High Spec Computerized Equatorial Mount : building from scratch.

[gajjer]

As an alternative to the Asduino, have a look at the ESP8266, which can be programmed with the Arduino IDE and has WiFi capability. Dirt cheep too and a fast processor.

Just a thought.

cheers

gaj

[FrenchyArnaud]

6 hours ago, windjammer said:

So I changed to a floating caliper type brake, much happier. 

I will look into this. Right now the only three things that are pretty well established as for the mechanical design are : 
1) no turret but a wedge (much easier to draw)
2) axis based on car bearings.
3) no shafts (except for the weights but they are not really part of the mechanics) 

And yes, those are the kind of thing I am thinking about. Specifically, these have a very interesting shape and price point : 
For BMW X5 E70, F15, F85 2006-2018 Front Wheel Bearing Kit

The fact that they are almost flush, with a very symetrical drilling design, and allow to be used as an axis for the counterweight shaft really speaks to me. Also, because the ABS support is almost straight I am thinking I might be able to glue a belt around it, dips on the outside to convert it de facto in the last stage of pulley.  A belt glued there would not be sujected to any traction but only to lateral forces and due to the pressure of the actual driving belt would very likely never move even with just a tap of double face tape! Let alone proper gluing with cyano or CT10. simple spacer on the visible part of the picture will make sure there is no friction across the support plate. And OBVIOUSLY these things are designed for non-existant play and immense load tolerance. So... yes, that's my idea. 
 

6 hours ago, windjammer said:

Makes one wonder if you need any shafts at all.  Just bolt two together at right angles and off you go (I simplify :)).  You must supply some sketches!

I don't intend to use shafts at all. I am thinking along the lines of a welded (with my about to be acquired welding skills 😁) square tubes squeleton on each of these bearing, one fixed to the wedge and carying electronics + stepper/gearbox on one side and carrying the other one with the second stepper/gearbox on the other side, itself carrying the second bearing, which will in turn carry the dovetail. So, no need for shafts at all. These will provide all the rigidity and articulations needed. Again, if they can carry a SUV at 120mph I doubt they would need help to carry an OTA at 15deg/hour...

6 hours ago, windjammer said:

Have you any ideas about getting rid of the meridian flip ?  Some GEMs seem to have managed that.  A neat trick if you can pull it off with your about to be acquired welding skills.

I did not go that far yet  On the bench, the firmware nicely goes to home and pauses before flipping which makes for very smooth operation so I was happy with that but now that you have mentioned it... I need to think about it

 

 

 

 

 

[FrenchyArnaud]

7 hours ago, windjammer said:

They look very interesting indeed.  Makes one wonder if you need any shafts at all.  Just bolt two together at right angles and off you go (I simplify :)).  You must supply some sketches!

Here is a very crude free hand representation of what I am thinking of at that stage (just did that in 45secs right now - i KNOW it's nowhere near to scale!)

But that might make what I said clearer. Basically, 2 welded "cages" assembled by one of these bearings,  nemas and gearbox within these cages, and the top one carrying the second bearing carrying the dovetail. So, a bit like you said, "just bolt them together" - not far from what I was seeing!

I am trying to figure a simpler way pier > RA assembly because I think there should be enough space to put a lot in the opening of the wedge and simplify a lot the part wedge+RA, doing it a single piece. Still unclear how to set az for PA, I am thinking maybe a second brake disk freespining on top of the one already existing on top of the pier. 

In any case, this sort of "open core"  assembly would take like an hour to build perfectly square to someone who knows about welding and grinding. It looks complex to me because I never did it but it seems to me the design is intresinquely very simple with very limited flex point and potential failure points. And, with proper gearboxes and 2" square tubes, could easily carry any conceivable OTA. 

 

[RobinH]

an alternative processor is ES32, same price as arduino, very compatible and uses arduino IDE, Wifi, and fast.  The only down side is it is wider so needs 2 breadboards. 

[windjammer]

>>Still unclear how to set az for PA, I am thinking maybe a second brake disk freespining on top of the one already existing on top of the pier. 

Yes, with a tangent arm to make adjustments...

 

[FrenchyArnaud]

@windjammer

Wow that is one serious piece of kit with a few interesting points for me. 

The most obvious one is that you have set your RA shaft 90deg of how i envisioned it. I mean that your assembly is parallel to the wedge, not at a right angle to it and now that i have seen it it makes perfect sense...

Also I note that you welded jack on that..Everything is bolted. That's an awful lot of bolts, some pretty complex. So 3 questions : 

Aren't you plagued by loosening bolts?

Does it have any effect on the overall rigidity? (the bores have to be bigger than the bolts!)

Where did you find these long bolts that you can bolt into at a right angle, I never saw these before.

Also, I considered these pillowed bearings, how to they fare play-wise? 

Also, most of it seems to be aluminium rather than steel, any particular reason? It sounds much more expensive 🙂

Is it an OnStep system too? 

[windjammer]

>>you have set your RA shaft 90deg of how i envisioned it.

Yes - no choice really, the bearing blocks are so large.  It is funny how are there are so many ways of arranging the same bits to make a GEM. Flange bearing blocks are very similar to your wheel mounts but would need a shaft, so I think you are onto something here with single ended assemblies.  The wedge could be made quite narrow with small format bearing assemblies - the wedge is the thing that gets caught up in the meridian flip, so the smaller the better.

>>Also, I considered these pillowed bearings, how to they fare play-wise? 

Perfect as far as I can tell.  No mount glitches I can trace to the bearings.  The wearing-in process is odd - they don't come aligned, so when you first put a shaft though a pair of them the blocks have to be very loose and gradually tightened down.  The bearings gradually (and noisily) move in the housings until they line up.  After a while of running the shafts glide freely.  I had to watch a few youtube videos of burly blokes using huge levers to adjust bearing blocks before I got my head round it.

>>Everything is bolted.

Yes, a bolt up top to bottom.  Actually, more properly a screw up!  That thing has more or less every thread known to man tapped into it, M1.4 to M16.  So the bolts and screws have spring or lock washers to keep them tight.  There isn't much vibration going on a big scale to work things loose.  Motor couplings and grub screws get a housekeeping check every so often.  The trick to a tight thread in soft metals like Aluminium is to go straight in with the 2nd cut tap and skip first cut and plug taps.  

On rigidity, the pillar could be better.  The 4 pipes are secured top and bottom with barrel nuts and its one of those things you make and find out can't be assembled!  I think 3 pipes might be better with cross bracing, or just get a big fat pipe.  But it works OK and bigger fish to fry.

>>Where did you find these long bolts 

Do you mean the long nuts ? e-bay is good for all of them.  The M8/M10 and M12s are usually used for suspended floors and ceilings, so very cheap.  The M16s I think are actually track rod adjusters from the auto world.  M6 down to M3 most often spacers in electronic assemblies.  The tapped right angle hole in them is the bit you have to do - they are very handy.

>>aluminium rather than steel, any particular reason?

Corrosion and weight if you don't need the strength of steel.  The Al bits are as bright now as the day they went on.  The steel parts and gears I brush with a thick gear oil every so often to keep rust at bay.  As you know there are huge forces working when the scope is loaded up and everything will flex at some point.  Most  bracketry is non critical so Al will do.  Steel right angle is the strongest stuff I know.  Motor and gear mounts are the most critical - if they flex they wind up like springs rather than turning the shafts, until the stored energy lets go and moves the shaft.  Not a smooth motion.

I picked up the thick wedge plates for 50quid at an engineering show.  Humungous pillar bottom plate was £70 from ebay (Droitwich Aluminium I think) - it was a hoot machining that.  The large round was an off-cut (!) from an ebay someone.

>>Is it an OnStep system too? 

No, too far off-piste for that - not really compatible with anything.   As you probably gathered from earlier posts I wasn't able to reconcile tracking resolution with a fast slew speed - so each axis has two motors: stepper tracking motor and a DC slew motor.  There is a servo that moves an idler gear to engage stepper or DC motor into the train as needed.  It has separate encoders on the axes but in the process of debugging my arduino project I found the knack of star hopping, even to invisible targets. Surprisingly easy to do and great fun, so the goto stuff and alignment faff is not attractive.

The only mount oriented computer stuff are the PHD guide cameras - ZWOs with their ST4 guide interface.  ST4 is very simple to interface motor drives to, just 4 active-lo lines for RA/DEC fast/slow and if nothing just default motion (RA) or stop (DEC).  With ST4 it has automatic adaptive PEC, an anti-vibration shutter and guidescope flex correction.  This is where all the action is in my view - getting 50kg to single and sub arcsecond rms in both axes and being consistent enough for 10, 15 and 20 minute exposures running all night.  Still working on the all atmosphere global cloud zapper...

I have some ideas on the wedge:

For the hinge I have an M16 rod running in M16 long nuts attached to the wedge plates.  Doing it again I would look at, say, 2 inch pipe attached with U clamps to the top plate and sitting on V blocks or similar attached to the bottom plate.  I would also put side reinforcements on the wedge to stop twisting.  For the elevation adjustment there is a very elegant arrangement going around using a pushrod and a dividing plate - don't know any details but looks solid.  

Looking at commercial mounts their profile is narrow so the meridian flip can be delayed as long as possible up to south and restarted as soon as possible after south.   With your wheel bearing assemblies you could get away with a narrower wedge than the plumb blocks, but a narrower wedge and thus a shorter hinge might be less stable.

Hope this is useful

Simon

[FrenchyArnaud]

55 minutes ago, windjammer said:

Hope this is useful

This is MORE than useful, there is more to take on from these 3 pics and associated description than in all of my mechanical research so far... If I dared I'd ask for an extra picture of the front of the RA assembly - basically the mechanics on the top part of the wedge to have a better view of how you geared/aligned motor, gearbox and shaft to the dec assembly. 

And if I was a real pickle i'd ask for a close up of your caliper brake... 

But as I am not a drag I will not ask 🤭

Sidenote: that's a very impressive bit of kit, seriously. How long to put it together? Your day job is in mechanics or engineering? 

 

[windjammer]

I will hunt out pics and post.  Check out this wonderful project posted here a few years ago - similar geometry to you I think

https://stargazerslounge.com/topic/347485-a-very-large-goto-gem-built-with-diy-techniques/

 

[FrenchyArnaud]

Well, it's on its way so now waiting... As this will be the most defining part of the mechanical asselmbly and might even be part of the gearbox, I am pausing design and waiting for it to arrive so as to work from actual, measured dimensions. 

[windjammer]

Here are some pics for the RA assembly and connection to DEC.  I will post on other questions tomorrow!  I hope this is useful.

1. internals of the RA shaft - M16 rod, lock nuts at LHS, and travelling nut at RHS.  Two rawl nuts placed face to face in between. Travelling nut has a screw in it that engages with the internal slot in the shaft shown in 2.  As you turn the M16 stud, the travelling nut moves down the stud and forces the leaves of the rawl nuts to open and bite into the inside of the shaft, fixing the M16 stud to the shaft,  Centreing disks at each end keep the stud centred in the shaft.

2. Interior view of shaft showing internal packing and slot to engage travelling nut.

3. RA final drive gear, packing pieces and DEC plate.  Packing pieces locate between gear and DEC plate to accommodate RA Jesus nut.

4. Machined final gear and packing pieces. Test stud in the gear shows where the RA shaft stud will come out.

5&6. Holes in the final gear to take a locating key for the RA shaft.

7. Bolting the RA gear to the shaft stud - 200 grunts.

8. Close up of attached gear and locating key.

9. Bolting DEC plate onto RA packing plate.

10. DEC bearings and test shaft attached.

Pics 9 and 10 give a view of the RA brake mark1 - similar to your idea.  A curved clamp, fixed at one side and a tensioning bolt the other side.  Tighten the bolt and the clamp bears down on the shaft.  The bearings did not like this at all - so made it floating (will describe later).

Simon

 

 

[gajjer]

If you do look at the ESP8266 or the ESP32, I would suggest either the ESP8266 NodeMCU variant or the ESP32 DevKit - they both come in a bewildering assortment, but those two have a good mix of I/O.

Also, I would recommend that you look at the excellent tutorials at https//randomnerdtutorials.com - very full descriptions and clear.

I have a small workshop with mini lathe and mini mill and I look on at what you are doing with envy.

Good luck.

gaj

 

 

[FrenchyArnaud]

On 02/04/2023 at 01:42, windjammer said:

Here are some pics for the RA assembly and connection to DEC.  I will post on other questions tomorrow!  I hope this is useful.

A LOT to take on from this and more good ideas I will shamelessly steal than I dare to admit 🤭 Really very useful and very much appreciated. 

 

My first axis was delivered today (front wheel bearing assembly bmw x5) and immediately several remarks. First : it's an absolute BEAST. It scales at 4.4kg - and I'll need 2, so with bolts etc, that's 10kg right there (22 pounds)

2) There is way more friction/stiction than I envisionned. It takes about 550g of force to get it moving. As the part where I will apply the drive is about 10cm in diameter, so 5cm away from the center, that is about 2.5Ncm required just to get it moving - I'll come back to this in a moment. 

3) There is ZERO play. At all. The thing is absurdly stiff on all axis and ONLY rotates. I am 101% confident it can be loaded as a cantilever with 100kg of telescope and never ever flex by the width of a hair. I am absolutely certain that any flex will come from the supporting frame and never from the bearing in a million year. You have to hold it to realize how stupidly heavy duty it is by our standards. 

4) the middle part, that rotates (in between the plates) has a perfectly cylindrical section of about 15mm and a diameter of 98.5mm. So i think I'll install a reversed 11mm gt2. I could have done the maths but i just sacrified on of my gt2 belts and, put inside out, that's exactly 160teeth with a neat seam! If the last pulley is a 12teeth that's 13.3:1 on top of the gearbox, or 10:1 with a 16teeth. 

I might go for gt3 but i don't have one here to test (the ratios obviously would stay the same)

5) the bearing is PERFECTLY smooth. It is impossible to detect any ball wobble, sticky point etc... 

6) The torque issue! 

Basic maths problem... Given a Torque T at the motor, a loss of torque L due to microstepping, a total distance D between axis and an efficiency E due to mechanical imperfections, what is the force ACTUALLY applied to this axis? 

T: 44Ncm at full step, 6Ncm step32 and 3Ncm step64. 

😧D : estimated 50cm

E: say 50% for safety

Gearbox : hypothesis 720:1

That gives for step 32: 

(6*720*50%/50cm) - 2.5Ncm of known friction = 41Ncm. 

At step 64 that's 20Ncm - 0.2Nm. 

Apparently you need a minimum of about 10% of the inertia moment to induce smooth motion of the bearing which means that at step 64 the heaviest load is around 40kg with a scope 1m long. (half of the length for the inertia moment) but 50% is the goal. 

So the system will be, even with a very low efficiency, vastly sufficient even for massive OTAs but step32 would eliminate any limit (given that I cannot conceive of have ever heard of an 80kg ota) and would give an upper weight limit at 50% torque (rather than 10)  of 16kg - and a bit more if the gearbox can have an efficiency of more than 50%. At 25% torque vs moment, that would be 30kg+ and should be enough for any conceivable "normal" OTA. 

So I am now looking at 925:1 @32usteps because that would give the torque reserve i want with a theoretical resolution of 0.11"/actuation and a payload capacity of about 30kg and up to 50kg.

720:1 @64 steps gives 0.07" but then I'd be limited to about 12kg for ideal torque reserve and 25kg at a push.

Besides, I can flash the microcontroller later for 64ustep even with the 925:1 gearbox which would give 0.05" if it proves needed. But 925:1+64usteps would give ridiculously low slew speeds in the region of .25deg per sec (0.5deg per sec is already stupidly long horizon to horizon, about 6 minutes) 

Anyway! 

Having the actual part in hand has answered a lot of questions and dramatically narrowed down the gear options. And crucially it confirmed it defo will be a mechanically sound concept. 

 

 

[windjammer]

Having metal in the hand concentrates the mind!  I need to think some more about the details in your post (more diagrams would help!) but moving up to 925:1 sounds sensible from a torque point of view, but as you say, reconciling the slew rate is a sod.  If I understand right, you propose 50cm between axis and centre of the gearbox - that sounds quite wide, so wedge would be wide also as belts are transverse across the wedge ?  If so, collision issues near meridian ?  Re weight - the DEC axis has to be counterbalanced, so that doubles (conservative) the load.  The bearings sound utterly solid! - and the frames, wedge and pier are going to be just as important to carry through the whole design.

You asked >>better view of how you geared/aligned motor, gearbox and shaft

1. shelf to bring gears and motor to centre height

2. motor shelf

3. 1296:1 gearbox, key chuck to disengage, 4:1 final drive coupling

4. stepper, slew DC motor, transfer gearbox, servo actuator

5. later developments

6. upgraded transfer gearbox (DEC).  RA is more beefy - cogging on the DC motor affects stepper 

7. DEC (RA similar) as was ca 6 months ago.  Pickup gear opposite the motor drive for angle sensor. Updated brake central between plumb blocks.

I'll post on the brake details next email.

Simon

[windjammer]

Details on floating caliper brake

1. component parts.  A clamp is in two halves that go around the shaft.  The clamp is free to rotate with the shaft until it meets a sprung stop on one side, and an adjustable fixed stop on the other.  The hinged side of the clamp carries two long studs that screw down to the base plate and serve as fixed stops to limit the travel in that direction.  The other side of the clamp has a foot that bears down on a sprung plate, with adjustable stops to limit the travel in the other direction.

The clamp is lined with oil soaked leather for a smooth motion.  The clamping tension bolt is adjusted for (1) just hold any imbalance in the axis and (2) just start to slip before the springs bottom out, and (3) not provide so much force as to strip the gearbox.  The bolt can be tightened down to lock the shaft completely if required. The clamp only constrains the shaft rotation and applies no other lateral or vertical forces on the bearings.

The purpose of the spring is to take up backlash from the straight cut gears - important for DEC guiding.  Approach the target in the direction that compresses the spring, and DEC guide adjustments into the backlash void will be powered by the spring relaxing.  So you get a nice star cross pattern, that works over a field of view or more. Not so important for RA which always guides in one direction, and in this setup the stops either side are adjusted for no gap.  

The springs do not help backlash when slewing outside of the spring range, but a fast slew rate makes this less of an issue.

2. RA close up. Fixed stop and sprung stop.

3. RA from the other side - spring stop

4. DEC brake

 Food for your design - will your hubs take a splined shaft ?  If necessary you could have a short single ended shaft coming out to take extra stuff like brakes.

Simon

[windjammer]

>>How long to put it together?

I bought the plumb blocks April '15, finished (ha!) mechanics March '16, and fabricated motor drives for first light September '16.  But it never stops accreting stuff - really needs a 'planned maintenance' period now to strip down, and do a clean rebuild. But astronomy gets in the way!

A few oily gearbox pics here to show you what are missing with a belt gearbox...

 

 

 

[FrenchyArnaud]

@windjammer
Well Simon this is a LOT to process 🤩I will be back to you shortly with about 624 questions (+/-25%), but one thing that immediately stroke me is this : 

The design speaks to me for a lot of reasons (and mostly because I drew something very similar before opting for 100% belts throughout) but I fell of my chair when I realized both motors are actually geared to PLASTIC cogs ??? Are they some special vynil or specific materal, because I discarded the idea of plastic cogs with no second thought, absolutely certain that there was no way in this universe that 4mm thick plastic cogs could transmit any sort of torque without stripping the teeth clean in seconds. And clearly I was wrong. Don't they wear at all, warp, fracture? 

[windjammer]

The forces at the motor end of the string are small - they feed into a 6000:1 gear train which does the heavy lifting.  You could stop the motion at the motor with your finger, at the other end it would have your arm off!  These gears are actually quite tough - a bit of aluminium U channel and you can make a tight little gear box for pennies.  The Mod0.5 compound gears come in 50:10 and 48:12 - if you need something with more grunt you can mix with metal gears.  The plastic Mod1 worm gears are also worth a look: