Go Back   Steve's Digicams Forums > Digital SLR and Interchangeable Lens Cameras > Canon EOS dSLR

View Poll Results: Have you experienced focus problems with Canon EF lenses?
No - My EF lenses are near perfection 9 64.29%
No, but I wonder sometimes 2 14.29%
Yes, but Canon fixed the problem 2 14.29%
Yes, and Canon failed to fix the problem 1 7.14%
Voters: 14. You may not vote on this poll

Reply
 
Thread Tools Search this Thread
Old Mar 16, 2005, 7:43 AM   #31
NHL
Senior Member
 
NHL's Avatar
 
Join Date: Jun 2002
Location: 39.18776, -77.311353333333
Posts: 11,547
Default

TDM_Canon_User wrote:
Quote:
If the primary is off by 1 degree, the other mirror will be off by 1 degree. Simple geometry.
So far we agree; However, how does a 1 degree offset translate into distances? Would a 1 degree deflection make the projected image 'miss' the AF sensor sensor instead which remained fixed in one position? :?

Simple geometry to the rescue - Just like a geosynchronous satellite which remains fixed in orbit a lens is fastened in place by the lens mount - Strong wind gusts can shift the antenna reflector, just like the mirror being flap into place:



... and from Canon own posting "light emerging from one elliptical focal point that is reflected from an elliptical surface must converge at a single point."

Remember this is not the 1st camera that Canon make - This AF mechanism has been around since my EOS-3 and has been through now 4 generations of dSLRs... Could it be such that this AF mechanism is so flexible that it could potentially increase the exposure time for the AF algorithm to work on while the main mirror (which is transparent to the AF) is still moving (or vibrating) during high frames rate???



peripatetic wrote:
Quote:
It seems to me that if the main mirror is not returning to the correct position that would affect both manual and auto focus.
Exactly!

When a mirror is not at its optimum 45 degree position the distance to the top 1/2 of an image is different than the distance to the lower 1/2. One can detect this with test charts - ie focus with the top part of the viewfinder and the bottom part will be out of focus (or vice versa)... The AF sensor which is hopefully oriented in the opposite direction may detect this variation in focus as well
NHL is offline   Reply With Quote
Old Mar 16, 2005, 12:13 PM   #32
Super Moderator
 
peripatetic's Avatar
 
Join Date: Nov 2004
Posts: 3,599
Default

OK that sounds like a test that could detect whether the mirror is indeed the culprit.

Tape a nice high contrast test chart or newspaper to a wall.
Make sure there is plenty of light.
Mount the camera on a tripod and use a spirit level to ensure you have a vertical camera back.

1. Manually focus using an etched AF zone above the horizon & shoot.
2. Set the AF to use that same single AF zone above the horizon and shoot.
3. Manually focus using an etched AF zone below the horizon & shoot.
4. Set the AF to use that same single AF zone below the horizon and shoot.

If the mirror is the problem we would expect to see one half of the image sharper for 1 than the other and the opposite side of the image sharper for 3. And similarly for 2 & 4.

Of course depending on the path of the mirror it's possible that both halves would be out, but as long as the mirror is rotating as it moves up they should not be equally out of focus - NHL do you have any idea about the actual vector of the mirror?
peripatetic is offline   Reply With Quote
Old Mar 16, 2005, 12:35 PM   #33
NHL
Senior Member
 
NHL's Avatar
 
Join Date: Jun 2002
Location: 39.18776, -77.311353333333
Posts: 11,547
Default

peripatetic

It's not that simple since the distance (top/bottom) is unequal -> what you'll see is front/back focus whether the mirror undershoot or overshoot the 45 degree axis

-> the true distance is now incorrect since the mirror is off the mid-point (ie you can only detect the difference in focus(top/bottom) and not necessarily that the picture is perfectly focused...
NHL is offline   Reply With Quote
Old Mar 16, 2005, 10:27 PM   #34
Member
 
TDM_Canon_User's Avatar
 
Join Date: Jan 2005
Posts: 90
Default



The attached GIF: The incorrect mirror angle changes the distance between the optical center of the lens and the focal plane. A bad aspect is that the change affects the second mirror such that the image shifts in the wrong direction. Thus the AF mechanism may not even see what the intended focus oject is. Hence anything not at infinity, but in front or behind the object might align with the focus area and be focused on.

The real problem though is the change in distance, everything can be adjusted to a sharp focus, but at a distance that doesn't match the optical path to the film or sensor. So the image captured is out of focus.


Note thetwo errors Dp and Df are identical -its just a sketch.
Attached Images
 
TDM_Canon_User is offline   Reply With Quote
Old Mar 17, 2005, 6:48 AM   #35
NHL
Senior Member
 
NHL's Avatar
 
Join Date: Jun 2002
Location: 39.18776, -77.311353333333
Posts: 11,547
Default

Your simplified diagram may work well for the D60 with a flat sub-mirror, but it ignores the critical elliptical secondary mirror in the 45-point AF systems: "This new mechanism works on the principle that light emerging from one elliptical focal point that is reflected from an elliptical surface must converge at a single point."





For example your diagram only shows 'one AF location', what about the other three used by the camera (which are split from that single point)? i.e. One pair detecting horizontal and another pair detecting vertical out of focus shifts - At least the horizontal pairs should have the same amount of offset even if the secondary mirror is flat...

FYI - http://www.a1.nl/phomepag/markerink/af_expla.htm


One can detect where the focusing point really is with test charts especially @ 45 degree!
NHL is offline   Reply With Quote
Old Mar 17, 2005, 9:17 AM   #36
Member
 
TDM_Canon_User's Avatar
 
Join Date: Jan 2005
Posts: 90
Default

Sorry NHL, you simply don't understand what an elliptical mirror does. It works in the exact same manner as a flat mirror, i.e., it bends/redirects light rays.

Bottom line is : if the distance or length of an optital path changes in any optical system, the distance changes. Period.

Take a class or read up on ray theory... the lights will come on.




TDM_Canon_User is offline   Reply With Quote
Old Mar 17, 2005, 1:58 PM   #37
NHL
Senior Member
 
NHL's Avatar
 
Join Date: Jun 2002
Location: 39.18776, -77.311353333333
Posts: 11,547
Default

TDM_Canon_User wrote:
Quote:
Bottom line is : if the distance or length of an optital path changes in any optical system, the distance changes. Period.
You are absolutely correct which is what I wrote as well: "-> the true distance is now incorrect since the mirror is off the mid-point (ie you can only detect the difference in focus(top/bottom) and not necessarily that the picture is perfectly focused..."

My point is the elliptical mirror cancel out the offset in the main mirror - It needs too because again quoting Canon "After light converging at the focal point is directed by a CMOS area AF sensor, it is again dispersed and passes through secondary image-formation lenses. There are two sets of such lenses, one acting on each eye, as well as upper and lower "eyes" not found in humans. The CMOS area AF sensor binds these four parallax images together"

Like this - or it would miss the AF target area @ the bottom of the camera:
Attached Images
 
NHL is offline   Reply With Quote
Old Mar 17, 2005, 2:38 PM   #38
Senior Member
 
Trout Guy's Avatar
 
Join Date: Jan 2005
Posts: 132
Default

NHL, your light blue text is killing my eyes:shock:

You guys are sure doing your research, keep it up

TG
Trout Guy is offline   Reply With Quote
Old Mar 18, 2005, 12:26 AM   #39
Member
 
TDM_Canon_User's Avatar
 
Join Date: Jan 2005
Posts: 90
Default

Latest update: April 30, 2005

Body no longer suffers focus problem - rapid fire for >200 frames "freed up" the mirror up and now the bodyis focusing properly. Canon was not helpful in resolving the problem saying it was low contrast causing the problem (Canon never offered and opinion as to why my cheap D60 focused well under the exact same conditions the 1DmkII was messing up).

During all this testing I discovered anoth quality control issue.

My 24-70f2.8l focus plane was wrong. It was repaired a year ago for an Error-99 and returned with a subtle but frustrating problem:the focal plane was canted at a diagonal. Real odd problem - but caused some great shots to be missed when shooting at f2.8.

Sent lens back to Canon in SoCal forrepair/adjustment and it came back unchanged (I got a document saying I should learn how to use my camera). I further documented the problem and returned it to Canon (trip 3).

Canon opened the lens and actually replaced several items this time. Being smarter, I immediately ran a few hundred test shots. Half were out of focus again. Only worse. This time the lens wasn't able to focus at infinity!!! At 70mm it focused fine. At 24mm it hit the focus stop! These Canon guys didn't even test the lens at the two end focal lengths.

Upset, I called Canon. A Canon FSC manager requested I send it back to the California location to ensure they can learn from the experience. The lens just arrived back last night. Well, they did fix the hit-the-stop problem at 24mm. 24mm shows close to the infinity mark when focused at infinity - cool!! But when zoomed to 70mm the lens says I am focused at 15 feet whenfocused at infinity. So much for the PAR-FOCAL capabilities it once had - and so much for ever hoping to use the distance scale for anything useful.

Monday, the lens is heading to an eastern US service facility... my hope is they may actually try to fix the problem. And I will be climbing the Canon customer relations ladder to see if anyone there cares people who invest $10,000 a year in their products are getting pissed at the lack of quality being provided.

I am holding off my purchases of any other Canon products...




Attached Images
 
TDM_Canon_User is offline   Reply With Quote
 
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 8:10 PM.