Why wouldnt I Power a Propeller staright from two 1.5v AA batteries

JohnR2010

I have been running my test circuit for two weeks now straight off of two AA batteries (No 3.3v voltage regulator). I'm working on a battery powered ZigBee project and I have both the ZigBee's VCC and the Propeller's VCC tied directly to the two batteries. Both the Propeller's data sheet and the xBee ZB SMT say they support 2.7 to 3.6vDC. In doing this I don't have to worry about any stray currents my regulator would eat up. When both the Propeller and the ZigBee are sleeping I'm pulling .010mA and they sleep 80% of the time so I'm talking more than a year of run time on two AA batteries! This has got to be too good to be true!

What concerns does the forum have with this approach?

Mike Green

Using some kind of ideal battery pack is a great idea, but what's available in terms of real batteries is sometimes far from ideal. If you look at the Duracell datasheet, you'll see that they start off at 1.6V (for 3.2V total ... good so far), but rapidly drop to 1.2-1.3V ... right on the brown-out threshold. If you look at NiMH cells, they start at 1.4V (for 2.8V total ... marginal but OK), but drop relatively quickly to 1.2-1.3V ... again right on the brown-out threshold. Duracell Ultra Lithium would work much better. They start out a little above 1.5V and stay above 1.4V for at least 1/2 of their discharge life. They're relatively pricey though.

Best would be to use some kind of boost switching regulator designed for this sort of use that uses two AA batteries and puts out 3.3V.

Tracy Allen

The internal resistance of a battery increases with age, so in your setup it will perform much better if you include a large capacitor to supply the pulses of current needed during the active periods. Here is a graph of voltage against time for an Energizer E91 AA cell.

Note that it a 100mW discharge is currents around 100mA. The voltage will stay much higher (as seen through the internal resistance) when the current is low. Note that low temperatures increase the internal resistance too.

Li-Fe-S₂ primary cells start out at about 1.7V open circuit, and hold up with low internal resistance for much longer. When they fail, they do so rapidly, so it is harder to judge end of life. Here is a graph comparing L91 to E91 AA cells.

The cells are very lightweight, and perform well over a wide temperature range.

Dave Hein

John, the circuit may draw very little current while it's sleeping, but the real question is how much does it draw when it's not sleeping. You said that it sleeps 80% of the time, which means that it's not sleeping 20% of the time. If it draws 100 mA when not sleeping the average current draw would be 20 mA. A pair of AA batteries rated at 2500 mAh would last for 125 hours, or about 5 days at 20 mA assuming you use a high efficiency switching regulator. To run for an entire year would require an average current draw of less than 0.3 mA. If the circuit is awake 20% of the time, that would only be 1.5 mA when awake.

JohnR2010

Yep I have noticed the differences between batteries and when I change out my current batteries I will try the Duracell Ultra Lithium. I have had my eye on the TPS61090 2A Booster from TI http://www.ti.com/product/tps61090 but I'm trying to keep the complexity and cost to a minimum. But I fear a booster is the route I'm going to have to take as I also want to use this device outside in cold environments and I know all bets are off when it gets down below freezing here in Illinois. Do you have a recommendation on a booster?

But for use indoors in a controlled environment I'm intrigued with the performance I'm getting so far.

Mike Green wrote: »

Using some kind of ideal battery pack is a great idea, but what's available in terms of real batteries is sometimes far from ideal. If you look at the Duracell datasheet, you'll see that they start off at 1.6V (for 3.2V total ... good so far), but rapidly drop to 1.2-1.3V ... right on the brown-out threshold. If you look at NiMH cells, they start at 1.4V (for 2.8V total ... marginal but OK), but drop relatively quickly to 1.2-1.3V ... again right on the brown-out threshold. Duracell Ultra Lithium would work much better. They start out a little above 1.5V and stay above 1.4V for at least 1/2 of their discharge life. They're relatively pricey though.

Best would be to use some kind of boost switching regulator designed for this sort of use that uses two AA batteries and puts out 3.3V.

JohnR2010

Tracy, thank you sir for the great input. I think the big Cap suggestion is an awesome idea. In fact I have some 5v very small (physically) 2.5f ultra caps I can give a try. I'm using the xBee to monitor VCC voltage so I can keep an eye on performance over time. Once I have the ZigBee power configuration cluster implemented I will get alerts when the battery is low.

This ZigBee stuff is so freaking cool!!

Tracy Allen wrote: »

The internal resistance of a battery increases with age, so in your setup it will perform much better if you include a large capacitor to supply the pulses of current needed during the active periods. Here is a graph of voltage against time for an Energizer E91 AA cell.

Note that it a 100mW discharge is currents around 100mA. The voltage will stay much higher (as seen through the internal resistance) when the current is low. Note that low temperatures increase the internal resistance too.

Li-Fe-S₂ primary cells start out at about 1.7V open circuit, and hold up with low internal resistance for much longer. When they fail, they do so rapidly, so it is harder to judge end of life. Here is a graph comparing L91 to E91 AA cells.

The cells are very lightweight, and perform well over a wide temperature range.

LoopyByteloose

Most rechargable cells perform poorly in comparison to similar primary (use once) cells.

But the Li Ion cell is likely to offer better performance than 1.5AA cells. The discharge curve is flatter, the battery has a higher density of stored power. The only problem is being certain to cease use before you go below the proper use voltage. That will destroy the battery, even just one event.

So a 18650 Li Ion with low voltage protection might be a good choice. But it is still a bit awkward. Most of the use is around 3.6 volts to 3.7 volts, and low voltage cut off is at 3.3 volts. But the fully charged cell can offer up 4.2 volts for a short period of time.

So no matter what you use, you have an S curve and the tail ends are where the problems lie. Can you get 100% power? Will the newer switching regulators, you get 90% power out of the cell, and it is regulated.

JohnR2010

Dave, good point on my math!! The current I draw when not in sleep mode is about 40 to 120mA by far most of the time it is down around 40mA. I actually have two sleeping devices. The Propeller is sleeping for 99.9% of the time and it is the power hog (if you call 40ma a lot of power). Basically the Propeller sleeps until something wakes it up and asks it to do something. The other device is a xBee ZB Pro SMT and it sleeps for 5 seconds at a time and wakes for less than 250mS to poll its host for data. When it does one of these polls I see a current jump of 2 to 3mA but I know it is pulling more I just cant detect it. I can also extend the poll cycle out to once every 28 seconds and still be within spec or even longer if I really, really, need to. My original goal was to get two to three months out of the batteries and I think that is doable with some tweaking.

Dave Hein wrote: »

John, the circuit may draw very little current while it's sleeping, but the real question is how much does it draw when it's not sleeping. You said that it sleeps 80% of the time, which means that it's not sleeping 20% of the time. If it draws 100 mA when not sleeping the average current draw would be 20 mA. A pair of AA batteries rated at 2500 mAh would last for 125 hours, or about 5 days at 20 mA assuming you use a high efficiency switching regulator. To run for an entire year would require an average current draw of less than 0.3 mA. If the circuit is awake 20% of the time, that would only be 1.5 mA when awake.

JohnR2010

Loopy, I can go the regulator route I have always done it this way in the past. But this is my first real low power (every milliamp accounted for) project and and on a whim I thought I would just give direct connect a try. I have stayed away from the Li Ion batteries because the higher voltages wont allow me to direct connect. But to your point and Tracy's on the power curve, I do understand I may get more out of the battery if I use a voltage boost circuit since I can still use the batteries as they dip. I'm going to have to do some testing.

Loopy Byteloose wrote: »

Most rechargable cells perform poorly in comparison to similar primary (use once) cells.

But the Li Ion cell is likely to offer better performance than 1.5AA cells. The discharge curve is flatter, the battery has a higher density of stored power. The only problem is being certain to cease use before you go below the proper use voltage. That will destroy the battery, even just one event.

So a 18650 Li Ion with low voltage protection might be a good choice. But it is still a bit awkward. Most of the use is around 3.6 volts to 3.7 volts, and low voltage cut off is at 3.3 volts. But the fully charged cell can offer up 4.2 volts for a short period of time.

So no matter what you use, you have an S curve and the tail ends are where the problems lie. Can you get 100% power? Will the newer switching regulators, you get 90% power out of the cell, and it is regulated.

jmg

JohnR2010 wrote: »

I have had my eye on the TPS61090 2A Booster from TI http://www.ti.com/product/tps61090 but I'm trying to keep the complexity and cost to a minimum.
..... Do you have a recommendation on a booster?

The TPS6109x has a nice high efficiency, but TI have been rather lazy in their specs.
Flyback (boost) converters cannot have Vo less than Vin, and yet they spec 5.5v Vin on a 3v3 out device.
There are SMPS that can tolerate Vin above and below Vo, but they usually use two inductors.

Some form of battery reversal protection is always a good idea

You may be able to use the LBO to switch in/out a diode drop on Vin, so that on brand new batteries, it drops Vin to keep < Vo, as the battery drops, the diode is removed. ( eg a P-FET body diode can do both Diode and removal)

evanh

Loopy Byteloose wrote: »

Most rechargable cells perform poorly in comparison to similar primary (use once) cells.

I'm gonna guess you are referring to common AA sized cells, ie: NiCd and NiMH. Please clarify performing poorly?

Dave Hein

JohnR2010 wrote: »

Dave, good point on my math!! The current I draw when not in sleep mode is about 40 to 120mA by far most of the time it is down around 40mA. I actually have two sleeping devices. The Propeller is sleeping for 99.9% of the time and it is the power hog (if you call 40ma a lot of power). Basically the Propeller sleeps until something wakes it up and asks it to do something. The other device is a xBee ZB Pro SMT and it sleeps for 5 seconds at a time and wakes for less than 250mS to poll its host for data. When it does one of these polls I see a current jump of 2 to 3mA but I know it is pulling more I just cant detect it. I can also extend the poll cycle out to once every 28 seconds and still be within spec or even longer if I really, really, need to. My original goal was to get two to three months out of the batteries and I think that is doable with some tweaking.

So the xBee is awake for 5% of the time, and the Prop is awake 0.1% of the time. How does that work if the xBee is receiving data while the Prop is sleeping? Does it wake up the Prop, or does it have a receive buffer that the Prop can read after it wakes up?

tonyp12

Lifepo4 batteries is great when it comes to operate around 3.2V, as 3.4v to 2.8v most ICs are very happy.


But I would also recommend going with a 3.8V Lithium battery + extreme low QC LDO.
I'm not a big fan of Microchip but this is one of the lowest Quiescent Current I ever seen at 1.6 μA.
http://www.mouser.com/ProductDetail/Microchip/MCP1700T-3002E-TT/?qs=%2fha2pyFadugJgcIR8OQT8eRAbM%252bVI1ZZ%2fdXeIVikR%2fKG3mDhJuVuJw%3d%3d

Tracy Allen

Lithium iron disulfide would breeze through those subfreezing temperatures in Illinois.

At low currents, there is very little effect of temperature on performance over the -40 to +60 °C range. I'd still go with a big capacitor in parallel to supply the pulses. These also have long shelf life, low self-discharge. 90% of capacity left still at16 years with room temperature storage.
Energizer has great technical reference materials. The above is from
http://data.energizer.com/PDFs/lithiuml91l92_appman.pdf
18 page guide with characteristics and applications.

I'm all for the rechargeable lithium chemistries too, especially if you can include energy harvesting for recharge.

Don M

Maybe consider this?

https://www.sparkfun.com/products/10967

Cluso99

If you are not tight on space, don't forget you can parallel a few pairs of AA's.

Lawson

tonyp12 wrote: »

*snip*
I'm not a big fan of Microchip but this is one of the lowest Quiescent Current I ever seen at 1.6 μA.
http://www.mouser.com/ProductDetail/Microchip/MCP1700T-3002E-TT/?qs=%2fha2pyFadugJgcIR8OQT8eRAbM%252bVI1ZZ%2fdXeIVikR%2fKG3mDhJuVuJw%3d%3d

Micrel has a few that are lower current and better behaved. (MIC5231 and MIC5232) The 3.3v MCP1700 will draw >30uA when the input voltage is below 3.3v. (the 1.25 and 1.5 volt versions of the MCP1700 are well behaved so make cheap uA references)

I'd just disable the brown-out detector and use an external 2.0v brownout detector. A TC54 series voltage detector would be my choice. I've booted the prop at 1.5 volts. I bet it still does over 40MHz at 2 volts. (see my signature for details)

The speed of the internal RC clocks are also quite voltage and temperature sensitive. With some fixed external timing source it's easy to infer the power supply voltage. (an RC circuit on a pin or external clock)

Marty

JohnR2010

Dave, the xBee is configured to be an ZigBee end device with cycle sleep and pin wake. The xBee's CTS Pin (p12) goes high when the xBee is done sending data and is in sleep mode. When this pin is high I stop all cogs (except the main cog), switch to internal slowwwww clock and do a pin wait on the CTS Pin. This is my low current mode and the propeller stays in this mode until the xBee has data for it. It knows it has data when the CTS line goes low for more than 50mS.

I think i need to mention I'm communicating with the xBee through the API interface and I'm sending ZigBee formatted packets. ZigBee is more than just a RF communications protocol it also defines several standard ways applications can communicate with each other. For example if you have an on/off device an you want to turn it ON that command is all contained in one packet. So all the Propeller has to do is receive this packet, interpret it, and then switch on a pin. The Propeller doesn't do the polling it is simply waiting to be woke up and give data. The xBee's Freescale processor does all the low level polling for me.

So while the Propeller is sleeping the xBee wakes up every 5 seconds and connects to its ZigBee parent to see if it has data queued for it. This checking for data is extreamly fast (~10ms) basically one packet out and one packet back to acknowledge that it got the poll request and then the xBee goes back to sleep. The Propeller remains sleeping through this poll process and the xBee repeats until a data packet is received. When an actual packet is received the xBee's Wake Host timer is started (WH command sets the timer, its not well documented) and the CTS line is set high. The high CTS pin wakes up the propeller and it switches back out to the 5Mhz crystal, starts up the cogs and waits for data from the xBee. At this point the xBee's Wake Host timer has ticked down and it starts transmitting data to the Propeller.

If there isn't data for the Propeller to process it stays in sleep mode. It could sleep for hours or even days. If the Propeller needs to send a packet out and the xBee is sleeping it can wake the xBee by pulling the Pin Sleep (Pin 9 on the xBee low).

Dave Hein wrote: »

So the xBee is awake for 5% of the time, and the Prop is awake 0.1% of the time. How does that work if the xBee is receiving data while the Prop is sleeping? Does it wake up the Prop, or does it have a receive buffer that the Prop can read after it wakes up?

JohnR2010

Don, I have seen that board before and I think there is a 500mA version as well. It is on my list of things to try. Thanks!

Don M wrote: »

Maybe consider this?

https://www.sparkfun.com/products/10967