Photolithography lenses + astro cameras = extreme macro

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Photolithography lenses + astro cameras = extreme macro

Post by Lou Jost »

Most of us have heard of Nikon's Ultra-Micro-Nikkors, and a few lucky people actually have one. These are very expensive lenses originally designed for making semiconductor chips. They are meant to project the image of a large "mask" (a chrome slide of a circuit pattern) onto a tiny silicon wafer coated with a photosensitive layer of material. The exposed part comes off and can then be treated in various ways. The process then gets repeated many times on new layers, transforming the wafer into a useful chip for an electronic device. They were made in the 1960s and 70s.

Since there is a very strong economic incentive to manufacture smaller and smaller chips, enormous amounts of effort and resources have gone into perfecting these lenses and increasing their resolution. The Ultra-Micro-Nikkors were soon overshadowed by Zeiss lenses and others. These are the Formula 1 race cars of the photographic world, able to do one thing very well, but somewhat impractical for most things, while consumer lenses are more like Volkswagens, reliable and versatile but mediocre.

These photolithography lenses are characterized by essentially perfect field flatness and corner-to-corner sharpness. To achieve this they have sacrificed correction for color aberrations, so they are best used with monochromatic light. Usually a given lens is optimized for a particular wavelength.

Reduction ratios are typically 1:5 or 1:10. For our macro purposes they should be used in reverse, so they become 5x or 10x finite objectives. The most commonly available examples, from Zeiss, have NA ranging from 0.20 to 0.35 and have image sizes of 125mm x 125mm. Thus a 5x lens has a FOV of 25mm x 25mm and a 10x lens has a FOV of 12.5 x 12.5mm. These are basically giant microscope objectives, with NA sharpness similar to Mitutoyo lenses, but with much larger fields of view. A 10x Mitu has a FOV around 2mm compared to 12.5mm for the litho lens.

The more recent versions of these lenses are telecentric, so they are ideal for stacking and stitching. Their aerial images are so big that the lens doesn't have to be moved when stitching. We can just move the camera across the aerial image, while the lens stays still. This will ensure perfect image registration across stacks.

Today I made my first test of a Zeiss photolithography lens. It was a Zeiss 50mm f/1.6 436nm lens with NA =0.28, mounted on a QHY163M monochrome camera for astrophotography, whose sensor can be cooled to -10 to -20 degrees Centigrade to eliminate sensor noise. I lit the subject with an undiffused strip of consumer-grade 440nm LEDs made into a ring around the lens; the results would surely improve with diffusion.

The working distance between the lens front (which I painted white) and the subject is about 12mm though difficult to measure exactly. The distance from the back of the lens to the front of the astro camera is roughly 443mm. I made black paper tubes to enclose the optical path between the camera and lens. The lens and camera are not directly connected. Tomorrow when there is daylight I'll post a photo of the set-up.

Here is a 100% crop of a single shot of a Vanessa sp butterfly wing, taken at -15 degrees C, exposure time 500ms. Of course at this NA, only a tiny part of the image is sharp.
Image
This is an image converted from a 16 bit .fits file with tremendous dynamic range.

Imagine this level of detail over a 12mm field of view!!!

Next step is to build a platform so that this whole assembly can be moved by a rail, for stacking, and build an x-y stage to move the camera so it can sample the whole gigantic image circle.

zzffnn
Posts: 1896
Joined: Thu May 22, 2014 1:25 pm
Location: Houston, Texas, USA
Contact:

Post by zzffnn »

Very cool stuff, Lou. I always enjoy reading your discoveries. Thank you for sharing!

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Thanks Fan!

Here is the set-up. The thing on the near end is the camera. It sends up to 20 frames per second to the computer; there is no on-board storage memory. The 436nm LEDs are lit.
Image

Image
The Tronde light is just for focusing help. The butterfly wing is glued to that vertical piece of glass about 1 cm in front of the lens.

The pictures above are of the Zeiss 10x 50mm. Here is the very similar Zeiss 5x 95mm with NA =0.20, next to a Mitutoyo objective for scale. I have much bigger ones but they are harder to set up and perhaps not worth the trouble.

Image

harisA
Posts: 513
Joined: Sun Jul 03, 2011 9:31 am
Location: Greece

Post by harisA »

Lou many thanks for sharing these information with us. Personally I would like to see how these lenses rendering colors (I know that they are optimized for single wavelength use) but there is always the possibility to be acceptable since you are actually using a tiny central segment of a huge image circle.
Also a real flat test target (like a semiconductor wafer)will be perfect to judge the performance of these lenses.

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

harisA, I will try that, but others have tested them and have gotten bad results with the 436 nm versions, though the 546nm versions appear to give decent results. At the moment my non-rigid coupling doesn't let me attain perfect parallelism or centration.

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

harisA, I will try the 5x photolithography lens (pictured above) on a color subject. That one might give better results. It still has a large image circle of 144mm.

Smokedaddy
Posts: 1951
Joined: Sat Oct 07, 2006 10:16 am
Location: Bigfork, Montana
Contact:

Post by Smokedaddy »

Very interesting Lou. What acquisition software are you planning on using? I know the Astro-photography folks are using software such as EasyCap, SharpCap, Sequence Generator Pro, etc., just curious.

On another note, I noticed there was a Zeiss S-Planar 1.4/75 m1:5nA=0.30 436nm on ebay.

-JW:

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Here is the 5x Zeiss 95mm with NA = 0.20. Sensor temperature about -12 degrees C. Single image, not stacked.
This is a 200% crop!
Image

Here is a slightly sharpened version with levels adjusted, 200% crop:
Image

The actual FOV is 25mm x 25mm, and the image is 125mm x 125mm. To capture most of the detail in this lens' aerial image, we would need a 1000 Mp sensor, or 8 rows of 8 MFT images stitched together.

Next will do a color version.
Last edited by Lou Jost on Thu Nov 16, 2017 4:44 pm, edited 1 time in total.

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Smokedaddy, sorry, I missed your post. I am using Sharpcap. I would really like to use Sequence Generator Pro because it has a temperature ramp program, but I can't get the program to work.

I am very disappointed with the software for manipulating .fits files. Basic functions like "undo" are often missing. Now I am using FitsLiberator to get the file into Photoshop as a 16 bit tif.

The eBay lens you mentioned is a very good one (there are several available now, including one for $1700 and one for $400). I got one of those a few months ago. I had to leave it behind in the US because of its weight, but will bring it next time I go to the US (along with several other 10-20 pound lenses!). That particular model has a nice tapered rear lens housing, which makes lighting easier than with the cylindrical flat-ended lenses I am testing in this post.

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

harisA, I just now hooked up a color camera to my assembly, and the results show so much LoCa (blue/yellow) that there is no point taking a photo. It is possible that a completely flat subject might work, but the halos around points of light are very intense and in my experience this would make it hard to use for stacking (even though stacking reduces the effect of LoCa).

harisA
Posts: 513
Joined: Sun Jul 03, 2011 9:31 am
Location: Greece

Post by harisA »

Hello Lou. Did you try 436nm or 546nm version?Just for reference I think you must post a single color image (no stacking)
thank you in advance

Haris

soldevilla
Posts: 684
Joined: Thu Dec 16, 2010 2:49 pm
Location: Barcelona, more or less

Post by soldevilla »

Hi. very interesting. Do you only take just one picture? My experience is that the result is much better when you take video, at least 1000 frames, and stack them with AutoStakkert. A deconvolution works wonders on the resulting image.
I like the idea of working in Black and White and I am going to try to fit a portafilter wheel to mount an RGB image, just as we do with the images of planets.

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

harisA, these two lenses are 436nm lenses. Here is a color version at 200%. Looks terrible.
Image

Lou Jost
Posts: 5942
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Soldevilla, there is much more light available here than in astrophotography. My individual pictures look virtually noise-free when the sensor is cooled down below zero. So I am not sure the extra complication is worth it. To make a z-stack in that approach, I'd need 1000 X 200 or so pictures, which would take a while. I do like that some of the astro stacking software automatically throws away outlier frames, such as those taken when a car passes and vibrates my house. That would be a nice feature for Zerene to have!!!

I set the camera to zero gain, and my shutter speeds are about 500ms.

How does the deconvolution algorithm work? Does it use point stars as a reference to generate the point-spread function for the lens?

soldevilla
Posts: 684
Joined: Thu Dec 16, 2010 2:49 pm
Location: Barcelona, more or less

Post by soldevilla »

You are right. I just tested my new camera QHY5III 178 (6Mpx with very small pixels for companions who do not practice astronomy) and an exposure with a well-filled histogram is enough.

I have tried a Componon 50 lens first. I have been amazed at the enlargement, but especially about the working distance, maybe 15cm.

Image

Image

Then I tried my x4 microscope and I expected more magnification, but no. I like the Componon. But since it was already mounted I have tried a trichromy. In order not to degrade the image and since I have mounted the filter wheel in the telescope, I have simply used another RGB set changing filters in front of the LED spotlight. The resulting color is very good. The result is not so much ... I have to focus between each channel and I had to do it by hand. But at least, it works.

Image

I use ASTRAIMAGE for deconvolutions. I simply modify the size kernel until I get the best result.

Post Reply Previous topicNext topic