[Arm-netbook] RK3399

Pičugins Arsenijs crimier at yandex.ru
Mon Jan 22 16:47:02 GMT 2018

> I agree on that. For now I have spent 50usd on getting
> an asus 7inch eeepc. I like the size of it. It has a full
> keyboard. It is bulky. Therefore more likely to have space
> for building in another mainboard.
> ...
> Then lkcl has this argument about reverse
> engineering a lot of computer devices. I do
> not understand it. Apart from the display all devices
> should be usb. Sound, touchpad, keyboard, ethernet,
> wifi, storage. Display should be hdmi. Rather I believe
> getting a battery and powersupply solution would be
> difficult.

I have some EEE PCs (2 working and 2 spares), and I
absolutely love them, I spent lots of time growing up
with them, and learned to properly use command-line
Linux while I was at it. My question - would you be
interested in re-making it for EOMA-68 or something similar?
I definitely am. EEE 701 has been thoroughly
reverse-engineered, there are schematics available,
as well as lots of interesting replacement parts. So, my
plan: we'll need to re-place the motherboard, with
something much more simple and much less
power-guzzling. We'll also need to mod the chassis
a little bit (to make an opening for an EOMA68 card).
That's the extent of what we'll need to do - we won't
need to replace anything in the upper half, and we'll
be able to reuse many parts from the lower half.
Here's how it goes: 

* The screen is RGBTTL, which is the same interface
that an EOMA68 card has on its pinout. If my
understanding is correct, it should be trivial to interface
to. LVDS is used internally (for decreasing EMI, if I
understand it correctly), and the upper half has a small
board with LVDS2TTL IC - SN75LVDS86A. So,
interfacing that to EOMA68 (specifically, the hardware
side) would simply require using a TTL2LVDS IC -
something like SN75LVDS84A, to convert EOMA's
RGBTTL into LVDS. We'll be able to re-use the
original LCD cable that way, too!
* The LCD panel also has I2C, which is likely used for
just interfacing to an EEPROM containing panel's
resolution. If it's not only an EEPROM and there's some
kind of control data going over there, it is likely sniffable
using a 5$ logic analyzer (using all OSS software, too!).
* I've got no idea if there's any kind of initialization
sequence sent over TTL, but I think it's unlikely. If so,
we might still be able to sniff it with a more advanced
analyzer, or try to get a panel datasheet&init sequence
somewhere (it is a popular panel, after all).

So, we should be able to re-use the screen and all the
surrounding parts - win!

* Connector pinout is known, and desoldering the
connector from the board is tricky but not too hard.
* It uses a fuel gauge chip accessible over I2C, indeed.
Thankfully, as I have two working EEEs (and three
batteries), we should be able to sniff enough
communication to determine which commands correspond
to which data. Alternatively, it's very likely the fuel gauge's
datasheet is available somewhere.
* Charging the battery and powering the system will be a
problem, indeed. Fortunately, the EEE's solution to this
is documented, and we could, at the very least, learn from
it - it's going to be an interesting journey for me, but nothing
that's not been done by somebody else already.

So, we should be able to re-use the battery - another big win!

Other things we can re-use:
* Camera is USB indeed, dead simple to interface.
* Touchpad is PS/2, but that's not a big problem, PS/2 to
USB converters are available cheaply.
* Keyboard is a matrix one - but the pinout is known (the
matrix isn't known though, but shouldn't be hard to figure
out). Again, matrix-to-USB converters are both available
cheaply, and DIY-able easily.
* Speakers and microphone should be easily re-usable.

Parts we won't be able to re-use:
* Original Intel CPU, south and north bridges... We won't
miss them, though =)
* Embedded controller - the part that controls the power
states, power button, LEDs, peripheral power (like,
turning off the camera or the SD reader when they're not
used). We can, and likely will, replace it with a
* SD reader - it's too tightly integrated in the motherboard,
even though it's USB. Desoldering lots of chips & support
passives, as well as the connector, would be tricky, so it'd
be best if we don't go that path. We might be able to use
the SDIO from EOMA68 connector, though (alternatively,
we could use that SDIO for WiFi). Also, we can simply
add a USB SD reader, yay!
* Ethernet - luckily, USB-ethernet chips are cheap and
accessible, so if we need to add Ethernet, we can.
* Audio - either an AC97 solution or a USB one will be
* MiniPCI-E interface - no PCI-E on EOMA, maybe
that's for the best, we don't need it for much (PCB layout
would be tricky, too). Though I have to admit that a full-size
WiFi would be nice to have.
* SATA and IDE - again, godspeed, and we absolutely can
use a converter chip if need arises.

> Lkcl says, should I succeed in building one, only I would
> benefit. I disagree. I would tell others how to make a
> similar computer, should they want to.

Totally agree. Once one does something, others can follow.
The EEE PC was popular enough so that lots of other
people can follow. Broken laptops and replacement parts
are still available, too.

> How to make the keyboard's matrix, I do not know.

Interfacing to a keypad matrix is very easy nowadays.
There are a lot of ready-to-go hardware and software solutions from the DIY keyboard community.

> https://www.cnx-software.com/2016/12/05/firefly-rk3399-rockchip-rk3399-development-board-launched-on-kickstarter-for-139-and-up/
> has a hdmi port. What should be the difficulty in connecting
> the mainboard to a hdmi display?

As a rule of thumb, using HDMI internally, or converting
high-speed interfaces into similar-but-not-quite high-speed
interfaces, comes with an increased power consumption.
You don't really want that - and it seems that EOMA68 should
work without it, allowing us to use the HDMI for an additional
display instead =)

So, here's my summary - it's doable. If you're ready to work
for this, we can have it done in half a year. I'll be more than
happy to help (and I know some people that could help, too).
However, there are a lot of parts that are small and easy, but
time-consuming, and time is something I don't have much of,
so expect there to be plenty of work for you =) Let me know
(maybe email directly) if you're interested and ready to start
working on this project together.


More information about the arm-netbook mailing list