Need to home to a middle position

[mwebster]

Greetings,

The synopsis: I'm trying to figure out how to home a linear actuator to a point approximately halfway between its two extremes.

Pertinent details:
The device is driven by a precision linear actuator with a movement per step of 1.5 microns (the linear acutator is driven by a stepper motor). Total shaft travel is ~25mm (~16000 steps).

The linear actuator we're currently looking at has an option for a built-in incremental encoder.

The complications:
If the device or control software were to lose power while inserted into its fixture, it could cause 10's of thousands of dollars in damage if the actuator were driven all the way to one stop or another when powered back on, so traditional homing to one of the stops seems to be out.

Potential brainstorming solutions:
Use some sort of home switch or optical gate to trigger when the home threshold is crossed. Not sure how to determine what side of home its on when power comes up though. Also not sure how accurate such a positional reference could be.

Use an absolute linear encoder, but I'm not sure one exists at this resolution scale that doesn't cost $$$$.

Use an absolute linear encoder of lower resolution to get it close to its middle position then a finer resolution rotary encoder on the stepper shaft to get it the rest of the way there.

All ideas and experience are welcomed ...

Regards,
Mike

 

[rhatcher]

I would suggest adding a N/O relay, hardware or software, to the power supply circuit to the actuator. This would require the operator to manually turn on the actuator with a push button.

For a power loss, the operator would have the option to check the actuator position and correct it if necessary before turning it on. For a normal start-up, the operator can routinely turn on the actuator as part of the start sequence.

 

[MikeHalloran]

I wouldn't involve the operator; very unreliable device.

Halfway is a bitch.
If one end of travel were unequivocally safe, you could just drive it 16000+ steps in that direction at powerup, and proceed from there; no transducer needed.

Given the halfway home position as a requirement, you need at minimum two switches.
- One switch is 'on' at halfway, 'off' everywhere else.
- One switch is 'on' on the 'left' side, 'off' on the right side.
The first switch tells you when you're at home.
If you're not,
the second switch tells you in which direction you need to drive, until the second switch changes state or the first switch does.
If you go 8000+ steps without finding home, you've got a big problem.

Okay, now that I think about it, the first switch is redundant. ... but the system could be fooled by a broken wire.

So you need two switches, that change state in complementary fashion.

... or, Now that I think about it some more, an absolute encoder is probably cheap relative to the damage you'd do from a crash, or what you'd pay to develop a homebrew system that was robust wrt broken wires and such.



Mike Halloran
Pembroke Pines, FL, USA

 

[jpolson]

Is there a potentiometer option? If not, try a linear potentiometer (home @5v if using 0-10vdc)? Or 2 of them as redundant safeguards? A 10k pot can have pretty darn good resolution over 1". Not sure how accurate you need to be at startup. Found some decent looking units by searching "linear potentiometer" on Google.

Does the process have some feedback to adjust the actuator position? Can you use that?

Is this plc control? If so, depending on how you write your position value you may be able to maintain the value during power loss.

I assume you will set the control unit (plc, pc, etc) to go into "standby" instead of "run" when it powers up? If so, can you manually set the position?


Joel Olson

 

[jraef]

I think your easiest option is an absolute encoder, but you don't say the resolution you need, just that each step is 1.5 microns. Is that how accurate you need to be? You obviously already have an incremental encoder, so what's the resolution on that? I doubt difference in cost between that and an absolute version can be as much as the engineering time you are expending trying to get around it in my opinion.

Cheapest using your existing components is probably Mike's suggestion of running it to one end into a sensor to reset it and then homing it to 1/2 of the known total travel.

"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376

 

[MikeHalloran]

The problem with 2 pots is that when they differ, you don't know which one is lying.
So you need 3 to do a majority vote.


Mike Halloran
Pembroke Pines, FL, USA

 

[jpolson]

"The problem with 2 pots is that when they differ,
"you don't know which one is lying.So you need 3
"to do a majority vote."

I thought about that while I was lying in bed last night. Not sure why I really thought a redundant system was necessary. I have never had one fail...so one should be just fine.

I was hoping nobody would bust me on it.

I still wonder what control system is being used. A plc data block should maintain the value of the count from the encoder through a power outage. So the controler should know where the actuator is after it is restarted. That seems like the logical place to start.



Joel Olson

 

[OperaHouse]

I would propose a simple gate type opto interrupter. Half the travel it would be on, the other half off. Feed that into a simple PIC A/D input and interpolate the voltage for center. If you have a periodic zero on your drive, feed that into the mix for a closer center. If you need better than that go back to the manufacturer.

 

[mwebster]

It's going to be driven using custom software on a PC. I suppose I could force write unbuffered positional data down to the hard drive. This might be a decent solution now that I think about it. The only thing this should miss is if the actuator was moved manually while the power was down.

The incremental encoder is built in to the actuator and the company doesn't have an absolute version. I may wind up searching for a solution from another company to get something like that.

Using a potentiometer or two for rough positioning is a good idea depending on how accurate I need the centering to be. I think it doesn't need to be that accurate for it to go "near" home at power-up to prevent fixture damage then to home it to one side or the other once it's confirmed that the device has been removed from the fixture.

I'm trying to minimize the mechanical rework. Unfortunately, my client has already done some significant mechanical fixturing before involving me in the process to do the electrical control / software and they failed to take this bit into account.

Considering the amount of damage that could be caused by this, at a minimum I'm going to push for an installation switch that lets me know when the device is inserted into the fixture.

My appreciation for all the feedback.

Best regards,
Mike

 

[niallnz]

You could try an MLDT or LVDT. Over this range both are extremely accurate, with fantastic repeatability. The're non contact transducers so don't wear out like a POT.

Cheers Niallnz

 

[rhatcher]

My suggestion is based on simplicity and safety.

I suggest installing a 'warning' reset button that activates every time power is applied to the machine or the power is restored after power failure. The operator is trained to confirm that the machine is safe to start before pressing the button.

The 'normally open', push to close, safety reset is very simple to implement in a software or hardware system and is very reliable.

The exception to reliability is human error as suggested by MikeHalloran. This is a valid concern. However, in this case I would argue that a person qualified to operate a machine that could cause $10k of damage because of a simple malfunction or improper start sequence should be able to be trusted to confirm the device position before pressing the 'warning' reset button. If not, then you have very poor operators running a very expensive machine (process).

From a safety standpoint, a power failure can happen unexpectantly at any point in time or at any point in a machine's operating cylce. Likewise, restoration of power will also happen at a point in time that is equally unexpected.

As a general rule, you do not ever want a machine to start any type of mechanical movement unexpectantly. The operator should always be required to turn on or reset the machine before a mechanical movement can take place. This is necessary to prevent injury, loss of life, and/or damage to equipment.

 

[KiwiMace]

What is the overdrive consequence of the stepper motor? Would it hurt it, or be painfully annoying just to drive it 16k steps forward and 8k steps back?

 

[fab1961]

you may keep the incremental encoder (of course provided it has enough accuracy for your application) and reference your axis at the middle of the stroke or any other point other than the mechanical limits:

1-install a cam and a switch, the cam is to be long enough to cover all the way up to the mechanical limit

2-to reference, if the switch is already made move away and look for the reference signal of the encoder (usu termed as Z or C), if it is not made then go in the opposite direction, look for the switch rising edge,reverse and look for the encoder reference
as in previous case

if the mechanics doesn't allow a mechnaical switch then you may use a proximity switch, the precision is given by the encoder signal, the only constraints are
- that either the switch or the proxy must have a repeatability better than one encoder turn to avoid aliasing
- that your system is fast enough to catch the encoder signal at the speed you are using for the reference movements

I hope I have understood your needs, this forum is very helpful to me and I try to contribute if I can