thinkpads.com Support Community Forum Index Bill Morrow's thinkpads.com Open Forum - The Original Thinkpad Support Forum
Follow ThinkpadsForum on Twitter
Goodbye Lorita, wherever you've gone..

If you've found this forum useful, please consider donating a dollar or two

It is currently Wed Apr 26, 2017 4:22 pm

All times are UTC-05:00




Post new topic  Reply to topic  [ 303 posts ]  Go to page Previous 16 7 8 9 10 11 Next
Author Message
PostPosted: Wed Feb 24, 2016 4:21 am 
Offline
Freshman Member

Joined: Mon Jan 02, 2012 10:05 pm
Posts: 72
Location: Brussels, Belgium
For what it's worth... I tried the mod in two SXGA+ screens using side-emitting strips from iCCFL and was very disappointed because of the visible LED pattern and uneven backlighting. Yesterday I attempted to mend things by repositioning the strips, and ended up completely removing the backlight bracket and just taping the strip against the waveguide. This was not the original plan, but I'm very happy with the results: bright, even, no pattern, and easy to adjust.
My preferred explanation is that the bracket is counter-productive for LEDs because it acts as convex mirror that focuses them into points. But it's just as likely that the better positioning of the strip that I was only able to achieve without the bracket outweighs the (real) benefits the bracket provides.

Anyway, it has encouraged me to do the mod with the remaining screen and I'm looking forward to hearing more about the pro solution from xiphmont...


Top
   
PostPosted: Wed Feb 24, 2016 8:55 am 
Offline
Sophomore Member

Joined: Sun May 26, 2013 5:20 am
Posts: 193
Location: Seligenstadt, Germany
xiphmont wrote:
The fittings done using a 5mil accordion-fold boPET spacer are currently the best performing in terms of the intensity/pattern-elimination tradeoff:

Did you fold that yourself? Like under a microscope?

xiphmont wrote:
wileE wrote:
I do desperately want that for my Tablet.

I assumed once I got good results, a few would be reserved with your name on them (along with flyingfishfinger and rthur).

Yes, please.


Top
   
PostPosted: Thu Feb 25, 2016 6:10 pm 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
wileE wrote:
Did you fold that yourself? Like under a microscope?

folded them myself yes, but no microscope needed. I just cut a strip roughly 1.5mm wide on a paper cutter, then fold it in a zigzag every cm or so. It goes in front of the LEd strip. The amount of shimming I put on the other side of the strip determines how much it gets compressed down.

I'm using boPET (aka Mylar) because I know it will retain the fold (as long as I cut with and not gainst the grain) and it's deflection temperature is way higher than than the LEDs will ever get to. Regular PETG clear plastic hobby sheets work, but have a really low softening temp, so I don't think they;ll stay springy.

I'm mostly restricted to working on this stuff on weekends right now, and I only have one more such weekend before being shipped off on a two-week work trip. I'll see if I can get a few strips out, and we can figure out the spacer details later.


Top
   
PostPosted: Fri Feb 26, 2016 8:59 am 
Offline

Joined: Sun Jul 19, 2015 4:59 pm
Posts: 9
Location: United Kingdom <please provide city>
I'm definitely interested in a strip or two even without the spacer, it doesn't sound like something too hard to make / experiment around with.


Top
   
PostPosted: Fri Feb 26, 2016 2:24 pm 
Offline
Senior Member
Senior Member

Joined: Wed May 19, 2010 11:21 pm
Posts: 601
Location: Pardubice, Czech Republic
Just installed the board xiphmont mailed me into an x61s.
This was my first time doing any type of screen modification.
It was very easy to install into my Samsung LTN121XJ-L07.

Before led the display was 139 nits; with led it is 303 nits.
Here is a visual comparison vs a stock x61 and vs an x220:
http://www.jesseo.com/kites/x61-vs-x61-led-mod.jpg
http://www.jesseo.com/kites/x61-vs-x220-led-mod.jpg


With NO BATTERY installed, I (usually) observe there are 5 distinct brightness levels.
Let's call those, from low to high, 1,2,3,4,5.
At level 1 and level 2 the display works fine.
At level 3-5 I get intermitent backlight outages which last about 3 seconds.

Here is some data about the outages.

Outages vs. backlight setting (% read from 'powertop'):

@100% = 22 failures in 100 seconds
@29.5% and 26.7% = 13x in 100 seconds
@23.2% and 20% = [see following list]
* first failure after 380 seconds
* next at 467 seconds
* next at 556 seconds
* (wasn't watching..)
* next at 764 seconds
* next at 862 seconds
* next at 960 seconds

@16.8% and 13.3% = 0x in 700 seconds
@ ~8% = 0x in 400 seconds

fwiw, The first two times I turned on the machine, post-install, the backlight didn't turn on.
I used two shims: 2mm + 0.5mm.


wileE wrote:
Dimming range is completely different.
Measured the steps with my Spyder3: http://imgur.com/9EIj00e 77KB pic warning.

How do you measure the dimming range with a Spyder3 and dispcalGUI? Is it a 'tool' in dispcalGUI or are you doing a full calibration for each level?

_________________
[debian] x24 [xiphmont led] x60s [libreboot, led] T400 [libreboot, P9600] T420 [Ivy Bridge i7 3612QM]


Top
   
PostPosted: Fri Feb 26, 2016 5:20 pm 
Offline
Sophomore Member

Joined: Sun May 26, 2013 5:20 am
Posts: 193
Location: Seligenstadt, Germany
Quote:
How do you measure the dimming range with a Spyder3 and dispcalGUI?

I use an old version: 1.7.1.6
There is a quick report option under tools. Takes maybe 1 1/2 minutes for a reading.

Quote:
With NO BATTERY installed, I (usually) observe there are 5 distinct brightness levels.

X60 mainboard? Not X61?

Something is very wrong with your machine if you get outages. Are you absolutely sure that there are no shorts? Maybe see if the strip works outside the screen?


xiphmont wrote:
folded them myself yes, but no microscope needed. I just cut a strip roughly 1.5mm wide on a paper cutter, then fold it in a zigzag every cm or so. It goes in front of the LEd strip. The amount of shimming I put on the other side of the strip determines how much it gets compressed down.

That is unexpectedly easy. Will try that on my T. Never occurred to me to try that.


Top
   
PostPosted: Fri Feb 26, 2016 6:22 pm 
Offline
Senior Member
Senior Member

Joined: Wed May 19, 2010 11:21 pm
Posts: 601
Location: Pardubice, Czech Republic
wileE wrote:
Quote:
How do you measure the dimming range with a Spyder3 and dispcalGUI?

I use an old version: 1.7.1.6
There is a quick report option under tools. Takes maybe 1 1/2 minutes for a reading.

I use the same version. Under 'Tools' I see the following, which is it?
    Report on uncalibrated display device
    Report on calibrated display device
    Verify calibration
    Measurement report...
    Update measurement report...
    Measure display device uniformity
    Check measurement file...
    Automatically check measurements
    Import colorimeter corrections from other
    Create colorimeter correction...
    Upload colorimeter correction
    Create synthetic ICC profile
    Create 3D LUT
    Enable Sypder 2 Colorimeter
    Show curves
    Show log window
    Show log window automatically
Quote:
X60 mainboard? Not X61?

Something is very wrong with your machine if you get outages. Are you absolutely sure that there are no shorts? Maybe see if the strip works outside the screen?

X61s board.
The display cable wasn't inserted into the display completly. I've inserted by, on level 4 and 5 the screen still goes blank.

I will look for shorts and consider removing the led to test further. The screen looks so amazing that I'm reluctant to remove the led.

_________________
[debian] x24 [xiphmont led] x60s [libreboot, led] T400 [libreboot, P9600] T420 [Ivy Bridge i7 3612QM]


Top
   
PostPosted: Sat Feb 27, 2016 12:26 am 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
rthur and wileE: Keep in mind the new strips don't hit 'Ludicrous Speed' without the tweaked driver. They'll work fine with the regular TLD1/TLD2, but they'll not hit the brightnesses I've been talking about. Given that... how many do you want?

jaspen-meyer wrote:
With NO BATTERY installed, I (usually) observe there are 5 distinct brightness levels.

You should have 15 roughly evenly spaced brightness levels. AC vs battery should make no discernible difference (regulation is better than .5% across all use cases)

jaspen-meyer wrote:
At level 3-5 I get intermitent backlight outages which last about 3 seconds.

The driver is faulting; it will shut down momentarily when something is way-out-of-spec wrong, eg, overheating. That plus the wrong brightness steps means there's a serious problem with the driver.

Dang. Something may have gotten damaged in shipping or installation (I sent you a driver already built onto an inverter board, right?). The only thing I'd expect to be particularly physically fragile is the ferrite coil housing. The other possibilities, as wileE mentioned is something shorting (eg, the LED strip shorting in the bracket) or static damage. This driver chip is particularly static sensitive on several inputs, though it should be reasonably well protected once it's on a board... hmmm. Worst case, I'll need to send another.


Top
   
PostPosted: Sat Feb 27, 2016 2:53 am 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
trmsw wrote:
For what it's worth... I tried the mod in two SXGA+ screens using side-emitting strips from iCCFL and was very disappointed because of the visible LED pattern and uneven backlighting. Yesterday I attempted to mend things by repositioning the strips, and ended up completely removing the backlight bracket and just taping the strip against the waveguide. This was not the original plan, but I'm very happy with the results: bright, even, no pattern, and easy to adjust.
My preferred explanation is that the bracket is counter-productive for LEDs because it acts as convex mirror that focuses them into points. But it's just as likely that the better positioning of the strip that I was only able to achieve without the bracket outweighs the (real) benefits the bracket provides.


It's not so much that it's a focusing mirror that it's a mirror at all. Anything that adds to the reflections that appear in the waveguide makes the patterning worse. But you're right that it's a liability, and you can do better without.

That said, the bracket plays a part in holding the screen together, and messing with it is deeper screen surgery than most people will want to mess with.


Top
   
PostPosted: Sat Feb 27, 2016 3:45 am 
Offline
Freshman Member

Joined: Mon Jan 02, 2012 10:05 pm
Posts: 72
Location: Brussels, Belgium
xiphmont wrote:
That said, the bracket plays a part in holding the screen together, and messing with it is deeper screen surgery than most people will want to mess with.

For sure - and it's a barrier to dust. Removing it wasn't the plan but a shim got stuck and in my attempts to remove it I twisted the bracket...
If you have the time I would be interested in one of the new boards.


Top
   
PostPosted: Sat Feb 27, 2016 9:05 am 
Offline
Senior Member
Senior Member

Joined: Wed May 19, 2010 11:21 pm
Posts: 601
Location: Pardubice, Czech Republic
xiphmont wrote:
You should have 15 roughly evenly spaced brightness levels. AC vs battery should make no discernible difference (regulation is better than .5% across all use cases)

I get 5 steps, sometimes 6 steps and once I counted 7 steps - in bios.
Same thing under an OS, counted during this:
Code:
xbacklight -set 0%;for i in `seq 1 100`;do xbacklight -set $i%;sleep 0.5;done


xiphmont wrote:
Dang. Something may have gotten damaged in shipping or installation (I sent you a driver already built onto an inverter board, right?).

Yes, you sent driver on an inverter board, baggie is labeled "IL x61 41W1468 INV" and there was the number 46 on the led's power cable.

xiphmont wrote:
The only thing I'd expect to be particularly physically fragile is the ferrite coil housing.

If the ferrite coil housing is the long rectangluar thing which doubles as the mounting structure, it was under strain when fitting inverter to lid and I also stuck ~10 cm of excess red and white led-power wire between it and the lcd.
The first time I turned it on the backlight didn't come on and I smelt the tragic scent of sizzling electronic stirfry.
I see a chared spot on the invertor board - will unmount and look closer.
Fwiw, when the machine's cold, level 2 and 3 cause blackouts but after it wams up 2 and 3 don't cause blackouts.

_________________
[debian] x24 [xiphmont led] x60s [libreboot, led] T400 [libreboot, P9600] T420 [Ivy Bridge i7 3612QM]


Top
   
PostPosted: Sat Feb 27, 2016 9:47 am 
Offline
Admin
Admin

Joined: Mon Sep 18, 2006 5:17 am
Posts: 17248
Location: Mt. Cobb, PA USA
First off: I have not read all the details.

So this may not be relevant, but I have done many LED-mods for 15" T60 machines.
The BoeHydis HV150UX2-100 screens already come with LED-backlight ex-factory.
To work in a Thinkpad, either the 30-pin connector on the LCD, or the LCD-cable's connector needs to be 180 degrees reversed.
The original IBM/Lenovo inverter is also modified by connecting 4 wires (Plus/Ground/Backlight/PWM) to it.
These wires then get connected to another 14-pin LED-connector on the back of the LCD.

Finally (and that may be crucial), I remove the transformer (IN :5-20V, OUT: 700+ Volts) from the inverter, because:
- it is no longer needed for LEDs
- eventually it overheats and could even burn out
- it extends battery-life by at least 30 minutes
For details, see http://theboardroom.info/led_mod.htm

HTH.

_________________
Lovely day for a Guinness! (The Real Black Stuff)

Check out The Boardroom for Parts, Mods and Other Services.
And don't forget the Motherboards! Refreshing webpages is advised!


Top
   
PostPosted: Sat Feb 27, 2016 10:26 am 
Offline
Sophomore Member

Joined: Sun May 26, 2013 5:20 am
Posts: 193
Location: Seligenstadt, Germany
jaspen-meyer wrote:
I see a chared spot on the invertor board - will unmount and look closer.
Fwiw, when the machine's cold, level 2 and 3 cause blackouts but after it wams up 2 and 3 don't cause blackouts.

Can you take some good pictures and post the links here?
Can you check with a ccfl + inverter to see how many brightness steps that does with your mainboard? Unlikely to be the cause, but to be sure is better.

@ RBS The mods we are doing here do not have have much in common with the BoeHydis HV150UX2-100 mod. Those screens are unique to the 15" machines and there is nothing comparable in 12" size.


Top
   
PostPosted: Sat Feb 27, 2016 11:25 am 
Offline
Admin
Admin

Joined: Mon Sep 18, 2006 5:17 am
Posts: 17248
Location: Mt. Cobb, PA USA
I only posted here because I saw references to inverters.
And AFAIK the X6x inverter has a trafo, which probably could also be removed (unless you use a LED-board instead of a modded CCFL-inverter).


Top
   
PostPosted: Sat Feb 27, 2016 11:48 am 
Offline

Joined: Sun Jul 19, 2015 4:59 pm
Posts: 9
Location: United Kingdom <please provide city>
RBS: Monty (xiphmont) has made a (tiny) LED driver that replaces the coil on the inverter board, which is what we're all mostly using now.

Here's a picture of the driver board if you're interested:
https://people.xiph.org/~xiphmont/thinkpad/IMAG0480.jpg (large pic warning)


Top
   
PostPosted: Sat Feb 27, 2016 1:39 pm 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
trmsw wrote:
xiphmont wrote:
That said, the bracket plays a part in holding the screen together, and messing with it is deeper screen surgery than most people will want to mess with.

For sure - and it's a barrier to dust. Removing it wasn't the plan but a shim got stuck and in my attempts to remove it I twisted the bracket...
If you have the time I would be interested in one of the new boards.

The boards I'm make are buck drivers for 8-10v led strips, and the strip/kit you have is a boost kit with a 30v LED strip. You'd also need a different strip to work with one of my boards. PM me if you're still interested.

[edit: oh and yeah, they're a pain to get back into place until you figure it out. I use a little 10mil plastic 'shoehorn' to guide brackets back into place. I'll get some pics of it.]


Last edited by xiphmont on Sat Feb 27, 2016 1:50 pm, edited 1 time in total.

Top
   
PostPosted: Sat Feb 27, 2016 1:48 pm 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
jaspen-meyer wrote:
If the ferrite coil housing is the long rectangluar thing which doubles as the mounting structure, it was under strain when fitting inverter to lid and I also stuck ~10 cm of excess red and white led-power wire between it and the lcd.

That's just a plastic shield [or do you mean the whole circuit board?] The X61/X61s inverters always have to be flexed a little to fit in place, so that's as expected. The ferrite piece I'm talking about is a dark gray square under the plastic shield, has a circular area inside the square with a '101' marked on it. If the square ferrite housing of that part cracks (or disintegrates) things won't work right. It's the only part on the board that I'd expect could take damage if abused. That wouldn't necessarily cause faulting though, just a messed up brightness range and lowered efficiency.

jaspen-meyer wrote:
The first time I turned it on the backlight didn't come on and I smelt the tragic scent of sizzling electronic stirfry.
I see a chared spot on the invertor board - will unmount and look closer.

I definitely need to see some detailed pics of that!


Top
   
PostPosted: Sun Feb 28, 2016 11:38 am 
Offline
Senior Member
Senior Member

Joined: Wed May 19, 2010 11:21 pm
Posts: 601
Location: Pardubice, Czech Republic
wileE wrote:
Are you absolutely sure that there are no shorts? Maybe see if the strip works outside the screen?

Yes, there was a short! The end of the led strip was touching the screen brace.
Now on battery power in xubuntu 14.04 I now get 15 levels! On AC+battery there are about 8 levels.

At level 9 of 15 dispcalGUI reports 469 cd/m^2.
The screen gets much brighter at levels >9 but the driver crashes (3 second blackouts) and I didn't test.
I haven't found where to run a test spanning all 15 levels in dispcalGUI (1.7.1.6).

After removing the plastic sheath from the bigger half of the board I didn't see any fried components,
the charred bit I saw just on the side of the green board. Will post pictures.

_________________
[debian] x24 [xiphmont led] x60s [libreboot, led] T400 [libreboot, P9600] T420 [Ivy Bridge i7 3612QM]


Top
   
PostPosted: Sun Feb 28, 2016 3:10 pm 
Offline
Sophomore Member

Joined: Sun May 26, 2013 5:20 am
Posts: 193
Location: Seligenstadt, Germany
jaspen-meyer wrote:
At level 9 of 15 dispcalGUI reports 469 cd/m^2.
The screen gets much brighter at levels >9 but the driver crashes (3 second blackouts) and I didn't test.
I haven't found where to run a test spanning all 15 levels in dispcalGUI (1.7.1.6).

If you have one of the old strips with 48 or 51 LEDs that is too bright at step 9. Your driver board might now be putting out way too much power and that cannot be supported at high brightness levels with AC.
If that is the case you might be frying your LEDs by using it further.

And there is no automatic brightness measurement in dispcalgui. You have to adjust the brightness and take a reading at each step.


Top
   
PostPosted: Sun Feb 28, 2016 3:41 pm 
Offline
Senior Member
Senior Member

Joined: Wed May 19, 2010 11:21 pm
Posts: 601
Location: Pardubice, Czech Republic
wileE wrote:
If you have one of the old strips with 48 or 51 LEDs that is too bright at step 9. Your driver board might now be putting out way too much power and that cannot be supported at high brightness levels with AC.
If that is the case you might be frying your LEDs by using it further.

And there is no automatic brightness measurement in dispcalgui. You have to adjust the brightness and take a reading at each step.

There are 51 LEDs on the strip.
The 469 measurement was a single how-bright-will-this-get test. For daily use it's running at around 250 cd/m^2.

_________________
[debian] x24 [xiphmont led] x60s [libreboot, led] T400 [libreboot, P9600] T420 [Ivy Bridge i7 3612QM]


Top
   
PostPosted: Sat Mar 12, 2016 1:30 pm 
Offline
Sophomore Member

Joined: Sun May 26, 2013 5:20 am
Posts: 193
Location: Seligenstadt, Germany
I have been using one of those Nichia LED strips from Monty in my SXGA+ X61 for a day now and I am delighted. It is like having a completely different screen.

Especially the color red is so much better (even than with a brand new CCFL). Imagine a dark muddy orange turning bright red like a tasty strawberry.
All other colors also look much cleaner and crisper.

The strips are also incredibly well made. Pic: http://i.imgur.com/EJm1YNB.jpg

And with one of those accordion folded platic strips in front of the LEDs I have no pattern at all. Pic: http://i.imgur.com/x0UPWuD.jpg
There is something like a shadow, a 6mm high horizontal stripe, near the bottom edge. That will be caused by the mirror surface in the bracket, but it so faint that it is only visible against white or light grey background. And only detectable if you know what to look for.

Brightness measured with my Spyder 3 (which always gives lower readings than Montys spectrometer) was at 540 nits when pressed against the waveguide, now 485 nits with a diffused folded strip in between.
That is crazy bright in an SXGA+.
Did a quick test in a XGA TN Toshiba screen and got 850 nits max brightness. The Toshiba Ultralight screen in the X6x s models should be similar in brightness. Black levels and contrast are of course horrible in those screens.

The brightness steps are well usable, but some less steps at levels above 300 nits would be even better. Pic: http://i.imgur.com/z7C54Gd.jpg

Color temperature is a little cold, but the colors are nicely balanced and the LEDs are really very well suited for AFFS screens. No real need for color corrections.

Tablet SXGA+ is next. That has a real bad mirrored pattern with one the ICCFL strips.

EDIT

This was the pattern on the X61T - pic: http://i.imgur.com/CJqjIXE.jpg
The machine was banished to serve as offsite backup because of this.

With the Nichia strip pressed against the waveguide I get this - pic: http://i.imgur.com/5xaagKk.jpg
Only a tiny hint of a pattern and very bright.

With a clear accordion folded strip in between, the max brightness again drops to 480 nits. But that eliminates any kind of pattern completely.

This screen has a slightly warmer color temperature than the one in my X61. Probably the LED strip has a different position in the bracket.
Brightness steps are different with the TLD1 driver - pic: http://i.imgur.com/jSzAZTo.jpg
More steps in the usual brightness range. Anything above 300 nits is rescue mode for using it in sunlight anyway. And that will quickly eat the battery - difference between step 2 and 15 seems to be about 4.5W.

My X60s SXGA+ just told me it will stop working if it does not get a Nichia strip too.
Just what I was afraid of. The chinese strips may be better than a dying CCFL, but none of them can compete with the kits Monty has made.


Last edited by wileE on Sun Mar 13, 2016 5:32 am, edited 1 time in total.

Top
   
PostPosted: Sat Mar 12, 2016 7:28 pm 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
wileE wrote:
I have been using one of those Nichia LED strips from Monty in my SXGA+ X61 for a day now and I am delighted. It is like having a completely different screen.

It sounds like the first shipped unit has passed an important field test :-)

wileE wrote:
And with one of those accordion folded platic strips in front of the LEDs I have no pattern at all. Pic: http://i.imgur.com/x0UPWuD.jpg
There is something like a shadow, a 6mm high horizontal stripe, near the bottom edge. That will be caused by the mirror surface in the bracket, but it so faint that it is only visible against white or light grey background. And only detectable if you know what to look for.

Yes, I see that same thing on my main testing screen too. I know how to kill that dark shadow with a little screen surgery, but I'm going to keep looking for ways to kill it without having to disassemble the panel. As you say, it's very faint, and only noticeable once the larger patterning problem is fixed, but it would be nice to fix that too.

wileE wrote:
Brightness measured with my Spyder 3 (which always gives lower readings than Montys spectrometer)

Right, I'm using a narrower field of view, which always yields higher brightness and contrast numbers. Both should be considered accurate, they're just measuring slightly different things (the Spyder is taking more off-angle light into account, my spectrometer only sees ~ directly ahead).

Quote:
Did a quick test in a XGA TN Toshiba screen and got 850 nits max brightness. The Toshiba Ultralight screen in the X6x s models should be similar in brightness. Black levels and contrast are of course horrible in those screens.

TNs are crazy-bright. Too bad everything else about them sucks in comparison :-) BTW, I haven't taken any spectrometer readings of the new strips in one of the TN screens, since it seems like kind of a waste. I would expect them to be much colder, maybe even purplish?

Quote:
Color temperature is a little cold, but the colors are nicely balanced and the LEDs are really very well suited for AFFS screens. No real need for color corrections.

Yes, color is slightly cold. Positioning can shift it warmer by 400-500K easily. In general, the closer the strip is to the waveguide, and the more carefully it's pointed exactly forward, the brighter and colder the light.
Also, the slightly diffused accordian will warm the temp about 150K. Heavier diffusers will warm it more.

Quote:
This screen has a slightly warmer color temperature than the one in my X61. Probably the LED strip has a different position in the bracket.
Brightness steps are different with the TLD2 driver - pic: http://i.imgur.com/TEHOxGS.jpg
More steps in the usual brightness range. Anything above 300 nits is rescue mode for using it in sunlight anyway. And that will quickly eat the battery - difference between step 2 and 15 seems to be about 4.5W.

The two brightness scale readings you took would be what I expect if the TLD1 and TLD2 were reversed... Are the tests labeled backwards?

Power consumption range sounds exactly right. A brand new CCFL eats about 4.2-4.5W at max brightness by my measurements but is less efficient at low brightness, so it doesn't swing quite as wide a range. At only 1W the LEDs should still be very bright.

Quote:
My X60s SXGA+ just told me it will stop working if it does not get a Nichia strip too.
Just what I was afraid of. The chinese strips may be better than a dying CCFL, but none of them can compete with the kits Monty has made.

Yay! I mean _ahem_ how terrible.


Top
   
PostPosted: Sun Mar 13, 2016 5:30 am 
Offline
Sophomore Member

Joined: Sun May 26, 2013 5:20 am
Posts: 193
Location: Seligenstadt, Germany
Quote:
Yes, I see that same thing on my main testing screen too. I know how to kill that dark shadow with a little screen surgery, but I'm going to keep looking for ways to kill it without having to disassemble the panel. As you say, it's very faint, and only noticeable once the larger patterning problem is fixed, but it would be nice to fix that too.

You do realize that you only see that if you really want to. It is so faint that it will never bother me.
The problem of uneven brightness distribution is much worse. All the SXGA+ screens I have ever measured were darker on the left side and much brighter on the right. I have no explanation for that. Maybe some pressure inside the screen. If the LED strip were twisted it should be visible through the changed color temperature.

Quote:
TNs are crazy-bright. Too bad everything else about them sucks in comparison :-) BTW, I haven't taken any spectrometer readings of the new strips in one of the TN screens, since it seems like kind of a waste. I would expect them to be much colder, maybe even purplish?

It is cold, but clean cold. Spyder says 8550K. No tint of any kind. That screen was cold with CCFL too, just not this much. Colors are much nicer with the Nichias.
But yes, it would be a crime to use one those precious strips in such a screen. The non existent viewing angles of those Toshibas are beyond description.

Quote:
The two brightness scale readings you took would be what I expect if the TLD1 and TLD2 were reversed... Are the tests labeled backwards?

Yes, of course, you are right. Going to fix that.

EDIT

Thank you again for making these awesome LED strips. Finished my X61T now and I love that machine again.

Used a 0.5mm plastic strip taped to the back and a clear accordion folded strip in front. The outer LEDs are 1/2 covered with black tape to avoid too bright corners.
And I had to use a small drop of glue to keep the strip from twisting when bending the cables into the needed position.

Max brightness is at 522 nits. No pattern at all. Color temperature is at 7344K, but it appears warmer than the X61.

Colors are as perfect as can be expected from this screen. Although the covered color space is smaller than on my external screen, the colors appear similarly well balanced.

Screen in bright sunlight - big pic: http://i.imgur.com/FWeorGY.jpg


Top
   
PostPosted: Sun Mar 13, 2016 3:42 pm 
Offline
Sophomore Member

Joined: Sun Nov 18, 2012 5:42 pm
Posts: 244
Location: Bay Area (Berkeley) , California
Looks absolutely amazing. Chris continues to astound me...

R


Top
   
PostPosted: Sun Mar 13, 2016 10:27 pm 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
wileE wrote:
You do realize that you only see that if you really want to. It is so faint that it will never bother me.

Perfection is usually unattainable. It can be fun to chase it though.

Quote:
The problem of uneven brightness distribution is much worse. All the SXGA+ screens I have ever measured were darker on the left side and much brighter on the right. I have no explanation for that. Maybe some pressure inside the screen. If the LED strip were twisted it should be visible through the changed color temperature.

Interesting, I had not noticed such a problem problem. I'm currently away on business for another week, so I can't check till I'm back (I did not bring my X61T this time). Do you notice it visually, or is it something that is more obvious in measurement? I can think of a few possibilities, but I'm most interested in whether or not it's been happening in my tests and I'd never even noticed.

As you say, I don't think it's twisting. My first thought is that I've been trimming away one of the plastic ridges that pushes the strip slightly forward on the right-edge-extreme of the channel, and you're not. On my own test screens I trim the plastic on the right more aggressively than I mention on my web pages, but I didn't think it made a difference or I'd have mentioned. Maybe it does. Or maybe the problem is still there and I never noticed it.

Quote:
The outer LEDs are 1/2 covered with black tape to avoid too bright corners.

Good thought. The CCFL is actually just a little shorter than the waveguide, and they compensated for that in the dot pattern at the edges :-P

Quote:
Max brightness is at 522 nits. No pattern at all. Color temperature is at 7344K, but it appears warmer than the X61.

Hm, maybe the front acrylic is affecting it a bit. BTW, I've found that whether you use the pressure-touch feature or not, the XGA-style glass/touch both warms the color temp slightly and has a way better anti-reflective coating.

Quote:
Screen in bright sunlight - big pic: http://i.imgur.com/FWeorGY.jpg

Hooray! I've been privately calling this latest revision the 'daylight' backlight, but I didn't want to say that out loud in case it did not live up to the moniker :-)

OK, I should probably start updating my pages with recent plans and advice then. I have parts left for another 60 kits or so and I think I may rev the driver boards to include a small trimmer potentiometer to allow in-the-field brightness range adjustment.


Top
   
PostPosted: Mon Mar 14, 2016 9:05 am 
Offline
Sophomore Member

Joined: Sun May 26, 2013 5:20 am
Posts: 193
Location: Seligenstadt, Germany
xiphmont wrote:
Quote:
Max brightness is at 522 nits. No pattern at all. Color temperature is at 7344K, but it appears warmer than the X61.

Hm, maybe the front acrylic is affecting it a bit. BTW, I've found that whether you use the pressure-touch feature or not, the XGA-style glass/touch both warms the color temp slightly and has a way better anti-reflective coating.

The front acrylic is part of the cause, also reduces contrast noticeably. Angle of the LED to the waveguide is probably slightly different.
I remember the touch front having a very distinct yellow/brown tint. That was very bad with CCFL, but might be fun with the Nichias.
Problem is getting one without bad scratches and other marks.


Brightness distribution is a really bad topic when you like chasing perfection.
Look at the measurements on the notebookcheck.net page first. They divide the screen into 3 by 3 fields for the measurement and call anything with brightness distribution of 80% and above good.

DispcalGUI has the option to divide the screen into 3 by 5 fields and generate a html report with the results. The additional 6 fields have the effect of always getting worse percentage results then the notebookcheck method.

Uniformity Check - HV121P01-101 - 2014-09-07 X60s - CCFL: https://www.dropbox.com/s/gbdzxy9hxd2i9r9/Uniformity%20Check%20-%20HV121P01-101%20-%202014-09-07%20X60s%20-%20CCFL.html?dl=0

Uniformity Check 1.7.1.6 — HV121P01-101 - 2016-03-12 X61T-Nichia LED_obsolete: https://www.dropbox.com/s/dwo3ssyj8svcbf5/Uniformity%20Check%201.7.1.6%20%E2%80%94%20HV121P01-101%20-%202016-03-12%20X61T-Nichia%20LED_obsolete.html?dl=0

Uniformity Check - HV121P01-101 - 2016-03-14 Nichia strip in X61T_final: https://www.dropbox.com/s/xoqrvtf7b4v8s1j/Uniformity%20Check%20-%20HV121P01-101%20-%202016-03-14%20%20Nichia%20strip%20in%20X61T_final.html?dl=0

The X61T final may look bad with measurements of 584 on the right side and 408 in the upper left corner, but it is not that much worse than the measurements with the CCFL.
That brightness difference is impossible to see in ambient daylight, same if you have normal room lights on, and only if you use a middle gray image and sit in complete darkness will you be able to notice it.

Quote:
Hooray! I've been privately calling this latest revision the 'daylight' backlight, but I didn't want to say that out loud in case it did not live up to the moniker

The screens with the acrylic front are usually very bad when the sun shines in them. Not a problem anymore. :D


Top
   
PostPosted: Thu Mar 17, 2016 6:58 pm 
Offline

Joined: Mon Feb 06, 2012 5:09 am
Posts: 19
Location: Velikie Luki, RU
Quote:
The problem of uneven brightness distribution is much worse. All the SXGA+ screens I have ever measured were darker on the left side and much brighter on the right. I have no explanation for that. Maybe some pressure inside the screen. If the LED strip were twisted it should be visible through the changed color temperature.

I had a similar problem when installing LED strips.
My display is very dark compared to your display. The Chinese have slipped me a bad display.
Backlight from xiphmont, works is stable and without problems =)


Top
   
PostPosted: Tue Mar 22, 2016 12:10 pm 
Offline
Senior Member
Senior Member

Joined: Wed May 19, 2010 11:21 pm
Posts: 601
Location: Pardubice, Czech Republic
I've installed the replacement driver in my machine and now I've got normal brightness levels.

WillE, here are the pictures of the cooked (?) board:
http://www.jesseo.com/thinkpads/ (x61-inverter*)

_________________
[debian] x24 [xiphmont led] x60s [libreboot, led] T400 [libreboot, P9600] T420 [Ivy Bridge i7 3612QM]


Top
   
PostPosted: Tue Mar 22, 2016 12:21 pm 
Offline
Junior Member
Junior Member

Joined: Wed Apr 07, 2010 2:46 pm
Posts: 274
Location: Jaffrey, NH
jaspen-meyer wrote:
I've installed the replacement driver in my machine and now I've got normal brightness levels.

WillE, here are the pictures of the cooked (?) board:
http://www.jesseo.com/thinkpads/ (x61-inverter*)


A fried C3 would certainly give you some odd behavior.

I can't think of anything you could have done to have caused that accidentally without noticing, eg, your worry about shorting the output should be harmless. Could you drop it in the mail back to me? My guess is it was an assembly defect on my part (errant solder? mechanically damaged cap?), and I'd like a better look.


Top
   
PostPosted: Tue Mar 22, 2016 4:01 pm 
Offline
Senior Member
Senior Member

Joined: Wed May 19, 2010 11:21 pm
Posts: 601
Location: Pardubice, Czech Republic
xiphmont wrote:
Could you drop it in the mail back to me?

I'll drop it in the mail tomorrow.

_________________
[debian] x24 [xiphmont led] x60s [libreboot, led] T400 [libreboot, P9600] T420 [Ivy Bridge i7 3612QM]


Top
   
Display posts from previous:  Sort by  
Post new topic  Reply to topic  [ 303 posts ]  Go to page Previous 16 7 8 9 10 11 Next

All times are UTC-05:00


Who is online

Users browsing this forum: No registered users and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Limited