BTest-4 Release Notes
The Meaning of BTest-4
There are typically one or two internal builds before an external beta test; the BTest-2 build, which was released externally, was earlier in the development cycle than usual and was correspondingly more unstable. However, we got exceeding useful feedback from the community from this early hardware release.
BTest-4 hardware is very close to production hardware. BTest-4 can easily be distinguished from previous beta machines because the plastic around the handle has been textured and the bezel around the screen is gray (this is to enhance use in bright sunlight).
There will be a small CTest build in mid-August 2007 to verify the production "recipe", but these machines will not see wide-scale distribution.
BTest-4 systems have one gigabyte of NAND flash and 256M of RAM. They use the substantially faster Geode LX processor, with a larger cache and better graphics processor. Interpreted languages such as Python in particular benefit greatly from the larger cache.
In some SKUs we are using an alternate battery chemistry: a lithium-based battery chemistry (LiFePo) has become available that has none of the safety or environmental problems of Lithium Ion, is less costly than NiMH (nickel prices have gone through the roof in the last 10 months), and can charge at much higher temperatures (important for use in some parts of the world that can be so hot as to exceed the temperature at which NiMH can be charged). The LiFePo batteries can be identified by large labels "Battery Sample" on the battery packs and are significantly lighter than NiMH. LiFePo batteries have somewhat higher capacity than NiMH.
The hardware specification is elsewhere in the wiki.
Software Release Notes
Library Release Notes
Hardware Release Notes
- The Geode LX: much higher performance (both CPU and graphics)
- The microphone has been moved to the left side of the machine; it is higher up on the bezel, further from the speaker. At least one of the vendors of microphones has been disqualified due to poor quality. We believe the microphone should work significantly better now.
- There are LEDs above both the camera and the microphone to indicate activity. If the camera is powered up, the LED will light (dimly from beneath the bezel). Similarly, if the microphone voltage is enabled (needed for the internal microphone and some external microphones), the LED will light. These are hardwired and cannot be overridden by software, so that children always know if the camera or microphone are active as a means of protecting their privacy.
- The audio amplifier's low-frequency performance has been deliberately rolled off to so that low-frequency sound components will not drive the speakers into distortion at frequencies the speakers cannot reproduce.
- The base was significantly strengthened, increasing mechanical robustness around the battery compartment, which had been the weak point of the design.
- The keyboard lights did not work well and added cost. We removed them. We have ideas on how to uniformly illuminate the keyboard cheaply in future generations but these ideas were too late for this generation of the OLPC system.
The BTest-4 hardware has been extensively tested to ensure that it works correctly in all modes for suspend, resume, and power management. A very few changes (resistor values on the LCD chain) and similar minor items will be changed for CTest and production that will slightly reduce overall power consumption.
A number of software issues remain open. Please note that the software does not yet have suspend on idle implemented; there remain numerous places where we will improve power savings.
Our master list of hardware and mechanical issues and status can be found in our trac system. Please check this list before entering new issues, but please do enter any issues you discover. The items below are the most significant from the list.
- Plugging in an XO with a low battery causes immediate poweroff and EC lockup
- This will be fixed in CTest.
- Power management
- Suspend and resume are mostly functional. A number bugs still adversely affect power consumption during suspend; these are actively being fixed. The two most irritating are: (1) the wireless firmware may not recover on resume; and (2) the screen may not come back on after resume. Please note that only BTest-3/4 systems will resume upon keyboard or touchpad touch. BTest-2 systems can be suspended by the power button, but the keyboard and touchpad cannot wake these systems. We do not plan to make suspend/resume work on BTest-1 systems: restarting the Altera FPGA in these early prototypes takes time and we lack the manpower to expend on the small number of BTest-1 systems in the field.
- Wakeup event is repeated continuously
- Ebook mode (aggressively suspending the system with the screen on) is latent in the software already (it can be turned on by touching a file), but awaits resolution of a conflict between the embedded controller and Linux causing continuous scrolling after resume. We should have this fixed soon.
- Tiny clicks from the speaker on resume
- A transient when power is applied on resume causes very quiet clicks from the speaker. This should be fixed in CTest.
- The OLPC Keyboard layouts
- are found in the wiki. Some minor changes have been made since B2.
Fixed Since Previous Beta Builds (1, 2-1, 2-2)
These problems have all been resolved in BTest-4.
- The keyboard may have problems due to a short cable
- The most irritating problem on BTest-2 is the fact that the keyboard may stop working properly, due to a cable 1mm too short and slightly too much insulation. Flexing the base of the system may pull the cable out of its connector. The field repair of reseating the connector (which requires disassembling the base assembly) is not all that difficult, but annoying.
- LiFe batteries incorrectly flash red light
- LiFe batteries are lighter and have "battery sample" on the back; if they flash the red "battery low" LED, removing and reinserting the battery will get the indicator back to correctness. EC code has been fixed.
- The battery charging algorithms have problems
- The old EC code shipped with some units had a bug which prevented startup with low battery and another bug which prevented charging with a low battery, and also high power usage by the EC, which would drain a battery if left unplugged in a matter of a day or two. Updating to current firmware should fix most battery charging problems; please report any further difficulties. Please read this bug report for more information. BTest-2 and later should also be able to "resurrect" batteries that are so discharged that they cannot even be sensed when put into a system by a new trickle charger circuit. Unfortunately, there are a few batteries which cannot be recovered, due to flash memory in the battery being set to the wrong value, so the battery cannot talk to the laptop.
- Inconsistent battery charge state indications
- Fixed Q2B81 and later firmware. The EC can't report proper charging or capacity information and is overly conservative on the amount of power remaining, as the hardware chosen for the battery pack to report capacity is incorrect; this is fixed in the BTest-3 build.
- Keyboard keys occasionally stop working, and sometimes the machine does not reboot.
- (ESD problem with the embedded controller.) This has not been seen on BTest-2 hardware. One of the most irritating and noticeable problems using BTest-1 systems is a static discharge problem that can lockup pins or entirely crash the embedded controller, causing some or all of the keyboard keys or touch pad to stop functioning (requiring removing the battery for a period to allow the EC to reset fully). We believe this has been fixed in BTest-2. We will therefore be very interested if anyone sees this problem in BTest-2.
- Internal Microphone
- The internal microphone is still mispositioned in BTest-2 (too close to the speaker), and build 239 does not have the codec properly configured to avoid feedback. One of the second source microphones is particularly bad. The BTest-3 hardware has the microphone on the other side of the screen.
- Touchpads are too "noisy"
- This is being mitigated with the addition of a Mylar sheet on top of each touchpad. The Mylar sheets improve touchpad performance for B-Test 1 machiness too. You can also recalibrate the touchpad, by touching all four corner keys at the same time (aka four finger salute), and should expect to need to when you change the laptop to and from power. The Mylar sheet looks like a plastic sheet -- you should not remove it, else touchpad accuracy will decrease. BTest-3 touchpads should be auto-recalibrating and have the sheet integrated.
- SD high-speed not working yet
- We now do support high-speed SD, but are not seeing very high transfer rates when in high-speed mode yet. Fixed in build 303.
- Hardware performance is not what it will be
- Fixed in the Q2B81 and later firmware. The B-Test systems run 33% slower when USB networking is in use -- the internal wireless device is connected over USB, so triggers this problem.
- The resistive sensor does not work
- (for use with a stylus for writing), due to a design flaw: This is fixed on BTest-2 systems.
- The arrow keys ("game keys") cannot be distinguished from the arrow keys on the keyboard
- This is fixed on BTest-2 systems.
- The screen is covered by a plastic film by mistake
- This was resolved in BTest-2, and a diffuser coating significantly improves the screen quality.
- Manufacturing data values are in the wrong order
- This is fixed in BTest-2 (and when BTest-1 systems' firmware is first upgraded).
Design Release Notes
Note that the design is frozen for FRS.
The hinge on BTest-4 allows for the screen to be tilted further back.
The BTest-2 laptops do not have textured plastic: BTest-4 has a sample of the texturing on the battery cover plastic.
- Other design issues are also under investigation
- Note that design is always an over constrained problem: it is often not possible to get every thing you would like in a design, though you can usually get most or all of what you value most.