[Arm-netbook] EOMA68 laptop battery management.
Luke Kenneth Casson Leighton
lkcl at lkcl.net
Wed Mar 2 12:42:49 GMT 2016
On Wed, Mar 2, 2016 at 12:15 PM, Jakub Kákona <kaklik at mlab.cz> wrote:
> I take look on the bq24193's datasheet and I think it is almost perfect IC
> for intended application.
good to hear. there are some variants btw - bq24192, 24191, 24196 -
it would be good if you could double-check which one is best. the
USB-OTG functionality @ 3A is *critical* because EOMA68's 5V power
rail is *REVERSIBLE*. basically you can either *PROVIDE* 5V power or
the CPU cards can be *PROVIDED WITH* power.
so i am using the USB-OTG rail of the bq2419x series to provide up to
15W of power to the *entire* laptop.
> STC3115 seems to by quite cheap but not the best solution.
> We currently have quite outdated, but still better battery gas gauging
> design based on BQ34Z100.
if you've got something better, _great_.
> Documentation is here
looks great. it's almost the right size as well.
> I think the main work should be the redesign of our LION1CELL01A module with
> bq24193 charger integrated in it. We should do it in KiCAD probably in form
> of new MLAB module (for testing). The exact PCB for Libre laptop should be
> derived from it then.
.... it's almost perfect as-is. if you can keep the bits which
should not go into the laptop on a snap-off area along the bottom of
the PCB then it's pretty much done already.
> Is current blocks/circuit schematic of Libre laptop somewhere?
honestly... you're the first person ever to ask. i'll generate some
schematic PDFs and point you at them shortly. i will also document
the pinouts of the various connectors, as that will be part of the
specification for the Power PCB.
> I do not
> understand how the higher voltages (+5V) or +12 (for LCD) are efficiently
> generated from battery?
the LCD is only 4 watts (despite being 15.6in), backlight is around
3W, and when run at only 30hz the LCD uses under 1W. a single simple
step-up converter to around 17-20v @ 150mA constant current is
therefore perfectly ok, from a 5V supply.
this is *not* a monstrous "let's burn the user's eyes out with a 20W
backlight and dazzle them with 5W worth of 3480x2000 pixels @120fps
needing 5GHZ worth of bandwidth".
so i've used a Silergy SY7201 (or equivalent), it's a SOT23-6, it's
got a power dissipation rating of 0.4W - it does the job.
+5V comes from the USB-OTG function of the bq24193. that can provide
up to 3A, which is perfect, and is exactly what's needed. so there is
no need for a 5V regulator.
the "normal" power output is battery-level (4.2v) and does *not* need
to be stepped up. it goes straight (solely and exclusively) to PCB2 -
the "Embedded Controller" board. this is actually a 3.3v part, so
there is a 3.3v regulator.
basically the EC board - which manages power, keyboard, touchscreen,
and provides RTC functionality etc - is the *only* thing which needs
continuous power, so that's run off of the bq24193's "main output".
the EC board is where the I2C interface goes [NOT to the EOMA68 CPU
Card!]. so it is the EC's job to perform "on / off" function. the
*EC* controls the CPU Card's power state. the *EC* tells the bq24193
to switch the USB-OTG power to 5V @ 3A. the *EC* listens to the
battery status (from the bq34z100).
basically, this is *not* a "standard laptop", it's an "embedded"
design. so there is no 5V rail, and there is no 12V rail: embedded
power ICs for single-cell applications can therefore be considered.
> Has this design higher efficiency than two balanced
> cells connected in series?
given that the maximum power required is only 15W, which is within the
budget of the bq2419x series, i am more concerned about simplicity of
design (and completing this product) than i am about running
efficiency. a 2-cell design is far more complex to design - apart
from anything, 2 cells would need to be sourced of exactly the right
size, and it took me months to find even the current battery.
later however, once this first laptop is out the door, i will begin a
2nd design which will need something like a 30 to 35 watt budget (more
USB ports, built-in SSD, etc.) and at that point a 2-cell or 3-cell
design is going to be needed. however, the implications for that
design decision are MASSIVE. the casework - which has already taken
well over a year to complete - would need a total overhaul.
More information about the arm-netbook