Designing new Router called Maximus

[peteeng]

Hi All- I've been researching timing belts and their profiles for linear drives. The Optibelt Design Manual is full of good info. For instance they make an ATL profile which has less clearance and a negative length tolerance. Linear drives need more preload than power drives so the belt is stretched more. The negative tolerance takes this into account. Go to their site and download the manual if you want... Peter

https://www.optibelt.com/en/products...t-alpha-power/

Edit - I have a solution for the spindle clamp. Not detailed yet but i think it will work, see image. Good enough to be able to move onto assembly of the gantry. Have to adjust some of the widths yet so it trams... Peter

[ger21]

It seems to me that after all of the analysis to get the stiffest machine possible, using belt drive negates everything, as they are probably the least stiff method of providing motion.

[mactec54]

Hi All- I've been researching timing belts and their profiles for linear drives. The Optibelt Design Manual is full of good info. For instance they make an ATL profile which has less clearance and a negative length tolerance. Linear drives need more preload than power drives so the belt is stretched more. The negative tolerance takes this into account. Go to their site and download the manual if you want... Peter

https://www.optibelt.com/en/products...t-alpha-power/

Edit - I have a solution for the spindle clamp. Not detailed yet but i think it will work, see image. Good enough to be able to move onto assembly of the gantry. Have to adjust some of the widths yet so it trams... Peter

The GT series Timing belts has the least amount of backlash of all other Timing Belts, using a Timing Belt as a linear drives no matter what type you use, is going to compromise your machine accuracy

[wmgeorge]

A timing belt may be fine for a laser machine but not for a router.

[peteeng]

Hi All - The initial price point for this machine will not wear the cost of the R&P. If someone can point me at a reasonable cost R&P supplier then I'll look at it. Currently very expensive in Oz. Eventually the machine will have both options. Kit machines are cost driven. As long as it has options then it will be fine. Everyone's idea of a "router" is different. Scoot my development machine has made signs, guitar necks and routing jigs, funerary plagues, boxes, trays, cast iron bath stands, car grills, rudders, wing mould plugs, marquetry and furniture. All manner of timber, plywood, mdf and plastic things and the occasional sheet aluminium object within acceptable tolerances for their owners and it uses 16mm T belts on all axes.

Maximus is aimed at the Maker, DIY Hobby market not the commercial market. Maximus is also an enquiry into what's possible, its a virtual project, instead of writing off an idea it needs to be explored and understood properly before being sidelined . Thanks to this forum I have uncovered lots of info that I would not have otherwise. Thanks for watching and commenting it's all appreciated. Thanks again Peter

[peteeng]

Hi All - To continue the belt discussion and put some numbers to it. I prefer numbers to conjecture. I looked up the stiffness of belts. Been looking for a simple table for these for a while. Optibelt you have to calculate them... So Gates has a chart excellent info, see attached. For Maximus I've been thinking about 25mm but from this I think 32mm is the go but 25mm would work (have to look at some accelerations soon as this will be the real decider) . The AT10 and the HTD8 have the same stiffness but the HDT is in 30mm wide. The N34 14mm shaft is 36mm long so as the 32mm pulley hub is 40mm wide it will just work.

Manufacturers recommend AT vs HTD for positional applications. I asked my supplier about ATL belts today but they cannot supply. The ATL series are considerably stiffer (1.57x) and more accurate if you can get them.

So the numbers go like this: 1) tabulate the specific stiffness of belts of interest, use these to calculate the load to stretch a 1000mm belt by 1mm. Turns out this is 122kgf for a AT10 32mm wide. I'm quite happy with this. My current machine is 13kgf so Maximus would be 122/13=13.7x stiffer which is very good I feel.

I also looked at the pulleys available. 15T is the min recommended but 18T is in stock. 18T gives 0.88mm per step at 200 steps/rev. If we ustep them at 4000 steps/rev this gives an accuracy of +/-0.04mm. If we use a 1:3 this is +/-0.013mm which is good. The mechanical resolution is 5% so mechanically can't get better then +/-0.04mm straight so again with a 1:3 this matches. So a gearbox of 1:3 or 1:5 would make it quite accurate and have heaps of grunt. The max belt service tension of the 32mm is 457kgf so I think were Ok with that. They have a service factor of 4x on ultimate.

Cheers Peter

Edit - can't seem to upload docs at the moment will try tomorrow.

Gates Mectrol technical library of Gates

to progress this a little further:
Calcs show we need a force of 20kgf to achieve our accelerations and I have a nominal tool load spec of 20kgf as well . Say we really need 40kgf (inertia, friction etc) Gates recommend a preload of 30% of the service load so a preload of 40/3=13kgf is needed. This is tiny on a 32mm belt so we'll have more than that which is fine. The preload stops the slack side of the belt jumping off the pulleys. So more is a bit better especially if we are trying to hold a position. Next is to put this all together in a more detailed calculation including the rotational bits, friction and sag. Then the gantry drive will be complete...

Say we use 20kgf preload then the stretch will be over 1000mm for a AT10 belt is 0.16mm. So when we set up the belt we mark 1000mm on the belt and stretch it 0.2mm if we can see this!! Maybe set up a dial gauge on the tensioner and measure the full length stretch like when setting a critical bolt by the elongation method.

[peteeng]

Here's the calculation sheet from the prior post still getting very slow response and errors - Peter

Yesterday while working with Scoot I measured its static stiffness. Was 7.45kgf/mm which is 0.07N/um. So Maximus will be 10/0.07=143x stiffer then Scoot. When I designed Scoot I had no stiffness figures. Scoot was designed as a minimal functional prototype so stiffness was not important at the time. But we have moved along from there...

[peteeng]

Hi All - I have done some preliminary calcs to size the motors for the gantry. They do not include the rotational inertias just the translational forces. The summary is that a direct drive N23 or N34 will do the job. A gear reduction is preffered to allow the motor to operate a bit faster hence smoother and to increase the accuracy of the system. The N34 motor due to the shaft size requires a big pulley. This makes the machine speeds a bit fast. I have selected a 14mm shaft but a 1/2" shaft may allow a smaller pulley. These are details to be figured once I get P2 to a better detail level. It's also not unthinkable to use a N17 with a 5:1 or 10:1 gearbox. I have picked 1500mm/s/s as the design acceleration. This is high for a hobby/maker/DIY machine. Scoot runs at 350mm/s/s and it does not seem slow. So back to P2's structure so we can figure out where the motors are going to go....

The N34 motor gives me around 50-25kgf and the N23 gives 44-22kgf (slow speed to mid speed) . They are similar due to the smaller pulley on the N23 making up the difference.. I need 20kgf to accelerate the gantry at 1500mm/s/s and 20kgf nominal tool load at a steady velocity. A 2;1 reduction would double the available forces and account for friction and other losses. Or I could go to a bigger N34 and kill it...A 1840oz.in motor will probably be cheaper then getting a gear box set up.Hmmmm Peter

[peteeng]

HI All- I put an enquiry into Gates Australia today about ATL belts. They do have some and the sales engineer is getting back to me with stock and specs. They can get belts on special order as well. Then I have to work through a distributor to get prices. Cheers Peter

[peteeng]

"All my life's in circles" the song goes- I've built up the gantry with the 20mm rails but turns out these make the gantry rail spacing bigger. This makes the effective spacing of the 20mm cars smaller then the 15mm version if I squeeze it down. Unintended consequences!! So I have to rebuild using the 15mm again. Round we go again.... The current concept is to allow easy access to the bolts for the gantry cars, may rethink that to improver compactness. But thats life for a designer... Peter

Edit - So I have replaced the Z axis bearings with the 15mm and changed over to the 4mm screws. Need to change the nuts as well! The tool axis is only 118mm from the gantry front. This is very close for this type of machine so I have achieved one of the objectives with the design. It will also be a bit closer when I get the 15mm bearings back onto the gantry. Next is to update the Z drive and I can then get serious with the gantry details. So many holes to check!! Takes nearly a week for me to do a hole check usually. But before I do a hole check I have to get a drawing stack and that's a way off yet. Peter

[peteeng]

Hi All - The mornings work includes building the large bearing beam and fixing a few constraint issues within the model. Developmental parametric models usually develop quirks as they get chopped and changed quite a bit. At some point its best to rebuild them to remove these quirks to create a "robust" model. The tool offset has dropped to 112mm which is great. I can see that a top drive may be possible now so I'll have a look at that. I use a process called the "resilient modeling strategy" look it up it's a good system. So were back to using 15mm linear bearings seems it was the right decision in the first place! So a bit of massaging to minimise the drive real estate and then we can get into the gantry details. Maybe even the gantry transverse drive!! I can see a couple of bolt access issues on the Z rail bearings some access holes will be needed...

[davida1234]

Hi Peter, is that just not shown yet or will these two pieces of X-carriage be separate as shown?

[peteeng]

Hi David - Currently they are separate. The initial FEA showed that separate worked so it saves a lot of weight. Plus the large bearing plate warped under load. So this way the large bearing plate is correctly flanged to prevent this. I'll be looking at this design point closer next FE run by doing some eccentric loads to check whether it looks Ok currently just because of a pure push/pull design load. There are two structural loops of the bearing 1) is supported via the Z axis 2) support via the gantry. Perhaps the conventional way (3 loops) is just overkill or redundant.

I've looked at the big picture and the up position looks good. The down is not low enough to do work on the apron so the bench sides will be lowered. Currently Z is 275mm travel. Last modal analysis the Z drive plate wobbled a bit and may require a support bearing on the top gantry rail. Next FE round we'll look at that as well.

I've decided to push thru and get the transverse gantry drive done before I do more FE work. I talk in long and transverse because I have two machines and their axes are different on each. I defined the axes by the way my laptop is oriented so the axes are wysiwig to the UCCNC and CAD screen.

Cheers Peter S

[peteeng]

Hi All - Gates Australia have responded about the ATL belts. They would have to get them from overseas and I have to put the enquiry in through their local distributors. Sounds way too hard so I'll go with AT belts from my usual supplier shame I was hopeful on that one. Peter

[peteeng]

Hi All - Trying to end the 25mm vs 32mm belt internal debate. The list price for 25mm belt is $117/m and the 32mm is $143/m so not much diff there. The pulleys however are 18T 25mm $9.60 and 18T 32mm are $25.70 so a bit of a diff when you have 6 on a machine.

The AT5 vs AT10 debate is easy. The AT5 is 25mm width max and its specific stiffness is 17600N/mm width. The AT10 Cspec is 37410N/mm width. So the AT10 is 2.1x stiffer. Problem with the AT10 is the recommended minimum pulley diameter is 56mm which makes the accuracy a bit big for my liking. The PCD of an 18T AT10 is 57mm, makes 1rev 179mm so one step is 179/200= 0.89mm and the motors are 5% tolerance so 0.89*0.05= +/-0.045mm say +/-0.05mm which becomes the mechanical tolerance of the system. When you use usteps these get corrected every time you come to a full step. So I use enough usteps to get a bit better than the mechanical accuracy, no point in going further unless the motion improves.

Prefer better then that which means using a gearbox. I looked around for a Nema34 400step motor with no luck. Prefer the N34 as it has a 1/2" or 14mm shaft so its strong and stiff. A 14mm shaft has a nominal 40kgf radial strength. If I use a 16T pulley then I get +/-0.04mm not much diff either. If I'm happy with a spec of +/-0.1mm then this is fine, or +/-0.05mm its fine but if I'm aiming at +/-0.01mm then gear reduction here we come.

Now if I drop to a AT5 16T I halve the tolerance but also half the stiffness and I think stiffness holds trumps as usual. Peter

[peteeng]

I looked at precision gearboxes. They have a backlash of 15arcmin. This works out at 0.002mm with the 18T AT10 as the output using a 10:1 gearbox. So the 10:1 will give me a +/-0.0045 mechanical tolerance with a 0.002mm backlash. I'll have to think about that. It's a N23 motor so we'd have up to 12Nm torque. heaps for the job. I have a couple here on my benchtop machine... Peter

I've had a think and the gearbox is the go (maybe). The 10:1 gives a tolerance of =/-0.005mm and the backlash adds 0.002mm which should always be within 0.01mm. Plus the software should account for the backlash. So onward... my only concern is midrange vibration. One of the units I have here vibrates on test, yet doesn't on a machine... Hmmmm Peter

Edit - I dropped the sides of the bench 100mm so now the tool can reach under the base, so back to detailing the mechanics.

[peteeng]

Hi All - I decided to check the inertia of a motor in regards to the torque required. Turns out on a large N34 motor it can be neglected at accelerations of 1500mm/s/s. That's good means the current calcs hold so far. Unless someone can find error in the calc attached?? So we just need to look at the translating masses. Peter

I've put some of Brevis's parts as the transverse drive to get a picture of how it looks. I'd like to do it the same as Scoot. This axis has been the most accurate and less fussy over the last 2 years of use. If I use the gantry tabs I won't be able to cater for N23 and N34 motors. But I also prefer not to use a bracket. Toss a coin. If I use the Brevis system I get a motor mount and a return pulley/tensioner in the same part. Plus I can have a N23 and a N34 version for Makers to choose from. I also have to consider the Light version which would use the smaller motor. If I use a bracket it's also easier to get all the pulleys and mounts in line, the better the alignment of the belt the better. So early days on this bit. Plus I have to look at how a rack will work. There is enough real estate on the back of the Z drive plate to mount this so that looks good.

More thought bubbles needed. Peter

[pippin88]

I've lost track of the width of the gantry.

If it is only 1400mm or so, a ball screw is fine. Just use 10mm plus pitch.

[peteeng]

Hi All - Spent a bit of time looking at the gantry transverse drive. Need to decide on the motor or motor gearbox very soon. Looks like I have to use a 25mm belt due to shaft lengths, even if I lathe the hub off the 32mm its still too wide for the shaft. maybe a 6mm overhang is OK... Or I build a custom gearbox with a suitable shaft. Prefer not to go that way so 25mm AT10 is it for the moment. Now we are into details the compromises will start.

Got quotes on 25 & 32mm belting and pulleys today looks good. Not many small 25mm AT10 pulleys in Oz though. Where I wanted to put the motor hence belt, cut into the Z drive plate so I had to reverse the motor spot. Not to bad. The belt is 30mm from the back of the plate which is a good dim for the anchor bracket. Next up is to fix the Z drive bearings and massage this area a little to lose some width.

I added a support bearing to the back of the Z drive plate. Last modal analysis this was the spot that vibrated so this will help there. It's easier to delete a part from the simulation then add one in to see the difference. So I'll check the torque calculations to clarify the motor size and then freeze the motor spec. Peter

[peteeng]

Hi Pippin - Yes this has crossed my mind. Better have a look at that. Likely to be around 1600mm Peter Just checked 1650mm needed. Peter