Designing new Router called Maximus

[peteeng]

Hi All - Time for a quick review. I think I'm at a tipping point in the design. I am designing a heavy duty Maker, DIY, Hobby kit machine. There are a few guiding themes:

1) Maximise stiffness. In the beginning I did not have a real stiffness value to work with but the forum provided that information and 10N/um static was selected. This is very stiff for this type of machine but we need to aim high
2) The design load has been 1000kgf to produce a deflection of >1mm see point No1. Plus it would be good to have an accuracy of better then 0.01mm. I use accuracy in a loose way as we need to talk about precision and accuracy carefully
3) Its intended to be a one sheet machine but the machine footprint is always a problem for this type of customer. Garages aren't big enough usually
4) To minimise the distance of the tool from the gantry to improve stiffness. Any cantilever is an issue with accuracy and vibration
5) Economy is key. Makers are happy to spend money on machines over time so provide a development pathway if possible. But the initial cost is always a hurdle
6) The Maker will build the bench in metal or timber
7) A little discussed point is the bearing arrangement. This is nearly an arbitrary decision but forms the foundation of the machine. The bearing arrangement was selected to allow easy access for install/removal of the Z axis. It has worked out that the gantry bearings will stay on to remove the Z axis. Sometimes it's good to be able to slide the Z Axis off in total. But that won't work here. So although the bearing arrangement seems trivial it's actually very important to realise a good machine. Started with 15mm cars went to 20mm and had to come back as 15mm were more compact and to achieve No4
8) I've reviewed most of these as I've gone along, now we are at a point where costs and parts availability kick in, compromises are starting to be made... Compromise is not used in a bad way, it's the inevitable process to getting the cost/performance balance as best as possible. I have full control over things I design but std parts like motors with set shaft lengths can create havoc with the wish list.

Round 2 - For me R2 has been about getting the Z Axis and transverse drive sorted and I think I'm close to that. R1 and R2 have been form finding exercises for the parts geometry and limits of the machine. Some call this the Design Space. So Maximus' design space is pretty well defined now. I have been pushing the design from the tool backward to the bench, now it's time to pull the design from the drives to the tool and do some reality checks. Cost and availability issues have kicked in so this means the design is much closer to reality.

Thanks to everyone that has inputted so far it's all been helpful. Peter

[peteeng]

Hi All - Been looking at the belt stretch thing. If I use an AT10 25mm belt on the transverse I can expect a max stretch of 0.3mm at 1500mm/s/s. This would be in air during a rapid so no harm done. If we are doing a very heavy cut at 20kgf then the stretch is 0.15mm and since this is a roughing cut at least 1-2mm from the finish I'm OK with that and if we are doing a finishing cut at 5kgf its 0.04mm or 0.029mm with a 32mm belt. The belt stretch is added to the machine deflection. I'm OK with these so far.

1000kgf = 1mm so 10kgf=0.01mm machine deflection

so finishing would be 0.005+0.03=0.035mm max error (0.0014") OK for a Maker machine and better then my competitors specs. eg shopbot full sheet machines have a no load accuracy of 0.08mm (using R&P) so the loaded accuracy will be >0.08mm. I'll model the pulleys and see how I can get them onto the motors. From prior work if I use a direct drive the mechanical error is +/-0.04mm so these tie together well. These all improve with a gearbox... Hmmmm...

I've seen much lesser machines claim to be much better then this so I think its OK to progress.

By the way this is in the centre of the travel. As the axis gets closer to the sides the stiffness doubles.

I could use a U drive to halve the belt length but I have a few issues with U drives:
1) Mounting two large motors/gearboxes on the gantry adds quite a bit of weight to them, this is acceptable if I had to. Placing the motors on the bench is easy for the wiring and easy on the motors
2) The U drive flexs the belt in reverse. If you want to break a paper clip quick bend it both ways a few times and snap. But this also flexs the cover and plastic suffers from viscoelastic effects and every time the belt runs over and back thru the U a little wave of plastic and energy is associated with this that contributes to belt issues. Hard to put a number on it but KISS says a single run is mechanically reliable and predictable.
3) The cost of idlers and associated brackets is more then the return run of the belt... $$$ count in the end for a production machine

Peter

edit - I found a chart about sizing belts and the size selected for the duty is way down at its bottom end of service so it will be very reliable as expected.

[coinman]

Peter... Hello from Oregon, USA I just wanted to say that I think you are doing outstanding work on this router, and I can tell you that many of us will support you in purchasing plans or parts when ever the project is completed......Thank you again for putting in these long hours to bring us this great stuff.. Really excellent work...

Peter from Oregon, the Coinman

[peteeng]

Hi All - A quick review of some machine specs from various web pages.

CNCrouterparts Pro 4x8 R&P drive
resolution +/-0.0127mm
repeatability +/-0.05mm
accuracy +/-0.127mm

Shopbot PRSalpha R&P drive
resolution +/-0.0127mm
accuracy +/-0.076mm

Maximus
I'll just use accuracy for now :

Using AT10-15T +/-0.042mm use a 10:1 gearbox and its 0.0042mm, resolution depends on the usteps chosen
Using AT10-18T +/-0.05mm use 10:1 gearbox and its 0.005mm

So I'm happy I'm in the ballpark with these. I also wondered what the belt stiffness looked like physically so I converted the specific stiffness to a rod size. The AT10-25mm belt would be a steel rod 1/8" diameter or 3.1mm diameter.

[peteeng]

Hi All - I have reduced the drives to two possibilities

1) a direct drive using a Nema34 motor. Same style as scoot which has given good service for the last 2 years. Scoot doesn't like cutting slow (under 1000mm/min) and due to the AT10 18T big pinion slow would be 1500-2000 in Maximus' case. But it would go fast!! who wants to cut slow?? The accelerations at 1500mm/s/s (Finertial 6kgf) are not the max load a nominal cutting force of 20kgf is which requires a torque of 6Nm so a medium size N34 will do the job or lash out on a big 12Nm one for the transverse and two mediums for the longitudinal axis.

a direct drive gives 0.05mm accuracy and a mid speed of 45m/min at 250rpm, very fast... I like it because it follows KISS and Scoot has worked really well. If you need speed this is the puppy...

2) Use a 5:1 gearbox and the accuracy becomes 0.01mm with some backlash of 0.002mm (15arcmin for the gearbox spec) even use a 10:1

speed range is 9m/min midrange at 250rpm so speed would be 18m/mm plus maybe

So now I have a question for you people out there. I have just started using gearboxes like these, are they good over a few years? https://www.omc-stepperonline.com/pr...2804s-hg5.html

My latest benchtop machine I have used these so don't have much time on them yet. The gearbox uses a NEMA 23 mount so that's good as I also have to consider the light version. Next is to design a motor bracket that will accept N34 and N23 motors for this application. Looks like I can get the 32mm wide belt to fit the motor shaft with a small 5mm overhang and turn off the hub. Have to decide if the 32mm vs the 25mm is worth it. 25mm is nicer on the shaft, Stiffness wins so 32mm it is.

Will get back to the Z axis drive now while my thoughts settle on these numbers... Peter

[peteeng]

Hi All - Working through some motor details. Those N34 suckers are heavy!! Will have to put an extra support on it somehow I think. Here's what the motor mounts will look like so far. The gearbox's look good at 1.5kg and the N34 starts at 3.4kg...Peter

[ger21]

I've got some similar chinese gearboxes for AC servos, but haven't used them yet.

I'd go with 10:1, which would give you 18mm travel per rev, which is a good number for stepper motors, giving you the best balance of speed and revolution.

What would the actual backlash be with 15arc min in the gearbox?

[peteeng]

Hi Gerry - Pulley is AT10 18T with a PCD of 57.3mm and a perimeter of 180mm. By the way the PCD of a belt is the centreline of the cords. So 15arcmin is 180mm/360deg/60min*15= 0.125mm. The gearbox sheet says the life is 20,000hrs we'll find out...

I'm thinking the 10:1 as well. Will have a play on the bench machine to get a feel for it. If I use a gearbox on the Z as well it won't fall under no power.

So what would the top rpm of a N34 be? Peter

I corrected the calc via Davids next comment

[davida1234]

If it is a circle with 57.3mm diameter, it has a circumference of 180mm.

If the domination is in arcmin, it is 1/60 of 1 degree.

One degree is 1/360 of a circle which, in this case, has a 180mm circumference.

Therefor, it is 180 / 60 / 360 = 0.0083mm backlash in linear movement.

Marketing accuracy and any additional backlash in drive train has to be added to that.

[peteeng]

Hi David - Thanks for the pickup, you are right to correct, thanks I did arcsecs. So the backlash is speced as <15arcmins. So backlash is 180/360/60*15= <0.125mm By the way you use the term "Marketing Accuracy" perhaps marketing optimism is a better term Peter

Hi Gerry- when I'm finishing a mould or plug I need speed, 1mm step overs means there's a lot of toolpath to cover!! so maybe the 5:1 is the go. No matter, I think the gearbox wins even though the backlash is big. Peter

[peteeng]

Hi All - this morning I looked at the transverse gantry drive mounts. To do the motor vertical or horizontal. If I go with the N34 motors they are a huge lump of metal and if I mount them horizontal on the front gantry plate I think they may vibrate. So I looked at them vertical on the rear plate. If on the front, the belt is 50mm from the Z plate, if vert its 68mm so the bracket will need to be a bit stiffer. Better finalise the Z drive first.

On Scoot they were on "ears" on the front plate. This reduced the part count and has worked well. But Max-H has to have motor options so the motor mount should be separate. Although my current thinking is to stay away from the N34 motor (too heavy, too fast in the direct drive mode) so N23s are good for the gearbox and and or a direct drive for the light version. Peter

[ger21]

So what would the top rpm of a N34 be?

Depends on the motors inductance, the voltage your running them at, and the quality of drives. Smaller motors spin faster. A 400-450oz Nema 34, rated at 5-6 amps, and running at ±60V may be able to get up to 1500 rpm.
A 1200 oz motor may only be able to reach 500-800 rpm.

[peteeng]

Hi Gerry ta for that - So my estimate of mid speed at 250rpm is low. If I use a speed of 0-700rpm means the tool velocity is 180mm*700rpm=126m/min a real zinger! Then /10=12.6m/min need faster then 12m/min so perhaps the 5:1 is the go at 24m/min.

Now I looked at the motor mount. One way up it takes the N34 and the other way it takes the N23/24 motors. I was wondering how stiff it was so made a model. Seems the shaft is the less stiffer object. This points at using the 14mm shafts. The model has a 4mm thick bracket with a N34 dia14mm shaft. The load is 4000N (400kgf) so I get results in the 1/100's of mm. The shaft deflects 0.09mm and the bracket deflects 0.02mm. The rated load of the shaft is 400N and my nominal tool load is 20kgf or 200N so I expect a shaft deflection of 0.0045mm at 20kgf tool load. This is a simplification as there are tight side slack side things to consider. My main aim was to see if 4mm was thick enough, was thinking of 5mm. So 4mm is OK. Peter

[peteeng]

Hi All - I have made some decisions about Maximus' drives, see summary calcs attached. I have decided the N34 motors are too big and heavy and will need a custom gearbox so they are out. This makes the mounting easier as I just have to work with N23/24's. stepperonline has 5:1 and 10:1 N23 motors but the 5:1 (56mm 1Nm) is a bit small, ideal to have the 76mm motor so I have sent them a note asking if they have a few in their back room. Its 1.8Nm so has 9Nm at 5:1 perfect. Nothing is ever perfect. Compromises have to be made. So 10:1 gives me target forces but its slow at 14m/min. But I don't do big moulds often where I need high speed finishing. But this will become more important as this sort of work picks up. So a 5:1 for the Z drive and 10:1 for the others.

I have a 10:1 here so will have a play and see how fast I can get it to spin. Next is to design the N23 mounts on the machine and tidy the Z drive. I have updated the bearings and brackets now need to massage it to make it minimum width. Getting close to drawings!! Oh yes then I have to work on the long axis... then check all the holes and access so we're a way off from quotes yet...Peter

[peteeng]

Hi All - I realised that when the Z axis pushes up (tool plunging) on its FK bearing it will flex the large bearing beam. This was one reason to include an extra support bearing on the rail. So I did a quick model to see what happens when I put 250kgf on the bracket. It deflects 0.28mm so I'm happy with that as the plunging loads will be much smaller usually. Plus it does not bend permanently! may make the bracket 8mm vs 6mm thick just to keep it in the Maximus maxim vein. Peter

[peteeng]

Hi All - I have massaged the drive width to a minimum, put in a dummy trans drive to see how it falls, adjusted and added a few bolt holes here and there. So getting ready for the final tidy. The transverse travel is down to 1130mm so I need to extend it back to 1300mm. Now I need to find my minimum flange length list from the sheet metal guys and check all the narrow flanges can actually be bent!! They can make exceptions but that costs time hence money. Try to keep everything as standard as possible. The large drive plate has a couple of welds on it to couple and stiffen the edges. Still lots of detailing to be done but the final form is very close. Peter

[ger21]

Why do you need a gearbox with a ballscrew? A direct drive 5-10mm pitch ballscrew typically works very well. Adding a gearbox would add backlash, and require a higher pitch screw.

[peteeng]

Hi Gerry - Does the ballscrew with direct drive backdrive under the weight of the Z axis when no power on (5 or 10mm pitch)? If it backdrives then I need to add a balancing mechanism to it. If I use the gearbox it can't backdrive. Peter

[ger21]

Hard to say. Depends on the detent torque of the stepper, and the friction of the seals on the bearings. And the weight of the Z as well.

But I wouldn't use a gearbox for the sole purpose of preventing backdriving.

[peteeng]

Hi Gerry - I'll have to calculate the backdrive forces then. Need to figure out how to deal with the backdrive before it happens. Peter

Edit - I looked up hiwin design data for the backloading and other references.
https://www.linearmotiontips.com/wha...s-it-important

A rule of thumb is if the lead is less then 1/3 the diameter it won't backdrive. So a 5mm lead with a 16mm screw won't by this "guide' but the backtorque at 30kg is 0.21Nm and the drag torque of the screw is >0.015Nm so I think it may. I didn't calculate backtorque as its complex and all the values are given in ranges. On Scoot I used a gas strut to stop the Z axis falling and that has worked well. So I won't design something but I'll make provision for bracketry in this area. Peter

Re: gas struts. I found that the specified thrust of the strut was way out. Had to take a scale to the shop and test each one until I found a strut I liked and on spec.

https://www.omc-stepperonline.com/st...204d-b200.html

I'll go with a braked motor. The backload with a 30kg Z axis is 0.2Nm the brake has 2Nm resistance so then it's a done deal. Plus I can use the 10mm lead, no bracketry and no permanent backload... Ok we can move along now...