Question about collimated display systems.

[Atomic_Sheep]

Hi Geneb, I didn't really pay attention to the resolution of the projector that you were using so I wasn't criticising your setup, I was just trying to figure out what aspect ratios were and how they fit on a spherical display. I just read the wiki page on aspect ratios and my wild stab in the dark in relation to my assumptions as to the relationship between viewing angle and resolutions proved correct or at least I think so. The question that I'm still mulling over in my head though is the whole issue of different circumferences at the top and at the bottom and how an image which is 800 pixels (or 853.3 to be exact) can fit on something that is lets say 100cm up top and 80cm down the bottom or does it simply get scaled due to the ?refraction?/?defraction?

In relation to building a 60 degree vertical field of view. After getting into the topic of cullimated displays just yesterday (just to give you guys an idea of how little I know and how totally not based on any experience my thinking is but)... I was thinking that the only real restriction that I could figure out that made the 60 degree version nearly impossible to make was just the nature of materials and the fact that it would be more difficult to get a nice spherical shape after applying the vacuum as there are essentially 4 points of contact... well... 2 if we're talking about the vertical plane... so in other words, with 40 degrees, it's not very far off the back plate that it needs to vacuum to but with 60 degrees, there's a much bigger space behind it so I was thinking along the following lines:

We know the properties of the Mylar (at least I'm guessing the specfications of the stuff would tell it's elasticity per square cm or inch or meter or whatever... I'm assuming that whoever designed it, knows such properties...). If that's the case, then let's assume the 40 degree version uses a stupid method of calculating the right size material i.e. we take the vertical distance between the two mounting points and the circumferences and then simply attach it, vacuum it and we're done. With the 60 degree version, due to the greater amount of stretch required (based on absolutely no experience remember)... the Mylar would stretch at a much greater range... if that makes no sense, then hopefully this will clear it up. If you stretch any material, the way I see it, it stretches unequally, in some spaces it stretches a lot, in some not so much and I'm guessing it stretches the most right in the middle of two points holding the material in any given lateral direction (I could be wrong here and it might stretch the most at the points of contact and least in the centre but I hope I've made my point clear). SO! What we need to figure out is how much stretch is happening at the different parts of the Mylar on the 40 degree one and then scale it up to the 60. In other words, we need to basically create a piece, which if vacuumed, will have similar amount of stretch at similar places to the 40 degree one.

In the image below, I'm showing my train of thought.

Mylar Shape.jpg

The smaller the vertical field of view, the more we can get away with a simple rectangular piece, the larger the vertical field of view, the more we need to make it bend. I'm not accounting for the fact that the circumferences are different at the top and bottom which would I assume, make the shapes a little bit more funky around the sides but I hope I made my point clear. Perhaps those more in the know might be able to correct what I'm saying and totally discredit it .

EDIT: Just found this in a patent document:

"It is known for typical wide-bodied cockpits that a 40 degree vertical field of view may satisfactorily be achieved with a spherical mirror of about 3m radius and a spherical screen. A vertical field of view of 50 degrees may be achieved with the same radius mirror, but the screen must become aspheric. For a 60 degree vertical field, not only must the screen become aspheric, but also the mirror radius must increase to about 3.3m."

[wledzian]

I don't have my raytracer handy, or I'd make some good illustrations for you. I'll include them in a later post.

If the design eyepoint is very near the center, the field of view is limited to about 40 degrees. This is partially due to material limitations - when the mylar is stretched from a conic shape developable from a flat surface into the spherical shape, the arc defined by about 40 degrees produces a stretch very close to the material limits of the Mylar. If you try to stretch more than that, and you pass over the peak of the stress-strain curve and the mylar pops.

There is also the choice in vertical positioning of the eyepoint. If you move the eyepoint downward, that same 40 degrees of arc can cover more than 40 degrees of vertical FOV, because you're physically closer to the bottom of the sphere. Because of the extreme off-axis eyepoint, spherical aberration comes into play more strongly, and the required screen shape gets weird. The volume in which the collimation effect is acceptable shrinks. The distortion band around the bottom edge also increases in width, just due to the varying amount of stretch. Since you're further down in the sphere, the entire sphere has to be enlarged in order to fit the cockpit structure. Also because you're so far down in the sphere, the shape of the unrolled film becomes more curved, leading to a requirement for an even wider sheet.

There is also the choice of horizontal eyepoint position. If you move your eyepoint towards the mirror, the usable portion of the mirror in front of you fills a larger vertical FOV, but the portions to the side fill a smaller vertical FOV. One solution for getting a big vertical FOV is to make a big mirror and sit closer to it. This works well for large commercial sims, as the cockpit structure and windows does a good job of cutting the FOV to the sides anyway. In fact, with each pilot shifted foward and outward from center, each pilot has the largest FOV just outboard of the main windows, just as in the actual aircraft. The screen design gets more challenging, but by no means impossible. Again, though, to fit the cockpit structure within the mirror, the mirror needs to be rather large. As I still haven't found a source of mylar in sizes larger than ~56" without requiring the purchase of an entire 10,000 lb production roll, that is the main limiting factor at the moment.

[mikesblack]

wledzian,
Sending you a pm.
Mike

[geneb]

Wayne & I got a lot of really good design work done this weekend - when I've got time to re-create the model in Inventor, I'll post pics. (I should note that he DOES know how to use AutoCAD! *laughs*)

The big win this weekend was a solid grasp of the geometric limitations of the simulator cab. Because we're limited to the size of mirror we can make with a 56" wide roll of Mylar, that kind of dictates what cockpits can be built for use in the system. Right now the tiny GA 2 placers are an easy fit. The Cessna 152 has got tons of room and the 172 will fit as well. Both wouldn't be good choices because they're high-wing and the mirror goes from 0 to 40 degrees, which is far less than what you can see out the side windows. Any one of the "standard" low-wing GA aircraft should work though (Piper, Mooney, & Beechcraft are the popular ones). Wayne even managed to squeak a King Air 200 into the space available. It would require a special 3D model in order to show the wing & nacelle on each side, but that's fairly easy to do.

We also were able to cost-reduce the framework quite a bit. This is a Good Thing(tm). My original design was to use 12mm Baltic Birch throughout with 3/4" plywood partial width doubler layers in specific places.

We backed off the 190 form to 180 in order to make it break down to three sub-assemblies that would be easier to manage - the design goal is to create no component too large to navigate the interior spaces of Wayne's house. The new design uses 7/16" OSB for the top, bottom and end plates and 3/4" plywood for interior screwing rails. Each segment will bolt to one another using 1/4-20 bolts. The center segment and "interior" ends of the left & right segments will be under-sized along the mirror arc in order to prevent any contact with the mirror as it forms. The segments will also use two 1/32" rubber gaskets in order to prevent leaks.

The segments will have multiple coats of paint in order to make them as leak-proof as possible. (A high vacuum will actually pull air through OSB and some other materials like MDF)

When I post pics, a lot of my rambling will make more sense I think.

g.

[Atomic_Sheep]

Wow, Quite disappointing in regards to the 56" maximum size. I've got a got a feeling that cullimated guys should come together and purchase the big roll and then use what we need and either try to sell the rest to farmers or something, or simply keep it and sell it as other guys decide to get into cullimated displays.

Sounds good geneb, can't wait to see your progress. I'm probably going to do something similar to your desing i.e. build it in 3 parts for some flexibility if moving is needed. I've just got one question, I can't remember reading your solution to the back of the frame i.e. when you create the vacuum, do you have some sort of flat backing that you have attached to the formers to which the mylar conforms when the vacuum is created or do you simply create a vacuum and use whatever shape the mylar naturally forms? I was thinking making the whole thing out of fibreglass but I then realised that fibreglass isn't exactly flat and finishing it off to make it flat would be a nightmare. I then thought of simply creating a large number of formers such as the ones that you have and simply sticking them together side by side and creating a solid wooden spherical strucutre type backing but then I realised it would be insanely heavy and totally not appropriate for me considering I want to in the long run create a motion platform. Carbon fibre would just be too expensive so that's not really an option either. Any ideas in this regard?

[wledzian]

When I say "full production roll", I'm talking the full, ten-thousand-pound spool.

The vacuum is used to form the mylar. It is not used to hold it to a surface. The volume behind the mirror is hollow. That being said, you can build your frame from whatever material will be strong enough to stand up to the loads without significant flexing. Fiberglass would be a good choice, if you're comfortable working with that material. The required stiffness could be achieved fairly easily by making it a foamcore structure.

[Atomic_Sheep]

When I say "full production roll", I'm talking the full, ten-thousand-pound spool.

Yep, I understand which is why I said we needed to pool resoruces to get it. Sure, that's a lot of people! But still :P. Not totally serious about that, I'll admit.

Do you think that the guys in the video (forgot which company they were from) who were doing the 777, do you think/know their mylar conforms to the backing material? I've still got a lot to learn about Collimated displays, I've only recently started reading about them so I don't know quite a lot of basic stuff. I remember there was mention of this in the earlier threads but I didn't find a subsequent response which I thought answered that question. I suppose the fact that there was talk of a feedback loop indicated that indeed, they were calculating some sort of optimal value for the amount of volume of air/pressure in the 'vacuum' to which the system tried to revert back to throughout the operation but I'm not sure.

[wledzian]

The mylar does not conform to a backing material.

The mirror frame is hollow. The mylar is stretched into final form by the vacuum. Except for the the edges where it attaches to the frame, the mylar does not touch anything.

For the prototype mirror, we used an adjustable bleed valve on the back of the frame. This was adequate for our purposes, as we haven't used the mirror for any long period of time. We've had to slightly adjust the valve each time, and this would be very inconvenient once the final version is built. A feedback device that operates purely on pressure differential may not work adequately, as mylar does return to its original form somewhat, but not completely, and has the odd property of slowly stretching back to its final shape over time. That is to say, if the vacuum has been turned off for a while and turned back on, it will require a stronger vacuum to draw it down initially, but will continue to stretch a little bit and require that the vacuum be relaxed a little as it stretches. Gene's got a concept for a contact position sensor based on a fine copper tinsel. I've got a concept for a contactless optical sensor based on phototransistors. Both concepts will modulate the vacuum pump and a small bleed valve in order to maintain the mirror at the proper position.

[geneb]

The cool thing is that both systems could co-exist and might even be able to work together. I suspect that the robot builders have a lot of experience with using photo transistors and could be of some help in perfecting the concept you want try out.

I checked out using the little ultrasonic sonar units the robot builder types use, but the ones I found didn't appear to have the resolution we'd need. I should check with the Arduino guys and see what they'd recommend for very precise distance-to-surface measurement. I'm going to need to find a solenoid valve too - that silly thing I bought from Surplus Center.com was waaaay to tiny to use for anything. Although, if the pump has a check-valve in it, we might not need the secondary valve - the pump itself would prevent vacuum loss when it was turned off.

Lots and lots of fun stuff to think about.

g.

[Jordan Farmer]

Hi guys, just to have my say on the vacuum side of things! Im a farmer, as you may have probably guessed, and use vacuum pumps on a daily basis for the milking machine. We have now got 2 robotic milking units, which use a variable pressure pump, which basically pumps ata much higher rate while pulling the desired vacuum, and then when it has the pressure, it slows down, until it will just maintain that pressure, its also extremely quiet, i often dont realise its going, so n a small box, it would be almost silent.

This may help you, it may not, but do look into the vacuum pump solutions that us farmers use, theres been alot of development into this area, and im sure you will benefit from it!

Jordan