved temperature difference.

                                                                           

                                                                          Put your finger on the chip, when it becomes uncomfortable, swtich fingers. I'm guessing that the second or third finger won't have any trouble staying on the chip. Your finger is now acting as a heatsink. At first the thermal capacity of the package was heating the finger beyond "comfortable", but after getting rid of that heat, if your finger is able to dissipate the heat, it's less than 1W.

                                                                           

                                                                          Or at least, with my fingers its less than 1W. That's what I measured once.

                                                                           

                                                                          The power on the 'pi shares between the 3.3V regulator, the soc and the lan chip. And input power is about 2.4W, so at most 1W for the lan and soc and a little less for the 3.3V.