Date: Sun, 4 Jan 2015 11:30:35 +0100 From: Bernd Walter <ticso@cicely7.cicely.de> To: Adrian Chadd <adrian@freebsd.org> Cc: Bernd Walter <ticso@cicely7.cicely.de>, ticso@cicely.de, "freebsd-mips@freebsd.org" <freebsd-mips@freebsd.org> Subject: Re: USB stability problem on AR9331 with stable/10 Message-ID: <20150104103035.GK3265@cicely7.cicely.de> In-Reply-To: <CAJ-Vmo=NHV-smV-iTKJTRKL%2BcN7mT_qbfUCmDJq8h6_aXDJ9OA@mail.gmail.com> References: <20150103023713.GB3265@cicely7.cicely.de> <CAJ-VmokgPvoUftwPCg4%2B6VGkAFTz7Gi-%2BFFAQkQYO=NO2LssQg@mail.gmail.com> <20150103114302.GD3265@cicely7.cicely.de> <20150103140918.GF3265@cicely7.cicely.de> <20150103214818.GJ3265@cicely7.cicely.de> <CAJ-Vmo=NHV-smV-iTKJTRKL%2BcN7mT_qbfUCmDJq8h6_aXDJ9OA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Jan 03, 2015 at 10:39:49PM -0800, Adrian Chadd wrote: > On 3 January 2015 at 13:48, Bernd Walter <ticso@cicely7.cicely.de> wrote: > > On Sat, Jan 03, 2015 at 03:09:18PM +0100, Bernd Walter wrote: > >> > >> I will swap for another board now. > >> Reason is that I have one board running on a OS version, which is about > >> 1.5 years old and it shows this problem only from time to time. > >> I even use it to copy the rootfs from nfs to the stick because of the > >> endian issue with mips. > >> It is very hard to say when this happens, because copying usually works, > >> but once you put some additional load, like compiling, or in this case > >> creating host keys. > >> Now this board with recent stable/current shows this problem extremly > >> often. > >> I want to rule out that I put a broken board into my test mix, because > >> it was a new one. > > > > Ok, I can't tell for sure if there is some hardware related influence, > > but I'd tested 3 Dragino MS14-P (likely same batch) and retested with one > > Carambola 2, which all showed the problem sooner or later. > > Hm, I wonder which particular AR933x revision they're using. The one on all the Dragino is marked: AR9331-AL3A PPW718,006C 1324 KOREA The Carambola is shielded and soldered shut. With the modules I can remove the cap since it is designed to be removeable, but this is the devboard and they added some solder at some points. I can unsolder it if required. A single module, which I'd bought at the same time, has the following: AR9331-AL3A PPU680,001C 1316 KOREA No sure if it has exactly the same chip on the devboard. The Carambola however seems to be running more stable. I've managed to build pkg 2 times of 3 tries, while the Dragino was never able to complete the build in countless tries. > Are you providing them enough power? It is very different in both cases. The Carambola devboard has a single micro-USB running both power and USB-console-UART, so I power it using a self powered hub, which has a switching mode 3.5A rates supply. The devboard has a 5 to 3.3V switching regulator running the module and it passes the 5V to the USB port. The 1.2V is somehow generated on the module. The Dragino has a wide range 9-15V input with 3 regulators. An RT8250 for 5V and 3.3V each - the 5V is used for the USB, the 3.3 for the rest. Don't know how the 1.2V is generated, since all I can find is a coil. Then there is a 1117 5V, which I can only guess that it is used for the moduleslot, which is intended to fit some arduino IO module. I run the board using 12V from a decent lab type bench supply with current limit at 0.5A, which is rated current and well below what it really used. But to be honest - I've already wondered about the power myself, since the only electrolytic cap is for the 1117, while the RT8052 use ceramic only and high rated ceramic have a high DC bias loss plus the RT8052 is not a very high frequency chip. So I just checked and the power rails look fine under a scope. -- B.Walter <bernd@bwct.de> http://www.bwct.de Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150104103035.GK3265>