vlaiv

Interestingly enough - flats would help as well. Although it is small sensor - there appears to be some level of vignetting as well.

Here are a few "teaser" images of new contraption. This is more proof of concept and to get the feel of how smooth 3d printed gears will be: I've yet to add bottom part of housing that will hold bearings for shafts and of course couple of knobs to be able to operate it. Then it is 3d printing time

That might be an option - but I'd still need gear set - only not as reduction, but rather overdrive. I don't like the idea of having three things stick out with draw tube . I barely accepted these two. But in all seriousness - would threaded rod need lubrication? I like the idea that these linear rods don't need it if I'm using PTFE bushings. There is option for threaded rod to be stationary inside casing, and nut to be attached to draw tube - but then we get limited travel - draw tube can only move the same distance nut is able to move inside casing. With above approach - we can have more travel. I did look up usual travel of refractor focusers, and most are limited to say 90mm (that is actually on longer side of things - most are 60-70mm), so yes, that might be an option. Now that I'm thinking about it - I sort of like the idea. Might design one of those next. There is a way to extend draw tube travel by "hiding" part of the mechanism inside telescope body - same way draw tube is inserted in telescope body deeper than the focuser itself when focuser is fully racked in. No reason not to have threaded rod extend on that side.

I think that @Victor Boesen has 102mm F/7 (tecnosky version) on AZ6T and can offer some insight in how it holds up?

In that case - it looks like it is FPL-55. Gear80 looks identical to this TS offering with FPL-55 glass: https://www.teleskop-express.de/shop/product_info.php/info/p5964_TS-Optics-CF-APO-80-mm-f-6-FPL55-Triplet-APO-Refractor-with-Certificate.html (except for blue color touch).

Well, maybe motorized focuser would hold it, but I'm doing manual version and I don't think that rack and pinion will be able to hold it in place, at least not "loaded" with 2" diagonal and an eyepiece. I might be wrong though. My next step is to design and add rack and pinion. I already printed few "test samples": First one is 9 teeth and it did not come out quite well - there was some strange warping in the corner of the rack and teeth on pinion are not that great. Other two are much better: middle one is cycloidial profile, 11 teeth for pinion (1mm module) and 24 degrees angle on herringbone pattern. Top one is involute profile, 13 teeth for pinion (again 1mm module) and 30 degrees angle. In any case, there will be additional reduction stage of about 3:1 - so that might help. For the time being, I'll be using 608 bearings for the test and 5mm threaded rod as shaft for these (I'll design and print threads so all bits can be threaded onto the rod and held in position with m5 nuts). Btw video is encoded as mp4 video, and I just drag&drop it into post like I do with images.

Just a small update. Linear rods work exceptionally well. In fact, as predicted - they work too well Motion is very smooth, silent, there is zero play in any direction and setup is rather stiff: first_video.mp4 Problem is that motion is too smooth - there is no enough friction to hold assembly in position under its own weight - let alone when adding diagonal and eyepiece: second_video.mp4 Does anyone have idea what would be the best way to implement friction brake of sorts to stop this?

Here is what I had in mind for automatic stretching. I'll work with example as it might be easier to understand (I might lack proper terminology to explain things precisely). Let's say that we have 16 bit data to start with. We need to determine three numbers in order to derive histogram transform: 1. Background level. Say that we identify background level to be around 900ADU 2. Standard deviation of pixel values around that background level (we identify pixels that we think are background and we compute standard deviation for sample) - say it is 20ADU 3. Maximum brightness value that won't saturate in the image - let's say we have nebulosity peak value at say 17000ADU. Stars can saturate - we don't care about those as they don't contain too much information, but nebula core does. Now we have couple of equations to solve. First is our background level - which we can set to say 8% of 0-1 range. So we know that 900ADU maps to 0.08 Next, we want background noise to be virtually invisible. We ensure this by saying that +/- 3 standard deviations around that background value (3 here can be parameter) will fall into range that we can barely detect. We can barely detect say 0.5% difference in output value - or 0.005 out of 0-1 range. 900 + 60 = 0.08 + 3 * 0.005 900 - 60 = 0.08 - 3 * 0.005 We have our two points for linear part. Now we have to connect that to a power law. First we define general shape as base^(x*a + b) Where X is our ADU value. We now need several equations to form our system of equations to solve. First condition is that for 17000 we want to have output value of say 0.96 (again - this value can be tweaked - but that is brightest part of target). so 0.96 = base^(17000*a + b) Second equation is that we want to "connect" above linear function to our power function at some point - let's say that is after those 3 standard deviations. 0.095 = base^(960*a + b) Third requirement is that we have gradient of both linear function and our power law be same at that point so we have "smooth" joint. Gradient is dy / dx so 6 * 0.005 / 6 * 20 = 0.005 / 20 = 0.00025 Now we find first derivative of our power law that will be base^(x*a+b) * ln(base) * a And that must be equal to 0.00025 for X = 960 so we plug in numbers for third equation: base^(960*a+b) * ln(base) * a = 0.00025 Now, if we can solve this set of equations for base, a and b - we have our histogram transform that will satisfy our initial conditions.

I'm not entirely sure that I'd buy into that FPL53 claim 115 with FPL51 already has excellent correction and I don't really see the point of going with FPL53 with that aperture at F/7

I design threads in FreeCad by doing following: Create sketch that is in plane perpendicular to threads. Sketch should contain metric profile - it just depends if you want to do internal or external thread and if you want to apply additive or subtractive operation. Both can be designed with both - but it will impact your profile. Say you want to do internal thread with subtractive method - then you will use nut thread profile - which means tooth top will be 1/8th of a pitch. You will add some tolerance to both major and minor diameter (I use +0.1mm per radius, or 0.2mm per diameter). Once you've created profile (which is trapezoid with 60 degree sides - have a look here: https://en.wikipedia.org/wiki/ISO_metric_screw_thread, I don't bother with rounded bits as tolerance takes care of that) - you use subtractive helix on geometry that you already prepared (which is just some cylindrical shape). There are a few youtube tutorials on how to do threads - both in freecad and fusion 360 - maybe have a look at those?

I wonder why is it so expensive, given that it is FPL-51 triplet that usually goes about £1000 less: https://www.altairastro.com/ts-optics-photoline-115-f7-ed-triplet-refractor-10171-p.asp

If I get that right - it is very similar to the non rotating helical design that I printed: Draw tube: Focuser housing: (thread that is visible at the bottom serves to screw in lens cell - it is not the part of focuser) Focuser knob (you'll have either belt or large ring instead of this): And finally - top part of the housing:

I've printed aperture mask for Samyang lens (although 85mm version) that uses filter thread. If you can print fine pitch (like 0.75) - then I see why not. I simply use 0.12mm layer height and design a bit more tolerance into the part and it works. Sometimes it needs to be screwed in all the way few times to "clean up" the thread, but after that it's rather smooth.

Just had an idea: How about 2GT timing belt instead of rack and pinion? It would certainly need reduction with 16T pulley as that gives 32mm per revolution, but then again, same can be done - another 16T paired with say 32T or 40T pulley?

I just got a meter long linear bearing rod that I cut into pieces (18cm x 2 for now). SKF PTFE bushings also arrived. Both are 6mm and fit perfectly. In fact - fit is too good. Linear rod glides like nothing is there I'll probably need to design some sort of tension breaks into system to allow focuser to hold some weight without slipping. First step will be to design prototype rig for sliding and 3d print it. Some interference fit testing first for both bushings and rods into 3d printed frame. I already printed some herringbone rack and pinion. Due to 3d printing precision - I can only print components that will move something like 40mm per turn (1mm module, with pinion having 11-13 teeth) - and that is too coarse. Will need to include 3:1 or 4:1 reduction stage into design. That will come in prototype stage 2 - sliding rig + rack and pinion part. When I get that right - rest is easy, I just need to add 55mm aluminum tubing (easy to source, very cheap) and print casing for the lot.

I've seen that, and I must say - its one of the nicest looking designs that I've came across. For me, probably most difficult thing to source would be 3x8x3 bearings. They seem perpetually out of stock locally. Anyways, I was thinking more in line with refractor focuser and rack and pinion or lead screw type.

@windjammer What are you hoping to achieve with any given function? Maybe we can define a set of requirements and then simply find function that behaves like that? Here is an example, that might be useful for automatic stretching. Background level and standard deviation is calculated. "First part" of our stretch function will position background level at some predefined (or maybe parameter) value - like %8 or 0.08 (if we observe 0-1 range) - function itself will be linear in 0 - 0.08+3*stddev range - having that slope. From that value on we "connect" it to power low with such properties that: 1. power function at 0.08 + three standard deviations matches in value to linear first part 2. gradient at 0.08 + three standard deviations matches gradient of linear first part 3. power part reaches 1 at saturation value (again some value measured from the image to be saturation value)

That is much more serious than I was prepared to start with At first, I was thinking I should go with 60 or 70mm, but then I decided to "go all in" and went for 80mm one

as far as I can tell - both lenses are marked for orientation and spacer is included. If not, it will be a little trial and error with artificial star. I've seen few cases where lens was not optimized with spacing and changing it improved spherical aberration substantially, so it's worth checking out anyway.

As long as you apply RGB ratio in the same way - it will preserve physical color.

Have no idea, but I'll post my findings as soon as I get the chance to inspect them. Given the holiday seasons, I expect them to arrive next year (at least 30 days shipping). That will give me plenty of time to design and print focuser and order aluminum tubing locally (or maybe I should wait and measure lens themselves? Not sure, never designed a lens cell before).

Just ordered 80/600 achromatic pair from aliexpress . I hope that if I plant it correctly and water it every day - it will grow into nice looking refractor by next year

It is actually basis for DDP - digital development process, one of first stretching functions developed to mimic analog film: http://www.astrosurf.com/buil/us/iris/iris5.htm where a would be coefficient * level and b = level

You'll have the same problem if you go OSC + Mono route. However, its easily correctable if you do proper color calibration, which you should for both mono+RGB filters and OSC. That way you'll get exactly the same color from both methods (or very very very similar with barely noticeable differences).

What do you mean by this?