What's new
What's new

OT: Repeatability of limit switches

teemfan93

Cast Iron
Joined
Dec 5, 2015
We are working on a design where we are hoping to use limit switches to define our home positions. We need pretty good repeatability for defining this and don't want to use hardstops to achieve it. Are there any switches you have used in the past that you would recommend? Thinking something like what is used in a tool setter may work?
 
Any reason you can't define your home as "last index pulse before hitting the limit switch"? That tends to be more repeatable.

Agreed, only most CNC's use first strobe pulse after home switch comes off the ramp. The ramp initiates the slower feedrate. Coming off the ramp is when the control looks for the strobe.

Most cheap junk CNC's like lasers and routers use a low cost proximity switch for home location. Very crude, but it works within 20 thou or so.
 
I would recommend the micro switches. I have seen them used in a number of devices to detect the end of motion. I can not give any number on their accuracy but can say that how they are used/mounted can contribute to that accuracy or the lack thereof. You can see a number of different brands here.


Many of these micro switches have a arm which extends from the basic switch. These arms can be plain or can have a roller at the end. These arms are made of spring metal and will allow a certain amount of over travel after the trip point is reached. This can be good if your mechanism does not stop immediately, but it can also decrease the accuracy as the arm acts like a lever which magnifies by a factor of 2 or more the uncertainty of the trip point.

For the best accuracy, it is better to use the basic micro switch, without any spring arm at all. Then, to provide for the over travel, either mount the micro switch on a spring loaded mount or add a spring to the moving element which contacts the activator button. In either case, the external spring should be stiffer than the internal one in the switch. This allows the switch to activate first and then the over travel allowance kicks in so neither the switch nor the moving part is damaged.

If the standard micro style switch is not accurate enough, there are switches that are sold as "limit switches" and I would assume, due to their price, that they are a lot better. But hang on to your hat when you see the prices.

 
DeVlieg used physical limit switches 75 years ago and guaranteed +/- .0002. I've seen 80% .0000 and 20% .0001. I've never seen .0002. And the machine was already 50 years old at the time
 
Tool setters are just really nice limit switches. With limit switches speed is important. Different speed can result in a different perceived location
 
Many of these micro switches have a arm which extends from the basic switch. These arms can be plain or can have a roller at the end. These arms are made of spring metal and will allow a certain amount of over travel after the trip point is reached. This can be good if your mechanism does not stop immediately, but it can also decrease the accuracy as the arm acts like a lever which magnifies by a factor of 2 or more the uncertainty of the trip point.
Many have longer arms which make them more accurate. You can also use a cam on the moving element to activate the roller on the sensor which allows for overtravel. Put the hard stop just past it. Call around, the make plently of fancy sensors (Turck, Banner, Baumer, etc.). A basic prox switch should get you .010" or better.

 
My 2 Dyna Mectronics 2800s use limit switches to home. I usually run up on the switches a couple of times to warm them up, then home. I then jog into them a couple times to verify the over travel alarm repeats. These switches do repeat within .0001. If you can find one the design would be worth copying.
 
My 2 Dyna Mectronics 2800s use limit switches to home. I usually run up on the switches a couple of times to warm them up, then home. I then jog into them a couple times to verify the over travel alarm repeats. These switches do repeat within .0001. If you can find one the design would be worth copying.

That sounds really impressive until you google what a Dyna Mectronics 2800 is.
 
That sounds really impressive until you google what a Dyna Mectronics 2800 is.
I thought limit switches were being discussed. The machine is not very fast and tool changes are done manually. On the plus side when you press on the spindle housing an indicator shows less deflection that a Bridgeport shows with the same pressure. Well, I guess that is not saying much though.
Back to the switches, the one I had apart could be made in a shop without too much difficulty. I just turned the machine on and bumped the x switch 5 times in a row. The machine stopped at -.0004 each time.
The switch is made from approximately 3/4 inch square stock and about 2 inches long.
 

Attachments

  • 2800Switch.jpg
    2800Switch.jpg
    270.5 KB · Views: 9
  • AHHAScreenRS.jpg
    AHHAScreenRS.jpg
    275.4 KB · Views: 9
Last edited:
DeVlieg used physical limit switches 75 years ago and guaranteed +/- .0002. I've seen 80% .0000 and 20% .0001. I've never seen .0002. And the machine was already 50 years old at the time

Yes, but the Devlieg sequence for touching off the limit switch is pretty complicated. They don't just run the table past until the switch clicks ...

All my nc machines have done something like, rapid to first limit, drop back to a slow feed until the axis kicks the second limit, then go to the nearest null point on a resolver or marked point on an absolute encoder. The limit switch just got you close.

That may be overkill for what youre doing, but at least the two switches system has merit.
 
I must differ. It is basic lever theory. The switch activation button is the first thing from the pivot or fulcrum. It has a SHORTER lever arm. The end of the arm is at a greater distance from the pivot of fulcrum so it has a LONGER lever arm: for the sake of discussion, lets say it is twice as far away.

Now, if the error at the switch activation button is +/-0.001", that is at the SHORTER distance from the pivot. The error at the end of the arm will be TWICE that amount because the longer lever arm means there is twice as much travel as at the shorter lever arm. So the error becomes +/-0.002" or twice as much as the basic switcher's error.

While these positions could be reversed in order to get more accuracy, the manufacturers of the micro style switches do not seem to do so. At least I have never seen one like that. If you know of any, please do share that knowledge.



Many have longer arms which make them more accurate. You can also use a cam on the moving element to activate the roller on the sensor which allows for overtravel. Put the hard stop just past it. Call around, the make plently of fancy sensors (Turck, Banner, Baumer, etc.). A basic prox switch should get you .010" or better.

 
It is really fun answering these questions where the OP does not give any better requirements beyond "pretty good repeatability" or something of that sort. I know I am not the only one here that thinks in terms of +/- followed by actual, definite numbers.

And then, there's price. "Oh, that's great, but it costs $xx,xxx. Our budget is $0.25. Or vice-versa. Just about every thing I have ever done in my professional life has had a budget.

I keep telling myself not to answer these posts, yet I still do. Probably a character fault. I hope it's not fatal.
 
It is really fun answering these questions where the OP does not give any better requirements beyond "pretty good repeatability" or something of that sort. I know I am not the only one here that thinks in terms of +/- followed by actual, definite numbers.

And then, there's price. "Oh, that's great, but it costs $xx,xxx. Our budget is $0.25. Or vice-versa. Just about every thing I have ever done in my professional life has had a budget.

I keep telling myself not to answer these posts, yet I still do. Probably a character fault. I hope it's not fatal.
I am here and following along with all of the responses. Right now a lot is still in flux so it's hard for me to give a specific requirement of X. We are currently using optical sensors similar to the Keyance ones mentioned previously but the customer is pretty adamant they want mechanical limit switches. Pricing is not too much of a concern within reason -> switches in the $10-20 range are fine, but I am not going to be able to justify the $150 switches with micron repeatability.

I found a few through digikey and dug up the spec sheets. Omron makes some plunger style with repeatability of +/-0.2mm at around $8 a switch. I would like something closer to +/- 0.05 or 0.1 but it's close and we can likely make it work. I would prefer to avoid the lever style switches as something always seems to happen while you're building the unit, bump or bend the lever and your home is now offset.

I am also planning to take some of the responses to our EE/SW to chat through homing strategy.
 
It is really fun answering these questions where the OP does not give any better requirements beyond "pretty good repeatability" or something of that sort. I know I am not the only one here that thinks in terms of +/- followed by actual, definite numbers.

And then, there's price. "Oh, that's great, but it costs $xx,xxx. Our budget is $0.25. Or vice-versa. Just about every thing I have ever done in my professional life has had a budget.

I keep telling myself not to answer these posts, yet I still do. Probably a character fault. I hope it's not fatal.
totally agree, we also need to know how much time the machine has to establish a true zero. Ideal way is to touch the moving cutter on an acoustic sensor, but that costs serious time and money.
 
Last edited:
I would prefer to avoid the lever style switches as something always seems to happen while you're building the unit, bump or bend the lever and your home is now offset.
Perhaps you could touch off a part feature to find the part zero rather than the machine zero? This way part setup is less critical. Optical is another option.
 
The switch activation button is the first thing from the pivot or fulcrum. It has a SHORTER lever arm. The end of the arm is at a greater distance from the pivot of fulcrum so it has a LONGER lever arm:
You are correct, my bad. You get more movement at the end of a longer level, but not more accuracy. Thanks for clarifying.

I've seen lots of home switches done with a basic prox. I believe the encoder takes over after that as EG describes. Depends on lots. I was getting .0005" repeatability on a tool touch pad I redesigned on a CNC router which was two pins touching, one spring loaded that would move then the tool touched off. I don't think the control even did a skip function.
 








 
Back
Top