Sansa Fuze+

[Molot]

I want to buy Fuze+. But I will do that only if one of the following is meet:
1) Rockbox will be available quite soon, at least in an unstable version.
2) There will be any way for me to contribute to making rockbox available.
Sadly, I hardly know anything about microcontrollers. I can debug, can write in C, can test. Can learn coding. Can try to buy particular subversion of the device.
So should I buy it?

[[Saint]]

At this stage I probably still wouldn't buy a Fuze+ in the hope that Rockbox might one day be available for it.
No one can, or ever will, give a projected release time as there are just too many variables.
It could look promising today, and tomorrow a roadblock could be found that completely stalls development.

Also, what is your definition of "quite soon"?

Quite soon, relative to some other ports starting to unstable, could easily be a year or more.


[St.]

[Molot]

I was not expecting any promises. Sometimes a project needs just few minor tricks to get basic functionality - I only wanted to know if it's the case. But I really do not expect it is (sorry if I sounded like I was - language barrier, it seems), so there was point 2 - is this work at a point I might be in any way useful? Or get useful in a month or two?

[pamaury]

From now on it's hard to tell. LCD is working which is good point, USB can easily made to work, the physical keys work. The storage code is standard SD and the on-disk layout is documented so normally it shouldn't be difficult to get it work. The unknowns are still:

• Touch pad: I've started to reverse engineer what I think is the touch pad code but I have no doc and I'm really not sure that it follows the synaptics MEP protocol. Looking at the synpatics website, there build lots of custom modules so this module might even be specific to the Fuze+ !
• Sound: although the SoC includes a DAC and ADC, there might be some technical details so we'll see

The sad point is that I will not have too much free time starting from mid-may until the beginning of June (except for DevCon), after that I'll have much more time I hope.

[pamaury]

I have some news regarding the touchpad code. I think it uses the RMI protocol from Synpatics. On there website one can find two specification: http://www.synaptics.com/developers/manuals
But on the internet there is another difference PDF ! I still can't tell what is the difference (probably more device types). I'll try to write some code to read the interesting registers.

EDIT:
I made some tests and it's indeed a RMI device. It reports itself as "1533", RMI version 0.1 but rmi specification 3 matches.
It has one function: 2-D TouchPad sensor (0x10) with 3 sensors. It's quite late so I did not dumped all the registers yet to see its capabilities.

A funny fact: the fuze+ OF has both a software and hardware implementation of i2c but it uses the software one with the hardware pins! I don't exactly know why but I myself used the generic rockbox i2c software implementation since it's not worth using the hardware in this case.

[pamaury]

When reverse engineering the LCD code for the second kind of lcd, it became clear to me that the two share a pretty close register set. Looking on the internet, I found that Sitronix produces a ST7783 lcd controller, a datasheet for the ST7781 is available and matches. So I guess the two LCD kinds are:

• ID code 7783: ST7781
• ID code 9325: ILI9325

[Molot]

Fast questions:
1) Do you own both types of device?
2) If not, would you get any benefit if I'd buy the other kind and test stuff for you?
3) If so, how can I tell in the shop what kind is it?

Last AND least - which one is better? Both seems to have the same resolution and up to 18bit color, if I found right datasheets.

(Yep, I am willing to help however I can with my very limited abilities)

[pamaury]

I only own one Fuze+ and as suggested by metaphys previously in the thread, this might be a regional choice. Currently we don't even know if there exists Fuze+ shipped with the 9325 kind LCDs, only the OF analysis suggests it Furthermore you probably can't tell the difference except by tearing down the device (and even then it could not be obvious). I think both are equivalent.

[Molot]

Most regional choices I've encountered in electronics was like:
1) USA & Canada
2) European Union
3) Rest of the world
With just minor tricks around.
I'm in the European Union. Maybe it's worth checking... I'll try to investigate. I kinda doubt I'll find out, but no harm in trying.

Addendum:
I have mailed Sandisk technical support and got answer. They assured that there are no regional differences between currently manufactured players, parts used are the same all around the world, especially there are no technical differences between America and European Union products, despite different warranty terms & periods. Spare parts are told to be the same all over the world.
However they did not say a thing about changes in time and I don't know how to ask this to get answer I couldn't just go to them saying "hey, help me to crack your device!", you know.

As I see it now, the only way to test the "time" option is to obtain old and current original firmware and see if there are two drivers in both. If so, then Sandisk is not perfectly honest with us and there was always some kind of variants. If not, then one is just later. In the second case we'll need to see if you have the older version. In that case I might buy new one and hope to have the screen you don't, and test stuff for you.

[pamaury]

I think the safest option is wait and see. The code is there so if it doesn't work people will show up and say "hey it doesn't work!"
On the good side, I've played a bit with the touchpad and it should be matter of a few hours to have something basically working. I let the most motivated read the datasheet but basically the touchpad handle "gestures" itself: tap, double tap, tap and slide, ... So one can let the device do the work and we can just translate those event and map them to virtual keys like left, right, up, down, enter, return and play/pause. Later on, there will be some work to handle slide as a button repeat but the most important is to have the device usable.
The code is in the SVN so if you have a fuze+, feel free to play with! I'll try to implement a nicer debug screen soon. They I'll move on to the storage part.

[metaphys]

Is it the files on the same location as for the LCD's test or is it somewhere else...
Yeah I feel quite stupid on this one... 

[pamaury]

No, I committed the code to the SVN so if it's not the case you need to get the rockbox code using SVN, build the fuze+ bootloader and then use some tools to download the code to the device:
http://www.rockbox.org/wiki/HowToCompile
http://www.rockbox.org/wiki/SansaFuzePlusPort
I'll try to provide a binary if it's too much work but I can't do it on a regular fashion.

[metaphys]

Ok, wait I will try to find my way through those link - if I can't, I will tell... 

[kerobaros]

Currently compiling the toolchain to test this out. Question: does installing your bootloader to the Fuze+ overwrite the Sandisk bootloader? Is there a way to boot the original firmware again afterwards?

[pamaury]

When booting in recovery mode, nothing is written to disk, your device will boot normally afterwards. The recovery mode is not an "installation", it's really just a way to run some binary without installing anything.