So my EE-S screen arrived and now, especially using a viewfinder magnifier, it becomes plainer why I've found it so difficult to manually focus. When shooting a test chart at f1.4 about a meter from the lens, the viewfinder is showing accurate focus when infact the focus is 70mm behind the focus point!
Am I adding (need to find a way to buy, not so easy outside the US) or removing shims to fix this problem and with such a huge focus shift, am I going to end up successful?
Possibly in backwards? Either that or your viewfinder needs calibration. Back focus is corrected by adding shims (from memory). It will do nothing for focus shift.
Beni wrote:
So my EE-S screen arrived and now, especially using a viewfinder magnifier, it becomes plainer why I've found it so difficult to manually focus. When shooting a test chart at f1.4 about a meter from the lens, the viewfinder is showing accurate focus when infact the focus is 70mm behind the focus point!
Am I adding (need to find a way to buy, not so easy outside the US) or removing shims to fix this problem and with such a huge focus shift, am I going to end up successful?
Any advice would be most appreciated.
If that lens has a focal length of 50 mm, you need to add 0.18 mm worth of shims. Depending of course, on the accuracy of your measurement and choice of reference point for the distance from the lens.
Beni wrote:
Focus shift was the wrong phrase, sorry. The viewfinder is in correctly. I understand it needs calibration, the question is which way.
You can remove the shims and see if the problem gets better or worse. If it gets worse you'll need to get some shims to add to the ones that are already behind the focusing screen.
I tried putting tape on the edge of the EE-S focusing screen to simulate shims, it sent the focus even further behind where it should have been, I'm going to try removing washers but at present I don't have a small enough screwdriver to remove the bit infront of them, I'll try and buy a set of screwdrivers tomorrow.
Incidentally I have two 5D's, both with viewfinder calibration as wildly out as each other, canon, there are only 3 mechanical functions that a camera uses to create a photo. Shutter, aperture and focus. Is being able to see accurate focus that irrellevant in today's world that they bother so little to get it accurate? There are shims in the body, did they take a handful and chuck them in without even checking? It's not that the AF is so reliable, took 6 trips to canon to get that right and even so one body is less accurate than the other.
I've printed off the appropriate page from the canon parts catalogue for the 5D, I'll take it in to the local camera repair guy and ask him to order them in for me and hope he doesn't gouge me too much.
Please keep us posted, Beni. My 5D is out of calibration too, but I'm getting front focus (viewfinder image is back focused compared to the sensor).
@hooligan: Perhaps there are thickness differences between production batches of the focusing screens? I should put my Ee-A screen back and test if the focus is correct again, because I'm not sure if the front focus problem was there before.
Put a shimmer is very normal for changing focus screen. I did that for both of my D300s with Katzeye and D700 with Canon S-type screen.
Basically, there are separate path for view finder in focus and sensor capture in focus. (I should say three, one to viewfinder, one to image sensor, and last to AF sensor) If the distance between first and 2nd are different, you need add shimmer to compensate that. AF fine tune is for distance between 2nd and 3rd.
It is a very easy task, if you back focus, you put shimmer (first) before screen, if front focus, you remove metal shimmer pre_installed, use a macro lens without any focus shift is best to check focus.
Toothwalker wrote:
If that lens has a focal length of 50 mm, you need to add 0.18 mm worth of shims. Depending of course, on the accuracy of your measurement and choice of reference point for the distance from the lens.
?
The referenced link to the list of 20 or so shims for the 5D is precisely what one needs to fine tune the accuracy of the viewfinder. When my 5D came back from Canon, it had shim # xxx-012 and was .005 inches thick. The next thick or thinner shim was +/- .0005, or 1/2 of a thousandth inch. That's a very small thickness difference. Never the less, I tried a thicker and thinner one, but the Canon Service did a good job, as I couldn't get it any more accurate. The thickness of scotchtape is about .002".
Given the number of shims in the complete set indicates just how much variability there is in the focus screen fit and the mirrorbox lightpath.
In a similar situation. Recently acquired a 5D and just ordered an EE-S screen.
I wanted to add the folloing to the original thread with the part numbers for the washers, but it is closed.
The source to buy the shims in Australia is partsovernight.com.au . Unfortunately the cost for each shim is between $4 and $5, so the whole set would cost more than $50 + $10 shipping.
Toothwalker: can you share how to calculate the thickness of the shim required based on the focal length of the lens, distance of the test object and amount of front/back focus? It would be nice not to have to order the whole set but approximately calculate the required shimming and add a few thin shims for fine tuning.
OpticalFlow wrote:
Toothwalker: can you share how to calculate the thickness of the shim required based on the focal length of the lens, distance of the test object and amount of front back focus?
Yes, please share with us your calculation.
If you don't want to purchase all shims, order just 2 or 3 of the thinnest and maybe one that is thinner than your current one. They are made of brass or bronze so they should be precise in their thickness, so stacking 2 or 3 shouldn't be a problem.
JimBuchanan wrote:
Yes, please share with us your calculation.
f = focal length
dv = object distance according to viewfinder image
ds = object distance according to sensor image
h = required shim thickness
(add shims when h is positive, remove shims when h is negative)
h = f*f/(dv-f) - f*f/(ds-f)
In Beni's case we have
f = 0.050 m
dv = 1.00 m
ds = 1.07 m
and h becomes +0.00018 m.
dv and ds should be measured from the front principal plane of the lens. If the location of this plane is not known you should avoid a test setup where the object is close to the lens, because that could result in a large relative measurement error.
