06:33 | se6astian | joined the channel | |
06:49 | ApertusWeb8 | joined the channel | |
06:53 | ApertusWeb8 | left the channel | |
08:33 | sasha-w | joined the channel | |
08:36 | sasha-w | left the channel | |
12:15 | sasha-w | joined the channel | |
12:19 | sasha-w | left the channel | |
13:06 | Bertl | morning everyone!
| |
13:30 | se6astian | hello!
| |
13:31 | se6astian | I played around a little with the spherical joint idea
| |
13:32 | se6astian | and came up with an assembly that does not rely purely on material friction
| |
13:32 | se6astian | here is an animation
| |
13:32 | se6astian | http://www.youtube.com/watch?v=f3Fw4lJ-Jl4
| |
14:02 | se6astian | OK gotta go, let me know what you think of this
| |
14:03 | se6astian | left the channel | |
16:29 | aombk | joined the channel | |
16:58 | se6astian | joined the channel | |
16:59 | se6astian | Hello, I am back
| |
17:34 | Bertl | se6astian: looks really interesting
| |
17:35 | Bertl | what I still do not understand, in which range will those adjustments be?
| |
17:35 | se6astian | hi, great
| |
17:35 | se6astian | the animation shows 3° in 4 directions
| |
17:35 | se6astian | what range we really need... no idea
| |
17:36 | Bertl | the thing is, yesterday we talked about 0.01mm precision
| |
17:36 | Bertl | so that would be 5um
| |
17:36 | Bertl | now, if you assume an M2 screw
| |
17:38 | Bertl | the pitch is 0.4mm and 0.25mm for the fine spacing
| |
17:38 | Bertl | i.e. a turn of the screw will change the center by at least 100um
| |
17:40 | Bertl | and the other interesting question would be, how much should the total/maximum adjustment be?
| |
17:40 | Bertl | because if we want to adjust within let's say 150um total, there are simpler designs
| |
17:41 | se6astian | maybe we can talk to other camera makers at IBC to investigate these ranges
| |
17:41 | se6astian | All I know is that Elphel started with a non calibrateable/non tiltable sensor frontend
| |
17:41 | se6astian | then did a simple tilt correction one with 3 setscrews
| |
17:42 | Bertl | okay, and final question: how do you access the screws?
| |
17:42 | se6astian | then a full featured one with rather complex mechanics with springs and moving parts
| |
17:42 | Bertl | I mean, where are the screws and where is the sensor?
| |
17:42 | se6astian | and then after that one another redesign to compensate for material expansion due to heat
| |
17:42 | se6astian | so they definitely underestimated the problem in the beginning, something I want to avoid
| |
17:43 | Bertl | interesting
| |
17:44 | Bertl | btw, you can easily make that adjustable from outside, with a few simple tricks
| |
17:44 | se6astian | the screws would be access from inside the camera (holes in sensor frontend pcb?) by the engineer calibrating the camera, should no be required to be recalibrated by the customer
| |
17:44 | se6astian | oh, what kind of tricks?
| |
17:45 | Bertl | you remember my .stl with the slits for fastening the joint?
| |
17:46 | Bertl | make them six (if you like) and let them go to the base of the mount instead i.e. extend the thin sperical part to match the 4/6 corners of the base
| |
17:47 | Bertl | i.e. basically you screw system but screwed on from the outside into the base
| |
17:47 | se6astian | http://blog.elphel.com/2010/06/elphel-eyesis-camera-optics-and-lens-focus-adjustment/
| |
17:48 | se6astian | hmm, difficult for me to imagine, could you do a quick drawing? ;)
| |
17:49 | se6astian | Elphel: "focus/tilt adjustment module is capable of compensating for ±0.15 mm"
| |
17:49 | se6astian | this was not enough and they had to redesign the parts completely
| |
17:49 | se6astian | they use small sensors and small lenses
| |
17:49 | se6astian | we use big sensors and big lenses
| |
17:50 | se6astian | http://blog.elphel.com/2011/02/eyesis-4pi/
| |
17:50 | se6astian | "Insufficient range of the sensor board tilt adjustment to the lens optical axis. It needed to be just about twice larger to be able to compensate for most sensor chip misalignments"
| |
17:51 | se6astian | another redesign lead to this: http://blog.elphel.com/wp-content/uploads/2012/10/393-m12.png
| |
17:52 | se6astian | Model NC393 CAD rendering with M12 (S-mount) lens and thermally compensated sensor front end
| |
18:02 | Bertl | http://vserver.13thfloor.at/Stuff/AXIOM/lens_mount_tilt2.stl
| |
18:02 | Bertl | haven't modeled the screws or done any animation, but I hope you get the idea
| |
18:04 | Bertl | basically 3 screws would be more than enough to perfectly define the position
| |
18:05 | Bertl | but there is no harm in having six, only for adjustment you should remove 3 of them first, then do the adjustments and after that, screw the other 3 back in
| |
18:07 | Bertl | but as I said, for 10um adjustments that won't do the trick, because you will not be able to get that right
| |
18:08 | Bertl | (because screw pitches are at lest 100um but likely more around 300-500um to have at least some stability
| |
18:10 | Bertl | if you need um precision, you need to use some kind of lever or gear construct
| |
18:15 | Bertl | hmm, you can probably get away with a dual pitch screw system for small adjustments :)
| |
18:16 | se6astian | ok looking at your stl now, I am not sure I understand it right yet... wouldnt the tilt of the front part make the screw holes move away from the holes on the base so the screws wouldnt fit anymore?
| |
18:17 | Bertl | yes, but that would be the same as in your design
| |
18:17 | Bertl | i.e. you need to have flexible counterparts or slits or moving nuts, etc
| |
18:19 | se6astian | in my design the moving part has no holes for the screws, they just push the flat surface in place
| |
18:21 | Bertl | okay, I obviously missed that
| |
18:24 | Bertl | but in that case, you can simply turn it around and have the screws on the outside
| |
18:24 | Bertl | doesn't really matter, does it?
| |
18:25 | se6astian | true, yes
| |
18:26 | Bertl | unfortunately the dual pitch trick won't work with this
| |
18:28 | se6astian | dual pitch?
| |
18:29 | Bertl | well, you can make a cylinder with let's say a 0.5mm pitch on the outside and a 0.4mm pitch on the inside and you get a virtual screw with 0.1mm pitch
| |
18:30 | Bertl | this is used for microadjustments, but obviously requires a fixed counterpart
| |
18:32 | se6astian | ah I see
| |
18:34 | Bertl | but I'm still not sure I completely understand the problem
| |
18:35 | se6astian | me neither :)
| |
18:35 | Bertl | do the sensors have different glass thickness across the sensor?
| |
18:35 | se6astian | cause in the concept so far we have shift as well as tilt when moving the mount which is not good
| |
18:36 | Bertl | with shift you mean distance changes?
| |
18:36 | se6astian | I think the problem is that the sensor socket will not be 100% plain and the pins not pushed in 100% the same distance
| |
18:36 | Bertl | because the spherical design will definitely keep the distance constant
| |
18:37 | se6astian | and then again the image sensor pushed into the socket will not be 100% pushed in on all sides
| |
18:37 | se6astian | and the PCB itself might not be 100% plain either
| |
18:37 | Bertl | that doesn't matter
| |
18:37 | Bertl | assuming that we keep the/a socket, the pcb and socket is of no relevance
| |
18:38 | se6astian | I dont mean the FFD, I mean the center of the lens being pushed out of the center of the sensor when we tilt it
| |
18:38 | Bertl | the sensor should be tightly screwed to the lens mount
| |
18:39 | Bertl | that should give a fit within 10um or less
| |
18:39 | Bertl | and the sensor socket/pcb doesn't change that in any way
| |
18:40 | se6astian | if we maintain this design with the sensor being attached to the lens mount that problem should be radically minimized yes
| |
18:40 | se6astian | you had some concerns about that design though if I remember correctly
| |
18:41 | se6astian | also it would mean we need to redesign the lens mounf if we change the sensor
| |
18:41 | se6astian | or we make the lens mount a 2 part assembly
| |
18:41 | Bertl | yes, but they are mainly about pressing the sensor glass agains the metal mount
| |
18:42 | Bertl | i.e. I'd introduce a plastic foil or something similar between sensor glass and lens mount
| |
18:42 | Bertl | to absorb shocks and protect the glass
| |
18:43 | Bertl | (or ceramic in our case)
| |
18:43 | se6astian | ok
| |
18:43 | Bertl | assuming that we do the stacked design
| |
18:44 | Bertl | I'd screw the sensor pcb from the back to the lensmount (with spacers)
| |
18:44 | Bertl | after the sensor was screwed into the lensmount (with foil)
| |
18:44 | Bertl | so basically foil goes into the lens mount, sensor ontop of foil
| |
18:45 | Bertl | screws fix the sensor, pcb is placed onto the sensor pins (with spacers on the sides)
| |
18:45 | Bertl | and finally pcb is screwed against the lens mount
| |
18:45 | Bertl | where I'd make the lens mount cover the entire front of the camera
| |
18:46 | se6astian | sounds good, we just need to take foil thickness into account when we create the lens mount, but that should do it
| |
18:46 | Bertl | so the lens mount is sensor specific, yes, but it conencts to the rest of the camer block electrically and mechanically
| |
18:46 | Bertl | *camera*
| |
18:47 | se6astian | I d say we see with the prototype how plain the sensor alignment will be
| |
18:47 | se6astian | and then we decide if we need to take further steps
| |
18:48 | Bertl | probably
| |
18:50 | Bertl | let me ask, in what way would a constant shift of, let's say 0.2mm affect the lens mount?
| |
18:51 | Bertl | (regarding the comment about the foil thickness)
| |
18:53 | se6astian | the foil moves the sensor 0.2mm further away from the lens right?
| |
18:56 | Bertl | yes, for example
| |
18:57 | se6astian | so if the lens mount is designed for FFD of an nikon f mount (46.50 mm) it will actually be 46.52mm
| |
18:58 | Bertl | okay, and what will be the effect?
| |
18:58 | se6astian | your lens will be slightly out of focus
| |
18:58 | Bertl | you won't be able to precisely focus on infinity, I guess
| |
18:59 | Bertl | but everything else should work just fine, only with different focus settings, no?
| |
18:59 | se6astian | for slight deviation it could mean that when you turn the focus ring to 1m the focus will actually be at 1.5m, in worst cases it would mean as you say that you either cant focus on infinite, cant focus on near objects or cant get anything in focus at all
| |
19:00 | se6astian | everything else will work fine
| |
19:01 | se6astian | a fun thing with the Elphel cameras ( I might show you soon) is that the c mount thread can be used to change the FFD of the lens by just unscrewing the lens
| |
19:01 | Bertl | and how much effect has a single turn?
| |
19:01 | Bertl | (which will probably be about 0.5mm)
| |
19:01 | se6astian | this means you can basically shift the focus plane to the glass of the lens or even to the innards of the lens glass
| |
19:02 | se6astian | you then see tiny hairs and dust that are stuck inside the lens :)
| |
19:02 | se6astian | we can try it then how much one turn does
| |
19:05 | Bertl | okay
| |
19:08 | se6astian | I can take it with us to Brussels
| |
19:08 | se6astian | or we do it when I visit to bake the PCB
| |
19:16 | Bertl | both is fine for me
| |
19:56 | aombk_ | joined the channel | |
19:57 | aombk | left the channel | |
20:26 | aombk | joined the channel | |
20:28 | aombk_ | left the channel | |
20:40 | se6astian | ok time for bed for me, gnight!
| |
20:40 | se6astian | left the channel |