Above: The three screws (red marks) holding the sensor plate inside a Canon PowerShot S95.
.
A year ago I noticed that photos from a PowerShot A2200 camera I converted to infrared were blurry around the edges (research note from November 2013). This was disappointing because I had replaced the IR block filter with a blue BG3 filter made of optical glass. Photos from DIY converted IR cameras are expected to be a little fuzzy when the IR block filter is replaced with film or filters made from gelatin or polyester, but I expected this glass conversion to be different.

In the last year I have converted eight different models of PowerShot by replacing the internal IR filter with a piece of polyester or gelatin filter. None of these cameras makes photos with edges as poorly focused as that BG3 A2200. It sometimes takes a few tries to get the focus right, so I always take test photos and often have to readjust the screws holding the sensor plate behind the lens.
.

Kite photo taken by a PowerShot S95 with the IR filter replaced with Wratten 87 polyester filter. In all photos, focus was poor on the left side.
.
I lofted a converted PowerShot S95 with a kite (photo above) before I noticed that the left side of every photo was poorly focused. Adjusting the focus of the S95 is very easy because after the rear shell is removed only one more screw must be removed in order to expose the three final screws holding the sensor plate. So I adjusted just one of these screws at a time and took some test photos without even replacing the rear shell. It took three tries (adjust, take photos, examine on computer, repeat) to get it right.
.

Photo from an unmodified PowerShot S95 (top) and another S95 with the internal IR block filter replaced with polyester Wratten 87 filter (bottom). After a few adjustments, focus on the converted camera is just about as good as on the unmodified one. These two cameras are part of this rig package.
.
Whenever I convert a camera now I first mark the position of those last three screws and count the rotations required to unscrew them so I can return them to their original positions. These screws press the sensor plate against springs or a rubber gasket and thereby determine the distance between the back of the lens and the sensor. I no longer convert certain models of PowerShot that require lots of disassembly to expose those last three screws (see list here). About a third of the time at least one adjustment is needed to get the focus right, and it may never happen if it's too much trouble to open the camera again.
.

Photo from a PowerShot A2200 converted to infrablue with an internal glass BG3 filter. Focus is poor near the edges.
.
I don't have much use for an Infrablue camera so I had sort of forgotten about the BG3 A2200 with the soft edges. To bolster my growing conviction that any converted PowerShot can be made to take well focused photos, I opened it up and adjusted the screws. Over the course of a week I made 11 adjustments, each time taking photos of the same distant forest edge to check for edge-to-edge focus at infinity. I sometimes had to wait for enough sunlight so I could always use the lowest ISO and a high shutter speed (1/200 to 1/400). The aperture was always wide open (f/2.8) which minimizes depth of field and makes the test harder to pass. This test simulates aerial mapping photography (a distant planar subject and small f-stop).
.

The blue channel from the infrablue photo above. All three channels exhibited the same pattern of better focus at the center than near the edges.
.
No adjustment brought the edges of the photos into focus. I finally gave up and replaced the BG3 filter with one of the IR block filters I had removed from a PowerShot. After two adjustments the focus was a good as new.
.

Photo from the PowerShot A2200 after I re-installed a factory standard IR block filter. The focus at the edges is not quite as sharp as at the center, but is substantially better than with the glass BG3 filter.
.
Although the focus at the edges is not quite as sharp as at the center, I think that is typical of cheap PowerShots. The lenses are optimized for photos of people standing in front of the camera. Few people closely examine the edge focus of their photos of sunsets and never notice the slightly soft edges. Let me know if you can take a photo of a distant scene with a PowerShot and get the edges as sharp as the center.
.

The red channel from a photo taken by the A2200 with a Wratten 15 (yellow) filter replacing the internal IR block filter. All three channels showed a similar pattern. The focus performance is about as good as it was when the IR block filter was in place.
.
I have no need for a normal RGB A2200, so I opened it again and installed a piece of Wratten 15 (deep yellow) gelatin filter. It took a few tries to get the focus right, but it now focuses about as well as it did with the IR block filter in place. I have all the test photos I took between adjustments, so I know that I have now adjusted those last three screws on this camera 25 times, each time disconnecting and reconnecting a ribbon cable so I could lift the LCD out of the way.

I hypothesize that the BG3 filter prevented good focus because the glass was too thick. I measured the thickness with calipers at 0.60 mm, and the factory IR block filter at 0.33 mm. The Wratten gelatin filter is 0.12 mm. I can' t think of another likely explanation.

Conclusions:

1. Adjusting the innermost three screws that hold the sensor plate in place can make a big difference in how the camera focuses.
2. It's not always easy to get the adjustment right the first time, so repeated adjustments are often required.
3. Some cameras make it easy to take test photos without reassembling the entire camera.
4. Optical glass filters that are thicker than the replaced IR block filter might prevent proper focus in some cameras.
5. Thin polyester or gelatin filters consistently allow good focus when the sensor plate is properly adjusted.