From: William Hermans <yyrk...@gmail.com> Reply-To: <beagleboard@googlegroups.com> Date: Tuesday, March 25, 2014 at 12:04 AM To: <beagleboard@googlegroups.com> Subject: Re: [beagleboard] Hardware watchdog for BBB
> I am still not sure why anyone would need / want all this complexity. > > Living offgrid, powered by solar panels which charge a battery bank, which > then powers our home via an Inverter . . . I am not sure why the same concept > can not be used on the BBB. > > 1) Power the BBB via a small rechargeable ~5V power source. > 2) charge this ~5V power source via AC mains, solar power, whatever. > 3) Monitor power on the charge input, and when absent send a message to the > kernel to shutdown / hybernate. > > Then, all you need is to make sure your power source can work a few minutes > with no input power applied. Perhaps even double this value for "safety". > > The way I see things, there is nothing to complex about all this at all. Not a problem when you are there to push the on button or reset your BBB when it locks up. What happens when there is no human intervention and the BBB is in some remote location? What happens during brown outs, power surges, power fluctuations, auto reclosure operations, power bypass, etc? What happens when you have 10K, 100K or even 1 Million devices running. Even a 0.1% failure rate will be a disaster. Cost is a primary factor so you cannot spend 10x on the power supply (batteries, solar panels; really?). Batteries are expensive and have a limited number of charge cycles, typically less than 1,000 cycles (less than 3 years). Actually, the circuit is even more complex because supercaps have a max voltage of 2.5 or 2.7 volts, so you have to stack them. Now you need an energy balance circuit to make sure all caps in series maintain an equal charge. During power fail, the voltage across these supercaps decrease, but you need to maintain a constant voltage, so you need a boost switching supply. The switcher cannot start until the supercaps have a minimum charge. I could go on, but yes the complexity is necessary to ensure a reliable supply. Your solution may be perfect for your requirements, but I think we are talking about a different operating environment. Regards, John > > > > > On Mon, Mar 24, 2014 at 10:41 AM, John Syn <john3...@gmail.com> wrote: >> >> From: <kb2...@gmail.com> >> Reply-To: <beagleboard@googlegroups.com> >> Date: Saturday, March 22, 2014 at 10:08 AM >> To: <beagleboard@googlegroups.com> >> Subject: [beagleboard] Hardware watchdog for BBB >> >>> HI, I have been working for a wile on safe power supply for BBB with backup >>> power provided by supercapacitors. In case of power failure there is just >>> enough time to safely and nicely shut down BBB. For some reason BBB does >>> not always wake up fully. I need hardware dogwatch. Did anybody design such >>> a thing? I was able to find some design for ardunio: >>> http://www.playwitharduino.com/?p=291. >>> Anybody has any experience with hardware dogwatch for BBB?? >>> Thanks in advance >>> Robert >> Hi Robert, >> >>> >> Developing a power supply that ensures a reliable shutdown down in the event >> of a power failure isn¹t a simple design. You really need to monitor the >> input power supply and the state of the kernel to determine when to remove >> and reapply power to the BBB. You have to consider the corner cases such as: >> 1. power failure could occur during the boot up sequence >> 2. power failure occurred, triggering a shutdown sequence and then power is >> restored during the shutdown sequence. >> With Linux, you cannot arbitrarily remove power during the boot up sequence >> and you cannot simply reapply power during the power down sequence. In the >> first case, when would it be safe to simply remove power to the BBB and in >> the second case, when would it be safe to recycle the power to the BBB. >> Currently there is no external info to determine the state of the kernel so >> you would have to add a kernel driver which will control a GPIO to signal >> when the kernel is in a safe mode (all volatile info written to non-volatile >> memory) and also monitor a GPIO used to interrupt the kernel when a power >> failure occurs. >> >> So now, you need an external state machine which tracks the input power >> supply, state-of-kernel and charge state of super caps. Timers are also >> required to ensure a proper power recycle. >> >> I hope I have covered everything you need to consider in your design, but >> perhaps others has some insights I haven¹t considered. >> >> Regards, >> John >>> -- >>> For more options, visit http://beagleboard.org/discuss >>> --- >>> You received this message because you are subscribed to the Google Groups >>> "BeagleBoard" group. >>> To unsubscribe from this group and stop receiving emails from it, send an >>> email to beagleboard+unsubscr...@googlegroups.com. >>> For more options, visit https://groups.google.com/d/optout. >> -- >> For more options, visit http://beagleboard.org/discuss >> --- >> You received this message because you are subscribed to the Google Groups >> "BeagleBoard" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to beagleboard+unsubscr...@googlegroups.com. >> For more options, visit https://groups.google.com/d/optout. > > > -- > For more options, visit http://beagleboard.org/discuss > --- > You received this message because you are subscribed to the Google Groups > "BeagleBoard" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to beagleboard+unsubscr...@googlegroups.com. > For more options, visit https://groups.google.com/d/optout. -- For more options, visit http://beagleboard.org/discuss --- You received this message because you are subscribed to the Google Groups "BeagleBoard" group. To unsubscribe from this group and stop receiving emails from it, send an email to beagleboard+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
