DIY Fork Mount for Widefield Imaging Rig

[Gina]

Pololu do a 128 microstepper  :- AMIS-30543 Stepper Motor Driver Carrier but that is delivery from USA.

[Gina]

I rather wonder if extreme micro-stepping would be accurate though bearing in mind that a microstep represents a pixel in the image and I want to use exposures up to a minute or two.  Resolution is 0.002° per pixel and 1m = 1/60 of 15° = 0.25° which is 125 microsteps.  So I need an accuracy of 1 in 125 (or 250 for 2m exposure) minimum.  Stacking multiple frames can cope with the frames not quite lining up as long as it's not too bad.  But if the mount doesn't track accurately during each exposure the image loses resolution.

Edited June 2, 2018 by Gina

[Horwig]

8 hours ago, Gina said:

Now to look again at calculations of accuracy etc.

To accomplish this resolution a further reduction is required of 0.05625 / 0.002 = 28.125 - say 28:1

Hmm... that's more reduction than the final drive 

What's wrong with doing a belt reduction in two steps? isn't that what Avalon are doing here?

https://astrogarten-shop.de/de/zubehoer/montierungen-zubehoer/avalon/avalon-muno-fast-reverse-montierung-synscan.html#lg=1&slide=1

(OK, they have three stages)

H

Edited June 2, 2018 by Horwig

[Gina]

You mean three altogether including the final drive?  Yes, I think that's what I shall do.  Looks like Avalon have four.  I have plenty of ball bearings and can soon design and make a 3D printed gearbox belt/pulley box.

Edited June 2, 2018 by Gina

[Horwig]

40 minutes ago, Gina said:

You mean three altogether including the final drive?

Having looked at it again, yes indeed, there are FOUR in total

[Gina]

The 80t GT2 pulley seems to be available only from the USA otherwise it's 60t (3.75:1 with 16t) so I've been looking at MotionCo who do a much bigger range than I can find on ebay etc.  Only problem is that the number of sizes with 5mm bore is severely limited.  They offer shims to reduce bore but I don't like that idea as it might produce eccentricity.  Some available combinations give recurring fractions which I don't like. 

It all boils down to just one combination viz. 15t and 120t giving 8:1 ratio.  Keeping to 20t for the final drive gives 25 x 8 x 8 = 1600:1 which is better than needed for a 200mm lens if I use a 0.9° stepper motor.  OTOH I don't think it matters having too much resolution except that it would take longer to slew to the target.  I could use a 1.8° stepper motor (which I have in stock) and still have better resolution than needed.

Anyone see anything wrong with this?  Or have any better suggestions?  Cost £54 - including precision 5mm SS shaft for axles.

Edited June 2, 2018 by Gina

[Gina]

With only 3 of the 15t pulleys in stock I've decided to go ahead and order those bits.  It would be typical if I delayed and then found them gone or at least two of them gone.

[Gina]

As for the Dec drive, I'm thinking 28BYJ-48 mini stepper motor with gearbox.

Stride angle approximately 5.625° / 64 = 0.087890625°.  It's a bit less than that as the gear ratio is slightly less than 64:1.  A pixel movement is 0.002° with around a 100mm lens (105mm).  I may go to 135mm FL lens so something like 0.0015°.  The Dec axis is only needed for slewing to the target and 10px will be near enough (no guiding) giving a required resolution of 0.015° in the Dec axis.

For the final drive a 120mm diameter disc with a 20t GT2 timing pulley will give 10:1 ratio making the motor resolution needed as 0.15°.  Single stepping already gives and accuracy of 0.08° - nearly twice as good as needed so half stepping, which gives higher motor power will be more than adequate.

Is this right?  My brain's overheating ?

LATER :- No it isn't!  Didn't think so.

Edited June 5, 2018 by Gina

[Gina]

The pulleys and belts for the 64:1 reduction drive arrived in the post this morning and look the business.  Next job is to design and print a box to contain it.  Plus to see how it will fit in with the main mount frame.

[Gina]

Think I shall re-check the calculation of the angular resolution at the imaging rig.

1. NEMA17 (or smaller) stepper motor with 1.8° stride angle will be used as reference.
2. With 16x micro-stepping this gives 1.8 / 16 = 0.1125°
3. Final drive is a 20t GT2 pulley and over 300mm drive disc (derived from the size of the fork)
4. If the drive disc is 1000 / π mm diameter = 318.3mm, the circumference is 1000mm.  (Circumference = π x D.)
5. Pitch of GT2 timing belt/pulley = 2.0mm so disc represents 500 teeth.  Ratio is hence 500 / 20 = 25:1
6. The reduction ratio overall is 8 x 8 x 25 = 1600:1.
7. Angular resolution at imaging rig is therefore 0.1125 / 1600 = 0.000070312° = 7 x 10^-5.

I'm now going to re-check the angular resolution needed to resolve the image to an accuracy of one pixel.

1. Pixel size for the ASI1600MM-Cool camera is 3.8µm.
2. Focal length of longest lens I have that I could use in a dual imaging rig is 200mm.
3. By geometry, 1px subtends an angle given by tan(A) = 1px and for these very small angles tan approximates to the angle in radians.
4. So angle in radians = 3.8µm / 200mm = 3.8 / 0.2 x 10^-6 = 19 x 10^-6
5. A radian = 2 x π degrees so angle in degrees = angle in radians x 2 x π
6. Angular resolution required = 19 x 10^-6 x 2 x π = 119.4 x 10^-6°
7. Since we are only thinking rough values we can call this better than 100 x 10^-6 = 10^-4 or 0.0001 degrees.

Conclusion :- Resolution is better than required.  In fact resolution with a 200mm lens is about 0.6 of a pixel.  For my starting point of 55mm lenses it will be a sixth of a pixel.

Edited June 5, 2018 by Gina

[Gina]

Now to check the Dec again which just doesn't seem right.

1. 28BYJ-48 stride angle is approximately 5.625° / 64 = 0.089°
2. The Dec drive disc can be up to 150mm diameter so let's make the calculations easier by choosing a ratio of 10:1.
3. Drive pulley is GT2 20t so the disc wants to have a circumference of 400mm (10 x 20t x pitch of 2mm).
4. A circumference of 400mm gives a diameter of 400 / π = 127.3mm.  That's fine.  Could be bigger if desired.
5. Angular resolution of 28BYJ-48 and 16x microstepping gives 5.625 / (64 x 16)° = 0.0055°
6. From calculations above, one pixel corresponds to 1.2 x 10^-4° for a 200mm lens.
7. We have a resolution of 0.0055° = 5.5 x 10^-3 (or 55 x 10^-4) which would therefore correspond to 55 / 1.2 = 46 pixels.
8. Sensor height = 3520 pixels so 46 pixels is 46 x 100 / 3520 = 1.3% of the height.
9. Since the Dec axis will only be used for slewing to the target, this is quite adequate.  With guiding it wouldn't be.

Edited June 5, 2018 by Gina

[Gina]

Now that I have confirmed the calculations I can continue with the design.

The RA reduction drive first.  There isn't room to have it in a line so will have to be folded back.  This is more compact and needs a smaller container too.

Here is a drawing of the pulleys and belts plus stepper motor.

Edited June 5, 2018 by Gina

[Gina]

A bit more detail of the reduction drive with pulleys, belts, bearings and stepper motor.  Belt tensioners still to add probably on the motor side.

Top view diagram showing the internal workings.

Top view showing the output shaft and ball bearings.

Bottom view with motor and bearings.

 

[Gina]

Belt tensioning wheels.

[Gina]

Here is the fork with RA drive disc.  I plan to print this on my Giant printer when it's back in action since it's too big for my Titan printer.

[Gina]

To fit on the Titan printer, the disc would want to be 290mm instead of 318.3mm and the strengthening webs cutting back to this diameter as shown below on the RHS.  The left shows the current webs.  Resolution would be reduced by 290 / 318.3 which is 91% - not too significant but printing time would be considerably increased due to the smaller nozzle.

[Gina]

This project may be resurrected in the not too distant future.  Not yet though as I have other priorities.  Nothing to stop me thinking, planning and tossing ideas around in my brain and here for discussion.  It goes with my micro-dome observatory project, also in sight of resurrection sometime in the next year or two.

I think I've pretty much decided to go for a single widefield imaging rig to reduce the size and complexity for my first attempt at making a mount and getting it working.

Edited August 18, 2019 by Gina

[Gina]

I'm thinking of a mixture of metal and plastic in the construction.  The most important aspect of a mount to be used for a widefield rig without guiding is polar alignment.  With good PA guiding is not needed and the motion is in RA only - Dec is only used to frame the image.  This results in a much simpler system.  The important thing is that PA is maintained and that is where the metal parts come in, to avoid the creep in plastic.  The other critical requirement is that the mount move in RA at the correct rate but since this is relative after the image is framed, it isn't affected by creep.

One requirement this mount has in being entirely remote controlled is that the PA also has to be remote controlled.  Although this should be a one-off process, to use manual PA would mean having a computer display at the pier to display the images.  Overall, I think remote PA will be the best option.  It just means two more remote controlled stepper motors.

[Gina]

I've been thinking about the main construction again and how massive (or not) to make it.  I already have a massive part construction using heavy duty pillow blocks and aluminium plates.  This seems like over-engineering with a vengeance for a payload of well under a kilogram but that's better than under-engineering and it's there.  Those pillow blocks are over 6" wide.

On 14/07/2017 at 19:48, Gina said:

Here are a couple of photos of the mount sitting on the pillar.  I'm using a scrap part as spacer to raise the bearing plate to about the right angle.

 

[Gina]

On 01/05/2017 at 19:48, Gina said:

I have a pair of very large pillow block ball bearings which I bought for the fold-down flap on my observatory but changed to some lighter weight stainless steel cased bearings.  I get the impression that to use these for the fork mount would be like "using a sledgehammer to crack a nut" but I wonder.  I have these and no current use for them and they fit some aluminium pipe I have.

 

[Gina]

Here's a photo of a redundant aluminium plate with the pillow blocks mounted on it.

[Gina]

That aluminium plate the pillow blocks are attached to is not thick enough - I can bend it.  I do have a thicker plate I can use.

[Gina]

This is the sort of arrangement I have in mind with a hinge between two aluminium plates.  The dome part can be ignored - the new dome will be bigger.  Sorting out the RA drive looks like being "fun"!!

On 19/06/2017 at 17:21, Gina said:

Small change and drive pulleys added.

[skybadger]

I'm looking at this picture and wondering why you are not considering ...

an English mount, with the dome shutter  axis being along the polar axis. 

A very short ra axle , with a big bearing to compensate (skywatcher travel block would do)

One of those offset single arm mounts. 

A coelostat type arrangement to feed a fixed camera where you just need to point a large flat at the sky. 

A very stubby alt-AZ and rotator . If you can do the drive for this you can do the rotator for the camera. Especially if fixed. 

A DC servo motor to drive it. A 2 rpm motor as source will give you awful slew speeds.

I'm considering use of the SW travel head a a coelostat base since it can do 0.5X happily. It just doesn't do slew at all.

Just some thoughts

Mike

 

[Gina]

That's an interesting idea but tricky for a dual imaging rig.  I think I'll stick with a standard design fork mount but I don't think I shall use a 28BYJ-48 stepper motor for the DEC drive as they are very fragile.  Alright for the light-weight job of focussing but ...