[sane-devel] Questions about the genesys backend and the corresponding frontends (Wolfson Microelectronics)

[Pierre Willenbrock]

Hello

Jean-Baka Domelevo-Entfellner schrieb:
> Following on my project to include support for the OpticFilm 7200 (by
> Plustek) to the genesys backend, I have some questions to the
> maintainer(s) of the latter:
> 
> 1) Are the AFEs taken into account for the moment all made by Wolfson
> ? The Genesys_Frontend stuctures are different for the UMAX, ST12,
> HP2400c, etc. I wonder why.

Currently, the only tested AFE for the gl841 part is the Wolfson WM8199.
The code for AFEs is in no way universally useable. It would be better
to add function tables for the AFEs, as done for gl646 chip and gl841 chip.

> 2) Concerning this same structure, I don't understand exactly the
> meaning of the four values put in reg and the three put in reg2.
> Apparently being register contents corresponding to extra registers
> (we already have the three gain registers and the three offset
> registers), it would allow for seven configuration registers for the
> AFE, which looks far too much for me. In my case the AFE is the
> AD9826, and it has only two 8-bit configuration registers.

The WM8199 has three registers for each of gain and offset. Additionally
it has six registers for various configuration settings, registers 1-3,
6, 8 and 9. Some AFEs use register 0.

> 3) At the same point, in "sign", correct me if this is not the sign
> bit (bit#8, i.e. the ninth of the field) of the offset for each of the
> three channels.

Some AFEs are able to invert their input channels.

> 4) There seem to be some mistakes (or things I don't understand :-))
> in the addresses given to the sanei_genesys_fe_write_data function, in
> genesys.c and genesys_gl841.c. For instance, for the red gain, I can
> see the address 0x28, which doesn't seem to match the Wolfson WM8192
> chip documentation (please indicate which is the exact AFE used, but I
> think the register addresses are pretty much the same for all WM
> chips).

I looked at my documentation and found register 0x28 to be the red
channel gain, for WM8199 and WM8192. The relevant info is this:

address description
101000  PGA Gain (Red)

hidden on page 23 for WM8192 and page 24 for WM8199.
101000(binary)==0x28(hex).

> 5) As I could have other questions in the near future, please allow me
> to write directly to you (dear maintainer(s)) in the future, cause it
> could become boring for the other readers of the list. Problem is I
> don't know who is in the current maintainer's team concerning the
> genesys backend (Pierre ? Stéphane ? Gerhard ? Henning ? Maybe I will
> do some kind of temporary restricted ml...)

You are welcome to write directly to me.

Regards,
  Pierre