The conversation on this page has been archived and is no longer active.

Powerbook battery exhibiting memory

Message #1 - Posted 2004/04/09 - Daniel Karrick Boothe

I have a powerbook 1GHz 15" which has started, much to my chagrin, to exhibit characteristics of the memory effect, which makes no sense. However, I have been using an application called X-charge to monitor the batteries charge and discharge patterns, and it quite clearly will discharge from 100% to about 80% and then drop without warning to 0% and go to sleep. Then, when charging again, it charges to about 20% and then shoots up to 100%. Needless to say, I am frustrated.

I have heard that the memory effect can be caused by the charge controller itself, in that it monitors the charge going in and out of the battery and thus knows how much the charge is based on that history. If that gets miscalibrated, it can wrongly detect the state of the battery. I have also heard the way to fix this is to deep discharge the battery, which basically means leaving it on sleep mode for several days until it actually dies, because I
can't disable the soft sleep on (what it thinks is) low battery. I left it on low battery sleep for about 36 hours and it didn't die, and being an engineering student, that was about the limit of how long I could afford to have my laptop out of commission. There was no change in the behavior of the battery, in fact it got slightly worse if anything.

Any ideas/theories, or even a way to disable low battery soft sleep would be welcome.

dan

________________
Daniel K. Boothe
Worcester Polytechnic Institute
ECE major, Class of 2005
dboothe@ece.wpi.edu
________________

Message #2 - Posted 2004/04/09 - John Johnson

Previously, Daniel Karrick Boothe wrote:

I have a powerbook 1GHz 15" which has started, much to my chagrin, to exhibit characteristics of the memory effect, which makes no sense. However, I have been using an application called X-charge to monitor the batteries charge and discharge patterns, and it quite clearly will discharge from 100% to about 80% and then drop without warning to 0% and go to sleep. Then, when charging again, it charges to about 20% and then shoots up to 100%. Needless to say, I am frustrated.

I think that it would be best to avoid the phrase "memory effect" if only because of the arguments about what it is and what sorts of batteries are subject to it. This is a Li-ion battery and does not suffer from 'memory'. It might (as you point out) suffer from miscalibration of the charge controller.

Before investigating that, I should ask how old you battery is, and how much you've used it. Your behavior is similar to behavior of a worn-out Li-ion battery. My experience with a Pismo is that people actually do get between 2-3 years use out of a battery, but that some heavy users can wear them out in less than 1.5 years.

AFAIK, all PB batteries are rated for 2 years or 500 charge cycles, whichever comes first. Where do you estimate your battery to be on this scale?

I have heard that the memory effect can be caused by the charge controller itself, in that it monitors the charge going in and out of the battery and thus knows how much the charge is based on that history. If that gets miscalibrated, it can wrongly detect the state of the battery.

Look for posts by Peter Renzland (who no doubt will also reply to this) to this group for similar problems. He's got a procedure for recalibrating your charge controller. Incidentally, if you do decide to follow this procedure, please make notes of the state of the battery before and after. There have been questions about how effective the procedure is and Peter is/was looking for data (I'm interested as well, but don't have anything like the stake in the results that he does).

Message #3 - Posted 2004/04/09 - Daniel Karrick Boothe

I think that it would be best to avoid the phrase "memory effect" if only because of the arguments about what it is and what sorts of batteries are subject to it. This is a Li-ion battery and does not suffer from 'memory'. It might (as you point out) suffer from miscalibration of the charge controller.

Forget I mentioned the words memory effect, call the symptoms what you will.

Before investigating that, I should ask how old you battery is, and how much you've used it. Your behavior is similar to behavior of a worn-out Li-ion battery. My experience with a Pismo is that people actually do get between 2-3 years use out of a battery, but that some heavy users can wear them out in less than 1.5 years.

AFAIK, all PB batteries are rated for 2 years or 500 charge cycles, whichever comes first. Where do you estimate your battery to be on this scale?

I bought my PB in May of 2003, so it is less than a year old. I have used it extensively during that time, however, since it is my main machine. Normally, I leave it plugged in most all the time. On average twice a week since september I take it with me to campus and
use anywhere from 50% to 90% of the battery. Since I am at less than one year, and a generous estimate of charge cycles might be 300 (really generous), I would be extremely disappointed if the battery were shot already.

Look for posts by Peter Renzland (who no doubt will also reply to this) to this group for similar problems. He's got a procedure for recalibrating your charge controller. Incidentally, if you do decide to follow this procedure, please make notes of the state of the battery before and after. There have been questions about how effective the procedure is and Peter is/was looking for data (I'm interested as well, but don't have anything like the stake in the results that he does).

I'd be happy to try out the procedure, and I'll do what I can to document any improvement.

dan

Message #4 - Posted 2004/04/10 - John Johnson

Previously, Daniel Karrick Boothe wrote:

[down the memory hole]

AFAIK, all PB batteries are rated for 2 years or 500 charge cycles, whichever comes first. Where do you estimate your battery to be on this scale?

I bought my PB in May of 2003, so it is less than a year old. I have used it extensively during that time, however, since it is my main machine. Normally, I leave it plugged in most all the time. On average twice a week since september I take it with me to campus and
use anywhere from 50% to 90% of the battery. Since I am at less than one year, and a generous estimate of charge cycles might be 300 (really generous), I would be extremely disappointed if the battery were shot already.

Ok, It's probably not shot then. You may use the machine heavily, but you're pretty light on the battery. One good way to check is to use the battery in a different machine (and use their battery in yours), if you have a freind with a suitable machine. That way, if the behavior disappears or travels or stays in one place, you've got a good idea of what's going on.

It might be worth performing general system maintenance as well: fix permissions, etc. Good luck with it.

Message #5 - Posted 2004/04/11 - Peter Renzland

John Johnson wrote:

Daniel Karrick Boothe wrote:

discharge from 100% to about 80% and then drop without warning to 0% and go to sleep. Then, when charging again, it charges to about 20% and then shoots up to 100%.

Yes, this is a characteristic pattern.

This is a Li-ion battery and does not suffer from 'memory'. It might (as you point out) suffer from miscalibration of the charge controller.

I have heard that the memory effect can be caused by the charge controller itself, in that it monitors the charge going in and out of the battery and thus knows how much the charge is based on that history. If that gets miscalibrated, it can wrongly detect the state of the battery.

Looks that way.

Look for posts by Peter Renzland (who no doubt will also reply to this) to this group for similar problems. He's got a procedure for recalibrating your charge controller. Incidentally, if you do decide to follow this procedure, please make notes of the state of the battery before and after. There have been questions about how effective the procedure is and Peter is/was looking for data (I'm interested as well, but don't have anything like the stake in the results that he does).

Thanks John. Yes, I'm here.

I had made my batmon program really fancy with speech and warning windows, but that resulted in jams just as important things were happening. So I've taken that out for now.

Daniel, have you ever calibrated your battery?
I recommend calibrating it at minimal discharge (trickle discharge). I suspect that heavy loads can spook the power management system into shutting down the battery prematurely. Batmon has some documentation on what to do. It also makes it fairly easy to keep logs for later analysis. It helps if you are a little adept at Unix. :-)

You can get it at www.renzland.org/batmon.

Let me know if you have any problems.

-- Peter

Message #6 - Posted 2004/04/11 - Carl Price

Previously, Peter Renzland wrote:

John Johnson wrote:

Daniel Karrick Boothe wrote:

discharge from 100% to about 80% and then drop without warning to 0% and go to sleep. Then, when charging again, it charges to about 20% and then shoots up to 100%.

Yes, this is a characteristic pattern.

This is a Li-ion battery and does not suffer from 'memory'. It might (as you point out) suffer from miscalibration of the charge controller.

I have heard that the memory effect can be caused by the charge controller itself, in that it monitors the charge going in and out of the battery and thus knows how much the charge is based on that history. If that gets miscalibrated, it can wrongly detect the state of the battery.

Looks that way.

Look for posts by Peter Renzland (who no doubt will also reply to this) to this group for similar problems. He's got a procedure for recalibrating your charge controller. Incidentally, if you do decide to follow this procedure, please make notes of the state of the battery before and after. There have been questions about how effective the procedure is and Peter is/was looking for data (I'm interested as well, but don't have anything like the stake in the results that he does).

Thanks John. Yes, I'm here.

I had made my batmon program really fancy with speech and warning windows, but that resulted in jams just as important things were happening. So I've taken that out for now.

Daniel, have you ever calibrated your battery?
I recommend calibrating it at minimal discharge (trickle discharge). I suspect that heavy loads can spook the power management system into shutting down the battery prematurely. Batmon has some documentation on what to do. It also makes it fairly easy to keep logs for later analysis. It helps if you are a little adept at Unix. :-)

You can get it at www.renzland.org/batmon.

Let me know if you have any problems.

-- Peter

There has been some wierdness with PB batteries since 10.3. See this site for a run down. My 17PB doesn't seem to want to hold a charge at all.

http://www.macfixit.com/article.php?story=20040405031237646

replace the "dot" with "." to reply

Message #7 - Posted 2004/04/11 - Peter Renzland

Carl Price wrote:

Peter Renzland wrote:

John Johnson wrote:

Daniel Karrick Boothe wrote:

discharge from 100% to about 80% and then drop without warning to 0% and go to sleep. Then, when charging again, it charges to about 20% and then shoots up to 100%.

Yes, this is a characteristic pattern.

http://www.renzland.org/x-charge.png http://www.renzland.org/x-charge-flipped.png

Note that charge pattern and discharge pattern correspond.

It looks to me that there are several factors here:

1. Voltage level (around 12.4) signals that the battery is fully charged. 2. Voltage level (around 10.6) signals that the battery is fully discharged. 3. Calibration estimates capacity by summing consumption over time.

Imagine a car. With a gas tank. This gas tank starts out with a known capacity, but with time and usage, the walls get covered with stuff (like clogged arteries) so that the tank can hold less and less gas. What you really want to know is how far you can drive with the gas you've got in the tank. You can't measure how much gas is in the tank, or what the current capacity is. You only have two hard indicators:

1. When filling the tank, if gas spills out, you know it's full. 2. When driving, if you feel sputtering, the tank is empty.

(compare with 1 and 2 above)

3. There is a fuel flow meter which can be used to estimate the tank's capacity by starting with a full tank, running till empty, and adding up all the fuel flow measurements.

If you have an idea of gas consumption (l/100km or MPG) then you can use the trip odometer to estimate how full the tank is.

I hope the analogy helps. In any case, back to the battery:

First we must realize that the discharge graph produced by X-Charge is only an estimate based on the estimated capacity. There is the *actual* discharge graph, which is different.

*................... "+" is the remaining charge, based on the ..x+................ computer's (incorrect) estimate of the capacity. ....x.+............. "x" is the actual remaining charge. ......x..+.......... (It's possible that the capacity was under- ........x...+....... estmated, but we are not as likely to notice ..........x....+.... that effect, and even less likely to be upset ............x.....+. by it.
..............x...+.
................x.+.
..................*.

Another thing that can happen is that a heavy load will cause a sharp voltage drop that triggers premature coma sleep. This is like getting a sputter when there still is gas in the tank by driving hard.)

From what I can tell, trickle-discharge calibration (sometimes) improves the accuracy of the calibration process.

But it's also important that whenever running on battery, that load be minimized and load surges be avoided.

http://www.renzland.org/batmon

N.B.: this is not a "product", but rather a tool for observing what's going on. It's constantly evolving, and getting to look pretty ugly by now. I should rewrite it in Python ...

There has been some wierdness with PB batteries since 10.3. See this site for a run down. My 17PB doesn't seem to want to hold a charge at all.

http://www.macfixit.com/article.php?story=20040405031237646

One thing that I have noticed with 10.3 is that re-setting of the battery's estimated capacity can happen while discharging, close to the bottom, or while charging, close to the top.

This is new. Before, it only happened after coma-sleep, about 2.5 minutes after being plugged in.

None of this necessarily rejuvenates anyone's battery, but it may help understand what's happening.

-- Peter

Need Help? Have a Question?

Looking for more help, comments, and answers?

Ask your questions on Ask Different. Ask Different is a community of Apple users ready to help.