I have had a Sony VPL-VW100 (Ruby) since they were introduced. My unit always had a 2 pixel misconvergence of green. The misconvergence was uniform all over the image. After thinking about it for a few years, I finally decided to do something about it last week. First I thought about an arrangement of an interference (dichroic) filter and mirrors. I figured that if I could split the green component of the beam out, I could redirect it using an adjustable mirror. I tried that, but quickly found out that this changed the projection path length of the green image vs the red and blue which changed its scale. No dice. I thought of an arrangement with and interference filter with two mirrors which would work, but was really complex, bulky and expensive.

Then I got the great idea of eliminating the path length difference altogether by putting the dichroic filter in contact with a regular mirror. This would eliminate the path length extension since both the dichroic filter and the mirror are at essentially the same plane. I bought 2 items fro Edmund Optics:

45336 Mirror, 1/4 wave Enhanced Aluminum, 50mm x 50mm square $ 65
47950 Filter 45 Degree Dichroic Green 50.0mm Square $ 99

To make the convergence device (please see attached pictures) I put the coated dichroic filter side of the dichroic filter directly on the mirror. I calculated the the 2 pixel offset at the projection distance was equivilent to a 12 um wedge on one side of the dichroic filter / mirror assembly. I cut two small pieces of mylar shim stock, and put them between the dichroic filter and the mirror on one side. Then I taped the two pieces together while applying some pressure using gaffer tape. This made the corrector which steers the green image 2 pixels to the left of the red and green pixels, bringing all 3 colors into alignment.

The way it works is that the light coming out of the projector lens (white light) is directed to the convergence device at 45 degrees. The green light bounces off of the dichroic filter, and the red and blue light pass through the dichroic filter and bounce off of the mirror, and back through the dichroic filter. Since the dichroic filter (actually a 'green' mirror in this case) is tilted slightly relative to the aluminized mirror, the green light is deflected by the right amount, which is controlled to give proper convergence on the screen. You could have a complex adjustable device, but I found the shim method to be very simple and totally robust.

You will notice from the picture that the convergence device is placed in front of the lens, and it directs the light upward to another larger mirror which sends the beam out towards the screen (a 'periscope' configuration). While the convergence device is made from high-quality 1/4 wave optics, the larger mirror is an inexpensive plate glass first surface mirror. This mirror doesn't need to be as large as the one I used, that was just one I had on hand. This is a crude prototype made from PVC sheets and gaffer tape. Next, I plan on making properly machined parts. The prototype proves the concept works well though.

The resulting improvement was nothing short of stunning. I had gotten used to 'ignoring' the mis-convergence, but I noticed it from time to time, and it was tiring. Once I eliminated the problem, the picture looks much better. Generally, I view HD movies in 2.40 : 1 format on a 2.40:1 screen, and it is a large picture. With this level of magnification, the mis-convergence became quite visible. I know that current Sony projectors have software convergence, which would work great if I wasn't stuck with the Ruby.

There are some problems with the technique. The most notable is that the largest off the shelf 45 degree dichroic mirror I could find was only 50mm square. This meant that I had to get the convergence device as close to the lens as possible, and I couldn't get a 45 degree angle without having the light hit the projector case. A large 75mm device would work much better, but would have to be custom made. The device works best for the 2.40:1 format. In 16:9, I still have some convergence error at the bottom of the image, but still its much, much better than it was. At 2.40:1 convergence over the entire picture is excellent. I would limit the use of this device to fairly long-throw situations - my projector is 14 feet from the screen.

If you use fixed shims to set the angle, it will work best at one zoom setting, and may over or under correct at other zoom settings. While not a deal-breaker, you should optimize for the zoom setting you use most.

The device has no effect on the color shading uniformity or sharpness of the image. It does attenuate the red and blue slightly, requiring that I dial out 10units of green using a custom color temperature setting on the Ruby.

All in all, I would say this project was a great success. It is a bit ugly having a periscope stuck to the front of my projector, but you learn to love it when you see the much improved image. And it avoids having to send your project in for service and having it come back worse than before!

Anyone wanting to use this invention is welcome to it, I'm putting it in the public domain.

Regards - Jim Browning

Wow great job!

It would be useful to see before/after photos of a crosshatch, which would allow a precise view of the amount of correction you achieved.

It would be useful to see before/after photos of a crosshatch, which would allow a precise view of the amount of correction you achieved.
I'll try to do that. Regards - Jim

I love those thread. I'm always impress by smart modification. I agree that the difference is night and days. Very smart trick!

I love those thread. I'm always impress by smart modification. I agree that the difference is night and days. Very smart trick!

Here is a comparison image. The projector was zoomed to fill a 2.40:1 screen, and the center, LL, and UR were photographed with the corrector installed, and without the corrector installed.

Note - The first .jpg image I uploaded was incorrect. I used the wrong image of the Center Uncorrected. I'm not sure why the Lower Left, Uncorrected image shows green to the right, it should be to the left. I must have the wrong image there. I took so many pictures it was hard to keep them straight. Suffice it to say that the comparisons do show the degree of correction, even if the LL uncorrected one seems to be reversed.



Regards - Jim Browning

Wow I am impressed! I had no idea you could do such things.

Well done! The cost is minimal, are there any downsides? What happens to ANSI contrast performance?

Amazing, the wit & talent on this board is always fresh and exciting.
Thanks for sharing and welcome to the forum:)

James: If you had the time, skill, and inclination, you could put together an only slightly more sophisticated version of this and market it. It seems like a great idea.

Well done! The cost is minimal, are there any downsides? What happens to ANSI contrast performance?

There are very few downsides. There is possibly a very slight widening of the image relative to the height, but it isn't much. THere is a small loss of light of the 2 colors which have to pass through the interference filter twice, which isn't much - 10 contrast points reduction (in green in my case) on the Ruby's custom color temp setting corrects that. I see no change in either the absolute black level or the ANSI contrast, but I haven't measured it with an instrument. Also, I see no change in the shading.

There is one benefit, when you redirect the light upwards, any stray light which doesn't bounce off of the second mirror continues up and doesn't get to the screen, so it eliminates surround glow around the screen. It does light up the ceiling above the projector a bit, but that's better than having it around the screen.

The technique only works for fairly simple constant offset convergence. It may be possible to stack 2 interference filters and have the capability to adjust 2 colors, but one of the filters would have to be on thin glass so it might not be as flat.

Regards - Jim

Amazing, the wit & talent on this board is always fresh and exciting.
Thanks for sharing and welcome to the forum:)


Thanks for the encouragement, Alan, I appreciate it. The AVS forum, and your submissions in particular have been a really valuable resource for me over the years.

Regards - Jim

James: If you had the time, skill, and inclination, you could put together an only slightly more sophisticated version of this and market it. It seems like a great idea.


I thought about that for a bit, trying to decide if it was a good idea to release this potentially patentable idea to the public domain. I decided that it really was the responsibility of the projector manufacturers to solve this problem, and from what I have been hearing, they are rising to the challange. So I decided there wouldn't be much of a market. If I got inundated with requests for the device, I might reconsider though.

Regards - Jim

Excellent job! I agree there wouldn't be much of a demand for this device outside the AV enthusiast crowd.

Some companies have been doing well with this type of thing (Panamorph).
Some have died off unfortunately (Immersive)

It would definitely be a small market. It is a really cool idea. Thanks for sharing.