Why I switched to LCD

[Mxsmanic]

When my CRT monitor finally failed, I decided to switch to an LCD
display. I had been debating the matter for months, ever since my CRT
showed the first signs of an impending demise. While no two people have
exactly the same requirements for a monitor, I thought I'd describe my
reasoning here so that others might benefit from it (just as I profited
from asking around a lot before making my decision).

I use my monitor for a wide variety of computer tasks, most of which
require sharp images rather than precise or vivid colors and contrasts.
However, I do a fair amount of work on photographs, which requires the
best possible image quality overall.

My reasons for choosing a LCD were as follows, in roughly decreasing
order of priority:

1. WEIGHT. It sounds odd, but this was a big concern for me. A good
quality, 20" CRT weighs easily 30-35 kg. The better the CRT, the more
it weighs. Now, lifting 30 kg straight up from the ground is no great
feat--but lifting a huge, awkward CRT with no handles and wrestling over
the edge of a desk and onto a table is a hazardous operation for anyone
who doesn't regularly manipulate heavy weights in this way, because the
slightest wrong movement can put eccentric stress on the spine for which
untrained back muscles may not be ready to compensate. I've seen too
many people injure their backs (sometimes permanently) in seemingly
innocuous operations of this type to be willing to risk it myself (I'd
have to move the CRT alone). So this was a major issue for me, and the
huge weight of CRTs (especially good quality, 20" tubes) really worked
against them.

2. SIZE. An issue mainly in combination with weight, above. The size
alone would fit on my desk, but trying to wrestle a massive cube that
weighs a ton really seemed like a problem.

3. AGING. CRTs age more steadily than LCDs, as far as I can tell. They
gradually drift out of adjustment and one must watch them constantly.
Sometimes they can be brought back into adjustment; sometimes they
can't.

4. GEOMETRY. LCDs operating at their native resolutions (1600x1200, in
my case) are extraordinarily sharp and have no problems with geometry,
convergence, rotation, magnetic fields, etc. Since I do a lot of work
with text and graphic elements that are very small (I try to profit as
much as possible from the available resolution with tiny fonts), this
clarity of the image is important.

5. OPERATING CYCLE. LCDs are less prone to burned-in images and the
like, so I don't have to worry as much about running screensavers or
using power-saver modes. I already have cheaper LCDs that I run
continuously, since they use very little power and then aren't much
affected by continuous operation (as far as I can tell).

There are some serious disadvantages to LCDs, although in my case they
weren't enough to tilt the balance in favor of CRTs, such as:

1. PRICE. LCDs are extraordinarily expensive, especially in 1600x1200.
The price increases rapidly if you want image quality sufficient for
photo work, or even approaching that level.

2. DEAD PIXELS. I don't consider _any_ dead pixels to be acceptable,
despite the whining and excuses made by manufacturers who are still too
incompetent to mass-produce defect-free flat panels.

3. IMAGE QUALITY. Even the best LCDs can't approach the best CRTs,
despite the fact that the latter are nearly ten times cheaper. This is
not too much of a big deal for any ordinary work, but for photo and
prepress work, it does make a difference. However, all things
considered, I just couldn't afford to go too far in the direction of
top-of-the-line photo quality monitors. Maybe next time (and perhaps in
years to come flat panels may eventually match CRTs, but I'm not holding
my breath).

Anyway, I ended up replacing an eight-year-old Sony Multiscan 20seII
with about 30,000 hours on it with an Eizo Flexscan L885 flat panel.
The Eizo is a medium-high grade of flat panel intended for CAD/CAM, and
has a very sharp image with pretty good image quality overall. It's not
the CG line of photo-quality displays, which I could not afford, but I
think it's a good compromise. It certainly looks very nice after
squinting at the CRT as it gradually failed (during the last 90 days or
so the Sony deteriorated rather sharply). But I can see that the LCD
doesn't have the very dense blacks and blinding whites that CRTs can
provide. On the other hand, it's astonishingly sharp (every pixel
clearly defined, even with VGA input).

So there's my $0.02. I hope time will prove this to be a wise decision,
and I hope by the next time I need a new monitor flat panels will have
improved enough and come down in price enough to make CRTs truly
obsolete--but that may take a long time.

 

[Bob Myers]

2. DEAD PIXELS. I don't consider _any_ dead pixels to be acceptable,
despite the whining and excuses made by manufacturers who are still too
incompetent to mass-produce defect-free flat panels.

I don't know whether you'll consider this just more "whining and excuses,"
but it's not a matter of incompetence. Making TFT-LCDs is very, very
similar to making very large integrated circuits, and like any IC process,
you simply cannot get to zero defects across the entire wafer - or in this
case, the entire mother glass. So setting the number and type of defects
that you consider acceptable directly controls the overall yield of the
process, and thereby the cost of the finished panels. Defect-free panels
are produced today - but for any customer to insist on receiving nothing
but defect-free panels reduces the number of panels that can be sold to
that customer, and hence necessarily increases the price they must pay.
There will always be this tradeoff to be made between defect level and
price, and it's up to the market to decide where the appropriate
balance will be struck between these factors. More and more customers
ARE concerned with pixel defects, of course, but the "zero-defects" level
of performance will always come with a higher price tag. Develop an
LCD fab process that has essentially a 100% yield of perfect panels,
and that will, of course, change - but keep in mind that the LCD industry
is currently being driven to reduce cost, which has meant constantly
bringing on-line newer and larger fabs, and each time you do that there's
a learning curve to come down in terms of yield.

3. IMAGE QUALITY. Even the best LCDs can't approach the best CRTs,
despite the fact that the latter are nearly ten times cheaper. This is
not too much of a big deal for any ordinary work, but for photo and
prepress work, it does make a difference. However, all things
considered, I just couldn't afford to go too far in the direction of
top-of-the-line photo quality monitors. Maybe next time (and perhaps in
years to come flat panels may eventually match CRTs, but I'm not holding
my breath).

The day when flat-panel technologies in general, and LCDs in particular,
match the all performance aspects of CRTs may not be that far off. This
is still a relatively new technology, especially at these sizes, and it
continues
to develop rapidly (the new developments right now, though, are
being driven mostly by the TV market, and can be expected to trickle
down later to the monitor market, which is now much more of a
"commodity" sort of place). In contrast, the CRT has had over 100
years of development behind it already (50 years in the case of the
standard "tricolor/shadow mask" sort of color CRT), and is safe
to consider "extremely mature."

Bob M.

 

[Mxsmanic]

I don't know whether you'll consider this just more "whining and excuses,"
but it's not a matter of incompetence. [...]

What irks me is manufacturers and salespeople and others who say that
dead pixels are a "normal" or "inevitable" consequence of flat-panel
manufacture, when they are neither. You can make defect-free flat
panels. Yes, it costs more, but it can be done, and it's getting more
economical all the time. Engineers in the 1960s would be awestruck at
the density achieved on ICs today--I'm sure they were convinced at the
time that it couldn't be done, too.

In other words, I wish manufacturers would concentrate more on
eliminating dead pixels, and less on making excuses for their presence.
Eventually someone is going to start making guaranteed zero-dead-pixel
panels at affordable prices, and then everyone else is going to have to
fall into line.

After all, nobody sells microprocessors in which one or two instructions
fail to execute correctly, or memory chips in which every millionth byte
is dead.

The day when flat-panel technologies in general, and LCDs in particular,
match the all performance aspects of CRTs may not be that far off. This
is still a relatively new technology, especially at these sizes, and it
continues to develop rapidly (the new developments right now, though, are
being driven mostly by the TV market, and can be expected to trickle
down later to the monitor market, which is now much more of a
"commodity" sort of place). In contrast, the CRT has had over 100
years of development behind it already (50 years in the case of the
standard "tricolor/shadow mask" sort of color CRT), and is safe
to consider "extremely mature."

I'll be very happy when LCDs match CRTs. I have no emotional attachment
to CRTs, but I still lust after their image quality in many respects.
Even this LCD screen that I just bought, which seems to be of excellent
quality, cannot get the dense, absolute blacks that a CRT can produce.
I suppose I can adapt to it by watching the black points carefully in
Photoshop, but I look forward to the day when flat panels will be just
as beautiful as CRTs. I don't know if there are any fundamental
obstacles to achieving this (?). They are two very different
technologies, for sure. I suppose the lack of a perfect black has to do
with the fact that liquid crystals are never completely opaque (or
completely transparent), no?

The flat panel sure whips a CRT for sharpness and resolution, though, at
equal image sizes and resolutions. Ditto for geometry. I hope the flat
panel proves to have a longer useful life than CRTs do, too, although I
suppose it's too early to say. The fact that the flat panels depend on
some sort of fluorescent lamp as a backlight (if I understand correctly)
worries me a bit.

 

[Bob Niland]

What irks me is manufacturers and salespeople and
others who say that dead pixels are a "normal" or
"inevitable" consequence of flat-panel manufacture, ...

They are unless the maker has a firm policy of trashing
panels with even one defect. If there are customers for
panels with 2 hot, 1 dead and 3 stuck pixels, are they
going to toss that money in the dumpster? Doubt it.

The industry is keenly interested in getting to Zero
Defects on LCD, because, despite the policies and
disclaimers, people still complain, and probably more
than a few send less-than-perfect LCDs back under RMA
even though the defects are "in spec". Some customers
may even do this as a "buyer's roulette", chancing
that the exchange unit will have fewer defects.

I recently bought a largish LCD, read the brand's
defect-pixel policy, and was fully expecting a few
duff pixels of the millions on the screen. Much to
my astonishment, there appear to be none. (Grayscale
tracking, on the other hand is awful, but ...)

Here's what may be a tip for improving the odds of
getting a zero-defect LCD:
+ buy direct from the mfr's NEW stock, or
+ from a first-tier reseller (like CDW), and
+ stick with the pricier brands

If, on the other hand, you want to improve the odds
of getting a panel near (or even beyond) the limits
for grungy pixels:

- buy from eBay (and the normal warnings apply about
buying anything on eBay that has "LCD" in the title)

- buy from the lowest price seller on pricewatch
(where DO those people get their inventory?)

- buy from the mfr as a refurb (when they get NTF RMA
units that merely have in-spec defect pixels, what
do you suppose they do with them?)

- buy the cheapest brand/model (Chances are the TFT
panel maker doesn't sell everything to the same
OEM customer. Guess where the lower-quality stuff
goes.)

Another way to ensure getting a ZD LCD is to pick one
out in person. Make up a CDR with .BMP images at your
intended resolution, one file each in black, white,
green, red, blue, gray, (plus some representative
screen shots of your normal use).

Take that CD to the local Walgetco. Ask if they will
agree to sell the display units. Load the images in
MS Paint and do a "View Bitmap" for fullscreen. Inspect.
When you find a display that passes your tests, buy it.

Yes, it would be nice if all LCDs on the market were
perfect. Unfortunately, there are people willing to
buy imperfect units, and, if only ZD panels were sold,
the already-high prices would be even higher.

 

[Mxsmanic]

They are unless the maker has a firm policy of trashing
panels with even one defect. If there are customers for
panels with 2 hot, 1 dead and 3 stuck pixels, are they
going to toss that money in the dumpster? Doubt it.

Then they should grade the panels.

For example, many broadcast camcorders have used CCDs identical to those
used in consumer camcorders. The only difference was that the broadcast
CCDs were selected for zero defects. The CCDs with defects went into
the consumer models.

The same principle could be applied to flat panels. Pay X dollars for a
"standard" grade of model ABC, with up to five dead pixels. Pay X+Y%
for a "perfect" grade of the same ABC model, with zero dead pixels.
With the right price points, you sell all the panels, and you satisfy
both customers who want zero dead pixels and customers who want the
lowest possible price for a given model.

The industry is keenly interested in getting to Zero
Defects on LCD, because, despite the policies and
disclaimers, people still complain, and probably more
than a few send less-than-perfect LCDs back under RMA
even though the defects are "in spec". Some customers
may even do this as a "buyer's roulette", chancing
that the exchange unit will have fewer defects.

I can't say that I blame them.

What I find interesting is that the display models at my local computer
stores never have any dead pixels. Apparently they are willing to sell
panels with defective pixels, but they aren't willing to put them on
display. How many do you think they go through before they find one
that's defect-free?

I recently bought a largish LCD, read the brand's
defect-pixel policy, and was fully expecting a few
duff pixels of the millions on the screen. Much to
my astonishment, there appear to be none.

I found one subpixel that's a bit reddish on mine, after initially
believing that it was flawless. I'd still prefer that it be flawless,
but as long as it doesn't develop any other defects, I guess I can live
with it for now. If I had bought a top-of-the-line CG model, though,
I'd expect zero defective pixels, no matter what the manufacturer says;
dead pixels just aren't acceptable for the most critical photo and
prepress work.

Grayscale tracking, on the other hand is awful, but ...

What is grayscale tracking?

Here's what may be a tip for improving the odds of
getting a zero-defect LCD:
+ buy direct from the mfr's NEW stock, or
+ from a first-tier reseller (like CDW), and
+ stick with the pricier brands

Do retailers just repack returned panels and resell them to less
demanding customers, or what?

- buy the cheapest brand/model (Chances are the TFT
panel maker doesn't sell everything to the same
OEM customer. Guess where the lower-quality stuff
goes.)

This sounds similiar to the grading system I mentioned above. I have no
problem with it, as long as it's up front and honest.

Indeed, there are some applications, such as rarely-examined servers in
back rooms, for which a cheapo LCD with a number of defects would be
fine. I just don't want to be staring at defects on my mission-critical
machines.

Another way to ensure getting a ZD LCD is to pick one
out in person. Make up a CDR with .BMP images at your
intended resolution, one file each in black, white,
green, red, blue, gray, (plus some representative
screen shots of your normal use).

Take that CD to the local Walgetco. Ask if they will
agree to sell the display units. Load the images in
MS Paint and do a "View Bitmap" for fullscreen. Inspect.
When you find a display that passes your tests, buy it.

I don't know how many stores would go along with that.

Yes, it would be nice if all LCDs on the market were
perfect. Unfortunately, there are people willing to
buy imperfect units, and, if only ZD panels were sold,
the already-high prices would be even higher.

They don't have to sell zero-defect panels exclusively; they just have
to make such panels an option for those who want to pay for them,
instead of forcing defects on everyone.

 

[Bob Niland]

Then they should grade the panels.

Not likely. I suspect they'd rather move to junking all
the imperfect panels. Openly retailing products by
quality grade has always been a tough sell. Lenox, for
example, sells factory second china in their factory stores
(mixed in with 1st-quality overstocks), and even with
the steep discount over 1st-quality at regular prices,
people still complain about minor nits.

Consider Sony TVs, they make no effort to make obvious
what you give up when you slide down from PVM, to XBR
to KVM (other than price).

What is grayscale tracking?

<http://www.dpreview.com/images/grayscale.gif>

What is supposed to be the lighter grays are distinctly
pink, and even separately adjusting the gamma for the
red channel can't completely fix it. Apparently, the
dynamic range of the subpixels isn't a full 8-bits per
color. A very non-linear color correction is required,
and that would likely result in some visible terracing
in some images.

Don't get an LCD for critical color/pre-press work
without knowing that it can do what you need. I'm not
sure that recent industry quest for speed (pixel
response times) has worked in favor of accurate color.

Do retailers just repack returned panels and resell
them to less demanding customers, or what?

Indeed, or what? Like I asked about those rock-bottom
obscure resellers on pricewatch: where DO they get
their merchandise? Same with those PowerSellers on
eBay, under names like "ultraliquidators" (not a
real userID), shifting thousands of units a month
and getting horrible feedback complaints.

Marginal but in-spec otherwise NTF RMAs have to end
up somewhere.

I don't know how many stores would go along with that.

I used to buy CD players that way, back when they were
really expensive, and there were big differences in how
they handled marginal and defective media. Some stores
will let you buy display stock, some won't. It never
hurts to ask.

They don't have to sell zero-defect panels exclusively;
they just have to make such panels an option for those
who want to pay for them, instead of forcing defects on
everyone.

Think twice about ordering LCD from newegg:
"Dead Pixels Policy:
Replacement or Refund for 8 or more DEAD PIXELS ONLY!"
on every LCD monitor page, regardless of brand & price.

CDW's LCD pages have no such disclaimer. Is newegg getting
different product quality? Beats me.

 

[Bob Myers]

What irks me is manufacturers and salespeople and others who say that
dead pixels are a "normal" or "inevitable" consequence of flat-panel
manufacture, when they are neither. You can make defect-free flat
panels. Yes, it costs more, but it can be done, and it's getting more
economical all the time. Engineers in the 1960s would be awestruck at
the density achieved on ICs today--I'm sure they were convinced at the
time that it couldn't be done, too.

I'd put it a slightly different way; as you learn to control your process,
you can make MORE defect-free panels. I don't think you can
ever reach the point where each and every panel on each and every
substrate is perfect, any more than you can get to a point where every
IC on every wafer is 100% functional. The difference with displays
(vs. ICs) is that many defects will still leave the display USABLE, although
not "perfect," at least in the eyes of many customers. So the question
remains where to draw the line as to what is "shippable." And yes, it
does cost more to sell defect-free panels as your standard product, since
that reduces your effective yield. And so you wind up looking not only
at how good your process is, but also what the competition is doing in
a given market. It is always POSSIBLE to sell defect-free panels by
doing a special sampling of ANY output, but there are problems there,
too - it costs more to do that extra screening, and the average quality
level of the "standard" product is now lower, with the "perfect" panels
having been cherry-picked out. So you also have to be aware of just
where THAT sort of thing positions you in the market.

The bottom line - ain't nothing as easy or simple as it should be...

Bob M.

 

[Mxsmanic]

Not likely. I suspect they'd rather move to junking all
the imperfect panels. Openly retailing products by
quality grade has always been a tough sell. Lenox, for
example, sells factory second china in their factory stores
(mixed in with 1st-quality overstocks), and even with
the steep discount over 1st-quality at regular prices,
people still complain about minor nits.

They could sell lower grades as different brands.

I mean, I assume that already a lot of different brands are getting
panels from the same fabrication facilities, no?

Consider Sony TVs, they make no effort to make obvious
what you give up when you slide down from PVM, to XBR
to KVM (other than price).

I suppose one could say that if you don't know, you don't need to pay
for the more expensive ones.

What is supposed to be the lighter grays are distinctly
pink, and even separately adjusting the gamma for the
red channel can't completely fix it. Apparently, the
dynamic range of the subpixels isn't a full 8-bits per
color. A very non-linear color correction is required,
and that would likely result in some visible terracing
in some images.

Wow, that sounds odd. Is this typical of all LCDs?

I just quickly tried a grayscale gradient in PS and it looked okay.
There is some posterization, which I had previously noticed, but I
assume that's part of the price one pays currently for having a flat
panel. I don't suppose anyone has true 32-bit color LCDs yet?

Don't get an LCD for critical color/pre-press work
without knowing that it can do what you need. I'm not
sure that recent industry quest for speed (pixel
response times) has worked in favor of accurate color.

I knew that this monitor wasn't ideal for photo work. It seems to work
pretty well, however. The photos I prepare are used by others on
systems that are even less exacting than I am, so to a certain extent it
doesn't matter that much.

Indeed, or what? Like I asked about those rock-bottom
obscure resellers on pricewatch: where DO they get
their merchandise? Same with those PowerSellers on
eBay, under names like "ultraliquidators" (not a
real userID), shifting thousands of units a month
and getting horrible feedback complaints.

Okay, but I'm thinking about more reputable retailers, not the
mail-order mysteries or eBay sellers.

Marginal but in-spec otherwise NTF RMAs have to end
up somewhere.

So does the retailer give them to someone else to dump, or does he try
to resell it as new to a different, unsuspecting customer?

The box to my screen had a hole in it, but nothing was damaged and the
screen works fine, so I haven't worried about it. Still, it makes one
wonder. Maybe I made the hole in transit, but I'm really not sure.

Think twice about ordering LCD from newegg:
"Dead Pixels Policy:
Replacement or Refund for 8 or more DEAD PIXELS ONLY!"
on every LCD monitor page, regardless of brand & price.

CDW's LCD pages have no such disclaimer. Is newegg getting
different product quality? Beats me.

I don't know.

But should one look more at the retailer, or more at the brand?

 

[Bob Myers]

They could sell lower grades as different brands.

Unfortunately, it doesn't work like that.

The LCD panels themselves are produced by a number of
fairly large companies, who sell those panels in volume to the
companies who integrate them in to the LCD monitors that
you buy. In other words, the panel manufacturers are not the
ones making the "brands" that you see on the shelves, but they
are the ones who ultimately determine what panels will be
available on the market. The monitor and system makers deal
with the LCD panel producers in very large volumes, and it has
never proven to be a good idea to try to separate those orders
into various defect grades, for reasons already discussed in this
thread.

That's close to the correct explanation, but not quite.
The basic problem is that the native response of an LCD
is not linear or "gamma-curve-shaped" (i.e., "CRT-like")
but rather a sort of S-shaped curve. Correcting this to be
linear, or better CRT-like, when only 8 bits/color of
control are available in the first place eats in to the available
accuracy of the gray level control as seen at the monitor
inputs, and can result in color tracking problems (as described
here) unless that correction process is done very carefully.
Even at best, the correction is very unlikely to provide
"perfect" color tracking AND a proper response curve
correction; getting there will have to wait for panels with
10 bits/color or better of effective gray scale.


Bob M.