[mythtv-users] Hauppauge PVR Questions

f-myth-users@media.mit.edu f-myth-users at media.mit.edu
Fri Jan 13 04:38:13 UTC 2006


I got mail back from Bryan about his results in comparing the 150 to
the 250; see the two messages below.

- - - Begin forwarded message - - -

Date: Thu, 12 Jan 2006 10:54:16 -0500
From: Bryan Mayland <bmayland at leoninedev.com>

f-myth-users at media.mit.edu wrote:
>     > Date: Wed, 11 Jan 2006 14:31:36 -0500
>     > From: "Michael T. Dean" <mtdean at thirdcontact.com>
>
>     > The 150 is a better card (it even has a potential to give a better 
>     > picture quality than the 250) and saves you money.
>
> bmayland (CC'ed) started a thread on 15 June 2005 showing dramatic
> differences in sharpness between the two cards, with the 250 winning.
> He also said he'd do some more experimentation; Bryan, did you come
> up any any further results?  (I'm particularly interested in RF-in,
> but I also use composite-in.)

    Yes.  Yes I did.  For reference, here was the image where I compared 
the two:
http://capnbry.net/~bmayland/fi/pvr150/pvr150-250.jpg

The problem in the second image is that the luma and or chroma comb 
error limit was exceeded, which then forces the card to go back to 
regular ole notch mode Y/C separation.  To fix this up, I use cx25840ctl 
to up the CCOMB_ERR_LIMIT from 80 (default) to 120.  Note that my 
PVR-150 is i2c device 1:
echo CCOMB_ERR_LIMIT=120 | cx25840ctl -s 1

The PVR250/350 also defaults to using a +5.5dB luma sharpening filter, 
this can be approximated on the PVR150 with:
echo PEAK_SEL=2 | cx25840ctl -s 1
echo PEAK_EN=1 | cx25840ctl -s 1
I find this to be too much on the 150, so that PEAK_SEL can be adjusted 
from 0 (+2.0dB) to 3 (+6.0dB).

I encourage anyone who wants to look into it to run their own tests to 
see what's best for them.  Settings:
PEAK_SEL=0-3
PEAK_EN=0-1 (default 0) Turn on luma boost (sharpening)
LCOMB_ERR_LIMIT=0-255 (default 20) Luma comb err limit before falling 
back to "complemenatry mode"
CCOMB_ERR_LIMIT=0-255 (default 80) Chroma comb err limit before falling 
back to "notch mode"
LLCOMB_2LN_EN=0-1 (default 1) 2 line luma comb filter enable
LLCOMB_3LN_EN=0-1 (default 1)
CLCOMB_2LN_EN=0-1 (default 1)
CLCOMB_3LN_EN=0-1 (default 1)
COMB_NOTCH_MODE=0-2 (default 1) What to do with notch data (0 = disable, 
1 = notch is chroma, 2 = notch is 50/50 luma/chroma, 3 = notch is luma

- - - Separator between forwarded messages - - -

Date: Thu, 12 Jan 2006 15:01:24 -0500
From: Bryan Mayland <bmayland at leoninedev.com>

f-myth-users at media.mit.edu wrote:
> Do you find that this tweaking, at least for your sources, gives you
> the same quality on the 150 as you were getting on the 250?  

    I kinda go back and forth on this and I'm not sure which I like 
best. Here's a cap from my 150 which I think looks great except for the 
interlacing and the tint is slightly too red:
http://capnbry.net/~bmayland/fi/pvr150/sample-news.png
Deinterlaced:
http://capnbry.net/~bmayland/fi/pvr150/sample-newsdeint.png

> (Though given the audio-level differences people are mentioning, 

    I wrote the volume control for the cx25840 in ivtv so I don't have 
this problem.  Apparently some cards are "to spec" and mine is not, so 
therefore what is right on my card (which I know is not right according 
the the datasheet) is not right on other cards. 

> and the
> current lack of closed-captioning reverse-engineering, I may stock
> up on 250's before they vanish, even if 150's or 500's might be a
> theoretically better idea...)

On the plus side, the 150 does run slightly cooler and draw an 
insignificantly lower amount of power.  :)

> Can I forward your response back to the list?  (I'm assuming the
> answer is "yes", since that's what you were trying to do in the
> first place, but I figured I'd ask just in case...)

    Please do.  I was trying to get it there in the first place, I 
didn't know I had to be a member to post and I haven't had time to go 
sign up just to send one message.

- - - End forwarded message - - -


More information about the mythtv-users mailing list