Display Driver and Display

We described earlier how color depth was related to the number of bits used to identify
the pels in a pixel (RGB discrimination). The dynamic range of the display driver
and display determines what can be shown on the handset—and hence determines the
properties of the downlink offered traffic.
Table 4.5 shows the progression from 1-bit to 24-bit color depth. Anything beyond
24-bit color depth is generally not discernible by the human eye, though as with highrange
audio products, this doesn’t mean people will not buy such products; in fact,
some video adapters and image scanners now deliver 32-bit true color. 3GPP has specified
a core visual profile that covers from 4 bits (grayscale) to 12 bits, but as we will
now see, display capability is rapidly moving toward 16-bit color depth.

The most favored candidate for digital cellular handsets to date are conventional but
highly optimized LCDs. These come in two flavors: reflective, which work well in
bright sunlight, and transmissive, which work well indoors. Products like the Compaq
iPAQ use reflective displays to meet the power budget constraint of a PDA that ideally
should be capable of running on two AA batteries.
Transmissive LCDs are the standard for laptop PCs. Laptops have a power budget
of between 8 and 10 Watts—along with a form factor and rechargeable battery to suit.
Digital cellular phones need to be well under a Watt to meet form factor requirements,
given existing battery densities. This means they are much closer to PDAs than laptops
in terms of power budget constrains. The Nokia 9210 provides a good benchmark for
a 2002 product against which future generations of display-enabled handsets can be
measured. The device supports 4000 colors.
The more colors a screen can support, the more light filters you need. The more light
filters you have, the bigger the backlight. The bigger the backlight, the more power you
consume.
The transmission display in the 9210 uses a cold cathode fluorescent lamp. It is positioned
right next to the battery and couples light to the display via a wedge-shaped
slab waveguide. The wider the prism (that is, the thicker the wedge), the better the
coupling efficiency and brightness uniformity of the display and the lower the power
consumption.
In this example, the waveguide wedge is 6 mm, which seems to be at present an
acceptable thickness/efficiency trade-off. The display can automatically adapt to
ambient light conditions. Flat out, the screen emits 100 candelas and consumes 500
mW. At the dimmest setting, it consumes 150 mW. This is, of course, in addition to the
existing baseband and RF power budget. Quoted figures from the manufacturer suggest
between 4 and 10 hours of use from a fully charged 1300 mAh lithium ion battery.
The resolution achievable is a function not so much dictated by the screen itself but
by the driver IC connections. The color screen is 110 × 35 mm, with a pixel density giving
150 dots per inch (dpi) of resolution at a pixel pitch of 170 μm. The response/
refresh cycle of the driver is 50 ms, which is sufficient for a frame rate of 12 frames per
second. There is no point in sending such a device a 20 frame per second video stream,
as it will be incapable of displaying it. The hardware bandwidth determines offered
traffic bandwidth and offered traffic properties.

In practice, the hardware in this area is moving rather faster than the software, but
it is nice to know that in the future, displays and display driver bandwidth will be
capable of supporting increasingly high-resolution, high-color depth displays sent at
an increasingly rapid frame rate. Table 4.6 gives some examples of present displays
available from Hitachi.
One rather unforeseen consequence of improving display quality and display driver
bandwidth is that as display quality improves, compression artifacts become more
noticeable; the quality of the display and display driver determines the quality needed
in the source coding and physical layer transport. Put another way, if you have a poorquality
display, you do not notice many of the impairments introduced by source coding
(compression), channel coding, and the highly variable-quality, occasionally
discontinuous radio physical layer.
While color saturation/color depth is reasonably easy to achieve with backlit displays,
it is significantly more difficult with reflective (sometimes as described as transflective)
LCDs. In an LCD, a single color filter covers each pixel. Transmissive backlit
displays use thick filters. Reflective displays use thin filters to allow the light to pass
into and back out through the filter. The thinner the filter, the better the reflective properties
but the poorer the color saturation. If the thickness of the filter is increased to
improve color saturation, the picture becomes too dark.
A reflective LCD from Philips makes one corner of the pixel filter thinner than the
rest, which means that light can pass easily, thereby increasing brightness. The rest of
the filter is optimized for color saturation.
So here we have another quality metric—brightness—that is directly related to how
the display hardware is realized. Additional metrics include uniformity and viewing
angle (usually quite narrow with LCDs). One problem with conventional displays is
the continued use of glass. Glass is relatively heavy, fragile and does not bend easily. A
hybrid approach presently being investigated involves the use of ultra thin glass
attached to a flexible sheet. Toshiba has recently shown examples of products that, in
the longer term, could provide the basis for foldable lightweight LCDs.

All displays, including flexible, foldable, and conventional displays, require display
drivers. The Digital Display Working Group is presently working to
standardize the digital display interface between a computer and its display device,
including backward-compatibility with existing analog driver standards. This working
group is also producing proposals for micro-displays (50-mm/2-inch diagonal size).
Consider that an SVGA LCD monitor needs to have an address bandwidth/bit rate
of 25 megapixels per second (25 million pixels per second). A QXGA cathode-ray tube
has an effective bandwidth requirement of 350 megapixels per second.
The refresh rate can be reduced by only refreshing the parts of the display that are
changing. This decreases the processor overhead in the driver but increases the memory
space needed. Even so, it is not uncommon in PC monitor drivers to encounter driver
clock speeds well over 100 MHz. These are power-hungry and potentially noisy
devices. Refresh rates in laptop LCDs are now typically 25 ms (the Nokia handset case
studied earlier in the chapter had a 50-ms refresh rate). Refresh rate obviously becomes
increasingly critical as frame rate increases.
A number of Japanese vendors are sampling display products (with chip-on-glass
display drivers) that are supposed to be capable of supporting 30 frames per second. A
present example is a Sharp 262,000-color 5-cm reflective display produced on a 0.5-mm
substrate, which is claimed to support 30 frames per second at a power consumption
of 5 mW per frame—small but efficient.
For backlit (transmissive) displays, performance gains include significant improvements
in contrast ratio and parallel reductions in power budget.
Pixel density is moving to more than 200 pixels per inch and contrast ratios are
improving from 50:1 to 200:1 or better (see Table 4.7). Power savings are being achieved
by using thin film transistors with latch circuits that hold the liquid crystal cell state at
the correct potential through the refresh cycle. Fortuitously, investment in LCD-based
micro-display technologies can be common both to digital cameras and 3G handsets
with digital cameras.
In effect, these are two related but separate product sectors, each of which generate
significant market volume. Market volume helps reduce component cost but also
tends to improve component performance through better control of component tolerances
on the production line. Digital camera performance drives user expectation of
how a digital cellular handset with an integrated digital camera will perform. The
problem is that the digital cellular handset also has to be able to send and receive pictures
and an audio stream over a radio physical layer that will typically consume several
hundred milliWatts. There is a balance to be made between memory bandwidth in
the handset and how much power to dedicate to sending and receiving image bandwidth,
which in turn determines the user experience and user expectations.

There are some other practical issues. Cellular handsets tend to be much more
roughly handled than computer products—for example, PDAs. All displays that use
glass are inherently fragile and don’t take kindly to being dropped onto concrete
floors. A very important present design consideration is how to improve the robustness
of high-quality displays. Using thin layers of glass bonded to plastic is one option. 127