There are so many posts that need to come out of this one...
I ordered a whole bunch of Chinese LED conversion kits. I tried lots of ways to mod them to work on Thinkpads, all of which worked... though many turn out not to work well enough, so don't use the mods on those pages for now :-(
Dissatisfied with all offerings, I designed and have been testing my own LED driver boards....
The LED strips that came with the Chinese kits all had terrible color rendering (low temperature, greenish, or both) on the spiffy AFFS screens the X61T is known for.
...so that's sucked me into the world of spectrometers, reverse engineering the protocols to use them on Linux (oh, yeah, I need to publish that), binning leds by hand with a makeshift integration chamber...
...and then of course all the software needed to take the SPD data from the assembled system and interpret the color data, then plot it, which I just finished today!
That contrast measurement is too low and I know why, the sensor is too close and unbaffled, so it's picking up IPS glow. Time for measurement jig rev 2....


Spectrometers under Linux
Date: 2015-04-30 01:22 pm (UTC)Re: Spectrometers under Linux
Date: 2015-05-11 10:25 am (UTC)The original ColorHug was a fairly typical colorimeter. It needed different calibration matrices for different kinds of backlights. It was remarkable primarily in that it was low cost and very well documented.
The ColorHug2 is built around a much more expensive sensor that claims to be a native CIE1931 XYZ 2-degree observer, which is awesome: XYZ is what I compute from the raw SPD data I'm getting from my spectrometer. My roomie here in Mountain View (Thomas Daede) decided to pick one up... and we've found that, out of the box at least, the data it returns is nowhere close to the XYZ integrations I'm calculating from the spectrometer data. We're not sure why. Using both the bundled software and the color management software in Gnome, calibrations look nothing like what I get from my spectrometer or a Huey Pro (both of which agree to within a percent, which surprised the heck out of me.)
Not sure what's up, we're still playing with it. Obviously a mail to Richard is in order if we don't figure it out.
no subject
Date: 2015-05-11 06:53 am (UTC)Some weeks ago, I came across the pages you linked above during a Google search and today considered to finally buy one of the proposed kits. Then I found this post during more googling.
Since you can't really recommend the kits/mods and have designed your own board, I'd be interested in getting one of those. Will you be selling those boards or publishing the design?
no subject
Date: 2015-05-11 07:27 am (UTC)I can just send you one of the boards I already have, I figured the first batch was for testing anyway. You can let me know how it works on a T61. BTW, do you have a TN screen or FlexView IPS? Matching backlight color can be tricky depending on the type of screen.
no subject
Date: 2015-05-11 11:15 am (UTC)It's a TN panel. I thought about swapping it for an IPS previously, but considering the age of the laptop and the difficulty of getting a screen that fits and getting it to work with the BIOS, it doesn't seem worthwhile.
no subject
Date: 2015-05-11 11:25 am (UTC)OK, TN is easier than IPS/AFFS. The IPS screens need a very cold backlight color, it's hard to find. Most of the strips out there will be decent on TN.