Equatorial Platforms, for free!

[wobblewing]

5 minutes ago, AlexK said:

The rod should be metal, you want to print the rack, which is much easier as you can print it on the side (most recommended for the threads but often needs supports if your parts fan is weak. Definitely a stronger filament will be needed, I would use nylon.

Yes the rod is metal. 

I don't have nylon,  so that's not free 😆 anyway i don't think much force is needed,  it's only a small table top dobsonian, not a Big Bertha 🤣  PETG should be OK,  I've got that.   I've used PLA on RC car gearboxes reasonably successfully.  Also a 3dp r/c spitfire in PLA, 3S Lipo powered! Very fast as a spitfire should be 😃

I've a fair amount of experience with 3DP, so I'm up on the construction techniques.

Edited January 16, 2021 by wobblewing

[AlexK]

6 minutes ago, wobblewing said:

Re-position the platform and press reset.  The motor stops when it gets to the end (max count),  so you soon find out when to reset it.  That's easier than accurate polar alignment.

It's only an evening toy used once in a while, i'd be amazed enough if i can get some good images.  My phone doesn't do long exposure!

10s is actually a decent exposure. Have you heard about stacking? Just a dozen of 10s exposures through the tiny aperture of my smartphone (Galaxy S8) reveal 9.5m stars already when stacked. With your 150mm Heritage you can do decent enough DSOs, but for that your tracking must be decent as well, as even the smallish Heritage has a significant magnification.

Not sure what you mean no accurate polar alignment required. You have to have it provided quite well on any EQ mount as even if you could adjust the tracking speed on the fly and even add the second motor to compensate for the declination error, you will need the third motor to compensate for the field rotation which might be an issue even at a 10 seconds exposure, and especially if using the smartphone in EP camera setup. Nothing drastic, though, just pointing the true rotation axis to the Polaris would suffice.

[wobblewing]

6 minutes ago, AlexK said:

10s is actually a decent exposure. Have you heard about stacking? Just a dozen of 10s exposures through the tiny aperture of my smartphone (Galaxy S8) reveal 9.5m stars already when stacked. With your 150mm Heritage you can do decent enough DSOs, but for that your tracking must be decent as well, as even the smallish Heritage has a significant magnification.

Not sure what you mean no accurate polar alignment required. You have to have it provided quite well on any EQ mount as even if you could adjust the tracking speed on the fly and even add the second motor to compensate for the declination error, you will need the third motor to compensate for the field rotation which might be an issue even at a 10 seconds exposure, and especially if using the smartphone in EP camera setup. Nothing drastic, though, just pointing the true rotation axis to the Polaris would suffice.

I'm aware of stacking but no experience of it yet. That'll be a trick to learn later!

Sorry what i meant was that i didn't think reseting the platform would be any more difficult than the PA anyway, so i don't see the problem with it.  It's only a case of moving the platform back and then pressing the start button

 

Edited January 16, 2021 by wobblewing

[wobblewing]

Yes i suspect the smartphone may be the weak link. This is where i have zero experience, so a learning curve.

Edited January 16, 2021 by wobblewing

[AlexK]

11 minutes ago, wobblewing said:

I've a fair amount of experience with 3DP, so I'm up on the construction techniques.

Sounds good! 👍

One last tip for now then: 
One of the tasks on a "compound" project like this (wood, metal, 3DP parts) is aligning the stuff together. For 3DP parts, you can print additional features dedicated for that. From trivial center/middlepoint markers/pilot holes all the way to long "rulers" which you align together before screwing parts down to the ply. Even a kid could do a nice assembling job with such visual aids.

[wobblewing]

Ha yes,  I've done lots of that.. that's the easy bit 😉

 

More the challenge is getting decent pictures, but as i understand it a 150/750 newtonian should yield some good stuff, once the sidereel correction is sorted

Edited January 16, 2021 by wobblewing

[wobblewing]

Btw you can print paper templates from fusion 360, so line it up on the wood for hole drilling etc. 

Edited January 16, 2021 by wobblewing

[AlexK]

4 minutes ago, wobblewing said:

Btw you can print paper templates from fusion 360, so line it up on the wood for hole drilling etc. 

That's a nice feature indeed. But you need an ordinary printer too. I'm trying to get away from paper as much as I could

The imaging is a huge subject on its own indeed. But even a mediocre EQ platform would allow quite decent planetary and moon images possible (they are all shot in the video mode, then each frame is stacked. You will be amazed how much details an ordinary deconvolution algorithm could retrieve from a few dozens of thousands frames with a smudgy blob dancing around

[wobblewing]

I already have a (paper) printer so no need to buy anything -free 😉 you can see where I'm going with this now! Haha

Edited January 16, 2021 by wobblewing

[wobblewing]

To close this thread:  

 

I finally finished it!  I ended up having to pay £2.30 for a stepper motor driver PCB (not Free 🙂 ).  Other than that, it's from the parts bin and made from an old shelf.

All 3d printed included the ball joints.  Driven by a stepper motor and worm gear pushing/pulling the platform (way more torque than needed, but this can be turned down on the driver PCB)... BBC micro:bit coded in Python... Powered by a 2S or 3S Lipo.... 

It actually works really well and I've got the accuracy spot on!

I had no issue at all with the potential error induced from pushing the platform following a circle profile by using a linear leadscrew.  As you can see the motion is as good as linear through its entire travel, so a fixed speed across it (without varying speed across its travel according to position) worked out fine.

I was going to replace the brass leadscrew 'nuts' with a split nut design so it's fast to reset.. but I found that it only took about 10seconds to spin be back by hand - so I didn't bother.

There are no limit stops with microswitches, I basically reset it to the start position and press a button to reset a counter.. once it reaches the count limit it stops and flashes the screen to indicate a reset is required (although it's pretty obvious of course!)

Edited February 23, 2021 by wobblewing

[Spile]

So are you going to be taking orders?

[Xilman]

43 minutes ago, Spile said:

So are you going to be taking orders?

Seconded.  I could well be interested.

[AlexK]

Congrats! It looks yummy!

The image though got killed for some reason. But the vid is OK enough (just maybe shoot it open so folks understand the inside design a bit too, it's more important than the drive itself).

Just 10 sec to rewind on high speed? Sounds good indeed.

I'm indeed puzzled why no correction required. It supposed to "rush" motion at edges. Perhaps, you haven't tried with really high magnification EPs on planets? OTOH, maybe you have a real invention here? That two points swivel hand of yours is indeed shortening in the front plane, as the elliptical projection of the cone edge moves it in the Y and Z directions. So X vector is shrinking. Could be just a lucky coincidence of curvatures, but it looks like a leverageable possibility. Some concise math is required to proof that.

Looking forward for (concept proofing) imaging results.

Edited February 22, 2021 by AlexK

[AlexK]

A couple of words on possible design improvements to try:

1. Get rid of steel wheels. Just solid rails and groves should suffice with that direct drive (at least for a lightweight telescope like that 130P). I would play with different filaments here.
2. Enclose the rod from debris (maximum enclosure box). That's a precision mechanism, which might stuck or even be destroyed with enough hard dust.
3. I can see the rod and motor moved from the front row right behind it. And operating horizontally to reduce the height. Also good for mechanics protection.
4. No need to follow the 2xSectors classical design here. It's just a mechanical compromise and becomes viable only for very heavy telescope . A simple cylinder wheel at (90 - Latitude) angle would work just fine here.

[wobblewing]

9 hours ago, Spile said:

So are you going to be taking orders?

Interesting thought... I only intended to make it as a one-off for my own use!  It was very cheap on materials, but I spent quite a lot of time on it...

I need to get a bit of experience on it first... I'm a newbie in terms of astronomy, so I don't really know what to gauge it against.. Also, I'm only doing afocal photography using my phone with a 3D printed 'phone to eyepiece' adapter, so I'm not sure if my photography setup is capable enough to really characterise how good the platform is... My engineering nouse suggests it's reasonable, but I may find foibles as I gain experience with it... I have a few improvement ideas, such as an add-on Polaris alignment 'sight'

Edited February 23, 2021 by wobblewing

[wobblewing]

4 hours ago, AlexK said:

Congrats! It looks yummy!

The image though got killed for some reason. But the vid is OK enough (just maybe shoot it open so folks understand the inside design a bit too, it's more important than the drive itself).

Just 10 sec to rewind on high speed? Sounds good indeed.

I'm indeed puzzled why no correction required. It supposed to "rush" motion at edges. Perhaps, you haven't tried with really high magnification EPs on planets? OTOH, maybe you have a real invention here? That two points swivel hand of yours is indeed shortening in the front plane, as the elliptical projection of the cone edge moves it in the Y and Z directions. So X vector is shrinking. Could be just a lucky coincidence of curvatures, but it looks like a leverageable possibility. Some concise math is required to proof that.

Looking forward for (concept proofing) imaging results.

Right now I've only got about 90mins use playing around with it..  Since my dob is only little with a FL of 750mm, I tried a 3.2mm BST EP looking at a star (thus x234 - approaching the practical limit of the scope) and it was holding in the centre of view really well as far as I could tell... the problem is it was quite windy and getting knocked around a bit - so more time will tell!

Also, only Mars is available at the moment and it's getting very small... so not much more useful than a star in terms of seeing how well it tracks.

In theory I'm sure there is some non=linearity to deal with, but as you can see in the video the pushrod is fairly straight and level across its sweep. the small amount of non-linearity doesn't appear to be noticeable, but maybe I'll find it at some point....  It certain has very precise control, so should be easy to do in the software if needed. I certainly couldn't notice anything in practice.

Edited February 23, 2021 by wobblewing

[wobblewing]

2 hours ago, AlexK said:

A couple of words on possible design improvements to try:

1. Get rid of steel wheels. Just solid rails and groves should suffice with that direct drive (at least for a lightweight telescope like that 130P). I would play with different filaments here.
2. Enclose the rod from debris (maximum enclosure box). That's a precision mechanism, which might stuck or even be destroyed with enough hard dust.
3. I can see the rod and motor moved from the front row right behind it. And operating horizontally to reduce the height. Also good for mechanics protection.
4. No need to follow the 2xSectors classical design here. It's just a mechanical compromise and becomes viable only for very heavy telescope . A simple cylinder wheel at (90 - Latitude) angle would work just fine here.

1) actually I tried 3DP rollers to start with, but found them a bit lumpy due to the FDM layering.  I switched them to skateboard bearings (had a set in my spares box!) and it made it run much smoother...  I also put an open bearing in the South pivot point as I found 3DP friction bearing a bit 'sticky', although the torque of the drive mostly overcame it... By using bearings the movement is very smooth indeed.  I printed the sectors to get the FDM layering along the axis of movement and then sanded it to get a very smooth surface to roll on... It's super slick and smooth.  I had more of an issue with stepper smoothness at first and had to use the 32 microstep/step option (DRV8825 driver). If using the basic 1.8deg step size, you could see the stepping affecting the image being too 'notchy' and the vibration got through the mechanism.  Simply increasing the step resolution fixed that (now 6400 steps per rev).

2).  Maybe, I only use it out in my garden on the patio and it shouldn't get too dirty.  

3)  Sorry not quite sure what you mean there.

4)  Yes I realise that, but I did it anyway when I was playing around with the idea, so I stuck with it!

BTW, the ~10sec rewind is by hand (not driven backwards).  The code is written such that when it gets to the end then the driver is disabled... So I just spin it back by hand (faster than driving it!), reset the counter and then press 'go'.

Edited February 23, 2021 by wobblewing

[wobblewing]

A few CAD images from Fusion 360... Ok, so I used white PLA instead of black 🙂

Latitude 52deg.  Adjustable levelling feet give probably +/-4deg ish.  Total movement ~125mm giving 82minutes of travel.

Edited February 23, 2021 by wobblewing

[AlexK]

Thanks for renders. Many lovely touches and good ideas!

The speed ramping is not drastic at that angle, but at x200+ you should be able to see it after like 15 minutes for sure. So it might be your arm drive angle compensating for the error even if only partially. The classic poncet is using a long pin on the top platform and a caret with the vertical slot for the pin. So the point of contact is moving in Z and Y as the worm slides it along the X. I have already mentioned how's that compensated in some mechanical designs. But maybe this is one more way around that?

Ahh, manually? That's a fun spinner indeed! 

[wobblewing]

It is definitely moving in Z and Y, whilst being pushed along X (the ball joint)... but I think it's such a small amount that it's insignificant.   We will see 🙂

I might have a go at some geometry calculations to see how much it is.

Edited February 24, 2021 by wobblewing

[wobblewing]

 Well, clear night tonight finally - albeit with a full moon!

So, testing the tracking out.. here's a hyperlapse video.. bear in mind this is ~22mins squash into 22seconds!

I can juuuuust notice the non-linearity, which equates to about one third to a half millisecond timing on a step..  Will fiddle with the code, but it's not really necessary as it exceeds the exposure length I can do with my phone up against the lens.

Not quite sure what the up and down is all about   I think my PA needs work... But the hyperlapse really amplifies it... in real time it doesn't seem like it moves at all.

 

 

Edited February 26, 2021 by wobblewing

[wobblewing]

A single 8 second sub, ISO 500. Looks kind of promising for an afocal effort.  Will DSS this lot.  My talent probably runs out here 😄

[AlexK]

Lovely vids!

On the second one I see it moving up consistently. That's the PA error.

The up/down wobbling is called periodic error. Caused most likely by the vertical "sectors" sliding sideways on the rollers (they are bending a bit in the axial direction when overcoming the friction perpendicular to rollers, then unbending by transferring the force in the direction of rotation). Just a hypothesis of course. To be sure it's better to film it in the meridian.

Edited February 26, 2021 by AlexK

[wobblewing]

I created a longer timelapse and it appears that I can see:

- PA alignment error (will do better next time).  Noted the other more accurate technique.

- a small amount of non-linearity as it approaches the centre of the sweep and the 'turnbucke' (ball joint thingy) goes horizontal - you get a tiny amount of speed drop requiring less than a millisecond of 'per step' adjustment.  I'll approximate it into a number of linear portions of the sweep.

- more significant I think is that the step signal of the micro:bit appears to jitter due to using micro python (interpreted code), it's not locked at exact 48ms per step as implied in the code - I can see this happening on an oscilloscope.  Maybe I should re-compile it in C, if I can figure out how to do it on the micro:bit.  Perhaps I should've used a RPi pico.

In any case it's not really necessary other than for interest, as I don't have the photography gear to increase exposure beyond 10 seconds at the moment.

Edited February 26, 2021 by wobblewing

[AlexK]

I'd go with the ESP32 vs pico. That way you can research what's going on remotely using BT or WiFi logging.