Makten
-
I've tried to use the old Summicron-C for Leica CL on my M8, but it doesn't seem to be very useful. Many here claim it to be very sharp and perfectly compatible with Leica M cameras. I beg to differ, or my sample is garbage.
So, what's wrong then? First of all, it focuses too close at all distances. If I stop it down it's not too bad since DOF will make up for it. But it's still wrong enough to throw the subject slightly out of focus at f/4 and two meters distance. Not acceptable.
Second; the corners are soooo unsharp, even at f/8. Even the borders, and this is on a cropped sensor!
Third; it doesn't quite reach infinity at the infinity stop of the lens.
Focusing and infinity should be possible to fix by grinding off a bit on the bayonet flange. But the corners must be something else. The funny thing is that my (now sold) CV 35/1.4 did exactly the same! Focusing was off, it didn't reach infinity and the corners were crap.
Because of this, I first expected the sensor or the bayonet of the camera to be out of alignment. But then my CV 28/2 is sharp to the corners at f/2 and reaches infinity.
What the heck is this? If it was the bayonet or sensor, I should be able to see the tilted image field, but there is no sharpness anywhere towards the corners at any distances with the 40/2.
An explanation could be severe curvature of field, which in combination with bad calibration of focus would throw the corners out of the image plane. I'd not be surprised if this was the case with the faster 35/1.4, but the Summicron shouldn't do this on a small sensor.
-------------------------
The CV 28/2 shifts focus when stopped down, so I'm about to sell it and get a Zeiss 28/2.8 Biogon instead. But I don't know if I dare. What if I'd get the same terrible problems with that? You can't get the lens here, and I don't want to buy one before trying now.
Any advice is welcome.
So, what's wrong then? First of all, it focuses too close at all distances. If I stop it down it's not too bad since DOF will make up for it. But it's still wrong enough to throw the subject slightly out of focus at f/4 and two meters distance. Not acceptable.
Second; the corners are soooo unsharp, even at f/8. Even the borders, and this is on a cropped sensor!
Third; it doesn't quite reach infinity at the infinity stop of the lens.
Focusing and infinity should be possible to fix by grinding off a bit on the bayonet flange. But the corners must be something else. The funny thing is that my (now sold) CV 35/1.4 did exactly the same! Focusing was off, it didn't reach infinity and the corners were crap.
Because of this, I first expected the sensor or the bayonet of the camera to be out of alignment. But then my CV 28/2 is sharp to the corners at f/2 and reaches infinity.
What the heck is this? If it was the bayonet or sensor, I should be able to see the tilted image field, but there is no sharpness anywhere towards the corners at any distances with the 40/2. An explanation could be severe curvature of field, which in combination with bad calibration of focus would throw the corners out of the image plane. I'd not be surprised if this was the case with the faster 35/1.4, but the Summicron shouldn't do this on a small sensor.
-------------------------
The CV 28/2 shifts focus when stopped down, so I'm about to sell it and get a Zeiss 28/2.8 Biogon instead. But I don't know if I dare. What if I'd get the same terrible problems with that? You can't get the lens here, and I don't want to buy one before trying now.
Any advice is welcome.
Last edited:
jaapv
RFF Sponsoring Member.
The C lenses use a different focussing mechanism, which is far less accurate than the M mechanism.
Erwin Puts:
Erwin Puts:
Furthermore CV, suffers from CFA, making it not very useful on a fully flat medium like a sensor. Film had a thickness which camouflaged such behaviour. The Zeiss is a fine lens on the M8.
Last edited:
jaapv
RFF Sponsoring Member.
They work but focussing is inaccurate...
Makten
-
I know that, but since the register distance should be the same, why doesn't the lens reach infinity (it's very close though)? It has recently been CLA:d (perhaps without the A).
Focus is also off just as much over the entire span, so it can be corrected for. I wouldn't really call it "inaccurate", but rather miscalibrated.
That makes more sense, and it could be the problem with the Summicron too. But then again I've seen pictures with the CV 35/1.4 that is sharp straight to the corner with an M9 at f/5.6: http://www.rangefinderforum.com/forums/showthread.php?t=83481
I really believe so, but I must make sure that it isn't the camera first.
ferider
Veteran
Since your 28/2 at f2 is sharp center to corner at infinity, camera sensor, lens flange and registration distance should be OK.
Ergo, if the Summicron is not sharp at the center at infinity, it is mis-collimated. Independent of the RF coupling and alignment.
Send it to collimation or shim it yourself, it is quite easy to disassemble.
-------
Test lens collimation and RF accuracy separately.
To test RF accuracy of your M8 I recommend a fast 50 or longer. The DOF of the 28 is too deep. Test at infinity and 1m. Also, your camera RF coupling arm might be bent which will cause it to align well at infinity for some lenses, but nor for others.
Roland.
Ergo, if the Summicron is not sharp at the center at infinity, it is mis-collimated. Independent of the RF coupling and alignment.
Send it to collimation or shim it yourself, it is quite easy to disassemble.
-------
Test lens collimation and RF accuracy separately.
To test RF accuracy of your M8 I recommend a fast 50 or longer. The DOF of the 28 is too deep. Test at infinity and 1m. Also, your camera RF coupling arm might be bent which will cause it to align well at infinity for some lenses, but nor for others.
Roland.
Last edited:
historicist
Well-known
Isn't it the cam profile that is different? If you look at the cam on the Summicron C compared to a regular M lens the shape is quite different.
I've used one on a M body with film and didn't have any problems, but maybe digital is more critical.
I've used one on a M body with film and didn't have any problems, but maybe digital is more critical.
Makten
-
I've almost never shot the Ultron wide open at infinity, but I've tested it at close distance. I guess the curvature would be there still if it had it.
The problem is that there are no shims to take away, so I'd have to mill the bayonet from the inner side (faceing in shooting direction).
I don't have such a lens. But I do have the Elmar-C 90/4, which is also for the CL and which also doesn't reach infinity and focuses too close.
Edit: I don't at all agree that the DOF should be too deep on the 28. You can clearly see things shift out of focus at several meters distance when stopping down to f/4.
Last edited:
Makten
-
The profile shouldn't be important as long as the surface moving the roller is at the right place at the right focus distance. I mean, the roller won't care if it's a straight flange moving it, or a slope. The slope just makes the lens easier to manufacture since there is only one helicoid, not two.
ferider
Veteran
Forget field curvature, etc.
Test sensor, flange alignment and registration distance by using the 28/2 wide open at infinity. Stars and tripod, for instance. Wide angles are very sensitive to film distance, so this is a good test.
That will rule out a set of possible camera problems.
You don't need to grind anything with the Summicron. You can separate focus helix from optical barrel with a spanner. You control the distance of the two with a shim (paper or aluminum).
Once you know the Summicron is well collimated, adress possible RF coupling issues.
Roland.
Test sensor, flange alignment and registration distance by using the 28/2 wide open at infinity. Stars and tripod, for instance. Wide angles are very sensitive to film distance, so this is a good test.
That will rule out a set of possible camera problems.
You don't need to grind anything with the Summicron. You can separate focus helix from optical barrel with a spanner. You control the distance of the two with a shim (paper or aluminum).
Once you know the Summicron is well collimated, adress possible RF coupling issues.
Roland.
Makten
-
Here's a shot with the Summicron at f/5.6 and infinity: http://img514.imageshack.us/img514/4995/l1001185large.jpg
Sharpness in the middle is OK, but look at the tree in the forground. Much sharper, indicating that infinity is not reached. And then look at the borders. Terrible! Though there are fairly sharp tree branches in the lower corners, closer to the camera. So I guess it's curvature of field that is the problem. Or part of the problem.
Sharpness in the middle is OK, but look at the tree in the forground. Much sharper, indicating that infinity is not reached. And then look at the borders. Terrible! Though there are fairly sharp tree branches in the lower corners, closer to the camera. So I guess it's curvature of field that is the problem. Or part of the problem.
ferider
Veteran
Compared to most other lenses, the Summicron has practically no curvature of field. Looks like front focus to me (lens is too far away from the sensor).
Again, at infinity, the RF coupling or alignment does not matter.
Open it up, and 90% probability you will find a shim that is too thick. That is if your camera passes the infinity test with your reference lens (the 28).
Again, at infinity, the RF coupling or alignment does not matter.
Open it up, and 90% probability you will find a shim that is too thick. That is if your camera passes the infinity test with your reference lens (the 28).
Last edited:
Makten
-
So, what could cause the curvature of field of my specimen then? If the image plane was tilted, it wouldn't have looked that way.
Edit: Or are you referring to results on film? The CV behaved almost in the exact same way. All corners unsharp if they were at the same distance.
ferider
Veteran
Sorry, Makten, in your picture, I see sharp foreground, and background that is less sharp (somehow covered by f5.6; hyperfocal distance is around 12m, BTW).
Last edited:
jaapv
RFF Sponsoring Member.
The throw of the C helicoid is far too short (120 degrees) to reach accuracy. It is simply not suitable for the M8/M9.
Makten
-
Sure, but this is still not about accuracy. I get the same misfocus everytime.
----------------------
Anyhow, it's pitch black outside here now (8 PM) so I can't do the infinity check for the 28/2. I'll forget about the Summicron for now and first try to confirm that the camera is OK.
Thanks for the advice so far!
Last edited:
ferider
Veteran
The lens is front focusing, with a rather flat field. Look at the plane of focus, center and corner (see attachment).
Roland.
ferider
Veteran
Maybe because you are close to the hyperfocal distance ? COC being larger in the corners ?
Makten
-
Maybe, but it's wierd. DOF is usually larger towards the corners due to the cosine^4 rule.
ferider
Veteran
I think you must have misunderstood something there .... Anyways, good luck with the 28 test.
Makten
-
What? The CoC will be smaller towards the corners because of the iris "tilting" off axis. I can't come to think of any other explanation in this case than curvature of field.
http://en.wikipedia.org/wiki/Vignetting#Natural_vignetting
http://toothwalker.org/optics/vignetting.html#natural