Thinkpads Forum

I've just finished an update that hopefully delivers on a few good suggestions folks have had...
I also added a new kit that several vendors just started carrying, which I've dubbed the 'Unboosty-Blue'. It's my new favorite-- electrically equivalent to the Unboosty, but the PCB layout is spaced just right to make a very clean/compact version of the continuous mod possible.

Main page is still here: https://people.xiph.org/~xiphmont/think ... ight.shtml
New kit is here: https://people.xiph.org/~xiphmont/think ... styB.shtml

xiphmont wrote:I've just finished an update that hopefully delivers on a few good suggestions folks have had...

Wow. Keep it coming, xiphmont. Your explanations of the various mods and options are great - detailed, yet clear.

Phil.

I was hanging out with my dad this evening and saw that he had a Spyder 2 monitor calibrator. I used it and it helped tremendously with my modded display. It isn't perfect, but it is a lot better than I was able to get by changing the settings manually.

Hello,

anyone oknows if this kit http://www.xccfl.com/product_info.php?c ... ts_id=4728 is ready-to-roll for an unflickering and bright use?

-or-

where can i get the unboostyB driver? trying whatever-search combo in google and no luck

thanks

Qing Dao wrote:anyone oknows if this kit http://www.xccfl.com/product_info.php?c ... ts_id=4728 is ready-to-roll for an unflickering and bright use?

That kit will easily at least give you double the brightness your screen had when it was new. But if you are sensitive to flicker, you will notice it.
I only see flicker with that kit on dark gray backgrounds when i dimm the screen. On Gnome3 startup background it is most noticeable.

But it has not bothered me enough to do the high frequency mod.

This could be the unboosty you are looking for:http://www.ebay.com/itm/255mm-12inch-la ... 1417429209

And if you order at xccfl you will probably get an email again asking for the details of your screen. Be sure to make clear that you are ordering for the X6x TABLET.

I thought part of the point of xiphmont's work was that the available backlight kits didn't give good color on the SXGA (tablet / AFFS / IPS) screens, no? I think I missed something here if this kit works fine in a X61T screen...

R

Thank you gents, for such a prompt responce.

Im not keen on the colors (256colors are enough! ), i need the light, my x61t is dimmed and yellowish, so i want to avoid the moment, when it will go dark.

wileE, one quick question, can it be controlled with the brightness buttons and in Linux without any issue/modifications?

Rgds,
Martin

The fancyboost kit you picked on XCCFL works without any modifications.
And it does not matter if you use Windows or Linux for the brightness adjustment.

And yes, the colors are not quite as nice as before, but the huge increase in brightness far outweighs this.

nakopla_plechovka wrote:Hello,

anyone oknows if this kit http://www.xccfl.com/product_info.php?c ... ts_id=4728 is ready-to-roll for an unflickering and bright use?

-or-

where can i get the unboostyB driver? trying whatever-search combo in google and no luck

thanks

That xccfl kit uses the Fancyboost. It plugs right in, no mods. The PWM runs at 60Hz (so some flicker at less than max brightness), and you can mod it to run flicker-free if it turns out you care. I'd say the kit is actually about 40% brighter than a brand new CCFL. It will be much more than twice the brightness of a CCFL that's a few years old. The color of the LEDs is rather yellow for an AFFS screen, you'll notice the color temperature is rather low.

The 'Unboosty-Blue' is my name for that specific kit, the vendors don't give them names Right now it seems all the vendors selling adjustable-brightness 9.6V (trim-in-threes) kits except for CreatAll are shipping the unboosty-Blue. CreatAll sells the regular Unboosty variant.

flyingfishfinger wrote:I thought part of the point of xiphmont's work was that the available backlight kits didn't give good color on the SXGA (tablet / AFFS / IPS) screens, no? I think I missed something here if this kit works fine in a X61T screen...

Correct. Still working on good color matches for the AFFS screens. I have a small stock of good LEDs right now, but no supplier to get more and nowhere I can send other DIYers to get some.

The stock kits are usually OK for AFFS, but you'll notice the color reproduction is much more yellow tending toward yellow-green. Some kits are close enough you only notice it when comparing side-by-side (but then it's really obvious

Do you have any pictures of what the good ones look like, or do they all look the same? Are there any distinguishing characteristics?

Thanks once again for all those details, i was reading in-and-out the page of xiphmont.

If there are any "proper" LED stripes, is it possible to make a spectral measuring (http://publiclab.org/wiki/foldable-spec) to know for what to look? With the proper color tempterature and-or wavelengths it should be no problem to achieve it using RGB LEDs

Rgds,
Martin

nakopla_plechovka wrote: If there are any "proper" LED stripes, is it possible to make a spectral measuring (http://publiclab.org/wiki/foldable-spec) to know for what to look? With the proper color tempterature and-or wavelengths it should be no problem to achieve it using RGB LEDs

I took a detailed spectrometer reading; it's earlier in the thread, link repeated here:

led-spectrum-affsmatch.png

The LEDs in question are normally constructed white LEDs, but a very cold white and likely made specifically for CCFL replacement. They do not have a color temperature; they're past the infinity point of the blackbody curve. The CIE 1931 chromaticity is approximately x=.25, y=.20.

Qing Dao wrote: Do you have any pictures of what the good ones look like, or do they all look the same? Are there any distinguishing characteristics?

Yes, I took a few photomicrographs; links were earlier in the thread, repeating here:

Top
Bottom
Profile
Internal (lens ripped off, unfortunately, chip ripped off with it)

The key identification points are the half-circle green dot, the internal trace structure, the very thin pads on the top with a separate piece of metal making up the pads along the side, the very wide pads on the bottom that crowd the green 'T' cathode marker, and a fairly thick profile.

I don't think they're particularly high quality LEDs (that package is low-rent), but they tested well, they're exactly the right color, and they're a fairly modern low-Vf junction so they're efficient.
Any LED that meets the criteria is fine, especially as I've not found a source for these.

Thanks. I'll keep an eye out for them.

xiphmont wrote:That xccfl kit uses the Fancyboost. It plugs right in, no mods.

No soldering, no removing components or adding them n bits of wire?. Just refit.

Dekks wrote:

xiphmont wrote:That xccfl kit uses the Fancyboost. It plugs right in, no mods.

No soldering, no removing components or adding them n bits of wire?. Just refit.

Correct. There's an _optional_ mod to eliminate the last traces of flicker, and that requires soldering. But the kit works just fine without it (using 200Hz PWM).

You still need to remove the old CCFL backlight and install the LED strip though-- so the howto page on that still applies.

xiphmont wrote:

Dekks wrote:
No soldering, no removing components or adding them n bits of wire?. Just refit.

Correct. There's an _optional_ mod to eliminate the last traces of flicker, and that requires soldering. But the kit works just fine without it (using 60Hz PWM).

You still need to remove the old CCFL backlight and install the LED strip though-- so the howto page on that still applies.

Cheers, might have to WTB some modded kits as a i probably shouldn't attempt soldering with lol. I'm watching ebay for a few broken X61 to try disassembling the screens as dry runs.

Well, dang.

Several of the mods are not working nearly as well on battery power as I'd hoped. The dimming range is off when running on unregulated battery power (they run brighter on battery then AC), but the truly unacceptable problem is that some the Unboosty and Unboosty-Blue kits flicker with CPU load when running on battery.

Some more detail: The way they're set up uses a DF6113 boost controller as a buck-down controller instead, which is a weird design the spec sheet doesn't even suggest is possible. The first batches I got seemed to work great anyway. Now I have a batch that really really doesn't. The first stage's current sense accuracy is all over the place, which isn't a problem when the second-stage error amplifier is correcting for it, but the error amplifier isn't usable in this design. The accuracy problem can be compensated with an additional resistor, but the correct resistor would be different for different batches, and it still doesn't fix the brightness range shift when on battery. *That* can be fixed by leaving the MP1010 in place and pulling a 5V reference voltage off of it, but.... and so on.

In summary, it's not really a huge unfixable problem, but I think this has finally pushed me past the point where a new design, made perfect for Thinkpads from the ground-up is easier, cheaper, and more reliable than trying to hack some random Chinese board that may or may not be the same in six months. So, I'm testing new designs now. The good news will be no more cutting, filing, and soldering cantilevered connections to 0402 under a magnifying glass! And I figured out how to mod onto an inverter board without removing the big preexistig controller chip, which should also make things easier.

Anyone here who already built one of the modded boards and was disappointed, ping me. I'll be more than happy to straight up send you a new controller board that works properly once I have them. I'll get a disclaimer up on my webpages as soon as I can (currently away at IETF without all my keys...)

With the assistance of this thread and your website-as well as the German Thinkwiki site-I just got finished installing an LED kit from iCCFL.com into a SXGA+ display from an X60 Tablet into my X61. I received a kit with the side mounted LED's, and using your instructions removed the CCFL tube, and installed the LED strip without any issues. I took your advice and installed the LED strip with the LED's pointed towards the acrylic backlight layer and with the entire strip at an angle so that the LED's would shine at a slightly downward angle. The fit was very tight, so no shimming was required.

I got lucky on the first attempt, as the backlighting is VERY bright and even, with no real hot or dark spots-it's the best results of any backlight replacement I have ever performed. My eyes aren't seeing any flicker, and I have a good range of dimming. Colors seems to be pretty close to normal, although I do think I see the color tint you mentioned. But honestly, it doesn't bother me one bit-to me it looks great.

One other note-I purchased the video adapter cable from eBay seller i.thinkpad, and it works perfectly with no issues. I don't have the eyesight to be able to solder such small connectors, so I was glad to pay someone else to do it for me.

Dumb question, but which kit exactly did you use? I've been meaning to try this for ages.

trmsw wrote:Dumb question, but which kit exactly did you use? I've been meaning to try this for ages.

This one from iCCFL.com: http://www.iccfl.com/product_info.php?products_id=11862

I myself thought about ordering the kit for quite a while, but the total price of $48 (including shipping) always drove me away. I am glad I finally went ahead with it, as the results are definitely worth the cost-especially for these somewhat hard to find 4:3 SXGA+ LCD's.

A quick update...

Designing an analog dimming LED driver that covers 40:1+ and doesn't let the brightness drift by more than 1.5% over a 10v-20v input range proved somewhat challenging! (Existing drivers that do this use PWM, but I wanted fully analog drive both to eliminate any hint of flicker, and because it's more efficient). I evaluated twenty driver chips over the past months...

The good news is that I just sent in two prototypes for a short fabrication run. We shall see soon if they pan out. They worked well when assembled point to point.

Proto 1: Based on the Texas Instruments TPS92510 buck driver, 93% efficient, 10v-24v input, dimming range of up to 8-335mA (40:1)
Proto 2: Based on the Richtek RT8450B multi-topology driver, 88% efficient, 10-24v input, dimming range of up to 4-320mA (80:1)

Both are 44.5mm x 8.25mm, will fit easily on any of the X6x inverters, are plug-n-play, continuous drive, and have a nonlinear dimming range with the brightness steps spaced more perceptually evenly. They're designed to work on T4x and T6x machines with bigger panels when configured for 500mA drive, though I've not any of those to test. I'm concerned machines with 3-cell batteries might be too close to the bottom end of the regulation envelope with the battery sagging down under 10.5v near full discharge. Not an issue with 4-cell models, they're never below 14V.

These new drivers are also more power efficient than any of the chinese LED kits I've tested, which all coincidently come it at right around 87%. If the TI board works out, that's almost an additional extra half watt of savings over the Chinese kits (in machines that idle at around 7W!)

Thanks for your dedication xiphmont.

This is one more step to the ultimate X6x series Thinkpad!

xiphmont wrote: Designing an analog dimming LED driver

Wow. Wish you success with that.

And of course interested in such a part. Maybe the chinese will live up to their reputation, and copy them quickly.

Just bought a lot of capacitors to do your high frequency PWM mod on the fancyboost drivers.

wileE wrote:

xiphmont wrote: Designing an analog dimming LED driver

Wow. Wish you success with that.

They arrived today, testing tonight!

And of course interested in such a part. Maybe the chinese will live up to their reputation, and copy them quickly.

I don't know how likely that is. The Chinese boards are all using generic drivers that are useful in a wide range of applications, so they're not perfect for ours. My boards probably won't work far beyond the Thinkpad requirements due to the size, dimming range, and analog requirements dominating the design tradeoffs, and they're using much pricier controllers ($2 instead of $.015).

Just bought a lot of capacitors to do your high frequency PWM mod on the fancyboost drivers.

The Fancyboost is actually a good design, not cheap at all, using a very nice controller chip. Its only drawback is the narrow dimming range. The HF mod works well.

Spent part of the night testing prototype #1... and it came through tests brilliantly. I did need to adjust one of the dimming resistor values, but that's not too surprising. Hooray! The boards are good, I don't need to toss 'em and try again! (I didn't use a electronics design package to design the boards, I just drew the PCB gerbers freehand, sent them off for fabbing, and hoped I hadn't made any mistakes).

The only disappointment is that the snubbing resistor I added to the design to improve regulation is still beneficial on the fabbed board. I was hoping that it was helping only because my early tests were using a solderless breadboard, and those really suck up over 100kHz. Oh well, it's still optional (the snubber reduces regulation variance with Vin changes by about 75%, but it also reduces efficiency by 1%. I'm *still* 4% more efficient than the best Chinese board I've tested).

The second prototype was mostly a hedge, but I'll test it too (coming up).

Did the high frequency mod on my fancyboost kits today. The flicker was really starting to bother me.
Much better now, although I lost 20 nits max. brightness.

Thank you for the guide.

wileE wrote:Did the high frequency mod on my fancyboost kits today. The flicker was really starting to bother me.
Much better now, although I lost 20 nits max. brightness.

Yes, that's about right. The minimum brightness is a little higher too; you lose a bit off both ends. You can get the max brightness back by bumping down the sense resistance value a bit if you want.
BTW, did you actually measure it, or is that just approximate? (I could add it to the howto page)

I made before & after measurements with my Spyder. I would never claim to be able to see a difference of 20 nits.

xiphmont wrote:You can get the max brightness back by bumping down the sense resistance value a bit if you want.

How exactly would I do that? Dont assume that I know anything about what these parts I solder actually do. Complete dummy where electronics are concerned.

Having problems with the PWM mod on older fancyboost kits. Flicker is still visible after the mod.

1. Kit from 2013. Has the same layout as in the picture from FryPpy. Was also ordered for X61T, and delivered for X61.

When still on the X61 board this kit had the problem that the LED did not completely shut off when the charger was plugged in. When connected to a strip with groups of 8 and 9 LED, only the groups with 9 LED would turn off.
This problem went away when I put the driver board on a inverter board that I modified myself. Never found out where the problem was.

Now I did the PWM mod, and the flicker is still visible, but also visibly faster. Not sure if it is better or worse now.

2. Kit from 2012. Has a completely different connection to the LED strip. And was delivered with a flexible LED strip with 36 LED.

Flicker was much less than with the 2013/14 kits.

Built an adapter to plug in the strips with 48 LED, and did the PWM Mod.

Some pics: (Warning - pics are larger then 400KB)
http://i.imgur.com/RpBuczJ.jpg
http://i.imgur.com/hOeOdAG.jpg
http://i.imgur.com/RKCVIoN.jpg

While not nearly as bad as with the 2013 kit, I can still see flicker in gray images when I dimm down halfway.

Any ideas on what might be going wrong? The mod works well on the two kits from 2014 I did before.