User:DanielDrake/XO Troubleshooting Battery

From OLPC
< User:DanielDrake
Revision as of 19:12, 16 May 2012 by 209.124.100.138 (talk) (Created page with '{{OLPC}}{{Translations}} This is the portion of the XO Troubleshooting Guide for diagnosing problems with the battery and power on an [[Hardware_spec…')
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
  This page is monitored by the OLPC team.
  Please copy/paste "{{Translationlist | xx | origlang=en | translated={{{translated}}}}}" (where xx is ISO 639 language code for your translation) to User:DanielDrake/XO Troubleshooting Battery/translations HowTo [ID# 270759]  +/-  

This is the portion of the XO Troubleshooting Guide for diagnosing problems with the battery and power on an XO laptop.

The Help activity and XO manual for release 8.2.0 has a "Battery charging" chapter [1].

Start Here

What problem are you experiencing with your laptop ?

The laptop won't run using the battery

If the laptop functions fine using a known good test battery, the battery should be replaced.

If the laptop doesn't function from a known good test battery, see Diagnosing Battery Problems.

The battery light is flashing red

If the battery light is flashing red, this usually indicates an error with the battery. Try a full reset. If the problem persists, then the battery should be replaced. ( For more information on the battery led see [2] )

If the replacement battery shows the same flashing red battery charge light, see common causes for lack of battery presence, below.

It is possible to obtain more information about the battery, and possibly repair it. See below.

The battery light is flashing yellow

This indicates the battery is being trickle charged. This is normal but only occurs for a very low battery. Leave the XO plugged in and normal charging will resume after trickle has finished.

Trickle charging occurs when the battery voltage is lower than 5.4V. This can be monitored with the watch-battery firmware command detailed below.

The laptop can't charge the battery

If the laptop can charge a known good test battery, the battery should be replaced.

If the battery is a LiFePO4 battery, with a yellow label, see the LiFePO4 recovery procedure.

If the laptop can't charge the known good test battery, see Diagnosing Battery Problems.

The laptop won't run using the power adapter

If the laptop runs fine using a known good power adapter, the power adapter should be replaced.

If the laptop can't run from a known good power source:

  • Does the battery charge LED indicate the laptop thinks it has power ? With a battery inserted, the battery LED should turn yellow or green within a minute of applying power. In this case, the laptop probably has a startup problem, not a power problem.
  • Otherwise, the motherboard needs replacement.
Suspect components are the fuse F1 (2A) and diode D21 (3A schottky) on the motherboard protecting the DC input. If the fuse was blown, check diode D123 (3A schottky) and diode D118]] (38V zener) to see if they are OK.

The laptop emits a high pitched whine when using the power adapter

The whine is usually the result of lower-than-normal input voltage or current from the external adapter, e.g. charging with the portable PV panel. The whine comes from vibration in the coils in the inductor -- if the whine is extreme, and goes on for a very long time, there's some chance of the insulating varnish on the coils being harmed, but for all practical purposes, the whine is perfectly safe.

Laptop abruptly shuts off before battery's capacity reaches < 10%

The reason for this may be that the Laptop fails to (fully) charge the battery. Please continue here.


Laptop recognizes the battery, reads its charge percent & voltage properly but does not charge it

Another behaviour is that Laptop does not work without power source (despite being battery fully charged). It would go to a state of power off, and when you try to power it on, power & battery leds will toggle continiously while holding power on button pressed. But, for this problem, battery loads sugar and works properly all its functionalities when AC plugged.

  • Solution: This problem fixes by replacing Q32 component, it's a transistor wich drains power to/from battery when charging/discharging.

The battery doesn't charge, yet the PV panel is connected and the sun is shining

In the first generation of XO's, the laptop's power management system would trigger a switch that would interrupt the powercables between the charging plug and the further interior of the XO when the PV panel would generate a spike voltage when the sun e.g. would come from behind the clouds. To close the switch again, you'd have to manually remove the battery for a moment and remount it. You would then hear the coils start zooming again and see the charging light come on again. This issue has been solved by e.g. by a Schottky diode or even more sophisticated PV power management technology that became standard integrated in the XO's.--SvenAERTS 11:36, 1 March 2012 (UTC)

Diagnosing Battery Problems

Charging of the battery, and interpretation of the front panel buttons, is under the control of a separate microprocessor in the laptop, the Embedded Controller (EC). In order to ensure that battery communications are being properly handled, we must first establish that the EC is operating correctly:

Is there sign of visible damage?

First, take a quick look at the battery, especially the black connector which contains the contacts that electrically connect the battery to the laptop.

Dried green crystals have been spotted around the connector on non-functional batteries on some occasions. This is from decomposition of the contact metal due to electrolysis in water. This may be cleaned with baking soda mixed with water, using a toothbrush. Wash your hands well afterwards. This may recover the battery, unless the contacts have become badly corroded.

Did the EC boot?

Follow the Full Reset procedure, and check to see if the battery LED flashes orange momentarily after reapplying one source of power. The flash is very brief, less than half a second, and occurs only once each time the first source of power is applied.

If it didn't flash at all, you probably have a broken motherboard.

If it flashed another color (red or green), then you probably have a problem with the LED or LED driver circuitry.

To check the EC is running properly, try removing the battery and powering up the laptop without it. Replace the battery after the laptop is running.

Does the EC recognize the battery?

When a battery is inserted into the laptop, the EC reads information about that battery from a integrated circuit inside the battery. If the EC does not receive valid information from this read, it will either assume that a battery does not exist, or report an error. The following tests tell if the EC is talking the battery correctly.

Does the battery light respond to changes in external power?

When you apply external power, the battery led should light up in one of four states:

  • solid green - indicates a fully charged battery
  • solid yellow - indicates a battery being charged
  • flashing yellow - indicates a battery being trickle charged
  • solid red - indicates a battery in dire need of charging
  • flashing red - indicates a battery error. See above and a detailed explanation below.

If the battery light does not indicate one of these states (e.g. nothing lights up) then the battery is not being recognized. If you get a solid battery light then the EC has successfully read the battery gauge. If the battery light stays unlit then either the battery is not recognized or the data read has confused the EC. Use one of the following methods to manually check for battery presence to determine the problem.

For more information on the battery led see [3]

Checking for battery presence

The following methods may be used to manually check for battery presence.

Using Linux

Linux has an interface to the laptop's battery subsystem. Its available via sysfs at /sys/class/power_supply/olpc-battery/. The value of 'present' indicates if the EC thinks a battery exists. (A value of '1' indicates the battery is present). To check this value use the 'cat' command in the terminal app or at a vtty:

cat /sys/class/power_supply/olpc-battery/present

Using OpenFirmware

You will need a developer key for the laptop. Open Firmware has command that will talk to the EC and read the current battery info. This command is watch-battery. At the Open Firmware command prompt do:

ok watch-battery

If EC thinks a battery exists you should see a single line that gets updated with various pieces of battery information. If a battery does not exist then that will be indicated with a 'No battery' rather than battery status info.

Using OpenFirmware and batman.fth

You will need a developer key for the laptop. Recent firmware versions have Batman.fth commands that can interrogate the state machine of the EC's 1-wire interface. This allows you to see what's happening on the 1-wire bus. The command for this is see-bstate. If a battery is not detected the numbers printed will cycle between 0,1 and 2. Numbers other than that indicate the EC has detected a battery.

NOTE: batman is incompatible with XO-1.75. Any repairs done with batman must be performed on a XO-1 or XO-1.5 laptop.

Common Causes for lack of battery presence

There are a number of possible causes for lack of battery presence. Here they are, in the order in which they should be checked.

Unplugged cable

CN21: The battery connector in the bottom part of the laptop

Check to see that connectors CN21 (shown in picture) and CN22 on the motherboard are firmly plugged in. CN21 has a second connector, hidden in the bottom part of the laptop.

Broken Wire

If no batteries are detected, a break or bad connection between CN22 on the motherboard and the data pin on the battery connector results in no 1-wire communication (seen as this problem). Check continuity of the cable with an ohm meter.

add pictures

Corrupt data in battery EEPROM

Normally if the data in the EEPROM has been corrupted a flashing red LED error will result. However, depending on the corruption it may be possible to confuse the EC. Use a known good battery to ensure that the problem is with the battery and not the laptop.

A battery in this condition can be detected by watching the output of see-bstate.

Incorrect speed setting in battery EEPROM

The gauge chip in the battery can support two different 1-wire speed modes, selected by a setting in the EEPROM. OpenFirmware only offers support for the low-speed mode, which is configured by default in our batteries.

If the battery EEPROM becomes corrupted in such a way that the speed setting is switched to high-speed, the system will fail to see the battery. watch-battery and see-bstate will behave as if no battery has been inserted.

This can be corrected with one-wire interface equipment, which is readily available and quite cheap. Some OLPC deployments have obtained such equipment and use it to correct problems such as these. Such devices are able to reprogram the battery EEPROM, removing the corruption and going back to low-speed mode. If purchasing such equipment, be sure to buy a device that supports high-speed more - not all do.

Firmware versions Q2F01 and newer for XO-1, and all firmware versions for XO-1.5 and XO-1.75 try very hard to avoid this condition, by detecting it before the EEPROM is written and enforcing low-speed mode.

If a large number of batteries are encountered in this state in your project, please contact OLPC; there is a possibility that a special EC firmware could be produced to reprogram batteries in this condition.

One-wire Communication Error

The sensor inside the battery could be malfunctioning or the output from the EC could be corrupted from a manufacturing defect. If the laptop cannot communicate with the known good battery, and you have already checked for an unplugged cable or broken wire, then the error is in the laptop motherboard and it will need replacement.

Battery LED is flashing red

Flashing red LED indicates that the battery subsystem has detected an error (for more information on the battery led see Power_management#Battery_LED). Open Firmware has a command that will read the error code from the EC. The command for this is 'ec-abnormal@'. At the Open Firmware prompt use the following to view the error code.

ok ec-abnormal@ .

Note the '.' at the end. This is required to print the error code.

Now run:

ok watch-battery

And hit escape. An error message will be printed. This message is a human-readable interpretation of the error code. See if the error appears below for potential solutions.

Wrong Pack setting (code 02)

Under certain conditions some settings in the battery EEPROM can get corrupted. This error indicates this has happened. It may or may not be permanent. Using batman and bat-dump-banks the contents of the EEPROM can be examined. This is a different condition than what bat-check-and-recover looks for. Please do not use bat-check-and-recover to try and fix this. Contact OLPC techsupport or the developers list for help.

LiFePo4 Overtemp (code 0b)

Battery got too hot. Turn off the system remove the battery and let things cool.

Bank 1 Invalid

This message can be produced when the memory in the gas gauge has become corrupted, causing the battery identification data to be lost. This fix should only be attempted on batteries that are labelled as Made by BYD Company Limited on the label printed on the internal-facing side of the battery.

This problem can be fixed with batman. Load batman, according to the instructions on this page, then attempt to fix the problem with:

ok batman-start
ok bat-set-life-byd
ok batman-stop

Now remove and insert the battery.

Note that these same set of symptoms can also be indicative of other non-ID problems, so this may not solve your problem. If the problem remains after attempting this fix, you will need to examine the battery banks and contact OLPC tech support.

Bank 0 Invalid

First, using a serial console, save the output from the batman command:

ok bat-dump-banks

This may be useful later. Plus, OLPC would appreciate seeing this information posted to the devel mailing list in cases where the repair worked.

There could be many causes of this error. One case found in Nicaragua had all bytes in bank 0 shifted left by one byte. The EC was then confused when reading a strange value from the status byte.

This was successfully solved with:

ok bat-fix-error-2

It should be safe to try this repair technique for any case of Bank 0 Invalid, but it may not be able to solve the particular problem you are seeing. After running the command, perform a full reset (disconnect AC power, remove and reinsert battery).

If that didn't fix your problem, you can try a more invasive Bank 0 reset. This is only valid for XO-1 and XO-1.5 batteries at this time, don't try it on a battery that originally arrived in a XO-1.75 laptop.

ok bat-reset-bank0

After running the command, perform a full reset (disconnect AC power, remove and reinsert battery). If the error condition cleared, the system will now be confused regarding the percentage of battery power remaining - you need to complete a full charge or a full discharge before things regain synchronization.

Battery fails to charge?

Checking the actual charge level

olpc-pwr-log is a Linux bash script that allows you to measure the amount of charge delivered to or extracted from the battery. It can be used to determine if the battery is performing at its rated capacity.

All-zero values reported by watch-battery

If the XO recognises the battery but fails to charge it, under the conditions that watch-battery reports 0V 0A 0C and trickle charge, the first thing to do is wait a good while (1 hour, to be safe). Sometimes these numbers stay at 0 for a good few minutes before the battery comes back to life.

If the numbers stay at 0, it may be worth testing the battery with a multimeter. The battery has three terminals that connect to the laptop, you need to measure the voltage across the middle terminal and the one closest to the center of the battery. If your multimeter probes are too big to enter the terminals, find a couple of thin paperclips (or another type of thin wire) and feed it into the terminals, then you can connect the multimeter to the clip.

For a good battery with some charge, expect a reading between 1V and 7V. If the multimeter reports 0V here, it suggests that there is a bad connection between the cells and the battery connector, or that the protection circuit has permanently failed. Both cases would indicate that the battery is permanently damaged and cannot be recovered.

Tips

Estimates

A battery is a chemical device whose characteristics change with age, history, and ambient temperature. The charge amount shown in the frame is only an estimate. Do not worry if your battery shows (for example) 96% charged, or does not reach 100% charge.

Minimizing battery depletion

Even when powered off (by choosing Shutdown from the Home view), the XO has slight power draw, losing about 2% charge a day. If you won't be using the XO for a while you can avoid this by removing the battery from the XO after powering off.

Software Tools

olpc-pwr-log

Currently this tool (a Linux script) is located at /usr/bin/olpc-pwr-log, source at http://dev.laptop.org/git/projects/olpc-utils/tree/usr/bin/olpc-pwr-log

To use olpc-pwr-log follow the following procedure:

  1. Enter olpc-pwr-log in a Terminal.

The script will wait until it detects a battery and then begin to log various parameters via the olpc-battery subsystem. If your battery is really low and the EC is using trickle charge rather than main charge then the script will appear to be still looking for a battery even though the yellow charge light is on. Once the EC finishes trickle charging it will report that a valid battery is present and the script will continue.

The output of the script is both to the screen and to a file. The file name is 'pwr-<date>-<gauge ID>.csv'. Gauge ID is a unique ID number that is contained in the gas gauge of each battery.

More details are in XO power draw.

batman.fth

Note: The q2e32 OLPC firmware pulls in some of the commands from batman.fth

batman.fth is a forth script that is available from http://dev.laptop.org/pub/firmware/scripts/batman.fth. To use it you must load it at an Open Firmware (OFW) prompt. Getting to an OFW prompt requires a developer key. To use:

  1. Place batman.fth on a USB key
  2. Insert the USB key into the laptop XO (while it is turned off)
  3. Start the laptop
  4. Enter Open Firmware by pushing the ESC key while the laptop is starting up.
  5. At the 'ok' prompt, type:
fload u:\batman.fth

The batman script may be stored on devices other than a USB key. It is loaded with the 'fload' command, which must specify the storage device where bathman.fth is located:

USB disk:
 fload usb:\batman.fth
SD card:
 fload sd:\batman.fth
NAND flash (root directory):
 fload nand:\batman.fth

Once loaded you run the commands in the script just like any other OpenFirmware command. If you turn off the laptop, you have to repeat the loading process.

Key tools in batman.fth

The following is a list of useful tools in batman.fth with a brief note how to use them. Some of these commands require that the EC be put into reset prior to running them so that the EC battery management system is disabled. The 1w-init command does this. After the 1w-init command is used all keyboard, touchpad and button activity is disabled. Including the power button. Any command requiring a 1w-init will require that any further commands be invoked using the processor serial port. To power off the laptop afterwards, you will have to remove both external power and the battery. (note: if you came here from LiFePO4 Recovery Procedure you only need to run the bat-recover command)

  • batman-start
Put the EC in reset and prepare the system so that Open Firmware can talk directly to the gas gauge chip in the battery. Newer firmware only suspends the battery interface and leaves the keyboard working. Required prior to some of the other commands.
ok batman-start
  • batman-stop
undoes what batman-start did.
ok batman-stop
  • see-bstate
Continually prints the EC 1-wire communication state machine index. Useful for determining if the EC can talk to the battery gas gauge.
ok see-bstate
  • bat-recover
Does PWM on the charge enable line resulting in a psuedo-current limited charge of the battery. Streams the battery temperature, currrent, voltage and accumulated charge register (ACR) to the screen.
Automatically runs batman-start.
Remove the battery to cause bat-recover to exit (with an error message).
ok bat-recover
  • bat-set-low
Zeros out various values in the battery's gas gauge EEPROM forcing the EC to resync with the state of the battery. Useful if your battery is marked as full when it really is not.
Needs batman-start.
ok batman-start bat-set-low

Linux battery subsystem

More information needed here