Adding cooling and filter wheel to a debayered 450D mono DSLR

[Horwig]

A little bit of industrial espionage for you Gina. This is what's inside a Starlight Express filter wheel. (Hope they don't mind).

The carousel is friction driven by the smallest geared motor I've ever seen. Filter position is calculated by the series of three bolts between every filter position, and the three hall sensors on the board, ic2, ic3 and ic4.

Don't know if this give you any new ideas.

Huw

[Gina]

Thanks Huw   Those gears in the motor are really tiny    Yes, I have seen pics of that FW   The Atik EFW2 I have drives the rim of the carousel which has a rubber tyre and the little DC motor is spring loaded against the rim.  It uses optical sensors rather than hall effect and magnets.  I shall be using optical on mine but only for the home position - counted steps will determine inter filter spacing.

[Gina]

I've had another idea to ease the non-integer number of steps problem - make my own timing gear with 70 teeth to match up with a standard 20t pulley giving 3.5:1 ratio and exactly half a turn of the motor per filter change ie. exactly 2048 steps.  No error accumulation as the wheel rotates   Timing gear teeth are a much simpler profile than spur gear teeth so easier to make a cutter for.

[Gina]

Of course if I omitted the Luminance filter and just used RGB (plus Ha) for galaxies I could get away with a 6 hole carousel which would make everything SO much easier   eg. 60t + 20t timing pulleys which I already have and a smaller filter wheel altogether.  I know there are some imagers who prefer just RGB unbinned for galaxies rather than L unbinned and RGB binned 2x2.

Another thought is that as this combo is designed to fit the Esprit 80ED scope for a wider field, am I really going to use it for galaxies (other than M31 perhaps)?  I use my MN190 with separate OAG, FW and 460EX for most galaxies.  That means I would use just the three NB filters though possibly RGB for star colour (if I bother).  I already have a 5 hole carousel for 36mm unmounted filters plus mounting and timing belt drive.

I'm beginning to think I'm making things unnecessarily difficult and giving myself extra work and problems for very little good reason   Particularly if the debayered camera turns out to be less of a performer than expected!  All this work for what is basically just an experiment

[Gina]

So...  The filter wheel part of this project is now subject to a complete rethink   Looking at the options considered so far then, by number of holes/filters :-

1. N/A
2. N/A
3. The smallest option with just NB filters - 3:1 drive is easy with standard timing pulleys - nice small and light FW.
4. Easy to make with holes at 90 degrees - any multiple of 2 drive ratio makes for an easy system - even 1:1 - also small and light.
5. Already have carousel - less easy to obtain drive ratio with standard timimg pulleys.
6. Allows NB plus RGB filters - easy 6:1 or 3:1 drive ratio.
7. Allows NB plus LRGB filters - would need custom timing pulley
8. Plenty of filter space - easy drive options but a big filter wheel.

So far I think I have ruled out options 7 and 8 - too big and/or too much work.

[Gina]

The 5 hole carousel timing pulley has 48 teeth - not really suitable as it means accululated errors unless specially corrected.  It ciould be changed for a 40t one perhaps matched up with a 16t motor pulley to give 2.5:1 ratio and half turn of the motor per filter position ie. 2048 steps.  Trouble is, the hub is smaller and wouldn't fitr the hole in the carousel which fits on the 48t pulley hub.  More awkwardness   Software error correction would be easier

If the motor pulley has 24t giving 2:1 ratio the step count for each filter would be :-

Position  StepCount  Integer   Increment

1.              0                 0     0 or 1638
2.        1638.4        1638     1638
3.        3276.8        3277     1639
4.        4915.2        4915     1638
5.        6553.6        6554     1639

[SnakeyJ]

I know size and max working diameters are an issue, especially working with 36mm unmounted filters, though I'm thinking that 8 is the perfect number to cover; LRGB, NB (Ha, OIII and SII) and ideally a dark filter.

To keep to the minimum diameter it would be possible (but difficult to engineer) a smaller double carousel of 5 x 36.   Two positions would have to be unoccupied hole to allow you to see through.    As I said not easy to engineer and would probably eat in to your back focus distance.

The other possibility might be to have a linear filter, driven by a simple rack and pinion.   This would be quite long, and would require a sealed track perhaps 300mm either side of the camera/light path.    In a straight ladder form this could be driven by a geared pinion.   This would be pretty simple and effective, though might look a little ponderous.     However, if made as a series of 40x40mm hinged plates, with some adaptation of the drive you could probably curve the feed tracks  to reduce the total width.

[Gina]

Here is a cross-section diagram of the camera with OAG and filter wheel using the 5 hole carousel that I already have.  At the middle is the 48 tooth timing pulley turned down both ends to fit in. 

[Gina]

I know size and max working diameters are an issue, especially working with 36mm unmounted filters, though I'm thinking that 8 is the perfect number to cover; LRGB, NB (Ha, OIII and SII) and ideally a dark filter.

To keep to the minimum diameter it would be possible (but difficult to engineer) a smaller double carousel of 5 x 36.   Two positions would have to be unoccupied hole to allow you to see through.    As I said not easy to engineer and would probably eat in to your back focus distance.

The other possibility might be to have a linear filter, driven by a simple rack and pinion.   This would be quite long, and would require a sealed track perhaps 300mm either side of the camera/light path.    In a straight ladder form this could be driven by a geared pinion.   This would be pretty simple and effective, though might look a little ponderous.     However, if made as a series of 40x40mm hinged plates, with some adaptation of the drive you could probably curve the feed tracks  to reduce the total width.

This project is designed to show whether a debayered DSLR with set-point cooling to -15 or -20C would make a uesable astro camera for DSO imaging.  The main reason for trying a DSLR is the large sensor to provide quite a wide FOV with the SW Esprit 80ED.  If initial tests are satisfactory I may go on the produce a combined mono DSLR plus FW and OAG with T2 thread rather than the M65 for the Esprit.

So, as mentioned above this first incarnation will be for the Esprit and be used with NB filters.  This is where debayering really provides the greatest benefit.  So my first experiment will be with NB filters only so a large FW isn't needed.  With that in mind I've decided to spend a minimal amount of time and effort into this first version.  Having a carousel already made for 36mm unmounted filters it struck me as a saving to make use of it unless it caused other problems.  With this setup the front FW casing has a maximum radius of 90mm from the centre of the mounting hole.  Now my lathe has a spcaing of 90mm from centre to bed so I could mount the plate and use the lathe to cut the M65 thread directly into the FW casing.  If the plate were to be bigger I would either have to provide an adapter for mounting or possibly fix the plate at right angles to the carriage and cut the thread with a fly cutter - I think that would be very difficult.

Two carousels in series would take extra optical path length and I'm pretty sure I couldn't get that in as well as the OAG.  Without the OAG it would be possible though the drive motors would have to be outside the carousel diameter and increase to overal dimensions.

The linear filter holder would be very difficult to engineer - I have looked into this idea in the past.

[Gina]

With a 3D printer in the offing capable of creating complicated shapes in plastic, this project is about to undergo yet another major design change.  You might say it's going in the melting pot (in the form of the extruder - plastic is melted and squirted into shape)

I'm proposing to print the frame for the camera, and most of the parts for the filter wheel and OAG.  The printed camera frame will replace some of the camera body parts and provide reasonably accurate alignment between the image sensor and the filter wheel casing.  I shall be able to print the carousel and have any number of filter positions within reason, the maximum size overall being the 200mm cube the printer can print in.

[Gina]

I'm thinking that I can "print" a structure that will fit inside the rectangular plastic frame of the camera - the part the goes to the mounting usually, that I have cut down.  This will have a flange to fasten to the inner side of the FW casing.  The idea is to define the distance from the sensor to the FW.  I shall probably want to cut down the original frame yet more.

Here are some photos of the skeleton camera that I hope will explain what I mean.  I plan to produce a SketchUp model of the printed part shortly.

[Gina]

Here is a sectional diagram showing the printed camera frame in purple (for clarity - it will be printed in black).

[Gina]

Here's a screenshot from a SketchUp model of the camera frame to be printed.

[Gina]

I can take the camera frame and extend it to seal up to the copper box and also to include the optical window making a complete sealed chamber.

[Gina]

I was working on the latest design of the camera frame when Win 7 crashed and I lost the changes so I'll do them again. 

Meanwhile, I've been learning more about SketchUp and worked out how to design a 7 hole carousel that I could 3D print.  Firstly, design one filter hole with ledge to hold the filter, then calculate a suitable pitch circle and draw a line from the filter centre to the pitch circle centre.  Select all and then using the rotation tool centred on the pitch circle centre and selecting the filter centre, press Ctrl to Copy, move the copy a few degrees and type 51.43 and return.  A copy of the filter hole is made rotated 51.43 degrees from the original,  then type 7x and return - 7 filter holes are drawn equally spaced throughout the 360 degrees.

I tried an initial move/copy of 360 degrees and division by 7 usung 7/ but that didn't work.  I guess that was because the copy was on top of the original and such copies on top of each other are counted as one by SketchUp.  Pity - would have saved calculating the one seventh of a revolution and typing that in (could probably use copy and paste from the calculator but I haven't tried that).

[Gina]

I have also found a gears plugin for SketchUp so I shall be able to include gears in my 3D prints, so providing a 7:2 gear ratio to drive the carousel from a stepper motor is no problem

[Gina]

I've been looking into the carousel design with a view to 3D printing it.  I have come to the conclusion that timing belt drive is going to be too difficult to implement and now looking at rim drive and optical sensing for filter stepping and homing.  I have also decided to provide a connection between  back and front FW casing plates in the middle of the FW.  So the carousel will be on a ball bearing which will fit onto the middle FW support.  Currently I'm using a 15mm ID x 24mm OD x 5mm thick bearing and have designed the carousel accordingly.  This is the way the Atik FW works except that uses thrust bearings and a spring.

Here is a screenshot of the ShketchUp model.

[Gina]

I've printed a carousel successfully (from the SketchUp model above) but I'm still uncertain about the drive - I'm not that keen on friction drive to the rim TBH.  I have been printing gears for another project using the "Involute Gear" plugin for SketchUp which seems successful, so spur gear drive to the centre of the carousel is now a definite option.  Here is a screenshot from SketchUp showing a 10t gear for the stepper motor and 35t gear for the carousel.  This gives half a revolution of the stepper motor per filter step.  The larger gear model could be added to the carousel model or I could just make the gear a push fit on the present carousel print.

[Gina]

Here's a couple of photos of the carousel with two filters fitted.  These just push into the holes and stay by friction - no fixing screws required   The almost matt surface produced by the roughened print bed can be seen in the second photo.  The shine on the top could be removed with sandpaper.

[Gina]

One advantage of a printed plastic frame to connect the sensor plate to the front panel is that the infill is 75% air and thus provides good thermal insulation reducing the amount of heat needed to be taken out of the cold zone.  I shall need to either allow air movement between the cold and warm parts of the dry zone or arrange sufficient room within the cold zone to put soilica gel.  The latter would clearly be best but is less easy.  Some way to change the silica gel for fresh/dry would be a good idea - I don't really want to have to strip the camera down to change desiccant when it gets "full" of moisture.  Of course, if the dry chamber were perfectly sealed the disiccant shouldn't need changing.  My Atik cameras don't have this facility.  With my last attempt at sealing a dry chamber the silica gel lasted a fortnight before the RH rose to the point that the sensor misted up at -10C.  I expect to do better than that this time but...

The design thinking continues...

[Gina]

I need to strip the camera down again to see how much cold zone space I can obtain for silica gel beads but I've been looking at the camera and with a new design of printed plastic frame I think I can get quite a reasonable amount of desiccant space.  I might even have a way of adding a plug to change the desiccant

I shall have to povide my own porous silica gel barrier or take some pieces of the bags and glue them to the plastic frame to retain the gel beads as the bags are too big to fit in as they are. 

[russellhq]

Do you think you could have a small sealed box with desiccant (possibly with glass window to detect colour change) located somewhere else and a small tube running between the tube and sensor chamber? Might be an alternative to having desiccant actually in the cold zone.

[Gina]

Do you think you could have a small sealed box with desiccant (possibly with glass window to detect colour change) located somewhere else and a small tube running between the tube and sensor chamber? Might be an alternative to having desiccant actually in the cold zone.

Anything connected by an air tube to the cold zone will automatically be in the cold zone - if air can move between desiccant chamber and cold zone it will carry heat from the desiccant chamber to the cold chamber.  Nice idea though

[Gina]

I've been looking at the new design for the camera inside frame and cold zone within the dry zone.  All my ideas so far have gone out the window because of the shutter.  This has to be in the light path (cold zone) yet I was having the shutter mechanism in the warm zone    Now apart from possible heat conduction, it's not sealed, of course, so the shutter can't be in both zones

I can see just two ways out of this - include the whole shutter mechanism within the cold zone or separate the sensor from the rest.  The latter would mean replacing the cover glass on the sensor in a very clean and dry environment.  I would need to make a box to contain the sensor and cover glass free of dust and dried.  Then remotely (or with sealed arms) lower the glass onto the sensor and glue in place.

Again, I seem to have taken on a job that's turned out to be more difficult than I thought!

[Gina]

I favour using the cover glass option.  I know one method of doing this is to provide two small holes and flush out the enclosed space with dry air or a dry inert gas.  I think dry air is quite do-able - I could force air through a bag of silica gel and through the sensor cavity.  The holes can be sealed up with any sort of glue.

The rest of the cold zone can be sealed and thermally insulated quite easily.  The cover glass will not steam up as it will be in the dry zone and the optical window that seals the dry zone from the environment won't steam up because the inside face is in the dry zone and it should be warm enough not to steam up on the outside.

This also means that the whole of the camera with several clear spaces can contain silica gel.  I could even make a hole in the copper box and glue in a window to see the silica gel beads.  A piece of Perspex or other clear plastic would do.