[linux-dvb] Re: mt352 and SkyStar2 (FlexCop IIB) oddities

[Christopher Pascoe <c.pascoe@itee.uq.edu.au>]

> > Attached is a patch against CVS that makes software I2C work, as well as
> > implementing the second policy.
>
> I had a closer look to your patch:
>
> +	unsigned char acqctl = 0x0B;	/* suggest spectral inversion on, force mode,
> guard */
>
> I think it should be set to 0x5B. If the FE looses the lock it should recover.
> Maybe one doesn't catch some wrong parameters but I think recovery is more
> importent.

I tested locally with 0x0B and all parameters explicitly set, unplugging
the antenna, introducing noise, etc, because the data sheet doesn't
explicitly state what happens in this case.  The mt352 regains lock and
continues decoding the stream just fine.

If you set it to 0x5B in the case when you *do* know all the parameters,
the chip will commence blind acquisition at loss of lock, and will
increase the time that it takes to lock again while it performs the search
(even though it will find the same settings again).

>  	switch (op->code_rate_HP/LP/constellation/hierarchy) {
> ..
>  		case FEC_AUTO:
> +			acqctl |= (15 << 4);	/* all parameters auto */
>
> Hmm, I think (5 << 4) is a better setting. But setting acqctl to 0x5B makes
> this unnecessary.

Oops, I diffed against an old version of my code.  My current code does
have (5 << 4) here, but the rest remains as is.

Regards
Chris
-- 
Christopher Pascoe
IT Infrastructure Manager
School of Information Technology and Electrical Engineering
The University of Queensland   Brisbane  QLD  4072  Australia
Web: http://www.itee.uq.edu.au/~chrisp/