Post by thread:[EE] Battery selection

Hi everybody, I've got a problem with battery selection for an RF device. The RF part is done by an XBee, the controller is a PIC. The XBee is specified for 2.8V-3.4V supply voltage; it runs in bursts of 20ms, during which it consumes 55mA. The consumption during the rest of the time is small: around 60uA (varies somewhat with battery voltage). I'm currently using a Tadiran TL-5955 Lithium battery <tadiranbat.com/prodpdf/viewpdf.php?datasheet=TL-5955%20Rev%20D.pdf>: 1.5Ah, max. pulse 150mA, 2/3AA size. I thought that this type of battery has a voltage that's rather independent of charge state until it is almost dead. But what I find is that the voltage under 55mA load dips below 2.8V much earlier in the lifetime than I expected, considering the battery capacity and the current consumption. (When it dips below 2.8V under 55mA load, it's basically useless from this point on, as the XBee doesn't work reliably anymore.) Another thing that is not good with this battery is that this type of battery seems to be difficult to obtain, as replacement (walk-in stores). Can somebody suggest a better battery? Better in the sense that with the given load I get to use a maximum of its capacity, that it is easy to obtain, and that it is not bigger than 2/3AA. As an alternative I thought about boosting the battery voltage for the XBee, during its 20ms of operation. I think a standard boost switcher would add significant components (I'm having a serious space problem) and require 2 PIC pins (currently none free), and a possibly simpler switched-cap voltage doubler would exceed the XBee max. voltage and still add almost as many components (if not more) and require 1 PIC pin. If I don't find anything better, I'm tending towards the switcher solution, but adding the additional components (inductor, switch, diode, capacitor, two resistors) won't be easy and probably will get ugly. Is there a simple solution that I'm missing for boosting the battery voltage during the 20ms XBee operation? Thanks for any input, Gerhard

On Sat, Feb 7, 2009 at 1:32 PM, Gerhard Fiedler <spam_OUTlistsTakeThisOuTconnectionbrazil.com> wrote: > Hi everybody, > > I've got a problem with battery selection for an RF device. The RF part > is done by an XBee, the controller is a PIC. The XBee is specified for > 2.8V-3.4V supply voltage; it runs in bursts of 20ms, during which it > consumes 55mA. The consumption during the rest of the time is small: > around 60uA (varies somewhat with battery voltage). > > I'm currently using a Tadiran TL-5955 Lithium battery > <tadiranbat.com/prodpdf/viewpdf.php?datasheet=TL-5955%20Rev%20D.pdf>: > 1.5Ah, max. pulse 150mA, 2/3AA size. > ok, that's about 3.3V open-circuit, no? So how about a capacitor to supply the current when the battery can't. Let's see, 55 mA for 20ms, with a maximum droop of (3.3 - 2.8) or 500mV. I = C dv/dt or. C = I / (dv/dt) C = 55 mA / (500mV / 20 ms) C = 2200 uF. got room for a 2200 uF cap? http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=P5116-ND > I thought that this type of battery has a voltage that's rather > independent of charge state until it is almost dead. But what I find is > that the voltage under 55mA load dips below 2.8V much earlier in the > lifetime than I expected, considering the battery capacity and the > current consumption. (When it dips below 2.8V under 55mA load, it's > basically useless from this point on, as the XBee doesn't work reliably > anymore.) A better first-order model for a battery is a voltage source in series with a resistor. As the battery discharges the resistor increases in value. You can estimate the resistance from the datasheet plots. (The internal resistance and load resistance form a voltage divider, which you can use to calculate the actual terminal voltage.) This is a better model than the simple "slowly varying voltage" model when you are trying to predict performance with a varying load. Regards, Mark markrages@gmail -- Mark Rages, Engineer Midwest Telecine LLC .....markragesKILLspam@spam@midwesttelecine.com

Gerhard Fiedler wrote: {Quote hidden}> Hi everybody, > > I've got a problem with battery selection for an RF device. The RF part > is done by an XBee, the controller is a PIC. The XBee is specified for > 2.8V-3.4V supply voltage; it runs in bursts of 20ms, during which it > consumes 55mA. The consumption during the rest of the time is small: > around 60uA (varies somewhat with battery voltage). > > I'm currently using a Tadiran TL-5955 Lithium battery > <tadiranbat.com/prodpdf/viewpdf.php?datasheet=TL-5955%20Rev%20D.pdf>: > 1.5Ah, max. pulse 150mA, 2/3AA size. > > I thought that this type of battery has a voltage that's rather > independent of charge state until it is almost dead. But what I find is > that the voltage under 55mA load dips below 2.8V much earlier in the > lifetime than I expected, considering the battery capacity and the > current consumption. (When it dips below 2.8V under 55mA load, it's > basically useless from this point on, as the XBee doesn't work reliably > anymore.) > > Another thing that is not good with this battery is that this type of > battery seems to be difficult to obtain, as replacement (walk-in > stores). > > Can somebody suggest a better battery? Better in the sense that with the > given load I get to use a maximum of its capacity, that it is easy to > obtain, and that it is not bigger than 2/3AA. > > snip > Thanks for any input, > Gerhard > How about a big capacitor... to act as a charge store and reduce the load on the cell when the 55ma is drawn? Somewhere around 500uF I guess will halve the load on the cell by supplying the current form the cap. Then again, I'm a hobbyist... ;-) Rolf

Rolf wrote: > Gerhard Fiedler wrote: >> Hi everybody, >> >> Can somebody suggest a better battery? Better in the sense that with the >> given load I get to use a maximum of its capacity, that it is easy to >> obtain, and that it is not bigger than 2/3AA. >> >> > snip >> Thanks for any input, >> Gerhard Sometimes, the inefficient soulution is the best, sometimes not. An inefficient solution that could maybe work in your case is a battery with higher voltage followed by a linear regulator. This ideas merits are low parts count and for the life of the battery, acceptable voltage output without sagging. Whether a battery of higher voltage exists is left as an exercise for the designer...

There are couple of hardware chips which can boost the battery voltage to 3.3V or 5V. Many MP3 which use a single AAA battery implemented this kind of technology to driving a few 4.0V white LED as back-light . For loads <80~100 mA, the input can go as low as 2.3V~2.5V. You can also use a 8 pin 12F PIC microcontroller as the core of a boost circuits if you can secure it working on the lowest input. EM Noise need be taken care of when switch mode boost circuits are used. Funny N. Au Group Electronics, http://www.AuElectronics.com ________________________________ From: Gerhard Fiedler <listsKILLspamconnectionbrazil.com> To: piclist <.....piclistKILLspam.....mit.edu> Sent: Saturday, February 7, 2009 2:32:18 PM Subject: [EE] Battery selection Hi everybody, I've got a problem with battery selection for an RF device. The RF part is done by an XBee, the controller is a PIC. The XBee is specified for 2.8V-3.4V supply voltage; it runs in bursts of 20ms, during which it consumes 55mA. The consumption during the rest of the time is small: around 60uA (varies somewhat with battery voltage). I'm currently using a Tadiran TL-5955 Lithium battery <tadiranbat.com/prodpdf/viewpdf.php?datasheet=TL-5955%20Rev%20D.pdf>: 1.5Ah, max. pulse 150mA, 2/3AA size. I thought that this type of battery has a voltage that's rather independent of charge state until it is almost dead. But what I find is that the voltage under 55mA load dips below 2.8V much earlier in the lifetime than I expected, considering the battery capacity and the current consumption. (When it dips below 2.8V under 55mA load, it's basically useless from this point on, as the XBee doesn't work reliably anymore.) Another thing that is not good with this battery is that this type of battery seems to be difficult to obtain, as replacement (walk-in stores). Can somebody suggest a better battery? Better in the sense that with the given load I get to use a maximum of its capacity, that it is easy to obtain, and that it is not bigger than 2/3AA. As an alternative I thought about boosting the battery voltage for the XBee, during its 20ms of operation. I think a standard boost switcher would add significant components (I'm having a serious space problem) and require 2 PIC pins (currently none free), and a possibly simpler switched-cap voltage doubler would exceed the XBee max. voltage and still add almost as many components (if not more) and require 1 PIC pin. If I don't find anything better, I'm tending towards the switcher solution, but adding the additional components (inductor, switch, diode, capacitor, two resistors) won't be easy and probably will get ugly. Is there a simple solution that I'm missing for boosting the battery voltage during the 20ms XBee operation? Thanks for any input, Gerhard

On Feb 7, 2009, at 7:33 PM, Funny NYPD wrote: > I'm currently using a Tadiran TL-5955 Lithium battery > <http://tadiranbat.com/prodpdf/viewpdf.php?datasheet=TL-5955%20Rev%20D.pdf > >: > 1.5Ah, max. pulse 150mA, 2/3AA size. > > I thought that this type of battery has a voltage that's rather > independent of charge state until it is almost dead. But what I find > is > that the voltage under 55mA load dips below 2.8V much earlier in the > lifetime than I expected, considering the battery capacity and the > current consumption. That does seem pretty counter-intuitive; it's a 3.6V battery after all... Have you confirmed the current draw/etc? It'd be more understandable if you're exceeding that 155mA spec. (and have you talked to tadiran?) Alternatives might include the CR2 "photo" batteries, but since those are 3V, you'd almost certainly need a boost converter to keep voltages above 2.8V. Or perhaps the Li-ion rechargeable CR2s, which are up near 4V (but are likely to have much less capacity, given that they're rechargeable.) And I think the CR2s are a bit fatter than an AA. (I don't know that there are are "popular" 3.6V consumer batteries.) BillW

Looking at the datasheet, the voltage does drop considerably with current drain, which is typical for long-life Lithium non-rechargeable batteries. They do not show a discharge curve for anything above 20mA, but the 20mA one looks like the voltage is about 3.2 instead of 3.6, so extrapolating to 55mA we would get 2.5V as the voltage at 55mA. I have previously used the LTC3400, which is a tiny 6 pin boost switcher IC, to boost one NiMH cell up to 3.3V for an XBee module. The entire footprint of the switcher circuit is about 1/2 inch by 3/4 inch. It works very well. It DOES dominate the circuit current drain when the XBee and PIC are asleep, but it gets down to around 2 or 3mA from the input side. It actually goes into a low power mode where it turns itself off and waits until the output capacitor discharges 1% of the nominal voltage. It then turns back on and recharges the cap. Sean On Sun, Feb 8, 2009 at 12:55 AM, William Chops Westfield <EraseMEwestfwspam_OUTTakeThisOuTmac.com> wrote: {Quote hidden}> > On Feb 7, 2009, at 7:33 PM, Funny NYPD wrote: > >> I'm currently using a Tadiran TL-5955 Lithium battery >> <tadiranbat.com/prodpdf/viewpdf.php?datasheet=TL-5955%20Rev%20D.pdf >> >: >> 1.5Ah, max. pulse 150mA, 2/3AA size. >> >> I thought that this type of battery has a voltage that's rather >> independent of charge state until it is almost dead. But what I find >> is >> that the voltage under 55mA load dips below 2.8V much earlier in the >> lifetime than I expected, considering the battery capacity and the >> current consumption. > > That does seem pretty counter-intuitive; it's a 3.6V battery after > all... Have you confirmed the current draw/etc? It'd be more > understandable if you're exceeding that 155mA spec. (and have you > talked to tadiran?) > > Alternatives might include the CR2 "photo" batteries, but since those > are 3V, you'd almost certainly need a boost converter to keep voltages > above 2.8V. Or perhaps the Li-ion rechargeable CR2s, which are up > near 4V (but are likely to have much less capacity, given that they're > rechargeable.) And I think the CR2s are a bit fatter than an AA. (I > don't know that there are are "popular" 3.6V consumer batteries.) > > BillW > > -

Sean Breheny wrote: > On Sun, Feb 8, 2009 at 12:55 AM, William Chops Westfield wrote: >> That does seem pretty counter-intuitive; it's a 3.6V battery after >> all... Have you confirmed the current draw/etc? It'd be more >> understandable if you're exceeding that 155mA spec. It seems Bill was right here. What seems to happen is that occasionally a bug in the firmware makes itself seen and the PIC talks to the XBee more than it should -- which drains the battery more than it should. Fixing this is my first thing to do, and possibly the biggest part of the battery problem is gone then. >> Alternatives might include the CR2 "photo" batteries, but since those >> are 3V, you'd almost certainly need a boost converter to keep >> voltages above 2.8V. Or perhaps the Li-ion rechargeable CR2s [...] Remains the battery replacement question, and you're right. The main reason for selecting this battery type in the first place was that I didn't find a more popular battery that satisfied my voltage requirements. If this should turn out to be a major deal, I'll probably have to bite the bullet and add a boost circuit for the XBee supply. > Looking at the datasheet, the voltage does drop considerably with > current drain, which is typical for long-life Lithium > non-rechargeable batteries. They do not show a discharge curve for > anything above 20mA, but the 20mA one looks like the voltage is about > 3.2 instead of 3.6, so extrapolating to 55mA we would get 2.5V as the > voltage at 55mA. At least in normal operation, it doesn't go that low. The voltage during the 55mA bursts (typically not more than one per second) is only ~300mV lower than the battery voltage with minimal load (around 60uA). > I have previously used the LTC3400, which is a tiny 6 pin boost > switcher IC, to boost one NiMH cell up to 3.3V for an XBee module. > The entire footprint of the switcher circuit is about 1/2 inch by 3/4 > inch. It works very well. It DOES dominate the circuit current drain > when the XBee and PIC are asleep, but it gets down to around 2 or 3mA > from the input side. Thanks for the tip; I'll keep it in mind. Also thanks to the others (the capacitor solution). It would have a space problem (it's marginal, but would be possible, with 4 caps of 6mm diameter in parallel) and it would have a leakage problem (the cap has a not negligible leakage current). But since the problem seems to have changed (see above), this option is for now at least out. Gerhard