How to keep water out of lenses, and make stepper motor AF crazy-fast - Olympus Technical Q&A
posted Sunday, June 9, 2019 at 6:46 PM EST
As I usually do, I stayed on in Japan after this year's CP+ trade show, and while there had a chance to visit Olympus' research and development headquarters in Hachioji where I sat down with some of their engineers. My interview at CP+ itself with Shigemi Sugimoto-san, Head of Olympus' Imaging Business Unit, was more focused on business issues and product strategy; this meeting was more about specific technical questions we had about Olympus' products, particularly their lenses, weather-proofing, and their latest autofocus algorithm updates. Read on for all the details...
Dave Etchells/Imaging Resource: Back in 2011, Olympus launched its ZERO coatings with the M.ZUIKO DIGITAL ED 12mm f2.0 lens. What’s the underlying technology there, and how does it differ from what came before? Are they standard dichroic coatings, and how much better are they than what you had before? And what led to the development of that improvement?
Shigeru Kato/Olympus: As you imagined, this is a general dichroic coating, or multiple coatings.
Dave Etchells: Mmm, mmm.
Shigeru Kato: What is our original technology is how thin we can make the layers, how well this can be controlled. That's our technology, which came from the microscope division.
Dave Etchells: Ahhh! Huh.
Shigeru Kato: And the benefit of the thin coating is the very low reflection rate. It reflects half the normal amount reflected by the usual coatings...
Dave Etchells: Half the normal? That's a big improvement! That's interesting. I need to study dichroic coatings more, but I know they use optical interference to get the light to cancel out?
Shigeru Kato: You know the brand name ZERO is the initials Zuiko Extra-low Reflection Optical coating: ZERO coating.
Dave Etchells: Mmm, mmm. Yeah, as I said, I will have to study to understand how making very thin layers helps.
Shigeru Kato: The benefit of the thin coating is that you can make more layers, which helps reduce the reflection. I don't know why. <laughs>
Dave Etchells: Ah, maybe you can specifically block more wavelengths. So instead of one notch, you can have multiple of those. <gesturing>
[Ed. Note: Anti-reflection coatings are a very complex topic. The simplest form would be a single layer of material with a different refractive index than the underlying glass, with a thickness 1/4 of the wavelength (or some odd-numbered multiple of that). Light waves of that specific wavelength that bounce of the front and the back of the coating will be 180 degrees out of phase, so will cancel each other out. But note that that's only for a single wavelength of light, and only for light that's hitting the lens head-on. Light having different wavelengths or striking the lens from an angle won't cancel out like that, but you can use other layers of other thicknesses and compositions to cover more of the spectrum. This was my first thought when Kato-san mentioned the benefit of thin layers meaning that you could have more of them. Actually, though, as I've been writing up this interview, I've discovered that it's not just air/glass interfaces that cause internal reflections, but the interface between different types of glass as well. Sometimes what you need to do is to build a stack of very thin layers of gradually-changing refractive index, to provide a smooth transition between one type of glass and another, in stacked lens groups. A gradual transition from one refractive index to another at a sub-wavelength level can almost completely eliminate reflections from the interface.]
Shigeru Kato: Yes, yes.
<laughter at my questions and discussion>
Dave Etchells: Yeah, I'm an engineer, but optics was always one of my weakest subjects.
Toshi Terada/Olympus: Your background is applied physics, right?
Dave Etchells: Yes, yes, right.
Shigeru Kato: Applied physics?
Dave Etchells: Semiconductor physics, yeah.
Toshi Terada: I don't like to say, but me too.
Dave Etchells: You too? I don't think I knew you were applied physics! Really, Toshi-san?
Toshi Terada: In the past.
Dave Etchells: Also the distant past for me, yeah.
Shigeru Kato: Me too.
Toshi Terada: Oh, really?!
Shigeru Kato: Three physicists.
Dave Etchells: Wow, three physicists; we need to go have a physics party next time!
Toshi Terada: Okay. <laughs> Why not? <laughs>
Dave Etchells: That's very interesting. And also, in a transfer function a zero means zero transmission, so you have more zeros in the function with more layers, neh?
Shigeru Kato: Mmm, sure.
Dave Etchells: We have a question on the sealing of the 12-200mm lens. You called it hermetically sealed in the marketing materials and the presentation. But I tend to think of hermetic as being completely gas-tight. Especially with a zoom that extends, you have to draw air into the body somewhere. So how do you keep the water from getting sucked in, and I guess what does the term hermetic mean to you?
Shigeru Kato: This is the zoom lens. That line shows the sealing gasket. The lens barrel moves in and out like this.
Dave Etchells: And there is a gasket?
Shigeru Kato: A wiper.
Toshi Terada: Firstly, yeah, here there is a gasket, right? And then on the lens barrel we have a gasket here. <pointing to the very front of the lens body, as in the image above>
Shigeru Kato: This first gasket just wipes out the water, it doesn't seal the internal parts.
Dave Etchells: Ah, ok. It's just a wiper, not a seal. Like a squeegee.
Shigeru Kato: Yes, that's right. And as you can imagine, we need a hole to balance the air flow. In this diagram, where it shows yellow under the zoom ring, there is a hole to allow the air out here.
Dave Etchells: Ahhh!
Shigeru Kato: It's covered with mesh, can you see?
Dave Etchells: Oh, very interesting. Kind of like how Gore-Tex fabric works? Holes that are too small for water, but air can come in. Oh, that's very...
Shigeru Kato: Yeah, it's a very similar material. This is part of the new lens. The air comes through the mesh, and finally goes out through this gap. <showing me a sample of a very fine gauze-looking fabric, with a mesh so fine that even small droplets of water can't pass through it, even though air can>
Dave Etchells: The gap at the front of the...
Shigeru Kato: Yes.
Dave Etchells: Yeah, hai. So it's that gap.
Dave Etchells: OK. And so really when you said hermetically sealed, you were meaning only water is excluded. Like I said, usually I think of hermetic as like a metal-to-glass seal, or something like that. But that's interesting. And the mesh has a very large area, so there's only a very low pressure differential. That way, the water is not forced through the mesh; it can just bead up and then drain off later.
Toshi Terada: As you can imagine, if it wasn't easy for the air to go out or in, it would be very hard to zoom the lens.
Dave Etchells: Oh, hard to move the zoom, yes.
Toshi Terada: Sure. So we need to make it easy for air to go through this part (indicating the area covered by the very fine mesh). That's the reason it's such a large area, to easily allow air to go out or in.
Dave Etchells: Mmm. Also, if you apply enough pressure, water will go through presumably, right? If you just had that material in a tube and you applied water under pressure, the water could go through with enough pressure.
Shigeru Kato: Yes, you're right. This doesn't perfectly block the water. That's the reason this part is not visible from outside the lens barrel. It's covered here. To avoid the water directly hitting the mesh.
Dave Etchells: Ah, if rain hit it directly, it perhaps could go through.
Let's move on to a question about the stepping motor, the autofocus actuator. A number of companies are using linear actuators now, and at least one of those competitors claims that only linear actuators can be fast enough for mirrorless cameras. To what extent is it true that stepping motor drives are slower than linear actuators, and how are you able to get enough performance from the stepping motor actuators in your lenses?
Hisashi Takeuchi: Autofocus speed is not defined by the speed of the motor. The first priority is how short the lens travel is during focusing. A short stroke gives quicker autofocus, as you can imagine, right?
Dave Etchells: Mmm. Mmm-hmm.
Hisashi Takeuchi: Of course, then it's more difficult to manage the control itself, but speed-wise...
Dave Etchells: Yeah, speed-wise, shorter is better, but then you also have to be that much more precise.
Hisashi Takeuchi: Sure. Under those conditions, the linear motor is quicker than stepping motor. That's true in the past, let's say, for example. <laughs> It's both a good thing and bad thing, right? And our R&D doesn't simply assume the linear motor is better than a stepping motor. It depends on the lens design and other elements. We try to choose the best actuator for autofocus [depending on the particular lens design]; we're not fixed to the linear motor itself.
Dave Etchells: Mmm, mmm, mmm.
Hisashi Takeuchi: We can minimize the speed disadvantage of the stepping motor with our technology, to achieve fast autofocus even with a stepping motor. With our new technology, I think there's not such a big difference between the autofocus speed using linear motor or with stepping motor.
Dave Etchells: Mmm! So you have made the steppers faster, and part of your technology is to be able to step more quickly?
Hisashi Takeuchi: Usually, stepping motors use open-loop control. But we are using the feedback-loop control. That's how the R&D team overcame the disadvantage of the stepping motor in terms of the speed.
Dave Etchells: Mmm. With the feedback loop, you're not depending completely on the structure of the motor to move or stop precisely.
Hisashi Takeuchi: Yes, sure. You know when using open-loop control, sometimes when we try to move two steps, the motor only moves one step.
Dave Etchells: Oh, you can miss or skip a step, and not realize it, yeah.
Hisashi Takeuchi: Yeah, that's right. So then, if we use open-loop control, we always have to be aware of this risk. It doesn't provide the full performance to the AF drive. We need to stay within tolerance for any condition. That's the reason: To avoid this failure, we cannot give the highest-speed movement with open-loop.
Dave Etchells: I understand.
[Ed. Note: This is very interesting indeed. It's a common issue with stepper motors that when you push up against the upper limits of their speed/power envelope, there's a significant risk of missing steps. One way around this is to add a rotary encoder to the motor's shaft, so the stepper driver can hit the motor with a fresh pulse if it sees that it's missed a step, or change the drive voltage and current to compensate for the higher-speed pulse train. This is what Olympus appears to be doing, and explains how they can extend stepper-motor AF drive to much higher performance levels than is commonly considered possible.]
Dave Etchells: So we're curious about the choice of the EVF on the Olympus E-M1X, and wondering why you didn't use a higher-resolution panel for it?
Tetsuya Toyoda/Olympus: Of course we know there are higher resolution panels in the industry. But when we choose the best performance of the the mirrorless camera's EVF panel, in terms of the refresh rate, as well as the panel quality -- the color or brightness -- on those aspects, we chose this panel as better than a higher resolution model.
Dave Etchells: Ah, that makes sense.
Tetsuya Toyoda: Of course, the resolution itself is also an important factor for the viewfinder. But we have to consider what piece is best including the speed, resolution and image quality. From those aspects, we selected the current one.
Dave Etchells: Hai. Just generally, with any sensor, there's trade-offs. As you increase resolution, you lose dynamic range, that sort of thing. How much can you push the resolving power of Micro Four Thirds, and will a lot of your continued improvement in image quality be around lens technology, or additional improvements in software and image processing? Or do you feel that the market segment that Murata-san [Ed. Note: Aki Murata, Vice President Sales and Marketing, Olympus America] identified in our interview at PhotoPlus last year -- outdoor moving subjects -- doesn't require single-shot resolution more than 20 megapixels?
Tetsuya Toyoda: This is like the story we mentioned for the EVF panel. The answer for the image sensor is that there are some aspects to consider for the mirrorless camera. Not only resolution, but also the readout speed, dynamic range, and color reproduction.
Dave Etchells: Mmm.
Tetsuya Toyoda: So yeah, we have to consider these aspects, not only the resolution. As you know, also, image sensor technology is improving year by year. But at this moment, our best balance of those aspects is at 20 megapixels. On the other hand, typical landscape shooters are asking more resolution. Not from sports shooters or for moving subjects; just landscape photographers are asking for more resolution than 20 megapixels. To me, answering that kind of demand, as you know we offer the multi-shot technology for high resolution.
Dave Etchells: Mmm, yeah. This is a typical question we get from some readers. Personally, I feel like the megapixel race is... it's like Americans always want a bigger number, whatever the number is, you know? But I don't know that you really need 50 megapixels for sports shooting, right?
Tetsuya Toyoda: Yeah.
Dave Etchells: I'm in agreement with 20 megapixels being a good balance, but the readers keep saying "More, more. More, more."
Tetsuya Toyoda: Yeah, sure. I understand.
Dave Etchells: A reader asked for more specifics about the different product lines. Sugimoto-san said all product lines will be continued, when I interviewed him recently. I was really focused on E-M5 line, but the reader was asking specifically about the E-M10, E-PL series and PEN-F. Will all of those lines be continued? I think a reader came back and said "No, the PEN-F will not carry on." So what is the situation, or what are your plans generally?
Toshi Terada: We cannot say more than that. As Sugimoto-san said, the existing lineups will be continued with new models, right? We cannot catch up on them just one by one: How is this product or this product, it's really difficult to say. If you want to ask us, all we can say is "No comment" or something. <laughs>
Dave Etchells: Yeah, yeah.
Toshi Terada: You can imagine some of the product planning is very important for the market. But if Sugimoto-san promised it to market, we will do it. <laughs> Right?
Dave Etchells: Ah, OK.
This was also a reader question in response to my interview with Sugimoto-san. Toshi-san said that the basic AF performance of the E-M1X was much improved. But the reader questioned that, saying that reviews so far haven't found that to be the case, and that DPReview in particular said that AF tracking seemed about the same. Is it a matter of that there are specific use cases that have been improved, and can you elaborate more on how AF has improved?
Hisashi Takeuchi: So our improved autofocus system performance, the point which has been improved is the autofocus tracking performance for moving subjects. We can say the situation of autofocus for moving subjects is very complex, with so many cases. For example, a subject that is suddenly moving, or changing its speed, or suddenly stopping. Right? We try to improve the continuous autofocus performance for moving subjects generally.
Dave Etchells: Mmm-hmm.
Hisashi Takeuchi: This time, we implemented a new algorithm for autofocus, especially for group targets.
Dave Etchells: The group AF point, yeah.
Hisashi Takeuchi: Yes, yes. With both five by five (25-point) or three by three (9-point). The algorithm that we implemented in the E-M1X is the center-priority algorithm for group targets. Then we dramatically improved the tracking performance with group target. Now we have the menu settings in the camera.
Dave Etchells: Ah, and then you can enable center-priority in the different groups separately?
Hisashi Takeuchi: You can select what type of group targets should be center-prioritized, and also the C-AF sensitivity.
Dave Etchells: Like how sensitive it is to small changes in subject distance?
Hisashi Takeuchi: Yes.
Yuka Iwasawa/Olympus: Subject motion.
Hisashi Takeuchi: For example if someone passes in front of your subject, the focus system can move to the new subject or stay on the original subject. That kind of sensitivity.
Dave Etchells: Ah, basically "stickiness".
Hisashi Takeuchi: Sure, stickiness, yes.
Dave Etchells: It sounds like the most significant thing is the optimization of center-priority. Is that on by default, or not on by default?
Hisashi Takeuchi: Now, the default is center-priority.
Dave Etchells: Hmm. I haven't read DPReview's article, but I think that as we have done sometimes, they just have a guy on a bike. That is the problem with testing autofocus, is there are in fact a lot of different scenarios the system could encounter, and it may respond differently to all of them. Two years ago, Ogawa-san [Ed. Note: Olympus Corp.'s Senior Executive Managing Officer, Chief Technology Officer and Director] told me how most companies have maybe ten scenarios, but Olympus has more than 15.
Toshi Terada: Sure.
Dave Etchells: And so just a guy riding a bike is one scenario, and maybe [in that respect, performance] is the same as the previous generation, but in other situations it might actually be much better. That is good information about the center-priority tracking improvement; I'll be sure to call attention to it in our writeup of this interview.
Another reader question was interesting. Apparently Olympus wrote a paper about stacked sensor technology back in 2015. So the question was why haven't we heard about that in a product. Is it still coming, or was it just an isolated R&D exercise? Where does the whole idea of stacked sensors stand with Olympus?
Toshi Terada: At this moment, we have no new comment for this. As you know, Olympus Corporation has several separate departments or businesses. We are related to the imaging company; we have imaging products and technology. And there is a department that takes care of more fundamental technology for all three business units. I think that this stacked-sensor technology is coming from that department. We will try to obtain information on that for you. But usually, these fundamental technologies need time to become a real product. I don't know [whether it's developed for] our imaging products or medical products, but yeah, that's a fundamental development or research.
[A reply received from Olympus R&D later, through Yuka Iwasawa: "The stacked image sensor technology, which was announced in 2015, was one of our research and development theme/topic. We are considering the application of these technologies that we acquired from this research to our business fields, however it is not likely to be applied to the imaging products at the moment."]
Dave Etchells: Well, I think we're out of time; thank you as always for all the information you shared!
(Group): You're welcome!
My summaries of these tech interviews probably all sound the same, because they always begin "this was interesting..." But this was interesting! :-)
While it's an older technology, I hadn't previously been aware of just what the key technology point with Olympus' ZERO coating technologies was. (It turns out the key capability was figuring out how to make both much thinner layers -- and many more of them -- in their anti-reflection coatings, thereby reducing internal reflections by a factor of two.) We also learned how Olympus gets so much higher performance from their stepper-motor AF actuators than conventional wisdom says they should. Finally, they answered the question of what specifically has been improved in the E-M1X's standard (as opposed to AI-based) autofocus algorithms, something that's puzzled both users and reviewers alike.
All in all, a great discussion. Given my passion for weather-resistance testing, I particularly enjoyed learning about how manufacturers (or Olympus, at least) manage to get air in and out of a zoom or external-focusing lens without also drawing in water.
What do you think, do you have any questions about any of the above? Leave them in the comments below; I'll watch the comments thread for this article closely for a few days, so I can respond to any questions you might have. (Or at least the ones I might have answers for.) ;-)
*Update note, 6/12/19, 1:25pm: I made a gross error, in identifying Hisashi Takeuchi at the top of the article as Kosaku Nishio (General Manager, Imaging Product Development, Department 3). Sincere apologies first to Mr Takeuchi-san and secondly to our readers. Besides comments by Mr. Takeuch not being attributed to him, several other replies were also attributed to the wrong person. The errors were entirely mine; I'll try to be more careful in the future, to make sure I correctly acknowledge all participants and correctly attribute all remarks! (And many thanks to Yuka Iwasawa of Olympus Japan for calling the errors to my attention.)