[mythtv-users] Hardware Purchase

[Devin Heitmueller]

On Tue, Apr 5, 2011 at 12:45 PM, Mark Lord <mythtv at rtr.ca> wrote:
> Nobody seems to complain about the opposite kind of misleading reviews.
> Funny, that.

It's a tough situation.  Many people post their experiences with
certain products, both good and bad, but it's often hard to
intelligently discuss such reviews without understanding the
environment and methodology (in particular when most users only have a
single tuner with no basis for comparison).  That said though,
statements like "tuner X performs worse than everything else" is a
little more concrete an assertion.

> I'm also a big convert to silicon tuners -- eg. the excellent XC5000 chip
> in the HVR-950Q, PCTV-800i, LG televisions, and other devices.
>
> Definitely a step up for consistency from hand-tuned can tuners.

I'm rather partial to the xc5000 (in the interest of full disclosure
though, I did a significant amount of Linux driver work on the chip,
including on the 950q and 800i).  That said, what tuner and demod a
given board has is often not a good indicator.  It's easy to screw up
the PCB layout in forms that effect performance (which as users we
have little control over), as well as random users submitting board
profiles and drivers for new devices that work in "ideal"
circumstances but are poorly optimized for other conditions.

Taking the HVR-1600 as an example, that particular device under Linux
had known issues related to ClearQAM reception for several years
(reception was off by over 3dB), even though it was using two fairly
mature drivers (the mxl5005s and s5h1409).  It was only when a
commercial customer did a consulting project with KernelLabs that we
took the time to analyze the issue and found some anomalies related to
that particular hardware design that required some "special register
programming" for both drivers.

And even that took on the order of 25 to 30 hours of
analysis/debugging to find the problem.  For most developers, it just
isn't worth the time to do the fine tuning.

It's all too often that a random user hacks together a board profile
based on the existing chip drivers, gets it to achieve a signal lock,
and then people think the board is crap because it doesn't behave well
in suboptimal tuning conditions.

Devin

-- 
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com