Had a go using motorcalc sheet but does not look right.

[Bob_series1]

Hello all, being trying to work out what motors i require and work back from there.

Used the Motorcalc sheet but the numbers do not look right or i can not make sense of the output. Can some help?
It will be a dual screw machine made from steel. I have inputted the Worst case data. I think. :eek:

My question is, how do I interpret this into the correct motor? since i was going to have two motors for the longest axis do i factor this in by halfing the max load?

Looking forward to getting some parts and starting a build log.

INPUTS
Load Calculations
Mass of Load 25 kg
Friction co-efficient 0.01
Cutting forces 5 N

Leadscrew calculations

Screw pitch 6 mm
Screw dia 20 mm
Screw length 1500 mm
Screw minor dia 18 mm
Screw efficiency 30% %
Screw Fixing Fixed-free

Max linear speed 1200 mm/min
Acceleration time 0.02 sec

Screw mass 3.676 kg
Screw inertia 1.838E-04 kgm^2

Total Inertial load 2.066E-04 kgm^2

Critical Speed 1486 rpm

Motor revs 200 rpm
20.944 rads/sec


Acceleration torque 0.216 Nm

Total Torque 0.240 Nm
at 667 steps/sec

Motor Power 5.030 W

Required Motor 7.55 W
per phase 3.77 W
(bipolar halfstep)

rotor inertia > 2066 gcm^2

[Bob_series1]

I should have also said that my budget for the electronic side of things is somewhere around the £400 - £450 mark. does my budget allow me to over engineer it, so i might be able to expand in the future? Would you recommend Saving more cash? The electronic side of this is my week point. Willing to learn though.

[irving2008]

Hi Bob,

A couple of thoughts... for 1500mm span, 1200mm/s is far too low, thats over a minute to get from one end to the other. Since screw inertia is the major factor on a screw that large you need to use a higher linear speed to ensure your motors can give you reaonable rapids... say 3000mm/min, you wouldnt cut at that speed but you need some headroom, you can reduce the acceleration to 0.1 for rapids calculation. Secondly, although it won't affect your motor choice, use supported-supported on a screw of that length and diameter - always worth spending more to get the mechanics right.

This gives you a motor needing about 10W output.. A suitable motor would be SY60STH86-3008BF (3Nm NEMA23, 3 stack) on 40v/4A drivers, wired bipolar parallel. You could go for a smaller motor on Y and Z

[Bob_series1]

Thank you for your comments irving. Looks like i did not set the speed and left it at the default value. Sill me. The one thing I had not even considered, is the speed I wanted it to move at. LOL

I will be supporting the lead screw but just wanted to put in a worst case. However, from your spread sheet it makes a massive difference to the power requirements. :surprised:

So, for 4 motors comes in at about £130. It is amazing at this point how many people go for either a driver board or PSU that can not drive them to their full potential.

So, i guess the next logical steps is to research the best buy when it comes down to the drivers and PSU. So much to learn!!

Thank you irving.

[Gary]

I also notice that you have selected a leadscrew with an efficiency of 30%.
Have you not considered a ball screw? A ball screw will be far more efficient and also will not wear as quick.
As an indication of the efficiency, a ball screw can be between 80-90% efficient, so for the same motor you will get a lot more thrust with a ball screw.
Allot of people buy a three or four axis driver board then the motors and find that the drivers just dont cut it on larger motors,
and in some cases they just start to smoke.
Look at the torque curves of the sy60 motor and you will see two torque curves at two different voltages.
This shows how important voltage is.

[Bob_series1]

since my design uses dual screws for the X axis cost would be my reason for using lead screws.

In the future i would also like to have a play with rack and pinion.

However, depending on the old cash flow i might be able to afford ball screws for the Y and Z axis.

Main thing for me now is to get some stuff ordered to get the project rolling.

[irving2008]

I will be supporting the lead screw but just wanted to put in a worst case. However, from your spread sheet it makes a massive difference to the power requirements. :surprised:



With big leadscrews the key thing is screw inertia and acceleration... a big screw operating at high cutting speeds needs high acceleration (short acceleration time) to be able to follow a cutting line.. and that needs torque AT THE REVS IN QUESTION... this is the thing that most people miss... its not bottom end torque you need, its torque at the top end... and as you say, you need to drive them optimally to get that.

The method of supporting the leadscrew does not directly affect the motor requirement, it affects the critical speed above which the screw starts to whip, putting sideways force on the nut and adding friction (as well as a nasty vibration and a poor cutting finish). But adding a better support configuration means you could go for smaller screws, and since inertia varies as the fourth power of the diameter even a small reduction in screw size can have a dramatic variation in motor requirement... or for the same motor get better performance... I cannot find a spec for a Tr20x6 screw, only 16x4 or 20x4 - given those two I'd go for the 16x4. You might also look for a 16x4, 2 start. (same as a 16x4 but the lead is 8mm so you get twice the speed, or more importantly, you need half the motor revs for the same speed). Anyway, its worth playing with different screw sizes and types (1/2-10 2 and 5 start ACME is popular in the US).

Note also if you are going for dual screws you need to consider how you will drive them... if you use one motor and a timing belt/pulley arrangement you need to add the inertia of two screws together plus that of the pulley arrangement to get the required power... in this case it will be more than twice the single screw arrangement.

[Bob_series1]

With big leadscrews the key thing is screw inertia and acceleration... a big screw operating at high cutting speeds needs high acceleration (short acceleration time) to be able to follow a cutting line.. and that needs torque AT THE REVS IN QUESTION... this is the thing that most people miss... its not bottom end torque you need, its torque at the top end... and as you say, you need to drive them optimally to get that.


Wow, all the hours spend reading the forum and i did not know this. I now understand why you recommended wiring the motors up in a bipolar parallel configuration. Looking at LeeR graph in the sticky, the torque is preserved much better high up the Rev range.

So, since my rails are going to use skate bearings i was going to use the skate bearings to support the lead screw. Turn the end down to 8mm on lathe, place skate bearing on, put thread on the end and lock with a nut+washer. Also, I have a plan at supporting the lead screw at the stepper end.
Wow, playing with the diameter of the screw on the spreadsheet makes a large difference. Will put some more thought into the screw size.



Note also if you are going for dual screws you need to consider how you will drive them... if you use one motor and a timing belt/pulley arrangement you need to add the inertia of two screws together plus that of the pulley arrangement to get the required power... in this case it will be more than twice the single screw arrangement.

So, the plan was to have one motor per screw. I would have the extra motor running on its own driver board and slave it in mach3. From reading a thread on CNCZone it was discussed that if your gantry is rigid enough then the your unlikely to encounter the screws loosing sync.

The build will be fully out of metal so it being rigid is not my worry it is the weight of the gantry.

I can see why people spend money replacing parts on their CNC machine if they do not do the maths.

P.S. That spread sheet is ace.

[irving2008]

Bob,

Thanks for your kind words.

With big screws running at speed you need to locate the leadscrew axially as well as radially, otherwise there is a lot of force on the motor. Skate bearings are pretty poor axially. You either need a pair of deep-groove bearings or thrust bearings, which are cheaper. The diagram shows a simple bearing block that will hold the skate bearing surrounded by two thrust bearings. The picture shows one of my motor mounts, you can just see the thrust bearing against the cover plate, which on mine is screwed in place though not strictly necessary.

Thrust bearings can be got cheaply from TechnoBots (http://www.technobots.co.uk/acatalog/Shop_Front_Miniature_Bearings_382.html)who also have an eBay shop.

[Bob_series1]

I understand what you mean, when you say there is alot of force on the motor. I did design a method to counter act this, however compared to your design it is over complex and thus more likely to fail. I like how yours deals with all the forces on all the axis in one neat solution. My solution was also bulky and thus not very good for use on the z-axis.

Going to order the bits and have a go at building some of them.

Is this your design or a common know configuration? I tip my hat to the designers.

Well that is most items covered for my machine then. I am going to order the bearings, screws and motors, start a build thread and once it is all together I was going to buy the Driver boards etc.. at the end.

Thanks for putting me in the right direction!!!!!!

[irving2008]

this particular one is my design but the concept is common. If I was to do it again for a bigger diameter leadscrew (as I may be doing) I'd not use 8mm skate bearings but a 10mm x 30mm double row angular contact bearing type 3200-2RS as this has good radial and axial support and is simpler to use (one bearing instead of three), but obviously the screw root diameter needs to be >10mm.

[Bob_series1]

My quick google search for that bearing showed a cost of £15 each.

The original design the cost for the bearings was
Thrust x2 @ £1.10
Abec 7 Skate bearig x1 @ £1
£3.20

So why is this bearing worth 4-5 times the cost?

[irving2008]

Because the quality of an angular contact bearing is much higher and they are much more complex... however you can get them cheaper on eBay (http://cgi.ebay.co.uk/3200-DOUBLE-ROW-ANGULAR-CONTACT-OPEN-BEARING-10X30X14_W0QQitemZ220292420908), though this is not a sealed bearing so that may have a bearing (excuse the pun).

If you can go to 12mm ID then here (http://cgi.ebay.co.uk/3201-5201-DOUBLE-ROW-ANGULAR-CONTACT-BEARING-12X32X15-9_W0QQitemZ200341458766)is one for £4.95

[Robin Hewitt]

Acceleration time 0.02 sec

That does sound a bit excessive :nope:

The place I came unstuck was the first time I encountered a G0 suddenly hammering the X axis in to reverse, the worst possible case when it comes to losing steps on a motor, even though you can ignore the cut pressure. My software just threw one in at the cut start, out of the blue, just for fun.

I eventually set an 'okay to hammer in to reverse' threshold, that exempted most normal cutting from accelerations, then gave it a quarter turn acceleration time on the G0's.

Setting the top speed is a bit suck it and see, but you can get an indicator by looking at the torque v speed curve for the motor to see where it starts to plummet.

I need to be a big comfort zone on the G0 speed. Having fitted new super 220V drivers to my machine it would probably be good for 25 mm/s plus, but I still have it set to 10 mm/s. I know it would work fine if I ramped it up, but I just haven't found a job yet where I don't mind risking the metal to find out :naughty:

[Bob_series1]

Should be 0.2 seconds (I think).

Yes those bearings look very good value. If you do go with them, I would be interested in a review.
For now I am going to go with your Mk 1 method, If I have permission to use your plans of course.



Robin, Whats a G0?

[Robin Hewitt]

G0 is the first G-Code, it signifies 'go to the following co-ordinates as fast as you can, doesn't have to be a straight line'.

Usually used to position the tool, cutting air.

G1 does a straight line and applies a feed rate.

G2, G3 are clockwise/counterclockwise arcs.

You can pick and choose from the rest, I ignore most of them.

[irving2008]

That does sound a bit excessive


Should be 0.2 seconds (I think).

The value of 0.02 seconds was taken from a document which I can't find now, from a major CNC manufacturer qualifying small, medium and large machines. The 0.02sec was the spec from the medium machine. Its possible that 0.02 is a bit aggressive, but 0.2 is def too slow.. The reason for using a time rather than an absolute acceleration is to make the acceleration dependent on the maximum speed, as thats where the torque is required.

I have worked out another way to estimate acceleration based on pocketing a circle, obviously to maintain feed rate as the radius reduces the rate of change of linear direction X/Y is directly related to increasing acceleration in the individual axis. This calculation shows an acceleration requirement approximately 1/2 - 1/3 (depending on the radius of the pocket) of that given by using 0.02Sec, suggesting 0.04 - 0.06Sec might be a better value. I shall be updating the sheet with the revised numbers shortly.


Yes those bearings look very good value. If you do go with them, I would be interested in a review.
For now I am going to go with your Mk 1 method, If I have permission to use your plans of course.I am going to use the £4.95 12x32x15.9 one in my rotary table conversion; although I could get away with an open bearing there, the bigger bearing makes the machining slightly simpler and i prefer the sealed one for peace of mind. Of course you can use my plans, the technique is common.

[Robin Hewitt]

Is that .02 seconds to go from zero to top whack?

I allow about 1/8 of a second ('ish), this is what it sounds like on a G0...


http://www.youtube.com/watch?v=bAoUW5bszRA

[irving2008]

Clearly the value for G0 is different than that for G1.... what value do you have for G1?

Maybe I should make the spreadsheet deal with different speeds and accelerations for G0 and G1 and estimate the motor power based on which is worst case... As stated the value of 0.02 will produce an overestimate on the motor size for large routers or table-based machines but isn't too bad on a small lighter machines

[Robin Hewitt]

Clearly the value for G0 is different than that for G1.... what value do you have for G1?

Now there's a question! I'd just wired up the new drivers and ran the last job on the computer for a look see. The X axis is in reverse.

Probably 1.5 mm/s, I usually cut between 1-2 but I might drop to 0.8 for a plunge.

When the cutters get long and spindly, even slower.

[irving2008]

sorry Robin, just to confirm - is that acceleration or cutting speed? If you really meant 1.5mm/s^2 thats 100 seconds to go from rest to 150mm/s cutting speed which doesnt sound fast enough to me! I think you meant 1.5m/s cutting speed, but what I was looking for was the acceleration you are using with that.

[Robin Hewitt]

I think you meant 1.5m/s cutting speed, but what I was looking for was the acceleration you are using with that.

That was cutting speed.

There has to be a top speed where it is okay to slam it in to reverse without losing steps.

No point applying accelerations below that speed so I don't. The 1.5mm/s feed has no accelerations.

Incidentally, slowing the axes down to maintain feed rate for arcs and slopes is easy. I started out playing gradients but it's not necessary. Took a while for the penny to drop.

At the sharp end of the program, you eventually step X, Y, neither or 'X and Y'.

If 'X and Y' you increase the delay before the next step by root 2.

Simples :naughty:

[irving2008]

Ah right, so your max cutting speed is 1.5mm/s = 90mm/min? That still seems for too slow to be a useful cutting speed, esp for alloy.. even with your high speed spindle. I was expecting 500mm/min+, i.e. something around 10mm/sec

Increasing delay between steps by root 2 on an 'X and Y step' is a neat approximation to maintaining a constant tangential linear speed in a curve. :)

[Robin Hewitt]

I was expecting 500mm/min+, i.e. something around 10mm/sec


Well, maybe if the machine had another ton or so of cast iron to keep the toolpath sweet :heehee:

I suppose I could try it but suspect on a quarter ton machine I'd get horrible graunching sounds from a fat cutter and bits of shattered tool whizzing past my ears from the smaller sizes.

Have you actually tried cutting metal? :naughty:

(Slight exageration there, whenever the tool breaks it never seems to do anything more than drop off the shank, never whizzes anywhere) :whistling:

[irving2008]

Well, maybe if the machine had another ton or so of cast iron to keep the toolpath sweet :heehee:

I suppose I could try it but suspect on a quarter ton machine I'd get horrible graunching sounds from a fat cutter and bits of shattered tool whizzing past my ears from the smaller sizes.

Have you actually tried cutting metal? :naughty:

(Slight exageration there, whenever the tool breaks it never seems to do anything more than drop off the shank, never whizzes anywhere) :whistling:Err yes... I don't know what you are cutting (stainless steel perhaps?), but 6082T6 on my small mill cuts fine with an 8mm cutter at a feed rate of around 5mm/sec (which is about as fast as I can turn the handles manually - 2 revs/sec) and I would expect a powered table to move faster than that.

Some rough numbers... 8mm cutter, 60m/min cutting speed for alloy = 2300rpm on spindle (I whack it up as fast as the belts let me which is 2580 I think)

The chip load for alloy is approx dia/150 for roughing, d/200 for finishing. Assuming roughing chip load is 8/150 = .05mm so the feed rate @2300rpm for a 4flute cutter is .05mm * 4 * 2300 = 460mm/min = 7.7mm/sec

The material removal rate is depth * width * feed rate = 1 * 4 * 460/1000 = 1.84cc/min

Power(kW) = specific cutting force * removal rate/1000 = 17 (for alloy) * 1.84 = 31W (~1/20HP) and the torque needed is P * 60/(2pi * revs) = 31 * 60/(6.28 * 2300) = 0.13Nm

Doing the same with a 50mm facecutter its 400rpm spindle speed, 533mm/min (8.9mm/sec) feed rate and slightly less power. I know this is too fast for me to do manually which is why my surface finish isnt as good as I'd like with the 50mm/4tip facecutter compared to the 25mm/2tip one.. I havent yet tried the obvious and taken two tips off the bigger cutter...

I've only broken drills and two 3mm endmills... and nothing whizzes anywhere.. just snaps off flush with the end of the collet!

[Robin Hewitt]

nothing whizzes anywhere.. just snaps off flush with the end of the collet!

Well at least we agree on that bit, same thing happens with larger sizes :whistling:

I did try an accidental fast cut once...

I was using CamBam and cutting an outline in 1/2" x 6" HE30 alloy bar. Rather than drill hold downs I used a CamBam facility to leave tabs around the bottom edge. 'Easy to saw out after and clean up with a file', I thought.

Didn't notice in the G-code that when it lifted and passed ove the tabs it was using G0's with the tool still 10mm deep in the slot previously cut.

I would have changed them to G1's if I'd spotted it, but I didn't.

Of course the groove forming cuts above had side loaded the tool and bent it. Riding free in the slot the tool found itself zooming through a slot it didn't exactly line up with.

Chip loading was minimal but the horrible graunching sounds resulted producing great gouges which were reflected in the clean up pass.

I even thought about hitting the abort button because I knew it still had 2 laps to go and I wasn't sure it was going to make it.

Maybe if I'd had 4 times the revs I'd have gotten away with it?

There was a problem with that.

The quill splines had been rattling when I was whittling a 5" x 1.25" iron bar so I added grease. Heavy clay loaded grease.

Shut it up a treat. Unfortunately I hadn't reckoned on the taper roller bearings being unshielded. Then I had a problem with the phase converter cutting out on over current when I wound the speed up.

As soon as I realised what was going on I flushed the quill through with engine oil until it ran oil coloured rather than black. Top speed again became an option.

I have been speed limited, maybe faster feeds will now be possible, I shall play.

OTOH 10mm/s still sounds very silly :heehee: