Michele Scotti

Next time make sure you shot a video - worm gear hobbing is fascianting. Would you reckon you had a similar set-up to any of these? https://www.youtube.com/watch?v=19jKlq8Ofd4&list=PLJtiP4DlrmJROzFYnk0rYQ9fnJPO9KyCn&index=16&t=0s https://www.youtube.com/watch?v=eKJe1JnvRc0 Hope not to detour the thread here....

Rather than a video let's see some maths behind this - it applies to any modern smartphone with Mpixels. At the minimum focus distance - usually 50mm or something- this is a clear picture of a caliper. A crop-up shows that 5mm takes 77pixels. This equals to 65um/pxl By the way in a video you wold appreciate sub pixel movement for sure. Out of curiosity this is what happens using the zoom (digital in this case) Reoslution here is 14um/pxl Not maybe the most accurate tool but it detects tiny movements very efficiently - it might be handy when checking mounts for slack

https://simplybearings.co.uk/shop/p88001/Budget-7000-Single-Row-Angular-Contact-Open-Ball-Bearing-10x26x8mm/product_info.html?backstep=1 2tons of axial load - i guess its enough

I've check most of jig elements -and specifically the upper and lower bearings- is in a pretty funny and maybe novel way. I pointed the smatphone at the area where movements would manifest and shoot a video while applying load. As funny as it sounds you can detect way finer movment than naked eye. I was trying to put some maths behind it...maybe later.- you can see sub 0.1mm for sure. I have few videos where the check is OK but can't find the one that was clearly showing a minimal movement in a different area of the jig that was later fixed. Bottomline - to my surprise- bearings are ok. Anyway if they moved the laser pointer wouldn't move as the pipe would not bend but rather tilt.

The quest for a stiff and reliable platform I've used some counterweights at the end of the board to evaluate the stiffness of the system. Not a good starting point... 0.9mm with something like 6Kg placed some 30cm outboard, just to exacerbate the flexure. I've then tighten all screws and added M6 bolts with inserts to cement the position of the wooden blocks. Better but still 0.6/0.7mm. Where's the flexing coming from?? Maybe from the element assumed to be the most rigid? Let's check the pipe... A cheap laser was clamped on the top of the pipe -see pic- pointing down. Some paper tape on the bottom to act as a 'screen' for the phone to record just under it. And there you go – the pipe flexes big time! Movement -subtle-is caused by pushing on the board and releaseing - laser dot moves 1 or 2mm. Smoking gun! So the pipe that fits perfectly in the bearings and we made all the wooden blocks around is not up to the task. Shall we throw it away? Nah! The pipe needs stiffening on the vertical plane where it meets the angle grinder – so next step is to insert a 5mm flat bar that is as wide as the ID of the pipe. Possibly a few welding points at the ends. For a structural application this would be gruesome borderline criminal – in this case it really needs to work under small stress.

I’d suggest a different approach: fixed bearing housing after lapping. The way I see it, is that you have a run-out from the shaft to tollok, tollok to sleeve, sleeve to worm gear I and finally ID to the generated teeth. Actually it’s not all – your shaft has a nice press-fit brass thread. All of the mentioed interfaces might be very precise but they all come with few hundreths mm of run-out or clearance. Lapping will take care of all of them. I have to be honest – never did it myself just because I’ve recently stayed away from worm gears in my telescopes. Also I reckon it is only mandatory for Dec axis. It’s maybe a lengthy process (and messy?) but not that complicated. Let me know what you think about. I suppose the shaft material is not treated/hardened. You could tap an M5 and add an extension – belt&pulleys are pretty forgiving to run-out errors. I would never suggest that for a gear coupling. What is the ID/OD of the bearing and the width of housing? 12x30x8?

my 2 cents - from previous picture the housing and bearing general dimension doens't look bad - I'd suggest to switch to tapered bearings with a (simple) system to pre-load one of the bearing. That will kill any lateral movement. Is the BH fixed or does it have a system to cope witht he inevitalbe run-out of the wormgear?

I trust you realized that all the examples you brought are 'dynamic' at a certain extent - even your swivel chair. It totally makes sense to have more 'legs' to achieve a more stable set-up simply by increasing the area on which the CoG can fall on. One of the perks of a AltAz mount is that all loads are pretty well centered. The CoG of the 'OTA' rotates around on a 124mm circle - well within the 'triangle' set by the 3 rollers. I do not foresee any stability issue. To me one of the design aspect that I tried to optimized is related to the load transfer from the Alt bars thru the Az table then Az track to the rollers on the tripod and from there to the feet on the ground - which incidentally I haven't designed yet so they are not shown in any CAD so far. As a matter of fact the FEA we ran was on the worst case where one of the Alt bearing is placed half-way over 2 bearings

To start by stating the obvious: the tripod has 3 legs and at the end of each leg there's a roller. So this set of 3 points defines a plane - adding any additional supporting point will introduce an unwanted constrain. Chairs have 4 legs and they are 'stable' because they deform under load allowing a distribution of weight on 4 points. So 3 is actually the only choice for such structure. Class-meter teelscopes uses a different approach - they usually have 2 very accurate flat rings that are part of a hydrostatic bearing where pressurized oil keeps the 2 rings apart. Hope this answers your question?

I might actually resort to sand blocks, eventually! At least they'll provide smooth transitions although not going to do much for the overall run-out. "Using maths to analyze "a length of string" is a common fallacy." what do you mean with this specifically?

I try to bundle up answers - in first place there's no such thing as a silly question. Only asnwers can be silly. Also, I totally share with both of you the fact that the likelyhood that all of this is going to be a big mess is very high - I'm not shooting for first time quality. I'll learn by mistakes along the way. There's a specific reason though why I'm going down this routet. And it's about the spirit of this project - this is a pilot project and an (ambitious) blueprint for a large, attainable imager that should be potentially built around the world without access to expensive/porfessional/dedicated machinery. A go-kart shaft and a couple of bearings are fairly common and inexpensive. Surely we might fail - then I'll reconsider what to do. Anyway - I took 2 measurements every 15deg - orange and grey lines. I applied a sinusoidal correction to simulate a tilted shaft being straighten-up - in blue. The result after such correction is the amber line.....which is more than I expected! We are looking at a ca. 1.3mm P-V or run-out. Quite some material to remove.... Worst scenario if (likely) the steel left is too thin I'll bond another set of arches like the first time - it probably costs less than 100Euros

A lot of good points... Ref is the shaft and its bearing - the one that will stay on the Az table. I'd like to target a few tenths of microns of total vertical run-out with very smooth transitions - somehow the latter is more importand than the first. The pipe is a go-kart rear axle with 2mm wall thickness - its rigidity will be scrutinized due course. The idea is that it has to allow the generation of an ideal plane which is the new track surface. f it is slightly tilted - to a reasonable extent- compared to the table itself is not a big deal. What counts is that the track and the 2 Altitude ground bars are 'parallel' - this can be quite easilty checked and rectified with shimming under the Alitude bearing brackets. The Az table is fairly rigid - at least by design anf for the purpose- as it's the most critical element for the overall mount performance. The picture here is for the HUGE investment in a micrometer to test the accuracy of the surface and preliminary the stiffness of the jig.

The track currently is of unknown geometrical quality - hence I can't use it as a reference. Any roller used to help supporting will inevitably act as a "copy-carver" to some extent. The challange here is to be able to use he tube as a reliable axis to start with. That is the reason why I extended the distance between the 2 bearings in order to better cope with the bending moments. Clearly success in not a given here... However my way to evaluate how far we are from a solid jig is to check how close to the ideal situation i.e. how insensistive to loading is the distance between the track and the tool. . I agree that the grinding forces are not going to be massive but I'd like to start with the most solid set-up. In the meantime this is the set-up we came up so far. Added quite few reinforcements and a rudimental 'micrometric' adjustment of the head height....still working on that though

I see what you are saying and I reckon I have contermeasures in the design of the jig - btw as a flat 2D it didn't provide any hint of the width on the jig itself. Here are a couple of pics of the early stages of jig build - let me know if it makes sense to you - still I would need to run many checkes before giving it a go on the actual azimuth table Bottom side with track: Upper side with the flange to carry the second bearirng:

On to the challenging bits of the build....and the reason why Iwe didn't want to load the strucutre just yet. Background/assumption: looking at its Periodic Error an ideal mount minimizes the amplitude and it looks smooth – in mathematical terms this means that the higher order content of the Fourier Analysis. I would add that a desirable element is that the movement is repeatable i.e. it’s ‘smooth’ and without ‘random’ jitters all around the rotation the axis. To achieve a good tracking capability I’d start from the most precise execution of the components – however soon you realize that even so, errors in the order of few microns will show up in the image given the long focal length. A bit of math done some time ago showed me that 1 micron error on the azimuth track would generate a 0.22arcsec (arcsin(1/900.000) tilt of the telescope – which is comparable to the sensor pixel resolution. And of course there is no chance -for me at least!- to produce such an accurate surface. Let alone that all other elements of the system have a tolerance – for instance the roller bearings are specified at around 11microns. As a consequence, for a telescope to be an imager, autoguiding is a necessity. However a smooth repeatable PE is easier to be guided out. So what’s the plan? A ground track of course! How…? In the absence of a gigantic lapping machine my choice is going into creating a grinding machine on the table itself. Here is the sketch which is hopefully self-explaining. Enough to say that it makes use of an angle grinder -with a controlled play- and a controller to limit its speed. The tool will be a “diamond” cup like the one in the pic. Sure it will take long time to carry out the machining operation but that’s ok if it yields the result. Also the shaft is rotating into the same bearing that will stay on the azimuth table and that is used to center the "tripod". On a side note: the use of a central bearing is part of an attempt to reduce the random jitters as the rollers will always be running on a constrained radius with no chance to build slack -on a micro-scale. It’s possibly more difficult for me to explain than for you to guess it. Well, that's the plan.....

Can you elaborate on this? I understand that the flanged bearing and "boxed" approach holds the inherent value of offering solid sides for the fork on the RA axis.

No loading at all just yet as the SS track still needs to be machined. On top of that I'd like to properly sort out the position of the feet. I'm going got minimize the distance between the contact point of the bearing and the feet themselves. Ideally they should be just below so that the bending stress and its deformation is minimized.

Some update on the works on the bottom on the mount. The 'tripod' is now (almost) completed enough to be tried out on the (almost) completed azimuth table. The tripod has all bearings in place - at the end of each leg- to run on a stainless steel track which is placed on the bottom side of the azimuth table. It was just nice to see the whole thing spinning around.... What's left to do on this sub-assy? On the tripod it's the implementation of the drivetrain and sorting out a robust way to place the mount on 3 feet. The azimuth table just needs the track to be properly prepared to function as a generous bearing - I'll come to that in one of the next post. However, all of that was good enough for a first 'spin'. I uploaded the video on yt: (btw does anybody know how to resize the video preview window? If it's even possible...) Clear skies, Michele

Marcus, c'mon show us at lest a sketch! The good part of a forum is that you can 1) collect a lot of ideas that you can consider for your project and 2) gather a bit of that collective feeling of social push to carry out your project. Either way I'm sure you'll get a positive feedback regardless if you just sharing a sketch on this thread or start a new one Clear skies, Michele

Yup, I think you should start a thread as your approach is fresh and it can easily inspire somebody else. No worries for the topic - still relevant and thnaks for the compliment. Comments slowed down a bit - I wonder if people are waiting to see whether this is 'real' or not.....me too to be honest!

I have to say that it's a very intriguing approach to GEM - I'll dig a bit in your blog. I like the use of the bearing directly as part of the structure as well as boxing with Aluminium panels and rods.

I like this option when low profile - especially the fact that is low profile apart form the large head. And I like your GEM too! To you have a thread where you decribe its construction? Can you track consistently for long exposure?

An element of the assembly that I hold dear is the connection of the components. A bad connection of just one of the element will preclude the chance to achieve the modal performance of the mirror box and the telescope itself. The following is my interpretation of the joints we are extensively using on the pac-men and is works around to 2 main constraints: • the Pac-men material is a sandwich of foam and plywood • all the connecting element are thin walled Aluminum Wrt sandwich construction the solution was to we carved out the foam and replaced with solid plywood in the areas where fixings were expected. Here is the picture where you can see few of them - some of these inserts have other function such as the manufacturing and assembly process. They improve the performance of the sandwich too. Second challenge lays with thin walled profiles - from 1.5 to 2.0 mm. The load from the fixing tightening needs to be distributed to avoid sagging which spoils the ability to reach high bolt loads. Such deformation is obvious for thin Aluminium but it might be subtle yet relevant for steel as well. So how to deal with that? 2 things: a generous steel washer under the bolt head/nut and a purposely built spacer to transfer the load through the hollow profile. Here is a picture of the rear main bar. The option to bolt directly on the inner wall in contact with the pacman is far from ideal as it's cutting off some of the material from contributing to the stiffness. Now to the actual joint – as it works on friction 2 main aspects are taken care of: • Contact area treatment • Bolt load As per picture here below we’ve sanded down the surface to flatten it. Or better to secure that the contact pressure will be more evenly distributed. Bolt load is tricky with a wood sandwich especially when the screw is inserted along the layers plan. As we would have to undo the joint several times even during the assembly tuning, we rules out wooden screws. What we opted for instead are these metric studs for wood. They can accommodate a nice M8 or M10 nut. They come with an handy torx at the top. A pilot hole 0.5mm smaller than the screw shank is drilled and some epoxy added in the hole to have a rock solid anchoring. All of these allows high torque hence high “bolt” load which is transferred evenly through the profile and the pacman. As always, thnaks for your attention - if you have any comment or if anything is not clear -which is likely!- I'll be please to answer. Clear skies, Michele

On with the work on the mirror box. This is just an assembly dry-run to sort out all those details that we've overlooked while producing the single components. The main piece of work here is to fit some long M10 and M8 studs for wood, preparing the mating surfaces to maximise the contact area and to sort out the right washers. All of this is to be able to tighten the fixing so that the structure shows its potential and it's not wobbly just because a fixing is not up to the task. With this minimal mirror box concept the structure is as strong as its weakest element or joint. I attach a pic here and I've just uploaded a video on YT with a bit more details if you're interested:

I have to admit that we were quite stunned by the stance of it when we pull it together. Also, it was less than 10Kg so it felt weirdly light. When looking at that as just a sub-assy, the whole scope will look fairly ginormous. However my mind goes to the fact that the success of this project is not its size but rather its capability as an remotely controlled, robotizer imager - that's the big challange to me.