Jump to content

NLC-Banner.thumb.jpg.acb5ba835b9e8bf0718b90539633017d.jpg

DIY direct drive mount


EarthLife
 Share

Recommended Posts

Got magnets, other bits on way.

  • 3D printed magnet holder/frame
  • 150mm * 2mm laser off-cut steel disc for magnet backing (£2.80 each ebay)
  • 15mm linear rail support (£5 AliEx)
  • 15mm * 400mm hardened steel linear guide rod (£7 AliEx)
  • 24 N52 neodymium magnets 20mm * 10mm * 4mm (£15 for 50 AliEx)
  • 2 15mm I.D tapered roller bearings, will be preloaded (£6 each AliEx) - not yet shown
  • Coils, stator and stator bearings - not yet shown

Easy to scale up/down as required (bigger magnets for more torque etc).

Can be reconfigured to position the magnets on the outer or inner side of a steel off-cut tube as the rotor. rather than using the disc configuration.

 

image_2021-11-11_003800.png.f43823437f226fe9708be7fb66bf8cb3.png1858294025_DirectDriveRotor.png.69c93a31db59234b5dc58dbd1ce9a1ed.png

Edited by EarthLife
price error + a bit more useful info
  • Like 2
Link to comment
Share on other sites

150mm * 2mm steel disc (rotor part of the motor) with 3D printed magnet spacer and 2 layers of 24 magnets (48 in total) in place. The magnets we have here are only 2.9mm thick and only N35 strength, so used 2 layers. 5mm thick N52's on order to replace them when they arrive.

 

Rotor so far ..

519737614_RotorTop.png.120e3b8d0d328d56d2b39f71b4f81c2d.png

 1358840491_RotorBot.png.5f07347a92aad9660d598360686aa603.png

 

  • Like 2
Link to comment
Share on other sites

A simple experiment with a monochrome webcam up close to a 0.5mm graticule steel ruler and a piece of P1200 sand paper alongside ..

1628830022_P1200gritwithscale.png.0933d8bc1b7f4496ba8cb2821494620f.png

 

A circular piece of sand paper alongside a printed regular graticule scale (with a 0 degree index pattern as the starting reference) on paper or plastic film (on back of motor rotor disc) together with a simple high frame rate monochrome image sensor (maybe an optical mouse image sensor) on a raspberry PI board and tiny lens can serve as the position feedback sensor.

A decent 600 real DPI laser printer can print a 100um period scale, a 2400 real DPI printer can print a 25um period graticule scale onto paper or plastic film. The PI board can do a little bit of image processing to count the passing graticules and also do a small amount of sub-pixel image processing to compute the position to way below sub-micron levels. The sand paper half of the image can be put through a very simple neural network to extract the exact position (absolute angular position) - after it has learnt the sand papers image signature at each graticule.

 

edit: The attached image I scaled down, it's actually cleaner and sharper than shown (image pixel resolution from the webcam is around 10um from around 1cm away from the ruler/sand paper). Holding the camera steady enough whilst pressing the keyboard key to capture an image was a challenge.

Edited by EarthLife
  • Like 1
Link to comment
Share on other sites

In actual fact, due to sand paper being unique and fixed along it's entire surface, it can be used on it's own (without a coarse graticule scale alongside it) as an absolute position sensor, an extremely accurate one at that.

Once a single run of the strip of sand paper is recorded and learnt by say a neural network (or whatever method is chosen to recognize the unique grit pattern), then by just taking a single image from any point along that strip (or from a circular disc of sandy paper) using the image sensor and feeding that through the image recognition routine (whether that be a neural network or not or a mix of neural and other methods), an exact position can be computed, not just to camera pixel level, but way down in the sub pixel range (as is currently done with image guiding on stars).

The coarse printed linear graticule though will be there with the sand paper at least to start with because it greatly reduces the amount of required image processing, the coarse graticule from a printer gets you to within say 20um (though better really), at which point you bring in the sand paper sampled image to get you right down to the desired position accuracy you're after.

 

1172795648_P1200gritwithscale2.png.186f0c355c1f9515f0b3e4dcef133e57.png

 

This is what the close up images are actually looking at ..

 

1451554404_P600grit(2).thumb.png.42faec00c8f450aeac1e8e69ef1880bb.png

  • Like 1
Link to comment
Share on other sites

Just now, markse68 said:

This is a really cool idea but won’t you need to calibrate it against a very precise high res encoder as whilst you’ll have resolution, will you have accuracy otherwise?

If the scale is rotary disc then that is kind of calibrated from the start - if a disc has say 360 lines around it, one line per degree, then if it expands/contracts with temperature then it will still only have the same 360 equally spaced lines on it, just slightly different diameter than otherwise, angular resolution remains unaffected.

But calibration can be done by following a star through an evening, the stars position will be exact in the sky (they don't tend to go walk abouts), the telescopes position will be static (and known), the position sensors CPU clock will be accurate (GPS time locking is easy - around £8 for a board level GPS receiver and antenna). Using that known data/info from a single night a precise calibration can be done by the software itself., a one time operation  (unless you change something).

 

Link to comment
Share on other sites

Will you be able to track a star using the encoder you’re trying to calibrate though? How fast do you think you’ll be able to do this image processing and feed back the position to your motor controller? Do you think it’ll be able to cope with fast adjustments to motor position to handle seeing wobble and wind gusts?

Link to comment
Share on other sites

3 minutes ago, markse68 said:

Will you be able to track a star using the encoder you’re trying to calibrate though? 

If I use a ras-PI board as the encoder board that shouldn't be a problem.

 

3 minutes ago, markse68 said:

How fast do you think you’ll be able to do this image processing and feed back the position to your motor controller?

Time will tell, I usually get something going then once I have the code doing exactly what's required I then work on optimization etc to speed it all up.

 

3 minutes ago, markse68 said:

Do you think it’ll be able to cope with fast adjustments to motor position to handle seeing wobble and wind gusts?

I should think so yes, those effects are actually very slowly changing external events as far as an controller is concerned. We as squashy/squidgy living beings aren't quite so good/quick at reacting to sudden/random changes.

Developing new tech/prototypes  etc is often quite fascinating/interesting/challenging at times, it's what makes it so. I've been involved with developing cutting edge tech for a few decades now, so this is not really a difficult one to solved.

Link to comment
Share on other sites

1 minute ago, markse68 said:

Ok i’ll stop asking questions now then- sounds like you have a pretty good idea what you’re doing ;) Really looking forward to seeing this progress 👍

No don't stop asking questions, by all means ask anything you like, it's how we all learn from each other (a good development engineer never stops learning or asking questions).

An up and coming (and serious) problem I see coming our way (quickly too) is the young don't appear to have any interest these days in learning how things work or how to go about doing something ourselves or having any kind of hobbies etc. Skills are being lost at a high rate of knots it seems :(

An example .. The school two of my nephews went too (they are 18 and 21 now I think) had no computers or computer classes for pupils at all, - crazy !   result is that their only experience/learnings of computers is playing on their xbox's and finger twitching on their phones all day long, no science learnt what so ever during the easy learning phase of their lives, they have no real-world or science knowledge at all :(

There's a film I quite like called "Idiocracy" that often comes to mind, a fun film but sadly all too real if we're not careful.

  • Like 1
  • Sad 1
Link to comment
Share on other sites

A short length of 100mm*100mm aluminium tube (around £5) as the support frame (for testing at least) ..

 

100294933_DirectDrive3.thumb.png.f7e39021360b1fcea094425eed684658.png

 

Is the axis rod used on mounts normally of a typical diameter ?

Obviously the larger the diameter is the stiffer/stronger the axis is, but just wondered if a preferred size is tended to be used.

Link to comment
Share on other sites

Brilliant project! 
 

On ‘mass produced’ mounts  axis size is dependent on the intended payload, but as you say the bigger the diameter, the stiffer the axis will be. I wouldn’t go below a 1” diameter shaft and 1.5” would be better. Sorry for the imperial units.

Link to comment
Share on other sites

Reminds me of my wheel hub motor in my BionX electric bike.  I had it apart to replace the strain gauge and it sort of looked like that except that the magnets were in the periphery with radial coils.

From little acorns do mighty oaks grow.

  • Like 1
Link to comment
Share on other sites

The ASA DDM60 mount has 60mm dia axis shafts, while the DDM85 has, surprise, 85mm shafts. I have no info on the latest series of DDMs which are spec'd in payload, though the DDM160 had 160mm shafts.

Link to comment
Share on other sites

15 minutes ago, Tomatobro said:

Reminds me of my wheel hub motor in my BionX electric bike.  I had it apart to replace the strain gauge and it sort of looked like that except that the magnets were in the periphery with radial coils.

Yes that's an outrunner motor (magnets around the outside of the stator). I couldn't decide on whether to do a pancake (disc) or out/in runner. The disc type seemed easier to start with.

18 minutes ago, Tomatobro said:

From little acorns do mighty oaks grow.

They do don't they ;)

Link to comment
Share on other sites

6 minutes ago, DaveS said:

The ASA DDM60 mount has 60mm dia axis shafts, while the DDM85 has, surprise, 85mm shafts. I have no info on the latest series of DDMs which are spec'd in payload, though the DDM160 had 160mm shafts.

OK, that's quite big !

Could never justify £6000+ mount for a hobby grade telescope though. Am hoping to create a DIY'able design for around £100'ish for two DD motors and encoders.

Link to comment
Share on other sites

I have been wondering about the process of identifying the axis position.

It strikes me that it could be not at all dissimilar to identifying the location that a camera is pointed at from an image of the sky, by treating the bright points from the sandpaper as "stars", in which case perhaps much the same algorithms might be used?  Hopefully in the case of the sandpaper the search space would be considerably smaller.

James

Link to comment
Share on other sites

23 minutes ago, JamesF said:

I have been wondering about the process of identifying the axis position.

It strikes me that it could be not at all dissimilar to identifying the location that a camera is pointed at from an image of the sky, by treating the bright points from the sandpaper as "stars", in which case perhaps much the same algorithms might be used?  Hopefully in the case of the sandpaper the search space would be considerably smaller.

Yes indeed, though I want to ensure that the routines are fast enough to keep up with rotation. If the motors start rotating at a fast rate then the software only has to keep track of the coarse printed graticule scale, it's only when rotation slow down that the firmware will use the sandpaper pattern to fine turn the exact position computation.

I'm going to initially use a cheap 360deg plastic protrator as the coarse scale as they tend to have at least 1 line per degree, which is enough to get the position down to say an arc-min with a bit of processing. The protrator will be overlayed on top of a round piece of sandpaper.

 

Here's P600 grit (larger pieces than P1200) ..

1703824429_P600gritwithscale.png.893236023cf0c466a73965c2358e50ed.png

 

I just need to pick a suitable image sensor chip (SPI bus or some such serial) that can be configured to take images at a fast rate, a 128*128 pixel sensor should be good to start with I think, together with a fast LED or laser (0.5mW at most) light source as a strobe light for the sensor so that sharp images can be obtained as the motor rotates. So long as I can capture images fast enough to keep track of the coarse line scale it will be fine. If the firmware did loose it's position, it could use the sandpaper to regain absolute position rather than going back to the 360deg index mark. It's really just a basic DSP image processing task.

 

  • Like 1
Link to comment
Share on other sites

1 hour ago, EarthLife said:

es indeed, though I want to ensure that the routines are fast enough to keep up with rotation. If the motors start rotating at a fast rate then the software only has to keep track of the coarse printed graticule scale, it's only when rotation slow down that the firmware will use the sandpaper pattern to fine turn the exact position computation.

It's more complexity in the design obviously, but you could even have some sort of very simple coding for position (Huffman?) just to allow you to identify, say, which eighth or sixteenth of the wheel you're looking at to be easily able to cut down the search space for the sandpaper if you've lost track of where the graticule scale is.

James

  • Like 1
Link to comment
Share on other sites

Interesting!   I have for some months some items in my wishlist.  Time to dive in this project is what keeps me from ordering.   Last thing i remember was planning to stack two harmonic drives to  get to a high resolution…..  
 

Time is an enemy😣.  
 

im going to follow the road of your project
 

€ 139,35 | Harmonische Drive CSF-25 Reducer Lager 20*85*18.5Mm Voor Csf CSG-25 Draaikrans Zengkun China
https://a.aliexpress.com/_vTjDFw

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.