How to connect Wifi to Samsung's S3C2410? Re: Congratulations to you and FIC

Robert Michel openmoko at
Wed Nov 15 18:19:46 CET 2006

Salve Giovanni, *!

Your hint with the Marvell chip 88W8388 from the OLPC project is very

On Tue, 14 Nov 2006, Giovanni Corriga wrote:
> I took my references from Jim Getty's blog:
> Marvell is not in a position to open their wireless firmware as it is
> currently dependent on the third party operating system kernel that they
> do not own. A GPL Linux device driver for the Marvell wireless chip, the
> Libertas driver, still under development but also fully functional can
> be found in ourGIT tree.

But the 88W8388 has an own ARM7 CPU and several serial IO so it could be
added like the GSM part with a clean interface:
USB 2.0, SDIO, and SPI 
has some information about the 88W8388:

Marvell 88W8388-BDK1 
 88W8388-BDK1  AGW1P .2  0530 A2P  TW
The 88W8388 is an 802.11a/b/g WLAN SoC from Marvell. Embedded are an
ARM946E-S ( core (as per
the ARM Networking ( brochure)
and on-chip memory. Together with an integrated TCP/IP stack, this allows 
for off-loading the host processor of protocol processing. This SoC is 
 targeted at small-footprint devices, eg. cellular/VoIP phones, PDAs,
video game consoles, etc.
 The 88W8388 is 88W88305-derived, as indicated below (table reproduced

88W8388 a/b/g
Package option: TQFP (No TBGA) 132pin 8x8x1
CLK Type: 19.2, 20, 26, 38.4, 40, 44 COMS & Low swing sine wave
Interface: SDIO (SPI) USB 2.0 
Feature/Advantage: 8385 + TCP/IP termination NAND Flash I/F Audio Codec I/F

On an french page I found an enduser price for a 88W8388 of 20 Euro.

Is the 88W8388 on this photo:
at the right top corner close to the two wifi conerctors,
in the electrostatic shield rectangle, so with many pins
and solderable with a quiet hand? ;)

> We are having open firmware for the Marvell wireless chip developed by
> Meraki. I don???t know yet what license that code will be released under,
> though would expect it would likely be one or more of the MIT, LGPL or
> GPL licenses; but we???ll have to think through the usage cases and needs
> of the communities involved before we can make that choice.
> This new firmware will be distributable by anyone to anyone in source or
> binary form. The existing closed firmware blob will be similarly
> redistributable as soon as we finish working with Marvell???s lawyers to
> get the right language on the license for it.

That would be great e.g. to optimize powerconsumption of the 88W8388 and
to have a more trustable WiFi solution ;)

But I see some problem how to connect Wifi to the S3C2410.
(Maybe only because I'm not electrical engineer or a hardware

### External Wifi for Neo1973 ###
By thinking about to connect an external USB Wifi adapter to the NEO1973
I would expect that the protocol overhead of USB will slow Wifi
802.11b 11 Mbit/s netto 50 % brutto
USB 1.1 12 Mbit/s netto xy % brutto

brutto: USB 1.1 + 802.11b << 5.5 Mbit/s 

So comparing with USB 2.0 with Enhanced Data Rate (EDR): 2,1 Mbit/s
brutto not so much more - 
not the speed, the advantage to use Wifi would be
- option to use wifi accesspoints directly
- greater range
- higher power consumption.

### Internal Wifi for Neo1973 ###
The Samsung S3C2410 has no USB 2.0 like the OLPC mainboard
end even next generation SoC of the S3C2410 family has only
USB 2.0 for a device mode (S3C2443)- not for a host mode
- together with the disadvantage of a different pin layout: 
SPI is to slow - isn't it?

To have an additional USB 2.0 controler would be expensive,
complicated and power consumpting?

SDIO = SD Host/MMC (Multi Media Card) Interface?

So upgrade the Flash Memory up to the maximum of 1 GB
and give up the SD port for the benefit of FAST Wifi?

Other easy posibilities that could used for a
dual use ciruit board layout (mass market mobile without
and Neo1973 with Wifi)?

As I already mentioned - Wlan router with USB 2.0 like the Asus
WL-500g deluxe have very disapointing USB transpher rates - not
the broadcom System on a chip itself has USB 2.0 - there is an
addiotional controler and this is no real USB 2.0.

When there is a internal Wi-Fi solution for the Neo1973
IMHO it should be a real fast one and not one with a
disapointing bottle neck.

I think the potential of the S3C2410 worth a deeper look:

So again, I'm (still) no hardware hacker/developer, but I'm
shure that an open Linux smartphone becomes more populare
and powerfull for hackers and developers when as much 
interfaces as possible of the SoC S3C2410 get open 
solderpoints on the Neo1973 board ;)

-10-bit A/D
and connectors for memory upgrade hacks ;)

Hope that my thoughts could inspire hardware hackers and
the Neo1973 team  - and maybe someone with more skills
could get more out from the potential of the S3C2410.
Just some free lines would avoid to drill the board
to get to the great IO ports of the S3C2410 :)))
Or when no additional connectors would be possible,
when the ciruit board would be leave free of lines over
interesting connection balls of the S3C2410 - 
hardcore hacks could drill the board to get to this
interesting connection... :)))

Remember the power, the potential of the S3C2410,

smartphone = mobile PC + GSM/GPRS

it would be realy a pitty to leave so much communication power
of the S3C2410 unused 
- because of a traditional phone board design


More information about the community mailing list