Basic Description
The video hardware produces
a continuous signal on its output connector, except when it is in reset
mode, where the video outputs are held in a single state. The continuous
signal is required because the pixel information is only displayed for
a short period of time, and relies on the persistence of the phosphor glow
on the monitor as well as the ability of eyesight to perform automatic
averaging to appear to be a steady image. That signal is usually
output on multiple pins of the monitor connector, although it could also
be a TV compatible output. LCD displays use a similar technique,
although the timing is more advanced and depends on the specific type of
panel and its driver circuitry. The signal includes both the pixel
data that the monitor displays, as well as timing and "framing" information
that the video display uses to drive its internal circuitry.
The image's pixels are "scanned"
on to the screen from left to right, top to bottom, and is broken up into
"dot periods" or pixels, which is controlled by the "dot clock" for the
mode, which is the basis for all the other video timings. Each horizontal
"scan line" of dot periods is called a horizontal refresh as it "refreshes"
the information on the display in a horizontal line. Many of these
scan lines (the amount depending on the video mode), scanning from top
to bottom, make up a vertical refresh, also known as a "frame". There
are many vertical refreshes per second, where a higher refresh rate produces
an image with less flicker.
Timing Measurements
One of the important pieces
of terminology to understand is how timing is measured. These include
terms such as megahertz, kilohertz, and hertz. The first three are
a measure of frequency which is based on the term hertz (abbreviated hz),
which can be replaced by the term "Cycles per second." In video timing,
hertz is used to describe the frequencies of the timing signals, such as
when saying that the vertical refresh frequency is 60 hertz (or 60hz).
This means that there are 60 cycles per second, which means that there
are 60 vertical refreshes per second. Another case where hertz is
used is when saying the horizontal refresh rate, such as when saying 31500
hz, which means that there are 31,500 horizontal refresh cycles per second.
One abbreviation frequently found is the term kilohertz (abbreviated Khz)
which means 1,000 cycles/per second. For example, 31.5 kilohertz
means 31.5 x 1000 hertz, or 31500 hz. This is used to save writing
a few zeros and is a bit more concise. Similarly the term megahertz
(abbreviated Mhz) is used, which means 1,000,000 cycles/per second.
For example, instead of saying that a certain mode uses a 25,000,000 hz
dot clock, or saying that it uses a 25,000 Khz clock, it can be concisely
be stated by saying that it uses a 25 Mhz dot clock.
Similarly, the periods of
time involved in video timing are very short as they are typically small
fractions of a second. The terms millisecond, microsecond, and nanosecond
are useful for expressing these small periods of time. The term millisecond
(abbreviated ms) means one thousandth of a second, or 0.001 seconds.
In one second, there are 1,000 milliseconds. This is used to describe things,
such as the length of time a vertical refresh takes, for example a 16.6
millisecond vertical refresh means 16.6 thousands of a second, or 0.0166
seconds. In one second, there are 1,000,000 microseconds. The
term microsecond (abbreviated us) is used to describe something in terms
of millionths of a second, or 0.000001 second. For example the length
of a horizontal refresh could be 31.7 microseconds, or 31.7 millionths
of a second, 0.0000317 second, or 0.0317 ms. The term nanosecond
(abbreviated ns) is used to describe one billionth of a second, or 0.000000001
seconds. There are 1,000,000,000 nanoseconds in one second.
One circumstance where this is used, is to describe the period of time
one dot period takes. For example, one dot period could be stated
as 40 nanoseconds, 0.04 us, 0.00004 ms, or 0.00000004 seconds. In
each case, the most concise term is used, to provide a shorter, more concise
description.
Because the unit hertz is
defined using a unit of time (second), the period of one cycle can be determined
by division. The simplest example is 1 hz, where the length of the
cycle, by definition would be 1 second. For other values, it can be calculated
according to the following formula:
Horizontal Timing
From a monitor's standpoint,
the timing is fairly simple. It detects the horizontal sync pulses
on the hsync line, then based on the polarity, frequency, and/or duration
of those pulses sets up its horizontal scan circuitry to scan across the
screen at the desired rate. During this period it continuously displays
the signal input on the analog RGB pins/connectors. It is important
to to know the horizontal sync frequency ranges of the monitor, as well
as the acceptable width of the sync pulse for those sync ranges.
If the width of the sync pulse is incorrect, it can make the displayed
information too large or too small, as well as possibly preventing the
monitor from synchronizing to the generated signal. The acceptable
range of sync pulse width and polarity for a given frequency should be
given in the specifications for the monitor; however, this is frequently
overlooked by the manufacturer. It is recommended to contact the
manufacturer, otherwise the result has to be determined by trial and error
which can be damaging to the monitor's circuitry.
In addition to those that
control horizontal sync frequency and width, there are other horizontal
timing registers which tell the display generation hardware when to output
the active display, when to output the overscan, and when to perform blanking.
The active display is when pixel data from the frame buffer are being output
and displayed. This could also be overlaid by data from another source,
such as a TV or MPEG decoder, or a hardware cursor. The overscan
is the border to the left and right of the screen. This was more
important on older video hardware such as those monitors lacking horizontal
and vertical picture controls, and is provided for compatibility reasons
although current hardware typically reduces the need for this portion completely.
The blanking period is used during the retrace portion of the horizontal
display cycle which is the period in which the horizontal sweeps from the
right of the screen back to the left. Outputting non-zero intensities
during this period would end up being stretched, in reverse across the
end of the current scan line to the beginning of the next scan line which,
while interesting and possibly useful in a small number of circumstances.
would add a bit of blurring to the image. Blanking is signaled to
the monitor by sending zero intensities of the red, green, and blue components
on the analog lines.
In the display generator,
horizontal timings are specified by the number of dot periods they take.
The dot period is controlled by selecting the desired dot clock frequency
by programming certain registers.
Vertical Timing
Vertical timing is nearly
the same as the horizontal timing, except that it controls the vertical
movement of the display scan, spacing the scan lines the right width apart
so that they seem to form a rectangular image. The monitor detects
the vertical sync pulses on the vsync line, then based on the polarity,
frequency, and/or duration of those pulses sets up its vertical circuitry
to scan down the screen at the desired rate. It is necessary to know
the vertical sync frequency ranges for a given monitor, and the range of
acceptable vertical sync widths and polarities for those ranges.
The rage of vertical sync frequencies supported by the monitor are nearly
always given my the monitor's specifications, but like the horizontal sync
widths, the vertical sync widths are not commonly specified. Contact
the manufacturer, as attempting to guess the correct vertical sync width
can possibly cause the monitor to fail.
As well as being programmed
with the vertical sync frequency and pulse width, the display generation
hardware has other registers which control when to output the active display,
when to output the overscan, and when to perform blanking. In vertical
terms, the active display is the scan lines which contain horizontal active
display periods. The overscan is the border on top and bottom of
the screen and, if present, consists of one or more entire scan lines in
which the active display period is replaced with the overscan color.
The blanking is used during the vertical retrace, and consists of one or
more (usually more) scan lines in which the active display and overscan
periods are replaced with blanking period, making the entire line effectively
blanking. This prevents intensity from overlaying the screen during
the vertical retrace where the monitor sweeps the vertical back to the
top of the screen. Non-zero intensities output during this period
would be written in a zig-zag pattern from the bottom to the top of the
screen. In the display generator, the vertical timings are specified
in terms of how many horizontal sync periods they take.
Programming Considerations
For maximum flexibility,
video timings should be configurable by the end users to allow for the
specifications of their monitor. However, it is probably a wise idea
to maintain a table of monitors and their rated specifications, to allow
the users to select thieir monitor and determine whether or not the configured
video timings are within the rated specifications of their monitor and
warn the user about this. There is a distinct need for a comprehensive
and accurate software-usable database of monitor specifications in a platform
and video hardware independent form with sufficient information for a program
to both select timings for a particular video mode, as well as verify that
a given set of timings will function properly on the end-user's hardware.
This database should contain a human readable description of the monitor
make and model, as well as software parsable fields giving corresponding
ranges of horizontal and vertical frequencies and sync polarities for those
ranges if applicable, as well as a method of determining the acceptable
widths of horizontal and vertical sync pulses for a given frequency in
the corresponding rages. Framing information could be included in
this table in a frequency independent fashion, although this is something
that can be safely adjusted by the end user without risk of damage to the
monitor, thus it is preferrable to provide a method or interface for the
end-user to adjust these parameters to their preference.
Notice: All trademarks used or referred to on this page are the property
of their respective owners.
All pages are Copyright © 1997, 1998, J. D. Neal, except where
noted. Permission for utilization and distribution is subject to the terms
of the FreeVGA Project Copyright License.