Accelerating demise of Video Projector DLP dead pixels.

[BrianHG]

After googling around, beside replacing my DLP chip, does anyone here know of anything I can do?

This is the second time.  It beings with 1 pixel, the next day 2, the next day 4 and grows from there.
The fans still work correctly and everything else is ok.

Is there anything I can do before the dead pixel gets beyond usability?

The projector has been in use for just over 4 years and the optics on this one is so good, it would be a shame to loose is to newer garbage out-of-focus in corners junk available today.

[tom66]

I've long wondered what causes this -- do you have an original bulb fitted or is it third party?

A former acquaintance of mine noticed that their projectors with dead pixels were all ones that had the bulb replaced.  The suspicion was the non-OEM bulbs that the school board would pay for produce a higher amount of UV - maybe the quartz glass does not attenuate it sufficiently or the arc runs at a different temperature.  This UV damages the transistors on the DLP array (speculation, but it's a known phenomenon for other semiconductors).  It's not immediate, it just increases the normal rate of failure, but when one pixel fails the others are rarely far behind.

[Gyro]

I don't know what supply rails a typical DLP chip needs, but it could be that one of them has gone badly out of spec.

[BrianHG]

I've long wondered what causes this -- do you have an original bulb fitted or is it third party?

A former acquaintance of mine noticed that their projectors with dead pixels were all ones that had the bulb replaced.  The suspicion was the non-OEM bulbs that the school board would pay for produce a higher amount of UV - maybe the quartz glass does not attenuate it sufficiently or the arc runs at a different temperature.  This UV damages the transistors on the DLP array (speculation, but it's a known phenomenon for other semiconductors).  It's not immediate, it just increases the normal rate of failure, but when one pixel fails the others are rarely far behind.

Not the bulb as I measured the Lumens after switching.  I changed just the bulb, not the housing which contains the UV filter glass.  The problem begun after 3 consecutive really hot days last week and I've used the projector for at least 20,000 hours in full power brightness so far.  My best guess is that the paste between the DLP and it's heat-sink has finally dried out and now we are in a run-away avalanche situation.

(Yes I once owned a cheap 1280x1024 DPL which did die a month after switching the bulb to a third party one, which did appear brighter...)

[tooki]

I've long wondered what causes this -- do you have an original bulb fitted or is it third party?

A former acquaintance of mine noticed that their projectors with dead pixels were all ones that had the bulb replaced.  The suspicion was the non-OEM bulbs that the school board would pay for produce a higher amount of UV - maybe the quartz glass does not attenuate it sufficiently or the arc runs at a different temperature.  This UV damages the transistors on the DLP array (speculation, but it's a known phenomenon for other semiconductors).  It's not immediate, it just increases the normal rate of failure, but when one pixel fails the others are rarely far behind.

What would be interesting is a comparison between OEM and third-party bulbs. Without that comparison, it’s impossible to know whether the third-party bulb is actually causing accelerated failure, or whether it’s simply that dead pixels start to appear right around the same number of hours as it takes for a bulb to burn out.

[BrianHG]

I've long wondered what causes this -- do you have an original bulb fitted or is it third party?

A former acquaintance of mine noticed that their projectors with dead pixels were all ones that had the bulb replaced.  The suspicion was the non-OEM bulbs that the school board would pay for produce a higher amount of UV - maybe the quartz glass does not attenuate it sufficiently or the arc runs at a different temperature.  This UV damages the transistors on the DLP array (speculation, but it's a known phenomenon for other semiconductors).  It's not immediate, it just increases the normal rate of failure, but when one pixel fails the others are rarely far behind.

What would be interesting is a comparison between OEM and third-party bulbs. Without that comparison, it’s impossible to know whether the third-party bulb is actually causing accelerated failure, or whether it’s simply that dead pixels start to appear right around the same number of hours as it takes for a bulb to burn out.

Yes, some non-oem bulbs do burn too bright.  Only a few extra lumens condensed into a tiny few mm squared spot will be an extra hundred degrees or so.  Also, if the bulb doesn't have a good central focus spot, IE, the top left corner of the picture is too much brighter than the bottom right, this is where you may be concentrating too much heat, like almost double in that corner.  If you don't recognize this and bitch to those cheap Chineese 40$ bulb guys to replace the bulb, if yo keep on using it, you are guaranteed to kill you DLP within a month as the unbalanced heat stress will destroy your chip.

There are 2 factors in the 40$ reject bulbs which you may get sold as new compared to the 250$ authentic manufacturer bulbs.  Overall brightness and central alignment and focus of the output spot.  Sometimes, the 40$ reject is fine since it just barely failed the QC test, other times, you are playing with fire if you do not know what you are doing.

Laser illumination looks to be the way to go, not that the temperatures are low, but still much lower than what we have now with the mercury arc lamps and none or a fraction of the invisible colors in the optical spectrum.  Also, corner to corner illumination balance seems to be vastly superior, almost as good as a high performance LCD screen.  (We are talking about consumer grade projectors, I know the above 50k$ have always had uniform corner-corner brightness...)

[tooki]

What I find fascinating about laser projectors, as we’ve actually gotten them, is that they rely entirely on blue lasers and then phosphors to create yellow for the red and green channels. (For those who don’t know, single-chip DLP laser projectors use two color wheels: one with RGB (or more) filters, just like in traditional DLP projectors, plus another color wheel with one transparent section for blue, and the rest with the yellow phosphor.

Similarly, there are now high end car headlamps that use the same basic idea of a blue laser and phosphor (under the name “laser activated remote phosphor”).

In both cases, I wonder what the advantage is of the blue laser over using either RGB LEDs or blue LEDs and yellow phosphor. If someone knows, please share!

[AndyC_772]

Does your DLP projector have an 'eco' mode that runs the bulb at lower brightness?

It might be worth giving that a go, and see if the degradation slows. Either way, it'll give a clue as to whether the failures are related more strongly to time or to temperature.

I remember studying the workings of the DLP chip in some detail as part of my degree course. It's frankly amazing that they work at all, IMHO.

[tooki]

I remember studying the workings of the DLP chip in some detail as part of my degree course. It's frankly amazing that they work at all, IMHO.

Absolutely! I think the same can be said for many technologies. (For example, hard disk drives. It’s a miracle they are anywhere near as reliable as they are. And that’s accepting the fact that they’re one of the least reliable bits of IT/electronics gear we use.)

[tom66]

I wonder what the advantage is of the blue laser over using either RGB LEDs or blue LEDs and yellow phosphor. If someone knows, please share!

I couldn't say for projectors, but as I understand it the advantage in car headlamps is that because the light output is extremely coherent, it is possible to design a far more precise lens.  This allows for brighter light in the allowable area of the road, whilst reducing the glare for oncoming traffic.

In fact, this is the same motivation behind segmented headlamps.  A different way to solve the same kind of problem.

[bw2341]

In both cases, I wonder what the advantage is of the blue laser over using either RGB LEDs or blue LEDs and yellow phosphor. If someone knows, please share!

https://training.ti.com/etendue-how-brief-introduction-etendue-projection-systems

It's the étendue of the optical system. I stumbled upon this TI training video. In even more more simplified terms, the tiny exit aperture and narrow cone of light from the laser is much easier to work with as a light source. The large surface area and broad emission angles of LEDs would require larger lenses and larger DLP chips to achieve the same optical efficiency.

[ebastler]

For those who don’t know, single-chip DLP laser projectors use two color wheels: one with RGB (or more) filters, just like in traditional DLP projectors, plus another color wheel with one transparent section for blue, and the rest with the yellow phosphor.

Sharp/NEC have nice diagrams of the different projector types' beam paths, including animated versions for each of them: https://www.sharpnecdisplays.eu/p/laser/en/technologies.xhtml. Here's the single-DLP, laser+phosphor illuminated design:

[Berni]

I wonder what the advantage is of the blue laser over using either RGB LEDs or blue LEDs and yellow phosphor. If someone knows, please share!

I couldn't say for projectors, but as I understand it the advantage in car headlamps is that because the light output is extremely coherent, it is possible to design a far more precise lens.  This allows for brighter light in the allowable area of the road, whilst reducing the glare for oncoming traffic.

In fact, this is the same motivation behind segmented headlamps.  A different way to solve the same kind of problem.

I think the benefit of those laser phosphor headlamps is that the light emitting area is very very small. The phosphor does not preserve the coherence of light. But the small spot size does make it easier for the optics to collect all the light and direct it into a tightly controlled beam. Quite a lot of complication for just a light in my opinion.

In projectors i am guessing phosphors are used to reduce the cost of the unit by requiring only one high power laser and no alignment to other lasers. There certainly are projectors that use a RGB laser too. Most of the tiny portable ones do this (needs less optics so it is smaller).

There are OLED screens that use phosphors to produce the red and green from the blue LEDs. LCDs are also moving towards using quantum dot technology to convert a blue backlight into red and green.

[ebastler]

bw2341 has given the right answer above, I think: Coherence of the laser light source is not relevant here. Collimation (small divergence angle) of the laser is relevant, in combination with its small beam diameter. Actually, the product of beam cross section and divergence angle, known as étendue, is the key property.

As discussed in the short video linked by bw2341 (worth watching!), étendue is preserved in a perfect optical system, and can never get smaller in a real-world system. The core DLP element can only accept a limited étendue: Its mirror array has a limited size, and the tilt angle of the mirrors limits the angle under which you can illuminate them (while still being able to separate the reflected light from the incoming illumination).

Since the component with the smallest étendue limits the optical throughput of a light path, using a light source with a large étendue -- LEDs or lamps -- together with a DLP array with small étendue will cause light losses: At some point in the beam path one has to "throw away" some of the illumination light via optical apertures, to limit its étendue to what the DLP array can accept.

That's undesirable, of course. Lasers, and even the diverging light from a small phosphor area illuminated by a laser, have smaller étendue and are hence advantageous.

[tom66]

I wonder why they don't use a red phosphor and still have a colour wheel.  Presumably the efficiency or accuracy of the red phosphor is worse.

It still seems absurdly inefficient:  generate 100% blue laser light (already quite inefficient), then strike that onto a phosphor wheel to generate green and yellow, and then filter out that combined mess to make the colour you need.  A true-RGB single-chip projector with an RGB LED is surely the better implementation, once the optics get there...

[ebastler]

A true-RGB single-chip projector with an RGB LED is surely the better implementation, once the optics get there...

Well, it's the LEDs which would have to "get there", namely get to the required brightness but with a smaller étendue. Since the emission angle from the semiconductor substrate will always be large, the only way would be to make the substrate smaller, and hence the power density higher, I think.

(Clever LED packages with built-in domes or other collimation means don't help, the "étendue is always conserved" rule applies to them too.)

[mikeselectricstuff]

Might it also be something as simple as  with age, airways get clogged so fan cooling becomes less efficient.
the correlation with replaced bulbs could simply be related to run time