Mill knee driving with a stepper

[TrevorGParticipant @trevorg]

[TrevorGParticipant @trevorg]

Converting my ancient CVA mill to CNC requires driving a very heavy knee. As replacing the leadscrew with a ballscrew would be extemely difficult I have kept the ACME one. My initial attempt involved buying the largest 34 size stepper (11.2Nm) I could find and driving directly via a solid coupling.

At speeds between 60 & 240 mm/min it works fine. However, when I reduce the speed to 0.6 or 6 mm/min I get problems the drive begins jumping and goes nowhere. I am not sure what the cause of the problem is. Clearly I am going to need some changes but I would like to understand what is going on first. Any ideas?

I am aware that the solid coupling is part of the problem as I can feel that the load varies a little throughout a revolution. However, replacing with a flexible coupling would make the stepper stick out even further than it does already and I don't think it would solve the problem completely.

My thoughts at the moment involve using a faster stepper via a toothed belt drive giving a reduction of 3 to 4 times. There are a lot of options here and I could easily waste a lot of time and money on the wrong one.Thoughts?

Trevor

[Anonymous]

Most likely due to static friction versus dynamic friction. In which case the Acme leadscrew and the highly loaded dovetails are the issue. Changing the drive arrangements would simply be hiding the real problem.

Andrew

[Ian PParticipant @ianp]

The high torsional rigidity of your the drive line is the cause of the missed steps and is one of the factors that needs to be taken into account with steppers.

Stepper motors (whether in single or micro step mode) move in a series of jumps. Regardless of the size of the jump (or step) the rotor tries to move instantaneously from A to B as the magnetic field changes. The magnetic field does change almost instantaneously and the rotor has to accelerate and decelerate to move from step to step. The laws of physics dictate that it cant actually do that.

If the rotor is rigidly linked to a significant mass it just cannot accelerate quick enough to get moving. Many home brew CNC systems have a toothed belt drive so the motor is only carrying the weight/mass of the pulley. Even though the belt is virtually backlash free it has enough 'give' to allow the stepper rotor to start moving at each step. Without some compliance in the drive between motor and load the easy way to ensure consistent performance is to use a bigger motor. (The sledgehammer to crack a nut technique)

Without knowing more about your setup I can only guess, but if your motor is the 200 steps/rev type does that give you enough resolution? If you incorporate a belt reduction not only do you get an increase in torque but very often you can make a much more compact arrangement as you are not stuck with the length of the motor protruding.

Ian P

[MuzzerParticipant @muzzer]

I wouldn't agree with Ian's explanation of stepper motor operation. Like any synchronous machine, the torque generated is proportional to the angular difference between the rotor and the magnetic field. When you step the drive, the field rotates 1/200 of a turn ie 1.8 degrees. If there is no resistance, the rotor will follow the full angle. If there is no movement, the rotor will generate a significant torque, hopefully somewhere close to the alleged "stall torque". Sounds as if you simply don't have enough torque.

Microsteps increase the number of steps but (if driven by a half decent driver) don't significantly decrease (or increase) the torque the motor can generate, despite a lot of misinformation to the contrary.

Although it isn't usually obvious, the speed-torque characteristic of most steppers looks rather like a constant power curve, so changing ratio from direct drive won't necessarily change the maximum speed you could drive the leadscrew at. Indeed, by gearing the motor down by 2 or 3 to 1, you would be able to increase the stall torque at the leadscrew and probably still get the same feed rate under load. As Andrew says, you are probably suffering from static friction (stiction?). This is probably one of the main reasons people change to ballscrews, apart from the backlash improvement.

Anyway, in terms of solutions, 2 come to mind:

• If you are intent on driving the knee up and down, you can reduce the load on the Z leadscrew by fitting one or more gas struts (springs) underneath to offset the weight of the knee assembly. You could get an adjustable gas spring if you wanted to play with the spring rate. This seems to be a common technique. Certainly neater than hanging a massive counterweight out the back of the machine. You probably need to leave a bit of weight on to preload the leadscrew and reduce backlash.
• The other solution would be to drive the quill rather than the knee. That's what I've done on my Bridgeport clone (see my albums under "Z-axis drive" or similar) and is much more commonplace with knee mills. Admittedly it limits the travel to 4-5" or so but if you need more, you can always move the knee up or down during the machining process.

Murray

Edited By Muzzer on 05/09/2014 17:28:25

[Ian PParticipant @ianp]

Murray

My description is based on my knowledge of stepper motors learned from reading about them rather than using them so is purely from memory.

You are completely correct in stating that when energised the rotor will develop significant torque. The torque is greatest when the fixed and moving pole 'tips' are aligned (in each step position) though it reduces slightly in between step positions.

Some stepper drive circuits use clever circuitry to boost the beginning part of each pulse in order to help overcome the load of moving the rotor from rest, which it is (sort of) between every step.

Regardless of anything else the electric current flowing through the windings in order to move it to the next step, the electrical signal is only present for a finite time (for each step). If for some reason the rotor has not moved in response to that particular pulse its too late do do anything about it because its an open loop system. Syncronous motors, which steppers are a type of, only run or move at the speed dictated by the frequency of the signal and nothing else. They don't accelerate in the normal sense.

I did once watch a demonstration of a stepper motor that would not move even though it was not coupled to any load, it did however have a flywheel on its shaft. With the flywheel removed, the motor worked perfectly.

I think Trevor problem would best be solved by incorporating a reduction stage.

Ian P

[Les Jones 1Participant @lesjones1]

Hi Trevor,
I think Murry's solution of driving the quill instead of the knee is the best option. If for some reason you have to stick with moving the knee then I think Ian's suggestion of a greater reduction from the stepper to the leadscrew is the simplest solution. I do not think Ian's comment that there is only a pulse of current for each step it true for all stepper drivers. I think some have the option to reduce the current from the rated stepper motor current after the initial step. (But I do not think they reduce the current to zero. ) It would be worth putting an ammeter in series with one of the windings to measure the holding current. (Note monitoring the supply current would not give a correct reading at the driver will use PWM to reduce the voltage to the coil to the value required to give the required current in the winding.) Another possible solution would be to use a system like the UHU servo. This emulates a stepper motor using a DC motor with an optical position encoder. At the motor shaft this behaves as closed loop servo. After a step pulse it will continue to try to drive the motor until it has seen the encoder move one position. You would need a motor that was capable of providing a lot of torque.

Les.

[Neil WyattModerator @neilwyatt]

I have no practical experience of driving steppers, but some genuine experts I have known claim that driving them is the darkest of electronic's dark arts.

If a single pulse is not enough to move the knee on its own, then you will always lose at least one step at the start of any continuous motion.

Such a large mass will take time to accelerate. Sophisticated drivers ramp up the pulse speed to allow for this, to avoid lost steps. Obviously if you were just using this as a 'power feed' rather than CNC then it would be unimportant.

My instinct would be to balance the load as much as possible – with such a big mill this won't compromise rigidity, and it will reduce leadscrew/nut wear as well.

Bear in mind that a simple weight on a pulley give a virtually constant load, few spring systems will.

Neil

[Neil WyattModerator @neilwyatt]

I have no practical experience of driving steppers, but some genuine experts I have known claim that driving them is the darkest of electronic's dark arts.

If a single pulse is not enough to move the knee on its own, then you will always lose at least one step at the start of any continuous motion.

Such a large mass will take time to accelerate. Sophisticated drivers ramp up the pulse speed to allow for this, to avoid lost steps. Obviously if you were just using this as a 'power feed' rather than CNC then it would be unimportant.

My instinct would be to balance the load as much as possible – with such a big mill this won't compromise rigidity, and it will reduce leadscrew/nut wear as well.

Bear in mind that a simple weight on a pulley give a virtually constant load, few spring systems will.

Neil

[Ian PParticipant @ianp]

Les

My explanation was gross simplification of how steppers are driven. There are umpteen methods and strategies used to get the desired performance for a particular application, mainly what I was trying to convey is the need to make an allowance for inertia,

I had mentioned that I had no experience of using stepper motor, by coincidence however, today I finished (well almost) putting together Steve Ward's indexer and I have it up an running driving a small rotary table.Most of the parts were from my scrapbox (I bought the PIC and PCB from J.S.) and the only part (I need?) to purchase is the 10MHz crystal. I have temporarily put in an 8MHz one. Am I right in assuming that this will have no effect on the actual number of steps and that the only thing that will be affected is the rotation speeds?

Ian p

[TrevorGParticipant @trevorg]

Thanks for the replies. Ians explanation fits my observations for a motor that is not microstepping. Once I add in microstepping the picture gets more confusing. Possibly because microstepping is implemeted in different ways in the publications I have read. It seems likely that the microstepping technique used by each controller manufacturer varies and needless to say they do not seem keen to explain how they do things. However the objective seems to be to reduce the torque ripple throughout the full step something that apparently they do with varying success.

For the record I am using a 200step motor with 8 microsteps. I tried cutting down to 2 microsteps but it made no difference.

The question is why does the motor run correctly if the step rate is fast enough. Can't be just stiction because the movement would never start. Possibly torque is sufficient to cause the rotor to overshoot and then pull back. This would fit well with the idea that inertial mass causes the problem. But then the problem would be changed by altering the limit current on the controller and in a test I just carried out increasing the current lowers the speed at which the problem goes way. If it is an overshoot problem I would expect an increase of current to make things worse.

It seems that my initial idea of using a toothed belt speed reduction and hence torque increase is still looking good. From a practical point it would be difficult to accomodate the 11.2Nm Nema34 stepper and I would prefer a Nema23. Using 4Nm motor and toothed pulleys from CNC4U would give a torque of 15.4 Nm. Speed would also be increased by ~ 50% but that is a side issue for me. What I am trying to decide is would this solve my problems or is there a better solution.I could use the 12.5 Nm Nema 34 from CNC4U via a toothed belt speed reduction and get a much higher torque but this solution is big and expensive. Bearing in mind I already appear to have a redundant 11.2 Nm Nema 34 whatever happens.

Can't argue with Neil about this subject being an electronics dark art. Especially when running the motor at the higher speed does NOT lose steps at start up. It was my first thought and I checked!

I must admit there is something about the idea of balancing the load with a weight strung out the back that makes it seem like a bodge. I am very keen to lift the knee so I am not ready to settle for just moving the quill.

Trevor

 

Edited By TrevorG on 05/09/2014 22:25:58

[Michael GilliganParticipant @michaelgilligan61133]

Trevor,

Just a thought …

It might be worth using a worm drive instead of toothed belts.

… easy to get a big reduction ratio in a compact space.

MichaelG.

[Les Jones 1Participant @lesjones1]

Hi Ian,
I can see no reason why it should not work with an 8 MHz crystal. If you have a problem let me know and I can try mine with an 8 MHz crystal. Neil makes a very valid point about missing steps at the start of higher speed motion. Trevor did not say when moving the knee at higher speed if it moved the exact distance it was told to move.

Hi Neil,
Balancing the weight of the knee with weights would reduce the static load but would double the mass to be moved. (Assuming it was fully balanced out.) I do not think gearing up or down with pulley system would change this fact. The best method I have seen suggested is using a pneumatic cylinder fed with a constant pressure air supply. This would require a reservoir with a volume many times that of the cylinder so the pressure did not change much when the piston moved in the cylinder.

Les.

Edited By Les Jones 1 on 05/09/2014 23:07:43

[Anonymous]

Posted by TrevorG on 05/09/2014 22:15:45:

I must admit there is something about the idea of balancing the load with a weight strung out the back that makes it seem like a bodge.

I'd class a counterweight as sound engineering; after all just about every lift and tower crane in the world has one. The bodge is mucking about with motors, belts and gearboxes, which don't address what is probably the real issue, but merely paper over it.

Never mind stepper motors; for a real black art look at RF, especially when you get into the GHz range.

Regards,

Andrew

[TrevorGParticipant @trevorg]

Hi Les,

The Knee does move the exact distance. Which is how I determined that no steps were being lost.

The effect of mass of the balance weight is a littlemore complex. While it is true that the mass of theoverall system has been increased one side of the system is lfted by the stepper while the other side is pulled down by gravity thereby reducing the effective mass experienced by the stepper. The effective mas will of course change with acceleration.

If only everything could be solved with a 10Mhz Xtal I have 10 in a draw.

Trevor

[Anonymous]

Les: I'm not sure that the increase in mass is significant, as the accelerations are pretty low. Let's assume we want to move the knee (say 100kg) at 250mm/min and we want to get there in one second, from rest. A velocity of 250mm/min is equivalent to 0.004167m/s. From Newtonian mechanics we know that:

v = u +at

The initial velocity u is zero so we get a = v/t

So the required acceleration is 0.004167m/s/s, or about 0.0004g.

We also know that F=ma, so that gives F=0.4166N

Of course the above analysis ignores the effect of friction, of whatever sort, and gravity.

Andrew

PS: If I've made a mistake I blame the bottle of home brew red wine that I'm working my through, home grown grapes too.

[TrevorGParticipant @trevorg]

Andrew,

Your maths/physics is OK. However, the assumptions… acceleration is 1 step not 1 sec so multiply by 250. As for the mass of the knee not sure of the actual weight but I suspect 100kg is rather low. Nothing on my mill is light!

It would be nice to accelerate more slowly unfortunately this involves running at low speeds which gets us back to square 1

Trevor

[MuzzerParticipant @muzzer]

The mass doesn't change with acceleration unless there is some form of warp in the space time continuum! And the addition of a counterweight will increase the total mass but reduce the static force the motor has to overcome to raise the table. That would be an improvement.

Andrew's sums are correct, as would befit a Cambridge Engineering PhD. Rather than trying to get your head around the physics, it's probably best just to get some stuff connected up and try it out!

It's a simple matter to accelerate a stepper motor. You start by sending it a low frequency pulse train (= slow speed) and steadily ramp up the frequency to a higher frequency (= high speed). You really aren't moving the table in sharp jerks – the inertia of the rotor and leadscrew ensures that the speed of rotation is reasonably smooth. The motor applies a force to the mass which results in an acceleration. Or not if there is not enough force to overcome the static and dynamic forces.

Murray

[John HaineParticipant @johnhaine32865]

Trevor, could we have some more information please? Your observation that everything seems to work with high feed rate with no missed steps seems inconsistent with the knee not moving at all at low rates. A high rate has to start somewhere after all!

• What software are you using for control?
• are you using the parallel port to feed the drivers or a motion controller? If the latter, which one?
• are you setting the speed through the max speed setting in the motor tuning config window (assuming Mach3) or through an Fxx gcode instruction?
• do the other axes behave properly?
• have you taken the motor off the feed screw and checked it works as it should at low speeds?

with more evidence it might be easier to make suggestions.

John.

[Les Jones 1Participant @lesjones1]

When I made the comment about the increase in effective mass I was thinking about its effect on one step of the stepper motor. Trevor, I do not agree that half the mass is moving up and half is moving down changes the force required to accelerate it at a given rate. Murray, The problem does not seem to be related to ramping up the speed as the problem does not occur at higher speeds. As the movement at higher speeds must start with a single step we need to know what is different in the way the stepper is driven between high speed (When it works) and low speed (When it does not work.) NOTE Trevor has confirmed that it does not miss a single step at high speed. I think it would be worth measuring the torque required to turn the knee leadscrew to see how this compares with the rating of the stepper. I also think trying to see what happens when the driver is given a single step pulse could be helpful. (Measure the coil current and note if it gets reduced from its initial value after a time. This will not be easy as the coils will be driven with a PWM signal. This would be need to be done with an oscilloscope to look at the duty cycle of the PWM drive to the coils.) Just out of interest what are your thoughts on my comment that the effective mass would be doubled no matter what the ratio of the pulley system was. ie If half the mass was used in the counterbalance then it would move twice the distance and hence move at twice the velocity.

Les.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by TrevorG on 05/09/2014 23:53:06:

It would be nice to accelerate more slowly unfortunately this involves running at low speeds which gets us back to square 1

.

… Which is why I mentioned the possibility of using a worm gearbox.

MichaelG.

[Ian PParticipant @ianp]

Posted by Michael Gilligan on 06/09/2014 08:47:53:

Posted by TrevorG on 05/09/2014 23:53:06:

It would be nice to accelerate more slowly unfortunately this involves running at low speeds which gets us back to square 1

.

… Which is why I mentioned the possibility of using a worm gearbox.

MichaelG.

A worm and pinion would work but the reduction ratio will probably be far too high. I suspect the best compromise will be in the region of 2:1 to 5:1 which is outside the worm and pinion practical range.

Ian P

[Ian PParticipant @ianp]

Les

I have a couple of questions (I know this is hijacking Trevor's thread a bit) but the 8Mhx crystal seems to work perfectly well. I do have stock of 10MHz crystal modules but as the PIC is configured for a basic crystal I am not sure it would accept a rectangular waveform (and on which input pin anyway!)

The stepper driver I am using has a 'Decay' configuration settings of between 0 and 100%. I don't fully understand what it does and what I should set it to.

Regarding setting the stepper current value, do I just set it to the value shown on the motor data sheet?

Ian P

[John McNamaraParticipant @johnmcnamara74883]

Hi Trevor G

A number Commercial CNC Machines particularly the bigger ones have counterbalanced Z axis carriages often using a weight and cable or chain. The weight is usually buried in the column so you don't see it.

Fadal for instance **LINK**

I made CNC router using toothed belts for the X and Y drive, it works well.

Some images here a couple of lines down.

**LINK**

Gear train From the above link:
**LINK**

You may have to resize the design to accommodate the weight of the Z axis on your machine. Note the motor shaft only had to carry an axial load via the coupling which drove a shaft supported in inexpensive flanged ball bearings clamped to the side plates. This machine works very well there is very little backlash.

If you can read a DXF file can send a cad file.

Regards
John

Edited By John McNamara on 06/09/2014 09:35:51

[Neil WyattModerator @neilwyatt]

Hi Ian,

Check the PIC data sheet, usually one of the two crystal connections will be usually suitable for inputting an external clock signal.

Neil

[Author]

Posts