Force Feedback on a Stepper Motor?

[ElectricalHammer]

Not totally PLC related, but I figured this would be a good place to ask:

I have an application where 7 small parts are being gripped and then held together to get welded. The final assembly is like a 3 footed chair stand, but only about 150mm in diameter: a central cone/cylinder, 3 legs, 3 feet.

Anyways, I have a coworker suggesting we use stepper motors because he thinks they will vibrate (jitter/dither) less when holding the parts against each other. Fair enough, steppers are cheaper as well. I was thinking with servos we would advance the parts to hold them together and then turn them off and rely on the motor brakes.

My issue is that steppers don't really have the longevity of a servo, and unless an encoder is added, they don't really have any feedback. Okay, we can add an encoder to a stepper, but one part of the poka yoke I was considering is after all 7 parts are loaded, I wanted to do a pre-check by advancing all the parts against each other until they hit a given torque (force feedback) to check that the parts are loaded properly. To the best of my knowledge, there is no force-feedback for stepper motors, is there?

 

[JRW]

Go with a servo

 

[GaryS]

I would go with a good VFD set up a vector drive running in torque control set it up for position holding

 

[kolyur]

My issue is that steppers don't really have the longevity of a servo

This has not been my experience. We use five-phase Oriental steppers and drives, and they are very robust when applied properly--low speed, high torque applications. We've had significantly less failures with these when compared to various brands of servo hardware in other applications. Just don't try to push the limits of the torque-speed curve.

Steppers are great when you need rock-solid holding power. Once the stator teeth are lined up with the rotor, it essentially acts like a mechanical brake. Compare this to a servo, which has to "give" a little in order for the controller to recognize the movement and respond with more current.

The disadvantages are as you mentioned: no position feedback and no force feedback. In properly sized applications you don't really have to worry about missed steps except in a crash situation, and there can be other ways to detect that. In some applications I've used a double shaft stepper and put an encoder on the back, but this is only to recognize when skipped steps occur and trip an alarm.

 

[ElectricalHammer]

This has not been my experience. We use five-phase Oriental steppers and drives, and they are very robust when applied properly--low speed, high torque applications. We've had significantly less failures with these when compared to various brands of servo hardware in other applications. Just don't try to push the limits of the torque-speed curve.

Steppers are great when you need rock-solid holding power. Once the stator teeth are lined up with the rotor, it essentially acts like a mechanical brake. Compare this to a servo, which has to "give" a little in order for the controller to recognize the movement and respond with more current.

The disadvantages are as you mentioned: no position feedback and no force feedback. In properly sized applications you don't really have to worry about missed steps except in a crash situation, and there can be other ways to detect that. In some applications I've used a double shaft stepper and put an encoder on the back, but this is only to recognize when skipped steps occur and trip an alarm.

Thanks for the ideas. Well, the Poka Yoke I'm thinking of is basically a 'home to torque' to check the positions of the parts before they get welded, which sounds a bit like the 'crash situation' you mentioned.

I was thinking for normal operation, when holding the part to 'weld', the servo would move to the contact position is was just in and then shut off, engaging the brake and holding everything in position without 'vibration' while the gas torches weld the parts together. Gosh, the more I talk about it I almost wonder if just a pneumatic cylinder with a precision micropulse transducer wouldn't be enough, maybe a rod lock just to hold things in position while welding.

The other aspect of this is that I'm talking really low torque... maybe 20lbs max being applied to the part, in case that matters. I'm thinking this is why the coworker suggested steppers unless I can find some very low power servos. Just FYI.

 

[arpus4KM]

Not totally PLC related, but I figured this would be a good place to ask:

I have an application where 7 small parts are being gripped and then held together to get welded. The final assembly is like a 3 footed chair stand, but only about 150mm in diameter: a central cone/cylinder, 3 legs, 3 feet.

Anyways, I have a coworker suggesting we use stepper motors because he thinks they will vibrate (jitter/dither) less when holding the parts against each other. Fair enough, steppers are cheaper as well. I was thinking with servos we would advance the parts to hold them together and then turn them off and rely on the motor brakes.

My issue is that steppers don't really have the longevity of a servo, and unless an encoder is added, they don't really have any feedback. Okay, we can add an encoder to a stepper, but one part of the poka yoke I was considering is after all 7 parts are loaded, I wanted to do a pre-check by advancing all the parts against each other until they hit a given torque (force feedback) to check that the parts are loaded properly. To the best of my knowledge, there is no force-feedback for stepper motors, is there?

Like other said, go with servo.

SMC makes some that specifically use a lot of force feedback for homing and operation. and they generally make them at a much cheaper price because they aren't a 1 size fits all.

 

[Peter Nachtwey]

I would do things much differently.
The stepper motors are OK as long as they can hold position and you KNOW what that position is. I would use pneumatic cylinders to apply the force. So the opposing cylinders are the stepper motor and the pneumatic cylinders.


You can't have opposing cylinders controlling force. One much control position while the other controls the force.

 

[GaryS]

I am going to have to the odd man on this one
Your best choice for this is still a good vector VFD

Let’s look at the stepper motor control system. By design the stepper motor system is to be used for positioning control if the motor can’t make the step because of the load torque requirements then it will go into an overload No torque control here it’s either go to the commanded position or go into overload. Not a good solution here because you will never actually get the commanded position.

Now let’s look at the servo system for a servo system a servo motor is required, a servo drive is also required and the servo driver are not controllers they just supply power to the motor according to the commands from the controller. So you will also need a servo controller and most servo controller do not include an operator interface (HMI) so you will need to have one of those.
So now you have 4 separate items you need just to get started and then you still have to setup and program everything.
For many years servo systems were used torque control to move the motor but the controller supplied the torque command based on the position and speed feedback. Newer servo systems do control speed using speed command but to do what you want you still need torque control.
And cost wise I can see the cost of a servo system could quickly exceed $6,000 for a very small motor and you still have to put it all together and program it.
As a side note: I found out recently the machine builders are going away from servo systems in favor of flux vector VFD’s they are much cheaper and easier to setup.
Anything you can do with a servo I can do better and cheaper with a good VFD

No matter what control system you use after you calculate the output torque you need, multiply it by 1.5 or more and use that number for you motor and drive sizing, remember that if your requirements require torque at zero speed the motor will generate a lot of heat you must account for it.

 

[parky]

I sort of agree with Gary, I have seen many positioning systems where over updated builds VFD's have replaced steppers/servo's.
At first when this post came out I was going to suggest loadcells then thought better of it as welding (I assumed elecric arc) can damage sensitive loadcells & amplifiers. however, after reading that this is gas weld it is another possibility but that depends on many factors like how they could be mounted etc.
I have used this in conjunction with a VFD not on welding but compressing pallets before strapping, basically the VFD ran at xx speed, a sensor detected when the press head was within xxmm of the pallet, slowed the VFD, stopped when the pressure reached a certain level. It was a special requirement of one of their customers normally the strapper would detect the top of the pallet and compress until a second sensor was made but the customer wanted a certain tension on the straps.
Others were a gear system where a bush was welded onto a gear, there were two systems both different in how they worked one used a linear transducer to detect when the parts were in place for welding, the other used an electrical connection so when in place there was a current (well voltage to detect the parts were together). no details as I did not program them originally, just did some mods for storing a few days production on good/bad parts etc.

 

[Peter Nachtwey]

If you guys want to get that expensive the VFDs are not the way to go. Hydraulics are the way to go. We have lots of experience and pressure force control. The advantage of hydraulics over any electric motor is that hydraulics require little if any energy when holding a force whereas a motor must have lots of current going through the armature to maintain a torque.


However, the OP didn't say his need for force was large so a stepper motor can offset the force of the pneumatics with the force/torque of the stepper.


We don't know how much force is necessary to grip the items the OP wants to grip.

 

[parky]

I always bear with Peter, he seems to be more of an expert than most.
However, it does appear to me that it is not the grip tht the OP is talking about but pressure of when the parts mate, which reminds me of some other system I have come across, very simple the parts are married but the grippers have a tension spring with a limit so as the parts touch the gripper is mounted on a slide & spring as the part mkes contact, the slide moves back as the gripper moves forward, when it hits the sensor you know tht the parts are mated & given a known pressure the compressed spring is under when the sensor is triggered hey presto you have a repeatable pressure holding the parts together for welding. (that is assuming regular ppms for replacing worn parts is done.
I think this would be an ideal, cheap & reliable way of doing it.

 

[ElectricalHammer]

I always bear with Peter, he seems to be more of an expert than most.
However, it does appear to me that it is not the grip tht the OP is talking about but pressure of when the parts mate, which reminds me of some other system I have come across, very simple the parts are married but the grippers have a tension spring with a limit so as the parts touch the gripper is mounted on a slide & spring as the part mkes contact, the slide moves back as the gripper moves forward, when it hits the sensor you know tht the parts are mated & given a known pressure the compressed spring is under when the sensor is triggered hey presto you have a repeatable pressure holding the parts together for welding. (that is assuming regular ppms for replacing worn parts is done.
I think this would be an ideal, cheap & reliable way of doing it.

Currently the manual process involves the operator loading all the parts into a fixture, checking, measuring, etc... then the parts are 'retracted' and one by one they are heated up by a gas torch until they are ready to 'stick'; at which point the heated part is manually advanced (fixture uses linear slides) until it makes contact with the other part (a hard stop currently controls the stroke) and a thumb screw is tightened at that point to maintain the position and pressure. As for how much force... no more than 20lbs, probably less than 10. The thumb screw is tightened with some force against the part present, but I'm sure the force changes as the hot working takes place. The precision of the completed part is critical though; were talking 0.03mm tolerances, so using basic prox switches isn't going to cut it; I'm thinking something like balluff micropulse transducers for position feedback if a servo isn't used.

 

[Peter Nachtwey]

I'm thinking something like balluff micropulse transducers for position feedback if a servo isn't used.

Now you are on the right track if you are doing linear motion.

The force you want is quite low and that is a huge problem. Normally I am dealing in tons but I/we have controlled down to 10 lbf before. Most of our applications are hydraulic with a pressure sensor on the cap and rod side. The force is then measure as Net_Force=Pressure_CapXArea_Cap-Pressure_RodXArea_Rod plus or minus friction. The friction is the problem as it is often WAY more than 10 lbf so a load cell is required instead since they measure applied force. The problem with load cells is that they have a half life of about 2 milliseconds in an industrial application but the OPs application appears to be almost lab like so I think load cells are the way to go.

I come problem with load cells is that people don't buy amplfiers that are fast enough for force control. Also, a thick rubber pad will add some compliance so the force will be easier to control. I just wonder how far the rubber pad can be from the heat.

I have given some hints as to where I would start but I still don't know enough to make a hard recommendation.

 

[parky]

I don't think you understand what I mean the prox is just to sense how far the slide/spring has travelled i.e. imagine the part is advanced, it touches the part but the arm still keeps going forward, the part is pressed up against the other part which causes the spring on the slide to compress effectively not moving the part but the slide moves back, hits the sensor so it knows the part is in contact with the other, at this point the motor/pneumatic whatever drives it stops the part is then at the pressure of the spring chosen to give the right amount then perhaps clam the position to stop it moving.
Did a similar thing on a pack tester the original was based on this, the packs had a film seal over the tray whis was domed due to the gas pressurewhen the inert gas was introduced during sealing, the rod & pad was moved sown onto the pack seal to a given position, if the seal was leaking or no gas then it would pass the sensor & this was deemed a fail, we modified it by using a spring, linear transducer & the sensor, originally due to different sized packs they had to manually alter the length the rod would travel before hitting the pack, using the spring & slider on the rod meant that any sized pack could be tested, there was a motor that drove the rod start position based on a selection i.e. recipe for each type of pack but the repeatability was very good, so in effect the rod pushed on the pack & the spring allowed the rod to stay in position while the main body of the tester carried on, the linear transducer depicted the amount of travel after the start of spring compression. if there was no compression due to seal broken or it suddenly shot forward i.e. seal breaks then it was deemed a fail, simple but effective.
If you like it was equivelent of ashock absorber but just a spring that upder a certain tension it made a prox this was the test pressure trials with different strength springs allowed even a reasonable tension over sizeable amount of spring compression I cannot remember the actual pressure but as an example it had to be within 1.4kg +- 0.2kg. without going past the sensor i.e. a pack that had a leak or no proper lidseal.

 

[BryanG]

until they hit a given torque (force feedback) to check that the parts are loaded properly. To the best of my knowledge, there is no force-feedback for stepper motors, is there?

I have a 3D printer and they level the bed by driving it up until it hits the print head. They look for an increase current taken by the Stepper Motor, so they are using force feedback. I have no experience or advice to offer otherwise, but that bit caught my eye.