lcd monitor manufacturer recommendation request

Tue Nov 24 01:37:58 UTC 2009

i thank you, sir.

in fact, mr. block, --you-- get a gold star.

Warren Block wrote:
> On Sat, 21 Nov 2009, Dag-Erling Sm?rgrav wrote:
> 
>> spellberg_robert <emailrob at emailrob.com> writes:
>>
>>> Person, Roderick <personrp at UPMC.EDU> writes:
>>>
>>>> You could get an LCD monitor that only has the 15 pin connection,
>>>> then you can choose refresh rate and play with modelines. At least
>>>> on the Viewsonic LCD I have which is 5 years old I had to do that to
>>>> get it working. That is if you really want to keep doing such
>>>> things.
>>>
>>> it isn't that i --want-- to keep doing this; it is that i --can-- do
>>> so, if i desire.
>>
>>
>> A CRT has an electrom beam that sweeps across the screen left to right
>> and top to bottom, and the horizontal and vertical sync frequencies
>> control how fast the beam moves.  An LCD panel does not have an electron
>> beam; it has discrete, individually adressable pixels.  If you insist on
>> hooking it up to an analog port, it will have to convert the analog
>> signal to a digital signal in order to display it, and you will get
>> sampling artifacts, aliasing etc.  I don't care how good you are at
>> writing modelines; you will never come up with one that looks better
>> than what you will get with a digital connection.
> 
> 
> For any technical use, an LCD should be used at native resolution.  But 
> I remember 800x600 on a 1024x768 LCD to be surprisingly good (the users 
> insisted).  Kind of a painted effect.  Not a problem for those 
> particular users, but they were not technical and PEBKAC applies.

i've been dealing with this for decades; i suspected that there was an acronym for it.

> What hasn't been mentioned yet is LCD panel types.  For many uses, TN is 
> fine.  For photography and other graphics, IPS shows a lot of detail 
> that TN can't, and PVA is somewhere in the middle:
> 
> http://www.pchardwarehelp.com/guides/lcd-panel-types.php

i had found something similar on wikipedia [ a source not known for its reliability ],
   but, this had a link to a list of ips types,
   which i found useful.

> Lots of information here:
> 
> http://forums.anandtech.com/showthread.php?t=39226

i spent most of sunday here.

in particular, it was interesting to compare manufacturers data with what is presented here.

it suggests that two of the old names [ hp, nec ] are still viable;
   i pays my money, i takes my chances.

[ fwiw, i like what i read about the nec 2490/2690; but, i think i'll try a hp 2475 first.
   being blue on one side and red on the other may not be as bad as it sounds.
   i might get lucky.
   compared to amazon, i'll have to see what new_egg wants for shipping.
]

> -Warren Block * Rapid City, South Dakota USA

i thank all of you for your respective efforts.
however, the un_spoken assumption, regarding formatting, comes after use.
i would have done the same thing; with use, new ways of thinking become "old_hat".

i am pleased to report that i have successfully shifted my paradigm:
   both were from the philadelphia mint; one 1980, the other 1992.
they had been on the left_side of my desk; now, they are on the right_side.

the paradigm_shift is to go from
   thinking in terms of "soft" pixels to
   thinking in terms of "hard" pixels.

an lcd panel has "hard" pixels; the number and arrangement are fixed.
in particular, the spacing is fixed.
a crt has "soft" pixels; it will display what the input provides [ up to a point; more later ].

with a crt, we think in terms of "filling" the display_area.
the pixel size and spacing are variable; 640x480 and 1600x1200 can be exactly the same size.
with a lcd, the pixel_density is fixed;
   the actual size of the display_area will be determined by the pixel arrangement.
i start by assuming a one_to_one mapping between "soft" pixels and "hard" pixels.

with a crt [ within bounds set by the manufacturer ],
   i can set the dot_rate and h & v frequencies and sync_positions,
   permitting me to control the size of the display_area and,
   independently, "stretching" and/or "squashing" the image in two dimensions,
   to my heart's content.
in particular, the h & v pixel_spacings need not be the same.
the display will be sharp, regardless, because the rgb_phosphor_dots are very tiny;
   where they are struck by electrons, they will glow.
there is no interpolation.

with a lcd, the pixel_spacing is fixed
   [ although not required,
       i suspect that the marketing dept will have an easier time,
       if the h & v spacings are the same
   ].
--something-- has to be given to each "hard" pixel, without regard to the arrangement of the "soft" pixels.

at this point, let us assume that we have
   a "high_end", "name" model, lcd_display_device, of "hard" dimension 1920x1200.
for simplicity, assume 24_bit [ 8x3 ] color rather than 18_bit [ 6x3 ].
i am going to treat this as a "black box", with a sophisticated embedded_controller running the show.
some of these things have half_a_dozen input_connectors, suggesting that
   the front_end is doing some dsp [ a separate processor ? ] so as to produce, for internal use,
   a time_sequence, at some rate [ e. g., 60, 75, 85 ], of frames,
   each of some "soft" dimension, translated to a 24_bit color.
i will treat this as a time_sequence of data_structures.

the main processor, then, has to decide how to map the pixels,
   from the "soft" array of the structure to the "hard" array of the installed panel.
the easiest approach is one_to_one, centered in the panel [ this --really-- appeals to me ].
other approaches are to fill the panel vertically or horizontally or both.
this will, probably, require interpolation; the resulting presentation may or may not be pleasing
   [ this reminds me of the two approaches to displaying 24_frame film on 30_frame dvd;
       the "clean" approach is to repeat every fourth frame [ great for single_stepping ],
       but, some create the fifth frame by mixing together the frames on either side
       [ this is a big thrill, if the film_frames straddle a scene_change;
           the effect is as psychedelic as a mid_1960s peter_sellers film
       ]
   ].
essentially, one time_sequence of data_structures is replaced by another time_sequence of data_structures.

i observe that the high_end models have many options for controlling the presentation of the input.
some of these models appear to be true "multisync" models that
   can handle any legacy_input that is thrown at them
   [ they all have to handle 640x480x60, with [ count 'em ] 16 colors ] and
   can format them in a variety of ways.
others appear to handle only "standard" or "common" legacy_inputs;
   i can configure my legacy systems, if necessary.

my new intel moboes [ not yet installed ] all have 15_pin analog connectors on them.
the "dg41ty" [ but, not the "dg31pr" ] has the 25_pin dual_rate dvi_d connector on it.
i think i'll get the hp 2475w, first, just to get some experience with this technology.
i suspect that i will, later, get one of those nec models.
of course, these choices are not yet final.
just for fun, i think i'll implement a dvd_player on one of the machines, just to see what happens.
not using any crt, as much, means that
   their eventual failure_dates will be, somewhat, farther in the future.

again, i thank you all for your responses.
in particular, i thank you, mr. block, for pointing me in the direction of much information.

happy thanksgiving.
happy everything.

rob