Rayfact 2x-5x lens tests

nathanm · Post by **nathanm** » Fri Jun 30, 2017 1:13 pm

I have a Rayfact 2X-5X lens on loan from Tochigi Nikon to test it. Tochigi Nikon is the subsidiary of Nikon that made the legendary Printing Nikkor lenses - and they still sell an updated version of the 105mm and 95mm.

Like the SK macro varon lens, it has a floating element which adjusts away aberrations at any magnification between 2X and 5X.

A lens designer tries to push aberrations off to conditions that are not typical. That is why if you put just any old lens on a bellows or extension tube it likely will not perform well - yes you can get close focus, but the quality suffers.

Microscope objectives solve this by being designed for a single magnification. Camera lenses sometimes do the same thing - the Printing Nikkor 105mm, for example, is optimized for 1X. It is fantastic there but the quality suffers at other magnifications.

A floating element lens gives the ability to adjust away aberration at any magnification.

The rayfact lens is expensive - I have been quoted about $17K. But it may be the ultimate lens between 2X and 5X. So I got one on loan to test it.

The lens has spectacular performance specs

http://www.tochigi-nikon.co.jp/en/produ ... Ver4.0.pdf

At 2X it has an NA of 0.133 and at 5X an NA of 0.167. That is better than the Mitutoyo objectives - their 2X is 0.055 and their 5X is 0.14.

An additional feature is that it has an 86mm image circle, which means that I can pretty easily fit the large sensor on my PhaseOne 100 megapixel digital back.

That even opens up the possibility of moving the back and taking stitched panoramas.

Beside being expensive, the lens has a focal length of 116 mm. Which means you need a LOT of extension. At 5X the sensor must be 600mm from the flange.

Here is my test rig

That is a Sinar 4x5 bellows which is at its maximum extension, but it wasn't long enough so the lens is mounted on a piece of 4"x4" square aluminum tubing.

nathanm · Post by **nathanm** » Fri Jun 30, 2017 1:28 pm

The first set of tests have gone incredibly well. The lens is essentially perfect at both center and edge of frame from 2X to 5X. No CA, no issues of any sort. I was going to post them but they are really boring, so I am not sure it is worth the trouble.

Of course the center and edge of my PhaseOne frame are still very far inside the image circle. Here is an illustration.

The blue rectangle is full frame 35mm (24mm x 36mm) the black is the PhaseOne (53mm x 40 mm), and the red circle is an 86 mm image circle.

Because they sell the lens for use with a line sensor, which can be 82mm to 86mm, they can't tolerate much quality issues at the edges. So I think that when they say the image circle is X, they really mean it, unlike a photographic vendor where they assume you can tolerate some fall off and issues with the corners.

Of course the image circle of 86mm is only at 2X. The image circle grows with magnification, so at 5X it would be huge - 215 mm.

My next test is to probe closer to the edge of the image circle, using a sliding back that lets me move the PhaseOne sensor inside the image circle rather than having it be centered.

Lou Jost · Post by **Lou Jost** » Fri Jun 30, 2017 2:36 pm

Very exciting, even if out of just about all of our price ranges.....One curious thing was the increasing NA as magnification increases. I guess that is because the entrance cone gets bigger as the subject gets closer to the lens front. What formula would the effective aperture follow?

nathanm · Post by **nathanm** » Fri Jun 30, 2017 3:09 pm

I am not sure what formula is used for the NA - I was just taking the numbers from the brochure. Because of the floating element this may not be something that a standard formula does well.

Here is a photo from the front.

You can see the stepper motor and gear that I am using to automate the floating element ring. The lens was designed with this in mind - the ring is already a gear. The SK macro varon is similar in that regard.

ray_parkhurst · Post by **ray_parkhurst** » Fri Jun 30, 2017 3:09 pm

Great initial report. I'm glad to see Tochigi Nikon continuing to innovate, and to produce such an excellent product. For various reasons this is obviously a specialized lens but having such a reference is really cool.

Seems the industrial market that is most relevant for our purposes is geared toward line scan inspection.

nathanm · Post by **nathanm** » Fri Jun 30, 2017 3:21 pm

Yes, the fact that Teledyne Dalsa, and others, make large line sensors with 16K pixels has definitely driven the creation of new lenses.

This is driven by a need in manufacturing industry to do quality inspection on things like circuit boards and LCD displays at high magnification.

Most of the line sensor lenses are designed for a fixed magnification, but manufacturers are making some, like this lens and others (the SK macro varon) which have variable magnification adjustment.

Which is nice because most of the other great macro lenses have been out of production for decades. It is great that some modern technology is going into this area.

Lou Jost · Post by **Lou Jost** » Fri Jun 30, 2017 3:29 pm

The unexpected thing about the NA is that the effective apertures calculated from them don't seem to increase with m as fast as one would predict from a simple lens on extension, or even a zoom lens, even with floating elements. Yet it sort of makes sense in terms of the entrance cone getting bigger with increasing m. Maybe there is an interesting principle involved. Hope Rik comes by...

Sager · Post by **Sager** » Thu Jul 13, 2017 11:57 pm

Interested to see a shot with the lens, i'm wondering how deep you will stack with the extra working distance and the motorized optimization of the floating element - the working distance must be incredible at that focal length. Definitely a lens to push to the extremes.

rjlittlefield · Post by **rjlittlefield** » Fri Jul 14, 2017 1:56 pm

Lou Jost wrote:The unexpected thing about the NA is that the effective apertures calculated from them don't seem to increase with m as fast as one would predict from a simple lens on extension, or even a zoom lens, even with floating elements. Yet it sort of makes sense in terms of the entrance cone getting bigger with increasing m. Maybe there is an interesting principle involved. Hope Rik comes by...

I'm guessing that you have not included pupil factor in your calculations.

I just now ran a quick calculation using Excel Solver to answer the following questions:

What values of nominal f# and pupil factor are required to match the published NA's as well as possible, using the simple formulas that f_e = f_r*(m/p+1) and NA = m/(2*f_e) ?
How good is the resulting fit?

To my surprise, it turned out that just having a nominal f# of 2.54 and pupil factor of 1.022 produces calculated NA's that match all digits shown in the values for all 8 magnifications published at http://www.tochigi-nikon.co.jp/en/produ ... Ver4.0.pdf .

So, the moving element for aberration correction may be in play also, but it doesn't have to be. A slight difference in the pupil sizes is enough to explain the numbers.

--Rik

Lou Jost · Post by **Lou Jost** » Fri Jul 14, 2017 3:43 pm

Thanks Rik, I knew you'd be able to figure out what was going on. I didn't realize the pupil magnificatiomn factor was so important in this calculation.

rjlittlefield · Post by **rjlittlefield** » Fri Jul 14, 2017 5:54 pm

Lou Jost wrote:Thanks Rik, I knew you'd be able to figure out what was going on. I didn't realize the pupil magnificatiomn factor was so important in this calculation.

Yeah, well, don't be too impressed. I was in a hurry and it looks like I took too much for granted in bothering to set up that fitting calculation at all.

When I looked again at this problem, what I see is that just plugging in the lens's nominal f/2.5 and an assumed pupil factor=1 also matches all digits of the published data.

For example:

A) 2X at f/2.5 on simple extension gives effective f/7.5 at the sensor, which implies f/3.75 at the subject, and 1/(2*3.75) = 0.1333 (repeating)

B) 5X at f/2.5 on simple extension gives effective f/15 at the sensor, which implies f/3 at the subject, and 1/(2*3) = 0.1666 (repeating)

What calculation was it that led you to say "the effective apertures calculated from them don't seem to increase with m as fast as one would predict from a simple lens on extension"?

--Rik

Lou Jost · Post by **Lou Jost** » Fri Jul 14, 2017 6:21 pm

I had used the sensor-side EA to calculate the NA.

rjlittlefield · Post by **rjlittlefield** » Fri Jul 14, 2017 7:47 pm

Lou Jost wrote:I had used the sensor-side EA to calculate the NA.

So, calculated as NA = 1/(2*EA), but plugged in a sensor-side number and interpreted the result as subject-side?

That sort of thing happens a lot, especially when working from formulas that try to be concise by not spelling out exactly what they're talking about. Tradeoffs, tradeoffs...

--Rik

Lou Jost · Post by **Lou Jost** » Fri Jul 14, 2017 8:06 pm

So, calculated as NA = 1/(2*EA), but plugged in a sensor-side number and interpreted the result as subject-side?

Yep, that's what I did. I only thought about the sensor-side EA.

Lou Jost · Post by **Lou Jost** » Sat Jul 15, 2017 7:43 am

Rik, what's the relaation between the sensor-side and subject-side EA? I have always assumed the sensor-side EA is what determines the amount of diffraction. Is that really the case?