Post by thread:[PIC]: self battery sensing

> I'm starting work on a project using a PIC powered by a battery. > > The idea is that the PIC will spend most of the time in sleep mode so the > batteries will last a nice long time - at least a year would be our target. > > The PIC will probably be something like a PIC16C711. > > We need some way of providing an alarm when the battery is low. i.e. a > voltage sensor that doesn't itself use much power. > > Are there any chips around to do this? > > Anyone got any ideas about how I could do this? > > Brian Gregory. Brian, I have a one-pin solution but need to know if the PIC is running from a regulated voltage, or if the PIC is running from the (ever decreasing) battery voltage? -Roman -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Brian Gregory wrote: > > Roman Black <fastvidKILLspamEZY.NET.AU> wrote: > > Brian, I have a one-pin solution but need to know > > if the PIC is running from a regulated voltage, > > or if the PIC is running from the (ever decreasing) > > battery voltage? > > -Roman > > I will probably be running the PIC from 5V from a LDO regulator. > The battery will probably be 9V Alkaline. Did you see Dan's suggestion? Basically have a high value resistor from bat + to the PIC pin. 560k will do nice. Use a small ceramic cap, like 0.1uF or 0.047uF etc from that pin to ground. Pull the pin low and then switch it to input and count time in software until it reaches a "1" level, about 1 to 2 volts depending which PIC pin you are using (schmidt input or not). This charge time will be dependant on the battery voltage, and since the PIC is run from regulated 5v it's input pin threshold will give a fairly reliable trigger point. You don't need an analogue input pin. For low power use hold the pin at logic high, 5v, so current will be 9v-5v = 4v/560000 = 7uA so not too bad. I imagine the PIC pin protection diodes will handle that 7uA, but if you have the space and budget put a 5.1v 400mW zener from pin to ground to keep everything in spec. I would do this if the product was for commercial sale. Heck I always have 5.1v zeners around so I would do it anyway! Check zener leakage, with the PIC unplugged, you may need a higher voltage zener to balance at such a low current, like 7uA. I am not sure how this circuit will hold up under battery measuring when the battery gets to 5v, but hopefully the user has seen the low battery indication and changed the battery by the time it gets under 7v or so. :o) -Roman -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Roman wrote: {Quote hidden}>Brian Gregory wrote: >> >> Roman Black <.....fastvidKILLspam.....EZY.NET.AU> wrote: >> > Brian, I have a one-pin solution but need to know >> > if the PIC is running from a regulated voltage, >> > or if the PIC is running from the (ever decreasing) >> > battery voltage? >> > -Roman >> >> I will probably be running the PIC from 5V from a LDO regulator. >> The battery will probably be 9V Alkaline. > > >Did you see Dan's suggestion? Basically have a high value resistor >from bat + to the PIC pin. 560k will do nice. Use a small ceramic >cap, like 0.1uF or 0.047uF etc from that pin to ground. >Pull the pin low and then switch it to input and count time >in software until it reaches a "1" level, about 1 to 2 volts >depending which PIC pin you are using (schmidt input or not). > Actually this was my "guess" as to what Roman's ckt was, as he wasn't saying at the time :). Turns out he liked my guess better than his own actual ckt. I hadn't really thought about it much, but Alice mentioned offlist that this dohicky is quite dependent upon the tolerance on the cap. Then of course you have to worry about calculations related to exponential charging waveforms, but I am sure the s.w. weenies got a way around exp/log computations. =============== >This charge time will be dependant on the battery voltage, >and since the PIC is run from regulated 5v it's input pin >threshold will give a fairly reliable trigger point. You don't >need an analogue input pin. Actually, as mentioned last week, my "preferred" ckt would use an A/D channel connected to a digital I/O pin thru 10K as well as to the battery thru a 5.1v zener. Pull the I/O pin low, and you get battery value directly, Vanal_in = Vbatt-5.1v. Also, when you pull the I/O pin high, the only current thru the zener is gonna be leakage - hopefully low. But you could probably deal with that by floating the I/O pin rather than pulling high. If you don't like floating tie a 500K from the pin to Vdd. ================= > >For low power use hold the pin at logic high, 5v, so current >will be 9v-5v = 4v/560000 = 7uA so not too bad. I imagine >the PIC pin protection diodes will handle that 7uA, Yeah, supposedly 25 mA - although I have put 75-100 mA into PIC pins in qual testing without frying the diodes. Also, if you look in the old AN521 appnote, Mchp shows a ckt where they connect 120VAC out of the wall straight into a PIC thru a 5M resistor. Gives me the willies, but they do it. They obviously figure the clamping diode knows its job. ================ but if >you have the space and budget put a 5.1v 400mW zener from >pin to ground to keep everything in spec. I would do this >if the product was for commercial sale. Heck I always have >5.1v zeners around so I would do it anyway! Bingo - see comments above about 120VAC to pin. =============== check zener >leakage, with the PIC unplugged, you may need a higher >voltage zener to balance at such a low current, like 7uA. > >I am not sure how this circuit will hold up under battery >measuring when the battery gets to 5v, but hopefully the >user has seen the low battery indication and changed the >battery by the time it gets under 7v or so. :o) Hopefully you track Vbatt and catch it before the ckt crashes. [hey Roman, I'll trade you some good stepper motor code for your milling machine - having but one of those and a 1 Ghz scope, and I'll die happy :)]. best regards, - Dan Michaels Oricom Technologies http://www.oricomtech.com ========================= -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Dan Michaels wrote: {Quote hidden}> > Roman wrote: > >Brian Gregory wrote: > >> > >> Roman Black <EraseMEfastvidspam_OUTTakeThisOuTEZY.NET.AU> wrote: > >> > Brian, I have a one-pin solution but need to know > >> > if the PIC is running from a regulated voltage, > >> > or if the PIC is running from the (ever decreasing) > >> > battery voltage? > >> > -Roman > >> > >> I will probably be running the PIC from 5V from a LDO regulator. > >> The battery will probably be 9V Alkaline. > > > > > >Did you see Dan's suggestion? Basically have a high value resistor > >from bat + to the PIC pin. 560k will do nice. Use a small ceramic > >cap, like 0.1uF or 0.047uF etc from that pin to ground. > >Pull the pin low and then switch it to input and count time > >in software until it reaches a "1" level, about 1 to 2 volts > >depending which PIC pin you are using (schmidt input or not). > > > > Actually this was my "guess" as to what Roman's ckt was, > as he wasn't saying at the time :). Turns out he liked my > guess better than his own actual ckt. > > I hadn't really thought about it much, but Alice mentioned > offlist that this dohicky is quite dependent upon the tolerance > on the cap. Then of course you have to worry about calculations > related to exponential charging waveforms, but I am sure the > s.w. weenies got a way around exp/log computations. > =============== Hi Dan, yeah the t=rc curve is very non linear but you have to imagine the main use of a simple circuit like this is with a cheap pic (12 series??) and one digital pin, for say a simple "low battery" threshold sensor, which is how many cheap commercial devices work, or indeed for a 9v battery it would be fine to sense say 9.2v, 9.0v, 8.8v etc so I imagine a lookup table with less than 20 bytes would do fine. When I used this system I just used it for medium/low battery sensing with two simple thresholds, which I just hardcoded after a few trial/error measurements. I didn't expect performance similar to ADC input but how accurate do you want your battery level sensing anyway? I have a $300 AU walkman that has a low/ med/high battery indicator. Works fine for me! :o) -Roman -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

>> Basically have a high value resistor >>from bat + to the PIC pin. 560k will do nice. Use a small ceramic >>cap, like 0.1uF or 0.047uF etc from that pin to ground. >>Pull the pin low and then switch it to input and count time >>in software until it reaches a "1" level, about 1 to 2 volts >>depending which PIC pin you are using (schmidt input or not). In practice that don´t work reliably. First, the guraranteed tolerance of the PIC high threshold is horrible, and even if it was: a cap with low enough tolerance is not cheap. I suggest using some cheap reset chip and resistive voltage divider or comparator + reference. Or to be cheap: use one single transistor to comapre regulated Vdd to batt voltage trrough resistive divider: a resiistive divider feeding the base of a PNP with emitter to Vdd, and collector with a pulldown resistor to GND. Four components, but much better guraranteed precision than RC-time solution. Function: When Batt voltage divided by resistor divider is lower than Vdd-0,6V the PNP start conducting and driving a high level into PIC. Both alternatives need Vdd to be regulated, BTW. Another solution i often use is to simply moinitor the voltage drop avross the regulator using a resistive divider and PNP sensing between input-output of regulator, and as long as it is higher than min specified drop for that regulator the PNP os on, feeding input to controller (though another resistive divider). Five cheap components and nice function. /Morgan -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.