Sizing lead screw and stepper motor

[peteeng]

Hi Craig - I think you will find that all ground screws start out as rolled screws. Rolling is a relatively fast process (vs cutting or full grinding) and creates an advantageous grain in the thread. Then its finally sized by grinding. Peter

[mactec54]

Hi Matec,
well Ok, I misinterpreted what you have said.

You may have noted that in the link posted earlier to screws from AliExpress in the advertising they claim 'C3 Level', and yet the prices are about 1/3 I'd expect to see a
genuine C5. I just find that a bit too hard to believe.

Bosch Rexroth do a C5 rolled screw. Well, its rolled and then lightly ground. Using the DIN standard, they claim 23um/300mm, but on the cyclic its 18um/2PI. It's the cyclic that falls well behind
what I'd expect from a C5 or C3 made to JIS standard. Bosch Rexroth reckon that rolling then grinding means you can get the quality but without the cost....although you wouldn't think so if
you compare their prices to say NSK or THK which are genuine ground types. Bloody Germans sure know how to crank up the price!

Craig

The AliExpress link are C3 Ground so you did get things confused. As Peter said some Roll the Screw then Grind this is faster and you get a better Screw and a lower price

[peteeng]

Hi Layton - I presume this is still a paper exercise at the moment so all ideas are still on the table. I'm looking at AC servos at the moment for a similar size machine. Servos give you a flat torque curve to 3000rpm and smoother motion which is attractive. But screws at 3000rpm need to be big. Looking at the permissible speed for 48" long screws: Dia16 - 1304rpm Dia20 - 1706rpm Dia25 - 2208rpm Dia30 - 2709rpm won't go further. So if I use a small servo with a 1:2 stepdown pulley I get 1500rpm and heaps of torque. This means a Dia20mm screw with a 10mm pitch gives me 15m/min top speed very good, or a 16 pitch gives 24m/min. My first machine used a belt and I did mould finishing at 20m/min. Now the reduction pulley also allows me to reflex the motor back into the machine footprint. This is attractive as next machine I want to incorporate a full dust enclosure. Usually with motors on ends of drive screws they stick outside the base footprint and the gantry end making this sort of thing difficult. You can still reflex the motors if they are steppers.

My aim is to look something like the "Cube" series by Datron. Dust and noise control are high on my must have list. Easier to sort this at the design stage then after...

https://www.datron.com/cnc-machines/m8-cube/ My aim is to look something like the "Cube" series by Datron.

Ideas for you to think about over more coffee.. Once you get your drive concepts settled you then need to work through your structures to take full advantage of the drives. What construction philosophy are you thinking of? steel, aluminium, concrete etc...Peter

[joeavaerage]

Hi Peteeng,
following the calculation I did earlier if you used a 20mm diameter screw with 10mm pitch and a 400W servo which are commonly continuously rated at 1.27NM at 3000rpm.
Just because the servo does 3000rpm does not mean to say you have to use it. For instance you might in order to avoid whipping choose to run the screw at 1500 rpm
with a consequent 15/min traverse, you could still direct couple the servo but just run it to 1500rpm. It does seem a bit of a waste 'to leave some performance on the table' as it were,
but its by no means the end of the earth, but it would save on the complication of pulleys etc. If you wish to invert the mounting of the servo to save on footprint then I assume you
are OK with that complication.

Craig

[peteeng]

Hi Craig - yep there's several ways to skin the cat and if your using a servo it seems a waste to have 1500rpm not used. You could go stepper hybrid and get to 1200rpm with some torque as well but only for rapids I expect whereas the servo with pulley would have good torque for cutting at 15 or 24m/min. Comes down to budget and what you are trying to achieve. For instance the Datron has a quoted top speed of 22m/min so I expect they are using pulleys. I don't see the reflexing of the motor as complicated if it solves some of the issues. You have to have a coupling so pulleys give you that. You could go the other way, use a 2:1 upspeed and use a stepper from 0-750rpm to get you to 1500rpm on the screw. If the stepper is a 3Nm then you have 1.5Nm at start and 0.75Nm at 750rpm. A little more if you use high voltage and hybrid... Have to work through the variables and the costs to meet the budget/performance needed... Peter

The flat torque curve and overdrive ability of the servo is the big attraction of servos, neglecting cost... The drooping torque curve of the stepper is always the tricky bit if you take it to its performance edges.

eg if you use the 400W with a 2:1 you get 2.6Nm everywhere that's quite a lot of torque everywhere. ie its 150kgf at any speed (say 0-15m/min) . More than enough for steel and aluminium cutting with a 10mm pitch screw. If its a mill use 5mm and that's a lot of grunt... Have to check buckling loads carefully.

[joeavaerage]

Hi,

The flat torque curve and overdrive ability of the servo is the big attraction of servos, neglecting cost...

Too damn right. I bought 750W Delta B2's for my mill, 2.4Nm cont at 3000rpm, but the max speed at somewhat reduced torque is 5000rpm. I have used that extra speed
and direct coupled to 5mm pitch screws gives me 25m/min. As it turns out my machine can do 25m/min but my nerve only goes to about 15m/min....so I dialled it back a bit!!

With steppers you always seem to be trying to extract a bit more......while with a servo it seems that often you 'leave performance on the table' just because you cant use it full potential.

Craig

[peteeng]

Hi - That will be Laytons next challenge to design a structure that can take full advantage of the drive system. The drive is the heart of the system the structure is the bones or skeleton, get both right and you have a ripper machine. Peter

[ardenum2]

Hi,



Too damn right. I bought 750W Delta B2's for my mill, 2.4Nm cont at 3000rpm, but the max speed at somewhat reduced torque is 5000rpm. I have used that extra speed
and direct coupled to 5mm pitch screws gives me 25m/min. As it turns out my machine can do 25m/min but my nerve only goes to about 15m/min....so I dialled it back a bit!!

With steppers you always seem to be trying to extract a bit more......while with a servo it seems that often you 'leave performance on the table' just because you cant use it full potential.

Craig

didn't your whole frame rock back and forth with that kind of speed? or is it bolted to the floor?

[Laytonrouter]

These are speeds I can't imagine using unless trying to polish a metal surface and not take all day. Anyhow I plan on going with 2 Nema23 steppers (425 oz/in) on the Y, 1 of the same on the X and Z, I think I can get away with sfu1610 on the X and Y as their accuracy is about .002" with 200 steps per rotation. With the driver's many multipliers one can only imagine the thousand of increments in between each step. Again, I think I can safely hit 150 to 200 ipm with an 80lb gantry. The X axis is only moving about 40lbs. if that.
Big jump in pricing if I go to 20mm screws, bearings and nema 34 motors. Would do if I thought I had to but right now I seemed to be convinced the above would do it and I don't want to greatly overbuild for now good reason. If anyone say "don't do it" I think I'll give it a try.

Many thanks all.

[joeavaerage]

Hi,

didn't your whole frame rock back and forth with that kind of speed? or is it bolted to the floor?

At the highest speeds and accelerations my machine does start to wander around, it's on heavy duty castors. It's quite hard case to see a machine that weighs 800kg
slowly but surely work itself across the workshop. Ive got it wedged in a corner at e moment and that's fine.

Craig

[peteeng]

Hi Layton - High speed machining techniques are now the norm. Every CAM system has HSM toolpathing. Plus to take advantage of high speed spindles you need fast feed rates. But move along to your next design area say the structure and sort that out, look at what CAM software you are going to use and see what it suggests for feeds and speeds etc and then review drives when you have the big picture together. Fast speeds do not make the machine move around, high accelerations do and they can be at slow speed.

Some machine makers recommend bolting the machine down, some discourage it. Peter

[Laytonrouter]

Hello Pete,
Your exactly right as I suspect I might eventually get into using high speed feeds and acceleration rates. I am using Linuxcnc which can handle quite high operating speeds due to it's real time operating system. One of my major design changes is to reduce the weight of the X axis gantry and also the height of the mass. Accelerating that along the Y axis is where momentum comes into play in a large way. Don't have it together yet but will send a pic when I do.

Thanks again

[joeavaerage]

Hi,

Your exactly right as I suspect I might eventually get into using high speed feeds and acceleration rates.

But you'll always want the highest acceleration you can manage because high acceleration is the best strategy for toolpath following.

I am using Linuxcnc which can handle quite high operating speeds due to it's real time operating system

I use Mach4 which can very happily drive a machine to hundreds or even thousands of meters per minute, its got nothing to do with realtime. All decent CNC software solutions can in theory
drive a machine much MUCH faster than any machine we have.

Accelerating that along the Y axis is where momentum comes into play in a large way. Don't have it together yet but will send a pic when I do.

What about the calculation I've already posted? With a 20mm diameter screw and 10mm pitch the inertia is 57% in the ballscrew, 23% in the stepper and 20% in the mass of the gantry.

If you really want to reduce the momentum then reducing the mass of the gantry is small-fry, reducing the momentum of the ballscrew is the big issue. For instance if you can reduce from 20mm diameter to 16mm diameter
the reduction in ballscew momentum is huge (16/20)⁴=0.4096. Now the momentum equation would be 34% ballscrew, 34% stepper armature and 32% in the gantry.

The momentum of a rotating cylinder, like a ballscrew, is proportional to the fourth power of its radius and usually totally dominates the momentum equation. Even with a 16mm ballscrew the mass
of the gantry is still only 1/3 of the total momentum.

Craig

[peteeng]

Hi Layton - I am reviewing the costs on my current build which is attached. Its a preproduction machine and I'm working thru costs and parts to figure out how to "productionise" it best. So I have the figures on my desk. Arn't you lucky!

Current material and freight costs amount to $4610AUD of which 27% is international freight and 7% is steppers and drivers. If I replace the steppers and drivers with servos it becomes $5356AUD so the servo motors go to 20% of the build and the freight then becomes 23%. The stepper system is $330AUD and the servo is $1076AUD. If I was trying to make money from the machine I'd go servos as this will return my money faster and make more $$$ into the future. If you do a value analysis I'd expect that we should put more than 7% of the budget into motors as they are the heart of the machine and are more than 7% important. Looking at hybrids a stepper unit system is $50USD and the hybrid is $112USD and the servo is $269USD. If you consider you will have the machine for 5 years then the servos will be the go (if you have the initial budget) because as you get better at machining you will go faster and faster and the steppers won't keep up with your progress. Thats been my observation with my customers and discussions on the forum. Linuxcnc will do you well. It has developed very well over the last few years and will continue to do so. I don't use it but I have a look occasionally. FreeCAD is a good fit to LinuxCNC. Peter

[mactec54]

Hello Pete,
Your exactly right as I suspect I might eventually get into using high speed feeds and acceleration rates. I am using Linuxcnc which can handle quite high operating speeds due to it's real time operating system. One of my major design changes is to reduce the weight of the X axis gantry and also the height of the mass. Accelerating that along the Y axis is where momentum comes into play in a large way. Don't have it together yet but will send a pic when I do.

Thanks again

The Realtime does not have much to do with how fast it can run, unless you are using a Mesa Card the Linuxcnc Kernal speed is slow compared to using Windows, CNC controls using Windows can use Realtime processing also.

[joeavaerage]

Hi,

The momentum of a rotating cylinder, like a ballscrew, is proportional to the fourth power of its radius

I should clarify this, because if you look at any physics text the formula for the first moment of inertia of a long cylinder rotating around its long axis is:

J=1/2 . m .r² where m is the mass of the cylinder and r is its radius.

Many will point out that is proportinal to the square of radius not the fourth power as I have claimed......but there is a catch, the mass of a steel cylinder is Pi. r². l.Rho
where Rho is the density of steel and l is the cylinders length.When you combine the square proportionality of the fist moment AND the square proportionality of mass that's when you get J is proportional
to the fourth power of r.

Craig

[Laytonrouter]

Have to say I'm not sure about the r^4. It seems your looking for the total mass of the cylinder regardless of solid or hallow, and the radius the mass occupies from the axial center. multiplying the density times the total volume times the radius gives you the same value. Guess I would have to work it out mathematically to understand it better. R^4 is a big jump in the scheme of things but one I will consider.

[Laytonrouter]

Pete,

Interesting pic. Looks like you have a longer than usual Z axis. the first table I build 4' x 8' had a 24 inch Z travel. I was doing a lot of styrefoam 3D models. I found it was much easier to do 6 inch sections and glue them together. A 24 inch Z travel has a angular limitation due to the angel between the tool and edge of the Z axis platform the spindle is attached to. Essentially you can't plunge 90 Deg. more than the length of the tool.
That being said I really think you are on to something with your horizontal X gantry platform. that is exactly what I am doing. My thinking is that that huge mass of metal high up in the air creates a lot of resistance to inertial forces. Not to get into the weeds too far, when you start the axis the weight way up on top acts like a loaded lever opposing the inertial movement. Put another way it increased the amount of inertia needed to start moving.

Well done, looks great to me.

[joeavaerage]

Hi,
thats why I posted the clarification....the mass depends on the square of the radius AND the first moment also on the square of the radius and that's why the first moment depends on the fourth power.
It also explains why choosing the right ballscrew diameter is critical, its easy to think bigger is better, but it can increase the inertia BIGTIME with a consequent loss of acceleration. I very near fell in the trap
a couple of years ago.

At that time I had not bothered to do detailed calculations about ballscrew inertia. I selected 32mm diameter 5mm pitch ballscrews for my new design.It was only later that I did the inertia equation only to find that my
acceleration has been restricted to 0.27g at the rated torque of my servos. As it turns out 0.27g is quite adequate but if I'd used 25mm ballscrews I could have had even more acceleration and 25mm diameter would in all other
respects been good enough. The inertia balance for my machine is 80% ballscrew, 12.5% the servo armature and 7.5% linear axis mass....and that axis mass is 150kg. So you can see that the ballscrew
totally dominates the inertia.....and the 150kg axis bed is but a small fraction of it.....useful to know!

Craig

[Laytonrouter]

The mass depends on pi * R^2 to determine the area of zero length. That times the length gives you the total mass and the simultaneous distance from the axial center.
The first moment of inertia I = 1/2 m.r^2 where m is the density x the volume. Essentially they both give the total weight and the distribution of the mass from the center of the axis. So I can see clearly you make a very good point about the diameter of the ball screw. Twisting a very heavy large diameter ball screw takes much more force than turning a narrower one. Given the numbers it appears the weight of the X axis gantry may not have as much bearing on the acceleration force needed as it is divided up into small increments by the threads on the ball screw. You may have a very valid point.

Thanks again.