Star Test of Canon 100-400mm f/4.5-5.6L IS Lens

Last year, I tested my Canon EF 70-200mm f/4L lens for astronomical imaging. Those results showed very good performance at 200mm and adequate performance at 70mm. This has been borne out by actual imaging sessions with the lens. This evening, I repeated a similar test with my Canon EF 100-400mm f/4.5-5.6L IS lens which I have been using for birding. My previous tests on test pattern charts showed promise that this lens might be comparable to the fixed Canon EF 400mm f/5.6L lens that my wife uses for imaging. Here is the MTF chart for the Canon EF 70-200mm f/4L lens:

Test Setup

My hope is to use this lens with the QSI-540ws for wide-field imaging. Since I cannot electronically stop the lens without a DSLR, I decided only to test the performance of the lens with the aperture wide open. In order to more easily capture color images to test for chromatic issues, I used the Canon 60Da instead of the QSI to gather the test images. I found that 300s exposures at ISO-400 gave a suitable histogram shape. I auto-guided during the tests to minimize tracking issues.

The DSLR and QSI have different fields of view. To show "corner performance" for both of these cameras, I am reporting the detailed star images for several regions of the images as shown below. The blue frame is the field of view of the DSLR. The smaller green frame represents the cropped region I would have obtained using the QSI.  The four yellow circles are the regions detailed herein. These show the center of the frame, the corner for both of the cameras, and the mid frame region.

I performed the imaging tests at four focal lengths: 100mm, 200mm, 300mm, and 400mm. I had intended to test each of these focal length with the aperture wide open. Unfortunately, I inadvertently left the aperture stopped down to f/5.6 on the 200mm test somewhat negating the value of this test for me.

Test at 100mm @ f/4.5

The following image shows the field of view at 100mm.  The bright star is Deneb.

Details in each of the four regions are shown below:

Test at 200mm @ f/5.6

The following image shows the field of view at 200mm. Note that the lens was unintentionally stopped down from f/5 to f/5.6 in this image. Introduction of the octagonal edges of the iris into the light path causes some strong diffraction patterns on the stars. I recall that for the Canon 70-200mm f/4L lens this occurred without significantly improving the shape of the dimmer stars. Personally, I prefer round stars.

Details in each of the four regions are shown below:

Test at 300mm @ f/5.6

The following image shows the field of view at 300mm.

Details in each of the four regions are shown below:

Test at 400mm @ f/5.6

The following image shows the field of view at 400mm.

Details in each of the four regions are shown below:

Observations

At the shorter focal lengths, this lens performed very well. The slightly stopped down image at 200mm was excellent though I will not be able to stop down the lens when using the CCD camera. I would say that this lens is close in performance to the EF 70-200mm f/4L at 200mm and significantly better at 100mm.

Whereas the EF 70-200mm f/4L performed better at longer focal lengths, this lens seems to perform better at the shorter focal lengths. I qualify this conclusion somewhat because of the fact that the elliptical distortions are oriented the same way across the full frame rather than being radially symmetrical. It may be that I am seeing a tracking issue rather than a lens issue. At 400mm, I also notice more pronounced asymmetry in the diffraction spikes and halo around Deneb.

Another interesting difference between the two lenses was "image shift". As I used the electronic focusing controls in Astro-Photography Tool, the image would shift slightly as the focusing motor changed direction. Fortunately, the amount of shift was just small enough that the star stayed within the view port of the Focusing Aid feature in APT. I assumed that this motion was related to the image stabilization mechanics of the lens though it was turned off for these tests. I recall that when I focus on birds with IS enabled, there is a noticeable image shift in the view finder. Coincidentally, the orientation of this shift happens to matches that of the distortion visible in the 400m images.

Th EF 100-400mm f/4.5-5.6L lens has a full-time manual focusing ring similar to that of the EF 70-200mm f/4L. However, I did not have the time to test the ease of focusing manually.

I think my next step is to retest explicitly on the QSI, test the manual focus, and verify more carefully tracking quality.

Update 2014-08-31

Based on a post on Cloudy Nights, it turns out that I can use the following trick to stop down the lens and still use the QSI:

• Attach the Canon 60Da to the lens
• Dial AV mode and set desired f-ratio
• Press the DoF preview button to force the iris to close
• Detach lens while holding the DoF button pressed

Looking through the lens, I can see that the iris remains despite being unpowered. Assuming this does not damage the camera or lens, this gives me one option for stopping down the lens.

Another suggestion was to use camera filter step down rings as an external stop. This is a bit challenging because the "a little bit of stop" translates to a wide range of aperture sizes across the zoom range. By way of calculating what I might expect, I used the imaged star fields to estimate the actual focal length of the lens at the four marked positions on the lens barrel that I used to perform the above tests. This calculation gives me: 102mm, 197mm, 293mm, 380mm. Yes, this lens is not really 400mm.

Checking one of my 77mm filters for this lens, the measurement refers to the outer diameter of the threads. The filter itself has a clear aperture of 70mm which is still wider than what can be seen through the fully open iris, 68mm. The clear aperture of a step ring should be close to the inner thread size.

The table below shows the necessary inner ring size to match wide open aperature and aperture reduced by 1/3 stop.

Marking	Focal Length	f-ratio	Aperture Open	-1/3 Stop	Nearest Ring	f-ratio
400	380 mm	f/5.6	67.9 mm	60.3 mm	77-62 mm	f/6.1
300	293 mm	f/5.6	52.3 mm	45.5 mm	62-49 mm	f/6.0
200	197 mm	f/5.0	39.4 mm	35.2 mm	49-35.5 mm	f/5.5
100	102 mm	f/4.5	22.7 mm	20.4 mm	-	f/4.5

As selected, I would be able to sequentially nest the rings. Another option would be to use a 77mm to 67mm step-down ring as a "holder" for circular cut-outs of the right diameter and construct my own set of external stops. I am not sure what matte material to use to minimize reflection glare artifacts.

Update 2014-09-01

I purchased a simple Pro-Master 77mm to 62mm step-down ring and did some more testing last night. This time, I used the QSI-540ws camera and performed focusing manually. The ease of focusing was about the same as with the EF 70-200mm f/4L. That is to say, challenging but doable. Again, I guided the mount and imaged the star field around Deneb for 300 seconds with a Luminance filter. I performed the test only at the longest focal length (380mm), first wide open at f/5.6 and then with the step-down ring attached.

Test at 400mm @ f/5.6

This first imaging test is performed wide open. Same frame regions reported as above.

Test at 400mm @ f/5.6 with 62mm Ring

The second imaging test repeats the test above but with the 62mm step-down ring attached to the front of the lens.

Observations

Either the tracking was better tonight, or the consistent stretching of the stars is somehow related to the DSLR. The stars are much rounder throughout the frame. In the corners there is evident "shuttlecock-shaped" coma but it is not too bad.

The usefulness of the step-down ring seems to minimal at only 1/3 of a stop. The symmetry and intensity of the diffraction spikes on the central star are improved. The halo size and intensity seems to be unchanged. The mild coma on the corner stars remains unchanged.

After performing these comparative test, I left the 62mm ring on and took calibration frames. The flat image field for this configuration looks like:

and has this profile along a centered horizontal slice:

For a real imaging test, I captured eight 15 min frames of the Pelican Nebula with a 12nm H-alpha filter, calibrated and stacked in DeepSkyStacker, then performed some basic stretching, noise reduction, and sharpening in Photoshop as well as reduction of halos on the two brightest stars.

QSI-540ws, Canon EF 100-400mm f/4.5-5.6L, 8x 900s, H-alpha