Category Archives: Boxee

An overscan fix for the Sharp LC-42SB45U television set when connected to a computer with a Linux operating system (Ubuntu, etc.)


This is an edited version of a post that originally appeared on a blog called The Michigan Telephone Blog, which in turn was reposted with the permission of the original author from a now-defunct Macintosh-oriented blog. It is reposted with his permission. Comments dated before the year 2013 were originally posted to The Michigan Telephone Blog.

If you bought a Sharp LC-42SB45U TV, perhaps because it was on a super great (and very lightly advertised) deal at Wal-Mart back in November, and then later tried to hook up a home theater PC to it, you may have been disappointed to discover that unlike most flat screen digital TV’s it doesn’t have a “pixel-to-pixel” or similar 1:1 pixel mapping mode. The result is that when you hook up a computer to one of the HDMI ports, there is a serious overscan problem — for example, if you are running Ubuntu Linux (or some other version of Linux) you won’t see the top or bottom menu bars, because they are outside the visible screen area. If you use XBMC or Boxee, you can go into that program’s settings menu and apply overscan correction from within the program, but most other programs and video players don’t offer an overscan correction option.

The problem is not that there’s no “Dot by Dot” setting in the Sharp TV — it’s just that it’s a (very) hidden option, and as far as I know, there is nothing you can do using the buttons on the TV or on the remote to make it appear (I’d be very happy to be proven wrong on this point; if there is some sort of hidden remote control key sequence that can make the Dot by Dot option always appear, I wish someone would spill the beans so we can fix this issue the right way). But with one small tweak in a Linux configuration file, you can make it appear, like so:

Sharp LC-42SB45U TV showing Dot by Dot option

My first approach to this came at a cost: I read that if you could send the Sharp a non-standard vertical sync frequency (refresh rate) a bit below the normal 60 Hz, the alternate View Mode would appear.  That did work, and in my non-scientific testing, I found that 59.55 Hz was about the cutoff point.  Anything above that, and you get the normal menu of View Mode options when you press the View Mode button on the remote.  Anything at about that or below,  and you get the View Mode options menu shown above. However, this was certainly less than ideal because of the non-standard refresh rate. I got started on that path after reading a forum post that suggested a custom ModeLine in your /etc/X11/xorg.conf file to give you a 1816×1026 display.  While this will work to fix the overscan, it also cuts down on the pixels available to programs, and makes things not quite as sharp (no pun intended) as they should be.

Now, the idea of using a custom ModeLine in your /etc/X11/xorg.conf file is not a bad idea, and the above-referenced post did contain some good information  (especially about disabling some unwanted Ubuntu packages that might cause your xorg.conf to be ignored).  So I tried the xorg.conf shown in that post, except I used the original ModeLine shown (which is correct for the Sharp LC-42SB45U as long as you don’t mind the overscan).  I then read in another forum post (on a different site) that someone had found that the Dot by Dot option would appear if the refresh rate were set to 59 Hz rather than 60 Hz.  However they were doing that on a Windows machine, not a Linux box, if I recall correctly.

But again, that had the disadvantage of a non-standard refresh rate.  I’ve read on several sites that the ideal refresh rate is 59.94 Hz (it’s very close to 60 Hz and is exactly twice the ATSC 1920×1080 progressive scan frequency of 29.97 Hz) so my goal was to get as close to that as possible. I then read that someone had actually accomplished this on a Windows box by changing the timing to something called “CVT reduced blank” (the procedure on a Windows box is to bring up the NVIDIA Control Panel, then click on Change Resolution, then Add Resolution, then Create Custom Resolution, then in the “Timing” section find the “Standard” drop-down box and select CVT reduced blank. Make sure the other settings look sane, click the Test button and go from there. Mac OS X users can do something similar using a program called SwitchResX — see Brian Semiglia’s comment in the Comments section for a link to instructions. The reason this doesn’t work under Linux is that the Linux version of the NVIDIA Control Panel doesn’t offer this level of functionality, and also, some might encounter this issue even if not using NVIDIA graphics). So my goal was to find a ModeLine that would do the CVT reduced blank but not use a non-standard screen size nor refresh rate. After searching the web, playing around with an online Calculator for video timings which I saved to a local drive and then hacked a bit to display four decimal points of precision on some key values, and generally spending more time than I intended, I came up with a working ModeLine.

First, let’s look at the original 1920×1080 ModeLine from the above-linked forum post:

ModeLine "1920x1080" 148.50 1920 2008 2052 2200 1080 1084 1089 1125 +hsync +vsync

If you change the pixel clock frequency value in a ModeLine (the 148.50 in the line shown above) you change the refresh rate, and if you change certain other values you change the other timings.  I cheated a bit and used Google to search for a working ModeLine that provided 1920×1080 at 59.94 progressive scan, and found one that was very close (59.93, actually) so I tweaked the refresh to give me exactly 59.94.  This is the final ModeLine I came up with:

Modeline "1920x1080" 138.5141 1920 1968 2000 2080 1080 1083 1088 1111 +hsync +vsync

Okay, so you may think it ridiculous to specify the pixel clock frequency out to four decimal places, but hey, it works! So, this is what I’m now using for an xorg.conf file (by the way, if any of the ModeLines in this article are truncated on your display, just keep in mind that the last two values in each line are +hsync +vsync — if you copy and paste any of the long ModeLines, hopefully you’ll get the complete line).  Bear in mind that I’m using this with an Acer Aspire Revo, so some of these lines are specific to the NVIDIA graphics chipset, but the principle of changing the ModeLine probably should work with this model Sharp TV even if some other graphics chipset is used on the computer:

# nvidia-xconfig: X configuration file generated by nvidia-xconfig
# nvidia-xconfig:  version 1.0  (buildmeister@builder75)  Tue Dec  8 21:04:28 PST 2009

Section "ServerLayout"
    Identifier     "Layout0"
    Screen      0  "Screen0"
    InputDevice    "Keyboard0" "CoreKeyboard"
    InputDevice    "Mouse0" "CorePointer"

Section "Files"

Section "InputDevice"
    # generated from default
    Identifier     "Mouse0"
    Driver         "mouse"
    Option         "Protocol" "auto"
    Option         "Device" "/dev/psaux"
    Option         "Emulate3Buttons" "no"
    Option         "ZAxisMapping" "4 5"

Section "InputDevice"
    # generated from default
    Identifier     "Keyboard0"
    Driver         "kbd"

Section "Monitor"
    Identifier     "Monitor0"
    VendorName     "Unknown"
    ModelName      "Unknown"
    HorizSync       15.0 - 75.0
    VertRefresh     55.0 - 76.0
    ModeLine       "1920x1080" 138.5141 1920 1968 2000 2080 1080 1083 1088 1111 +hsync +vsync
    Option         "ExactModeTimingsDVI" "TRUE"
    Option         "DPMS"

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    Option         "ModeValidation" "NoEdidModes"
    Option         "HWCursor" "false"
    Option         "DynamicTwinView" "false"

Section "Screen"
    Identifier     "Screen0"
    Device         "Device0"
    Monitor        "Monitor0"
    DefaultDepth    24
    SubSection     "Display"
        Modes      "1920x1080"
        Depth       24

Section "Extensions"
     Option         "Composite" "Disable"

This seems to work well on a Acer Aspire Revo running Ubuntu Karmic Koala (EDIT: and I’ve also used it under Maverick Meerkat), though I imagine it would work with other Linux distributions that use an xorg.conf file (including XBMC Live), however as far as I know this trick only works with the Sharp LC-42SB45U TV and no other model.  With this xorg.conf I don’t have to tweak the overscan settings in XBMC or Boxee at all. It works for me, but it may or may not work for you.  Standard disclaimers apply – I’m not telling you to do this on your setup, and if you break something, you own all the pieces, but from me you’ll get nothing more than perhaps a bit of sympathy. Don’t even think of doing this if you are not willing to assume any and all risks.

EDIT: If you don’t want to go through all the hassle I went through to calculate the correct ModeLine, you can run the cvt program with the -r option from the Linux command prompt, like this:

cvt -r 1920 1080

That’s for a 1920 x 1080 display. On my system this generated the following output:

# 1920x1080 59.93 Hz (CVT 2.07M9-R) hsync: 66.59 kHz; pclk: 138.50 MHz
Modeline "1920x1080R"  138.50  1920 1968 2000 2080  1080 1083 1088 1111 +hsync -vsync

You’ll notice this is nearly identical to the ModeLine I generated (the vsync is the opposite, though — don’t know if that would be an issue). What I’ve read is that you paste the generated ModeLine into your xorg.conf file and make sure you also have the line

Option         "ExactModeTimingsDVI" "TRUE"

in your xorg.conf (to force it to use your generated ModeLine) and that may be all you need. Certainly simpler than how I did it, but I didn’t know about the cvt program. (End EDIT).

By the way, if you want to hack that Calculator for video timings, just save the HTML page to your local hard drive, open it in a text editor and look for this section (it’s very close to the top):

function TwoDecimal(number) {
 return number;

Change that second line to


Then load the page into your favorite browser (with JavaScript enabled). That will display a couple extra decimal points on some of the critical values.

BETA Perl script for Caller ID popups when using Linksys/Sipura devices


This is an edited version of a post that originally appeared on a blog called The Michigan Telephone Blog, which was written by a friend before he decided to stop blogging. It is reposted with his permission. Comments dated before the year 2013 were originally posted to his blog.

Creative Commons License photo credit: bcostin

PLEASE NOTE: This  article has been updated as of December 30, 2008.  This now works with a Mac or Win32 computer (and Linux computers with libnotify installed or readily available, such as those running Ubuntu) and has been updated to reflect that fact. Also, please note that previous versions may have failed on devices/phones with more than two lines – this is (hopefully) fixed as of version 0.7.

If all of the following are true:

You have a Macintosh computer with OS X installed, or a PC with any 32-bit version of Windows installed (basically Windows ’98 through XP), or any version of Linux with libnotify installed

Growl icon

Image via Wikipedia

You have Growl (if you have a Mac) or Snarl (if you have a PC) notifications installed (EDIT: There is now a version of Growl for Windows but at the moment I only have an experimental version of the script for that – see bottom of this page for more information.  It MIGHT work with 32-bit OR 64-bit Windows 7 – feel free to test it).

You have a Linksys or Sipura VoIP adapter on your local subnet or home network and receive calls over it

You would like to see Growl, Snarl or libnotify popups on your computer when a call comes in, showing the caller’s name and number, along with the line that the call came in on and the time and date the call arrived (in case you are out when the call comes in)

You have previously run Perl scripts on your computer, OR are reasonably good at following instructions and problem-solving

AND you are willing to run a script that comes with NO WARRANTY whatsoever (if it breaks, you can keep all the pieces)

Then download this file (now at version 0.92), unzip it and read the Instructions.txt file in the folder appropriate to your computer.

This script is being offered under the GNU General Public License, so if you want to modify it to work on other platforms, you can do that under certain conditions (see the Instructions.txt file for details). Mainly, I’d hope that you’d contribute the modifications back (and please leave a comment on this article if you do that).

I don’t have any kind of regular web page up for this yet, for one thing it’s very rough (very little error-checking) and for another I’m very tired, having spent way too many late nights trying to get this to work. So this post will be more terse than most of my posts, but I think most everything you need to know is in Instructions.txt (and for Mac users, the “How to run at login.rtfd” file) inside the .zip file. Feel free to repost this information to other forums if you think anyone else might be interested.

For those Mac users that wish this were an app: I understand that there is an app called Platypus that allows Perl scripts (and any other types of scripts) to be converted to OS X app bundles. However, what it does not seem to include is any way to specify the command line options, or to load any missing Perl modules. So for now, this script will probably only be usable by those with sufficient knowledge to run a Perl script on their Mac. If I were a bit more knowledgeable, I’d build a preference pane to go in System Preferences, and then have the script read that for its configuration options. But I still have no idea how to make an app install missing Perl modules, particularly when OS X does not come with “make” installed until and unless the Developer Tools are installed (adding something like 3 GB of stuff that is mostly useless to non-developers to your hard drive!).

EDIT: I read somewhere that you can install make without installing the bloated Developer Tools package if you instead install Fink. Then, from a terminal prompt, you can type fink -b install make and supposedly that will do the trick. However, I am told that Fink has not been updated for Snow Leopard, but there is a make package in Rudix that should work with Snow Leopard (mind your paths – Rudix installs make in the /usr/local/bin directory and by default CPAN expects it in /usr/bin, so you may want to adjust the path during CPAN setup, or make a symbolic link in /usr/bin). Since I have not personally tried either of these I have not updated the instructions in the download to reflect this, but if it works you can skip the whole process involved in installing the Developer Tools.

Because this is a Perl script, it lends itself to custom modifications. For example, let’s suppose you have this script running on a Mac, and you are sending Growl notifications to the Mac, but you also have a home theater PC that runs XBMC and/or Boxee, and you’d like to send Caller ID notifications to it as well.  Assuming that Boxee and/or XBMC is configured to allow control via a Web interface, at a fixed IP address and port ( port 8080 in this example), you could add a line such as this to the script (this is all one line; select and copy to get it all if it gets truncated on your display):

eval {get "$displayname%22%2C$phonenum%20calling%20$lineid[$count]%2C15000%2C%2Fhome%2Fusername%2Fphone.png)"};

The above assumes that you have placed the icon file phone.png (shown at right — right click on the icon and save it) in the user home directory on the destination system (the one running XBMC or Boxee), and that you change ‘username’ to the actual name of the user’s home directory. Note that the icon path requires %2F in place of forward slashes (therefore %2Fhome%2Fusername%2Fphone.png really means /home/username/phone.png) Phone icon - right click and copy imageand this refers to the icon directory and filename — if you choose not to use an icon then leave that part out, along with the %2C that comes just before it.  If you are running XBMC or Boxee on the same system that’s running the script then you should be able to replace with localhost or The above line should be inserted just above the comment line “# Make output string in chosen format” near the end of the Perl script. Keep in mind that this won’t work if you don’t enable control via Web server in XBMC or Boxee, and make sure the port number matches the port in your added line.  Depending on the skin you use, this is generally accomplished by going to Settings, then Network (and in Boxee, then Servers). Then check “Allow control of XBMC via HTTP” (in XBMC) or enable the Web server (in Boxee) and verify the port number is correct.

Starting in Version 0.7 there is a minimal logging function, allowing all detected incoming calls (whether answered or not) to be saved to a text file and/or a comma-quote delimited file. I probably could support other simple formats, but don’t even think about asking for anything more complex (like a rather humorous friend of mine who asked for MySQL integration – considering that he knows how little knowledge I have about Perl programming, and that I have even less knowledge about databases, I’m sure he thought it extremely amusing to make that request). The one thing I really don’t like about offering these scripts in Perl is that it requires the user to know how to install modules from CPAN (or an alternative source if using Win32), but I barely know how to do this stuff in Perl and don’t know any other languages (well, except for QBASIC under MSDOS, but that’s even less compatible across platforms than Perl!).

Starting in Version 0.9 you can use a plain-text file of number-name substitutions, so (for example) if calls from a particular number always display a cryptic Caller ID name, you can change them to say “Uncle Bob” (or some other name if Bob’s not your Uncle, or it’s someone else’s number!). Read the sample config file to see the file formats. Note that the plain text file of number-name substitutions is a separate file, not a section of the optional configuration file, and also note that you must enter the numbers exactly as your VoIP provider sends them (in other words, if they send 8005551234 and you use 18005551234 or 800-555-1234 it will NOT match!).

Starting in Version 0.91 you can use a plain-text file of number-path/file substitutions, so (for example) if calls from a particular number are always from Uncle Bob, you can display Uncle Bob’s picture as the icon whenever a call arrives from that number. Read the sample config file for more information. Note that the plain text file of number-path/file substitutions is a separate file, not a section of the optional configuration file, and also note once again that you must enter the numbers exactly as your VoIP provider sends them.

Version 0.92 sets a rather short timeout on page fetches (still much longer than should be necessary to get the data), in an attempt to resolve a problem where very occasionally the script would just go into a coma, not exiting cleanly but still using memory and CPU cycles, without doing anything useful. I have been running this version for over six months now and have yet to see the script go into a coma, as it often seemed to do in previous versions.

(EDIT added September, 2010:) NOTE regarding EXPERIMENTAL version to work with Growl for Windows.  You should still download the main archive to get the instructions and such, but if you’d prefer to use Growl for Windows rather than Snarl, you can try this experimental version of the script. If you do try it, please let me know if it works as expected (and thanks to Andy Singh for his help with getting this working under Windows 7). Please read the Perl source code to find the module requirements (mentioned on or near line 15 of the script) as they differ slightly from the Snarl version.