It was fun and educational. :)
I'm sure we'll get it sorted - it's pretty much down to a mechanical process right now.
g.
Printable View
It was fun and educational. :)
I'm sure we'll get it sorted - it's pretty much down to a mechanical process right now.
g.
@Mike
Nice diagrams! Not being an optics guy, I'm puzzling over your comment regards the single viewer position and image distortion as you turn your head.
My understanding of a collimated display is that the image viewed is irrespective of position and the sight line from the "lookers eye" to the frame edge of the cockpit window describes the boundaries of the FOV. The image will not "change" if I move from the left seat to the right or get up and walk to the cabin door, just that portion of the image that I see.
If that is true then the purpose of your directions is only to establish a point of reference from which to start the design for mirror/screen geometry and relative placement. Is that correct? And distortion is a property of the projection system, lenses, warping software, geometry tolerances, and other artifacts when moving from the theoretical world to reality.
Perhaps should have asked via PM, but hopefully others will gain from this discussion.
JW
Collimation with a spherical mirror is good, but not perfect.
If the viewpoint is directly under the mirror center of curvature, looking left or right makes no difference. The vertical cross section through the viewer, mirror and screen is the same regardless of lateral viewing angle.
However, if the viewpoint has been shifted forward, perhaps to allow greater vertical field of view (or sideways to fit in a copilot), the vertical cross section including the viewer, mirror and screen is skewed when the viewer looks to the side. The optical paths are longer and the angles different. So, collimation is probably less accurate, and the image is more distorted.
Optimizing a wide-angle collimated display is a tradeoff between many factors.
@wledzian
Wasn't quite sure of your question regards numbers in the formula. Could you provide a bit of clarification?
JW
The 45 is obviously 45 degrees. Is that vertical or horizontal?Quote:
Originally Posted by castle
I'm assuming that 72 is inches, equal to 6 feet.
What do the other values represent?
The 45 is the vertical and 72 is the radius of curvature ( 6 feet ).Quote:
Originally Posted by wledzian
57.2957 is a radian. A constant used to convert degrees of arc to a linear value. Circumference of a circle is pi*diameter.
or 360 degrees = 3.1415 * 2 * radius
rearranging terms 360 / ( 6.283048) = radius or
the distance along the circumference of a circle equal to the radius is 57.29703 degrees of arc.
By the same token 180 degrees horizontal for a 6 foot radius requires 18.849 linear feet of mylar material (180/57.2957)*6
As you noted, this would be slightly off for something of a flattened arc and does not include the margin required to fastened the mylar to the surface.
JW
The material when flat takes the shape of a truncated cone which has been unrolled. There is significant curvature to the top and bottom edges. The impact is that the width of the required flat rectangular stock is much wider than the vertical chord of the mirror surface.
One very quickly runs up against limits of available material width.
@Mike.Powell, (cc)geneb:
One challenge Gene and I ran into was the vertical pre-stretch required simply to fit an unrolled cone to the 60° frame. We also tried using the vertical arclength instead of the chord; this fit the frame a bit easier, but would still require more horizontal stretching near the mid-lattitudes than is available by cutting across the horizontal chord. Something considered but not yet tried due to lack of time on Saturday was introducing an arc to the vertical edges to take up some of this gap. Even if we can eliminate some of the prestretch by this method, I'm still concerned that the uneven strain in the horizontal direction combined with near zero strain in the vertical direction will prevent the formation of a spherical surface.
In your research, have you come across any mention of ways to mitigate these issues during mirror fitting?
There are a few methods mentioned.
The horizontal arc of the framework can be extended with the vertical side made straight so no vertical stretch is required during installation. The extended mirror area is useless optically and is masked off.
The vertical framework sides can be made flexible so that the film can be installed with no stretch. The sides are cranked into a curve after film installation.
Temporary, straight frame extensions can be fastened to the frame for film installation. Once the film is vacuum stretched into shape, the vertical sides of the film are clamped to the curved frame sides, the film is trimmed, and the extensions are removed.
The frame sides can be extended with circular arcs which are on a plane which blends with the unstretched conic section of the film as shown below.
http://www.mycockpit.org/photopost/d...0692image1.JPG
This is from US patent 6050692. I think this has the best chance of success of any hobby approach. It appears to not require stretching during installation, only that the film be pulled taut.
The film tension is not uniformly distributed. I don't think you want it to be. The center horizontal band must stretch the most as it is at the deepest portion of the curve. The top and bottom edges have been cut to the correct length to be on the spherical surface without stretching, so you don't want any horizontal tension there. Vertical tension should be uniform across the whole surface, but of course it won't be at the sides, but the circular arc side additions provide a smooth transition from frame side support to active mirror surface.
I think this produces a satisfactory optical surface as long as the film stays in its elastic region. Once the film stretches beyond its elastic limit, it becomes necessary to prestretch horizontally at the top and bottom. I think this was a major issue in developing film mirrors for displays with vertical fields of view greater than 45 degrees. This probably means that DIY mirrors are limited to 40~45 degree vertical arcs if they are to be one contiguous piece. This may not be an issue due to the limitations on the width of mirror film stock.
Hey guys, umm, sorry to butt in here, ive read parts and im extremely confused, i dont understand this stuff, but, i saw on the Simquip website, that 2 collimating mirrors are for sale, i thought that this may be an appropriate place to mention it, if it hasnt been already ;) Heres the link anyway
Simquip Specials Page
Again, im sorry if i missed this before in the post!
Cheers, Jordan
Thanks, Jordan.
Both timely and appropriate, IMHO.
Weledzian,
Spent the last few days considering your diagrams and comments. I have also been reading the patents, and also Mike's additions.
While considering these many elements, the first and possibly most important is if I have the space required to do this. The thing I have done so far that gives me some hope is cut 6 curved pieces of plywood that is meant to fit just below the lowest point of my cockpit windows. This is intended to be the lower edge base for the mirror. My goal is to have this curve clear the entire structure to allow the rear windows to have full view. It turns out this curve is 5 feet 6 inches radius.
The arc radius will be greater than this 5ft 6, presumably 7 feet and as much vertical to allow for max 55 inch Mylar width limit. I'm thinking of 30 degrees. I'm considering as much adjustment as I can build in the better. I will use some tracking system that will allow adjustment of the mirror forward - aft and up and down, perhaps by several inches either way. I plan to do the same with the projection screen.
Beside the issues we are trying to better understand, e.g. clamping Mylar, I will need to be sure that I can walk through the narrow gap between the wall and the mirror and that the mirror can be set so as not to reflect the fuselage and but not set at such an angle to restrict passage.
Additionally, I was considering Mylar material. Do you know if there is much variance in quality? I was considering reflectivity and durability( scratch resistant).
Again, I want you to know how much I appreciate your help.
Mike
mikesblack,
No problem - this is fun!
I need a few more measurements:
I need to define the vertical axis of the screen and mirror. Where is your eyepoint with respect to the center of the 5'6" curved base?
I need to define the lower eyeline. Measured in a plane defined by the eyepoint and the mirror axis, what is the angle between horizontal and the edge of the lower edge base?
I understand that the quality of mirror Mylar does vary. However, I've had very good luck with 2 mil Mylar from Tap Plastics, and I've just received a roll of 1 mil Mylar from Nielsen Enterprises which also looks good. It's coated on both sides and is nicely reflective. I haven't unrolled it, but what I see so far looks to be high quality. Nielsen sells in 56 inch widths.
Mylar as a base material is strong. The coating is rather fragile. If you crease the Mylar, which happens very easily, you get a permanent mark. The surface also scratches very easily. Your best bet cleaning it is with a soft cotton cloth. A paper towel will scratch.
just found this document about plexiglas forming.
http://www.plexiglas.com/literature/pdf/135.pdf
and this one
http://www.iss.infn.it/cusanno/publi...io/mirrors.pdf
Gery
In addition, do not use cleaning products. If you need to wet the cloth use distilled water.Quote:
Originally Posted by Mike.Powell
Matt Olieman
@mikesblack
Some one can check my math, here are a few numbers...
For a base radius of 5.5 feet and a vertical arc of 30 degrees down from the equator requires a radius at the equator of
5.5 / cos(30) = 6.35 feet
increase the vertical arc to 45 degrees and the radius at the equator is
5.5 / cos(45) = 7.78 feet for the same base radius
Working backwards, for a radius at the equator of 7.0 feet if you go down ~38 degrees the radius at this point is 5.5 feet.
Hope this helps you with your sizing.
JW
@castle
your math is right, but the geometry is wrong. Your numbers assume the eyepoint is located at the center of the mirror sphere. In the case of mikesblack's cockpit, the eyepoint will be offset both horizontally and vertically, thus my need for this information before I can give an answer.
Got it! But for a side-by-side seating you would still have to account for the right seat and make the compromise that both seats would be a tad off and just use the center of the TQ for the horizontal point?? Then vertically, the distance from some "average" eye point to the top of the mirror assuming that is the equator.
Is my understanding correct?
As you said, this is really neat stuff. Just in passing, looks like a short throw lens for the projection system can be had for around $600 in the el cheapo models. Don't have the numbers as to depth of field, focal length etc. For the really good stuff the price is a bit steeper; from $1700 to $2500 per lens. in either case you still have to deal wth keystoning for off-axis and image merging at the boundaries, front and rear projection systems.
I'm trying to get some estimates for bending acrylic for those of us not as skilled as geneb with tools. Will keep you all informed...
JW
For the most part, yes. There is certainly accounting for the right seat. The center of the TQ does not have to correspond to the center of the mirror, and if space allows, should not correspond to the center of the mirror. Ideally, the eyepoints should be somewhat forward of the axis. This results in reduced distortion for both pilots, but leads to a larger mirror. For a cross-cockpit display on a 767 cockpit, the eyepoint is at least 21 inches off-center horizontally, assuming the mirror axis is directly between the pilots.Quote:
Originally Posted by castle
Vertically, the FOV is restricted by the optics. If you take a look at the raytrace plots I posted way back on pages 6 and 7, you can see that the lower eyeline in essence defines the upper eyeline by virtue of locating the lower screen edge, and that the upper eyeline does not necessarily correspond to the equator.
In Mikesblack's case, the design eyepoint is fixed by the cockpit. The horizontal offset is fixed by the 5'6" arc he's planning as his lower edge, and the lower eyeline is fixed by the angle to this arc. The mirror radius is a variable, limited by the 7' distance to the side wall, with the vertical offset defined by the selected mirror radius.
Once I have the limiting geometry, I can play with mirror radius to get the maximum vertical field of view. If the geometry allows, I may also be able to assume a larger base radius in order to move the axis rearward.
There are short-throw projectors available in the $700 range already that would likely work well for the rear-projected screen. No need to spend $$$ on aftermarket lenses. If you're talking wide-angle (fisheye) lenses, your wallet may be better off projecting onto a security mirror, at least as a first attempt. Keystone correction is best done entirely in software, as correction will be necessary and it's better to do it all at once, rather than run the image through two extra processing steps.Quote:
As you said, this is really neat stuff. Just in passing, looks like a short throw lens for the projection system can be had for around $600 in the el cheapo models. Don't have the numbers as to depth of field, focal length etc. For the really good stuff the price is a bit steeper; from $1700 to $2500 per lens. in either case you still have to deal wth keystoning for off-axis and image merging at the boundaries, front and rear projection systems.
Please do. The topic of forming the rear-projection screen came up on Saturday, and there was discussion of building a forming oven just for that purpose.Quote:
I'm trying to get some estimates for bending acrylic for those of us not as skilled as geneb with tools. Will keep you all informed...
JW
@geneb -
If it does come to that, I'd be more than willing to share the cost with you. It's looking like a screen for a 48" mirror could be formed from a single 4x8 sheet of plexi.
[QUOTE=wledzian;111483
@geneb -
If it does come to that, I'd be more than willing to share the cost with you. It's looking like a screen for a 48" mirror could be formed from a single 4x8 sheet of plexi.[/QUOTE]
Building an oven to evenly heat a 4x8 sheet of material isn't an insurmountable challenge, but could certainly be an expensive one. I think we'd be better off building a three segment screen and have the projector seams align along the screen seams. I'm game either way, but if we're going to build an oven & vacu-forming table to that scale, we'd better come up with a use for it after we pull a pair of 24" radius screens off of it. *laughs*
(Now if I was an Evil Scientist type[*], I'd be tempted to see how well you can vacu-form people...Muahahaha!)
[*]Oh who the **** am I kidding? Who wants to volunteer? *Evil Laugh*
g.
(mumbles unpleasant things about nanny-ware)
@wledzian
Ahhh! the fog is slowy lifting...
Since discussions seem to indicate that the projection screen is not a uniform sphere might it be easier to use a front projection system. Redifun builds a front projection system and that eliminates the problem of projector seams.
Makes the setup a bit more complex but doable.
JW
@geneb I think there still is a person at Thun Field that vacforms plexi airplane canopies. He has an oven big enough to walk in and the vacuum systems too. He is next to the prop shop and behind Atkins Rotary.
Andy Smith
I just got a reply from a professional visual system manufacurer. They told me that "low-cost" systems start at around half a million u.s. $'s. So sure is no option for a hobbyist ;) But at least now we know.
Gery
That's good to know Andy. When we get to that point I'll have to find out how much he'd rent time on it. I certainly wouldn't need him to build the form. :D
Gery, "low cost" and "collimated display" is like "military intellegence". It's an oxymoron of the most extreme kind. :D
Having said that, keep in mind that with a hobby system, perfection is the enemy of good enough.
g.
Hi wledzian,
The 5 foot 6 inch curved base may be too short in order to allow for adjustment. I may need to expand that by a few inches. I am using aluminum tracks to allow for front to back and up and down travel.
Eyes are 46 inches from the floor. More pictures and measurements to come.
http://www.mycockpit.org/photopost/data/690/031_2_.jpg
Mike
I think I'm going to move my bed to another location. If I have a chance of doing this correctly, I will most likely need to build the mirror without the restriction of needing regular passage.
Check out my latest photos.. The captains front window has a green mark and a string cross between the vertical and horizontal point of view. From that point to the wall measures 48". The height of my eyes is 45-46".
The upper edge of the side window is 53"and 31" for the lower edge. The angle of these windows is measured at 65deg.
I am not sure what I need to do to make sure the lower edge of the image from the mirror clears the side windows and fuselage while looking from inside the cockpit. Any ideas that you think I might consider?
Thanks again.
Hmm, Mike's question on clearances got me thinking.
Have DoubleCAD installed, but not ready to delve in, so got out my trusty old drafting set.
Starting with a 6 foot radius set the eye point down 3 feet and in 6 inches, think I got a workable solution for a FOV of +25 and -20 except
1) the top edge of the screen is 20 degrees below the equator
2) the lower edge is 60 degrees down
3) which provides 40 degrees of mirror to view or a width of ~ 50 inches for the mylar
4) the approximate radius of the projection screen is 4 feet
but the radius of the lower edge of the mirrot is 6*cos(60) = 3 feet which is inside the cabin :-(
Going to have to go back and reread Mike's tutorial in msg #78
@mikesblack
you might consider a projection screen that overhangs the top of the cabin and build a mirror accordindly
@mikesblack
Is your cockpit a to-scale representation of a 767? It looks to me like 5'6" doesn't fit.
I've got the sketchup model, I can trace the eyelines to locate the bottom edge of the mirror. So far, the best I've been able to match castle's geometry puts the eyepoint about 5" off-center from the mirror axis; the 767 eyepiont is at least 21" offset.
I'd like to give a more encouraging response, but I need to rebuild the geometry and play with it a bit, I won't be able to give it any real attention until Saturday evening, I'll let you know what I come up with.
I believe my cockpit is close to scale, but I'm sure there is quite some error considering that angles used have been created using multiple photos and not from plans. So while I believe there is error, and also a degree of asymmetry, I wouldn't think there was a significant difference. I'll elaborate a little and ask some questions to follow, so to make sure we both understand and are discussing the same things.
Can you explain what you mean by the "eyepoint offset" from the mirror axis? My understanding as of now is that the mirror axis is the center of the mirror sphere, situated in the center of the cockpit, located adjacent to the pilots' eyes, and set at 90 degrees to the cockpit floor. With this assumption, the Captains eyes are 1' 6" from this point.
The 5 foot 6 inch circle that I cut was based on what could fit just below the lowest edge of the cockpit windows, so that it would miss intersecting the rear windows. Other than that, it is an arbitrary figure. As it is now, this ring just fits as intended, but I believe in order to get more adjustability, up/ down or back and forth, I will extend this radius.
If there is any measurement, you need that I have omitted, please let me know and I'll let you know.
Thank you very much again for your expertise and your time helping me solve this. Please do take your time and convenience. Again, I must say I am greatly appreciative.
Thank you too Castle for your help and input.
By the way, moved my bed, so have tons more room forward of the cockpit, but still have to deal with those side walls. My wife has been great allowing me to do this, but I doubt she would be so if I suggested, tearing down these walls.
:lol:
767 Airplane Characteristics for Airport Planning
See section 4, page 109. In my early model, I used this information to derive an eyepoint. I found a to-scale model of a 767 in the Sketchup library and took a section cut through a vertical plane offset 21 inches from centerline per the Boeing document. From this section cut, I extended the upper and lower eyelines through the window edge intersection with the section cut and took their intersection as the eyepoint.
The mirror axis for a cross-cockpit display should be set at 90° to the floor at some point along the aircraft centerline. In a full-scale cockpit with the mirror axis directly between the pilots, the minimum offset would thus be 21 inches. If the axis is located behind the pilots, the offset would be larger.
So far, my optical calculations have been based on a symmetrical field of view, centered around the mirror axis. My cockpit will be single-seat, and the only horizontal offset I'm allowing is to account for my eyes being forward of my neck. In that case, the optics are consistent throughout the horizontal field of view. In a two-seat cockpit, these optics calculations only hold true along a plane defined by the eyepoint and the mirror axis, and in your case would correspond to looking 90° left. I haven't fully worked out the raytrace algorithm for off-plane viewing to see if the focal surface still lines up. The math is not undoable, I just haven't taken the time to do it yet.
As things are going, I think we may not be too far from having a functional prototype mirror, which should at the very least give us a good idea of the feasability of this project. If you can wait a few more weeks before cutting parts, I think we'll have a lot more to report.
Oh, and your wife must be very understanding to go along with having your simulator take up so much of the bedroom.
Thanks for pointing out this document. I understand the meaning of offset. I did re measure this point and rather than 18 inches as I had written, I think I am closer to the 21 inches as per the real aircraft.
If you could explain, I am not totally sure I understand this?
"My cockpit will be single-seat, and the only horizontal offset I'm allowing is to account for my eyes being forward of my neck. In that case, the optics are consistent throughout the horizontal field of view. In a two-seat cockpit, these optics calculations only hold true along a plane defined by the eyepoint and the mirror axis, and in your case would correspond to looking 90° left."
Me neither.Quote:
I am not totally sure I understand this?
The raytracings I've done to determine the screen shape are for rays in-plane with the mirror axis. A ray in this plane stays in this plane from the screen to the eyepoint, so my calculations are relatively simple. I cast a ray to the mirror, determine its intersection point and angle, and calculate the reflected ray. Doing this with several parallel rays produces an intersection point which defines a point on the screen. It doesn't matter that the eyepoint is offset from the axis, as the rays are still in-plane. Since I'm not worried about a second crewmember, this works well for me. A small offset to allow for my eyes being forward of my neck (the rotational axis of my head, assumed to be coincident with the center of the mirror) has no effect on the raytrace calculations.
For more significant offset, it cannot be assumed that the cast rays will be in plane with the mirror axis. I now have to cast the rays in a plane defined by the mirror center (not the axis) and the ray itself. This requires transforming the coordinate system to the new plane, casting the rays, determining the intersect, re-transforming the screen point to the original coordinate system, and finding a clean way to visualize the new screen points in relation to the in-plane surface.
Visualize if you will, the mirror around your cockpit. If the center of the mirror is directly between captain and copilot and you look directly to your left (as captain), you'd expect to be looking at a reflection of some point on the screen directly above your line of sight. On the other hand, if you look directly forward, you'd expect to see what is directly in front of the aircraft. This image, if properly located on the screen, will be directly over the cockpit centerline. Your line of sight is directly forward to the mirror, but the reflected ray is up and to the right, to intersect with the centerline of the screen.
The problem is that the collimation is not perfect. The directly-forward eyelines for the captain meet the screen slightly to the left of forward, and for the copilot, slightly to the right of forward. The offset is significantly less than if you were looking at a real image on a surface a few feet away, but it is real. The effect is lessened by the use of larger mirrors.
I haven't done the math yet, so beyond the concept I don't have an idea about the magnitude of the distortion or whether it will be acceptable or absolutely horrible. Once we have a working prototype, we'll be able to tell you more.
Thanks. Have to say that I'm very impressed by your level of understanding and detailed knowledge of this subject.
There is definitely quite a bit to it, and for me, without the good fortune of this site and more specifically folks like you who have been most generous with sharing knowledge, I would be lost and without hope of even being close to getting this done.
I am looking forward to the process, and even if not successful, I still think learning this has been worth the effort.
Cheers!
Your wife has to be one fine lady!
Attachment 4405
Bit fortunate here, have an extra stall in the garage, but it is a bit tight. Starting to build the shell and already seeing that it is not going to work if and when the mirror system is installed.
Attachment 4406
I jokingly suggested pehaps the wife would like to park her car outside and along with an icy stare she suggested maybe I wouldn't mind sleeping in my truck. ;-)
All kidding aside, gotta love them, amazing how they tolerate the kid in us and all our crazy hobbies
She says she loves me for that quality, of course that's until I start really start acting like a kid or perhaps I should say childish. Then she gets annoyed.
While I've gotten some prelim numbers that suggest you might be able to fabricate three acrylic forms for around $2K (does not include first surface mirroring) and wanted to use those on a vacuum system instead you would have to drill a gazillion tiny holes to suck up and stretch the mylar. So here is a crazy idea..
Going back to the idea of building a more conventional frame; use a multilayer approach for the concave surface. Base layer is strips of conventional peg board, cover that with a coarse cloth, apply a top layer of felt. Use two sided carpeting type in horizontal and vertical strips to hold the sandwich together and bonded to the board surface.
JW
Don't know if this vid has been shown before it's one year old. But I just stumbled upon it. You can see the mounting of the collimated visual from day 9 to day 11. Enjoy :)
http://www.youtube.com/watch?v=Wqq4-...eature=related
Gery
crashdog,
That was great. Thanks for bringing this up. I have been looking for something like this.
Fantastic,
Mike
Gery, FRIGGEN FANTASTIC!!!!!!
I haven't seen this one, good find and THANKS!!!!! :) :) :) :)
Matt Olieman