Limit Switches - optical / mechanical / proximity

[William McGuire]

For those of you using proximity switches, I have a question...

Are the distances of the switches very specific? As I see it, there is a 3/4 inch hole in the rail, but isn't there a rack directly below that hole? If so, does a 2mm range or a 4mm range switch just sense metal less than 2mm or 4mm direcly below it and cannot sense anything farther... such as the rack at 6mm? Or have I missed something?

[Gerald D]

That's right, a 4mm range proxy is oblivious to metal 6mm away.

The range is stated for iron(steel) getting within the range. For alu, the range is much less.

[Richards]

During the last few days, I've been playing with proximity sensors again. Much to my chagrin, most of what I've been experimenting with has already been discussed here in the forum. What's worse, is that I wrote some of the posts. (Oh well, when you loose your memory, everything is new and interesting - over and over and over again.)

Anyway, I've designed a small micro-controller card that accepts up to five NPN type sensors (either Normally Open or Normally Closed - as long as all sensors on the machine are of the same type). Since the micro-controller uses a computer program, the controller could be made to do almost anything necessary involving NPN proximity sensors.

The prototype card and program senses each proximity sensor and outputs a common active signal when any of the sensors goes active. It also outputs a separate output signal for each sensor.

So far, I've tested it with 12mm sensors with a 4mm range and 18mm sensors with a 5mm range, both types are shielded 3-wire sensors. I've also tested some two-wire unshielded sensors that seem to work as well as the shielded sensors. The standard unshielded 12mm sensors have a 4mm range. (To get that range with a shielded sensor, you have to buy the more costly 2X style.) All of the sensors that I currently have are Normally Open, but Fiero Fluid Power promised me four 18mm Normally Closed sensors next week.

Since it's well below freezing in the shop, I'll have time to run a few tests - if anyone has questions about proximity sensors that haven't already been answered.

[sailfl]

Mike,

I have a lot of those (senior) moments also. It is a frustrating but what can you do. Wait until the mind finds what it was looking for.

[smreish]

Nils,
Maybe now would be a good time to mention if your using Aluminum rail with the Vee rails you should use longer distance (extended distance) sensing proximity switches.
If you have steel rails, you may use any sensing distance units you want!

Happy new year.
...back to changing diapers and drinking an adult beverage. (it's amazing how well a diaper works as a coaster)

Cheers

[Gerald D]

Here is an un-shielded proximity sensor on the left and a shielded sensor on the right:

Proximity_Sensors.jpg

The "shielding" in this case refers to whether the sensing tip is surrounded by a metal shield or not. The metal shield is for the reduction of mechanical damage - it is not for the reduction of electrical interference.

For the MechMate, we are perfectly happy with unshielded sensors, which gives double the sense range of a shielded sensor. A 12mm unshielded sensor should have about 4mm range for steel (iron) and about 1mm range if sensing aluminium.

[sailfl]

Sean makes a good point. I had to go to 8mm sensors to be able to get my proximity switches to work. I could not get the 4mm to work on the X and Y.

[Gerald D]

How close did you try and mount the 4mm sensors to the alu? Theoretically you should have been able to mount at less than 1mm without the sensor actually touching the rail anywhere along its length of travel.

[sailfl]

I had problems with them being too close. Having painted the aluminum might had an effect on them being able to sense. They were so close that in some places they would rub. I kept having to adjust them so I just decided it was easier to get the 8mm sensors. They seem to be working great.

[Richards]

The standard distance for iron/steel for most flush/shielded 18mm sensors is 5mm instead of the standard 2mm for 12mm shielded sensors. Unshielded/non-flush sensors have 2X the sensing distance. 18mm is 50% larger, so they would require a much larger target hole. (I use them to sense the hex-head of a 1/4-inch bolt instead of a hole, so they work fine for my application.)

By the way, Gerald's method of always sensing the rail until a target hole is detected is a much more secure method, especially when the gantry lifts off the rails, which does happen to me occasionally when I get careless. An 18mm sensor would probably require at least a 1-inch (25mm hole).

The Balluff catalog, shows various models of the GlobalProx (inexpensive) non-flush/unshield sensors in 12mm size with sensing distances of 4mm, 6mm and 8mm. That same catalog shows various models of the GlobalProx non-flush/unshielded sensors in 18mm size with sensing distances of 8mm, 12mm and 16mm.

From AutomationDirect.com, they have the 4mm AM1-AN-2H 3-wire sensor for $18.50 (NPN, Unshielded - 4mm distance, quick-disconnect), the AM1-AO-2H sensor, 2-wire for $18.50 (NPN or PNP, Unshielded - 4mm distance, quick-disconnect). The 8mm distance models are AM1-AN-4H (3-wire) and AM1-AO-4H (2-wire) for $25.00 each. Quick disconnect cables are available in lengths of 2-meters to 7-meters for $9.25 to $14.50 each.

The 18mm sensors are the same price with sensing distances of 8mm and 12mm. Substitute AK1 for AM1 for the 18mm models.

I ordered some 2-wire sensors by mistake and found that they are probably the ideal sensor for this application, especially when you want to have complete flexibility in wiring your machine. They can be wired either NPN or PNP. They only handle 50% of the load current of the 3-wire models, but I limit the load to 10mA anyway, which is only 10% of their rated capacity. Other than that, I can't seem to find any practical difference to keep me from using the 2-wire sensors.

I have used the standard sensors (with wiring permanently attached) and the quick disconnect models. I prefer the quick disconnect. With cable lengths of 5m or 7m, I could permanently wire the cables to the machine, with one end going to the sensor and the other end going to a wiring/junction box. From the junction box, I could wire a multi-conductor cable going to the opto-isolator board in the control box. By doing that, repairing or replacing a sensor would only require unbolting the sensor from the machine and unscrewing the cable from the sensor.

The six-pin optoisolator chips that I like to use (4N25, 4N27, TIL-111) work well with 10mA current from the sensors, so I use a 1,000 ohm 1/2-watt resistor with 12VDC and a 2,200 ohm resistor with 24VDC. Resistance isn't critical. +/- 25% seems to work fine. The 1/2-watt size would be best, especially using Gerald's method of always sensing the rail. With a 24VDC power supply, a 2.2K resistor would always pull about 1/4 watt when sensing the rail. A 1/2 watt resistor would not get overly hot and a 1-watt resistor would be even better. With a 12V power supply, a 1,000 ohm resistor would pull about 1/8 W, so a 1/2 W resistor would be ideal. Usually, the 1/2 watt resistors cost about one-cent each, if you buy them in 100 quantity packages. Radio Shack sells them in packages of 5 for $1.00.

You'll probably find that a few 330 ohm, 1,000 ohm, 2,200 ohm and 4,700 ohm resistors will handle the majority of jobs when interfacing external electrical components to computer controlled devices. The 330 ohm size is ideal for controlling current to L.E.D.s when you use a 5VDC power supply. The 2.2K and 4.7K resistors are used as pull-up devices to ensure that the computer senses 5VDC when a device is disconnected or Off. As already mentioned, the 1K size can be used to limit current to about 10mA through devices when you use a 12VDC power supply and the 2.2K size works well when you use 24VDC.

The transistors/LEDs inside the sensors normally drop about 1.5V, but normally you can disregard that voltage drop, especially if you use optoisolators. The difference in sensing time is a fraction of a millisecond. Since you will normally use a fixed speed in your zeroing routine, the delay in optoisolator turn-on time will be a constant, so you won't need to even consider it.

All of the sensors that I've tried can handle a load of at least 100mA, so you could directly drive the coil of a small DC relay. However, relays cause electrical noise when they open/close, so I don't use them unless I'm switching high current AC loads. Normally, I build an interface circuit to drive DC devices or Solid State Relays to drive lower current AC devices. Electrical noise is hard to trouble-shoot, so I avoid using devices that create electrical noise.

All of the sensors that I've tried are rated to work with voltages ranging from 10VDC to 30VDC. I have tried using them at 5V, just to verify that they would NOT work at 5V. Sure enough, they don't work at 5V. (Shopbot supplied a 12V proximity sensor with their pre-PRS models that would work at 5V, but the sensing distance was critical. The luxury of having at least a 1/8-inch (3mm) gap between the sensor and the target, instead of the 0.5mm (1/50-inch) is so much better.)

[kanankeban]

I just finished reading all the thread post by post. But it seems that theirs no concenced solution am I right? Seems their are a lot of ways of doing the proximity switches...
Could someone point me out to the easiest way (schematic) for someone with poor knowledge in electronics.

[Gerald D]

Yes, this thread has rambled on far too long - a simple issue has been made to look rather complicated. any volunteers for a re-write?

[xkadet]

Fantastic thread everyone-- but I can certainly see how a non-electronic might be somewhat confused still at this point. A lot to read, as well and that's why a summary by one of you gurus would be most welcome and appreciated by the unwashed heathen among us --without electronics skills to speak of. I add just a few suggestions which you can blatently ignore if you want. Any one wanting more detail can read and read and read....

One main point that deserves a separate line of it's own, again:
Prox sensors aren't replacements for mechanical switches!!
They're completely different and functionality is much greater. You will waste your time trying to make them act like 2/axis mech swiches. One-per-axis is all you need, minimally for a reasonable prox switch setup on a 3 axis machine.

Come on guys, let's put a cap on this one. It's too good to leave hanging! And thanks again to all of you for putting forward all the effort and info. We who have benefited salute you.

What I'd like to see, personally:
1. A simple schematic using relays (one part number that works with 24v coil and right current draw will suffice) to a typical BOB with hole sensing proxs (again, one p/n that works will suffice 12-24V N0/NPN). Use as few abbreviations as possible for the unschooled of us. If a resistor is needed somewhere, indicate how to determine what R value to use.
If anything/wiring has to be shielded, please indicate so. Any wire lenght/ga limits on the sensor p/n you list above?? Is telephone cable ok?
Please don't use pin numbers on the BOBs, there's lots to choose from out there and a clearer picture is helpful here.
Wire colors on the sensors is good; brn, blu, blk.


Important note: Advice to the unshooled reading this, pull up the spec sheet on anything you're using, sensor, relay, and read the spec sheet. There's a ton of helpful info right there in condensed form.


If you want to get into And/Nand gate switching and that sort of thing... read it in the thread, it's in here. Just give something to the non electronically inclined that's simple enough to implement with out a lot of room for error. Yeah, I'm asking for a prox switch implementation for idiots.

Using the mech switches is discussed in the thread and is easy. Most BOBs show how to implement mech switches so this shouldn't be a problem. Just, as recommended, use switches with gold contacts so you'll have good contact. Hopefully hysteresis won't confuse the issue for you. This was never addressed in the thread that I found.

Use good advice and stay away from opto sensors. Nuff said. Want to know why? read.

Everyone interested and still confused should read the post above by Richard (AKA Mike) and the one below by Gerald D.

(Please don't ask another question without reading these posts


GERALD Posted in March 2008 the following GREAT Explanation:::

There was a good question in here last night, which the poster shyly decided to delete again:

"I am trying to wrap my head around the layout and wiring of either proximity sensors or the switches. I take it that the main purpose of these is to allow mach to be programed to set the router in a home position in all three axis's.

Okay if that is the goal, I do not understand wiring three or four sensors in series NO would work. Reason being, how does mach know which sensor tripped when they are all in series? "

Your car is parked in your garage and the garage door is closed. You get in the car, start the engine and you are now going to drive. Your name is Mach3, you are blindfolded. You (Mach3) puts the car in reverse and drives. Mach3 hears a bang, (the garage door is hit). Mach3 puts the car in Drive, goes forward and hears another bang (the front wall is hit). Mach3 being a normally bright person has figured out that the bang after reverse is the door and the bang after Drive is the front wall. Mach3 only needs one bang sensor.

On the router table, you can wire everything in series, to make one big sensor, because Mach3 is smart enough to know what it was doing before it hits this sensor, and therefore figures out which part of the sensor it hit.

And then, to give some perspective, this bit:
"I take it that the main purpose of these is to allow Mach to be programed to set the router in a home position in all three axis's."

The reason we manage to mostly get by without any switches is because we "home" the machine to the corner of the material lying on the table. After changing a cutter, manually jog the cutter over the corner of the material and set x and y to zero from the keyboard. Then jog the cutter till it just touches the surface and set z to zero. Then start the cut.

A home switch on the z-axis can never tell you the distance between the cutter and the material. Home switches on the x & y can only tell you the distance to the material if the material is located to the same spot everytime.

But there are times (not many) when we wish we did have proximity switches . . .
- When there has been a glitch (power failure?) in the middle of a sheet of material and we need to find the reference points again.
- When the cutter hits a clamp and the car jumps off the rail.

[Gerald D]

Keith, your second para: . . . .

One main point that deserves a separate line of it's own, again:
Prox sensors aren't replacements for mechanical switches!!
They're completely different and functionality is much greater. You will waste your time trying to make them act like 2/axis mech swiches. One-per-axis is all you need, minimally for a reasonable prox switch setup on a 3 axis machine.

. . . . has me confused.

I am of the opinion that prox switches can be replacements for mechanical switches. We need one or the other, not both.

[Richards]

I think that we're getting goals mixed up. The way I see it, there are three different 'goals':

1. Emergency Stop

2. Mechanical limit

3. Homing

An Emergency Stop (E-Stop) switch needs to kill both the power to the control box and the step pulses that are controlled by the Break-Out-Board. Two functions means two poles. A mechanical 2-pole switch is called for, i.e. one pole would open the coil on a self-latching contactor and the other pole would break the pulse-stream going from the Break-Out-Board to the stepper drivers. Because we would want the circuit to be as safe as possible, Normally Closed switches should be used, that way, we'll know immediately if a switch fails. Normally Closed mechanical switches can be connected in series, so, as many E-Stop switches can be mounted on the machine as safety requires.

A mechanical limit switch is used to protect the machine from a run-away axis. It usually only cuts the pulse stream from the Break-Out_Board. One function means that it is a one-pole switch. Again, to be safe, Normally Closed switches should be used. As long as Normally Closed switches are used, all of the switches can be connected in series. We really don't care which axis is in trouble, or even which end of the axis is in trouble. All we want to do is to stop ALL motion before something mechanical slams into something else.

Homing switches are used to detect when an axis is at a specific spot (relative to the "home" position of an axis). Repeatability is essential. Mechanical switches are NOT necessarily repeatable. Mechanical switches "bounce" (open and close a number of times each time they actuate); therefore, the time delay before a switch is activated is indeterminate. Proximity sensors work very well as homing switches. Because of the repeatability requirement, each sensor should be on its own circuit (series connections should NOT be used). Also, because of the repeatability requirement, mechanical relays should not be part of the circuit. Mechanical relays act like mechanical switches in that their contacts bounce just like the contacts of a mechanical switch. This is an excellent place to use an inexpensive micro-controller. In the 50 milliseconds that it takes for a mechanical switch to settle, even a relatively slow microcontroller can run 50,000 instructions. Homing routines are normally software controlled. Determining 'which' homing switch/sensor is active is a function of the software (program). In other words, the software might command the 'Y' axis to move in the (+) direction, then when a homing switch/sensor goes active, the program knows that the 'Y' axis has reached its (+) home switch/sensor position.

In the real world, both mechanical and solid state switches are used for homing an axis. If you use good quality switches and slow speeds, mechanical switches can do an excellent job. Personally, I prefer solid state proximity sensors simply because I've spent way too much time over the years writing de-bounce routines in software to compensate for mechanical bounce. In every case, a solid state proximity switch was more repeatable than a mechanical switch.

On my machine, I have one E-Stop switch that kills power to all coils on all contactors in the control box. I use the E-Stop switch when something bad is going to happen to me or to the machine. I know that when I press the E-Stop switch that the machine will lose its position and I also know that the part that I'm cutting might be ruined before the machine stops. The E-Stop switch is there for safety. I also have two proximity sensors that act as homing sensors as well as limit switches. When used as homing sensors, the software routine knows exactly how far from 0,0 the targets are when a sensor is detected. The limit switch function stops the machine via built-in software if either the X-axis or the Y-axis goes too far. The limit switch function acts much like the E-Stop function in that position is lost and the part may be ruined AND the power to the contactor coils stays ON.

One other point needs to be made: There has to be some type of switch/sensor over-ride or bypass. On my machine, software allows me to slowly jog an axis away from a proximity sensor. Without that software routine, I would need a mechanical on/off switch to bypass the limit switch/sensor.

[Gerald D]

Mike, Mach3 has a software routine for overiding a limit switch.

[xkadet]

you don't replace limit switches post design/install with a prox switch. Design for prox or mechanical, but they aren't strictly interchangeable, different method of application and operation.

Sorry to start up more confusion here.

[Richards]

Gerald,
I'm going to learn more about Mach 3. I promise . It has so many possibilities that will reward those who spend a little time studying.

Keith,
The problem with all electronics is that using simple, off-the-shelf parts is not always the best way to solve a problem - like limit switches. Relays have been used for decades in controllers of all types, but today's process control computers just don't like the electrical noise that a relay contact can produce, particularly when the goal is to accurately and consistently set the "home" position. Some users are starting to have issues with electrical noise. If they're like me, that's when it's time to read and re-read every post that talks about grounding. But, eliminating a cause of electrical noise is just as important as funneling that noise away.

My basic advice to anyone building a Mechmate would be to include a proper E-Stop switch but adding limit and/or home switches/sensors can be delayed. Granted, using a proximity sensor to find home is quick and painless, but it doesn't take hardly any time at all to manually pull the gantry against its stops and then jog to the 0,0 position. I ran my machine for about a year without attaching the proximity sensors. I hit the stops a few times, but no damage was done (except to my ego). The sensors make my life a little easier, but, if they stopped working, I wouldn't shut the machine down while I waited for replacement parts.

[sailfl]

Mike

After reading about some of the issues that ShopBot users experience with the controller software and that they can't do some simple things that Mach3 allows, I am grateful that Mach3 is what I have to use.

[xkadet]

Ok, I get it. Now we're left with just use steppers and forget the limit sensing unless you're willing to do the head-work to figure the further refinement of limit sensing. Seems like most of all that you need to do proxs is in this thread already, but no debounce discussion of length that I've seen. However, there's a great debounce "treatise" here for those still not so fatigued to continue this thread...
:http://www.ganssle.com/debouncing.pdf
you need a 7414 shmitt triggered hex inverter, couple resistors, a diode, a cap. The small circuit is spelled out along with values for the parts. Thanks to Jack for putting this together.

[Gerald D]

Keith, a couple of guys here use proximity switches driving quality relays to limit and home their machines with trouble-free success.........what value are you trying to add to that?

[Richards]

Keith,
Thanks for the information. Figure 1 on page 11 is a classic circuit that I've often used. It uses a simple 7400 NAND chip (which is the chip upon which most TTL logic is based). The 7400 requires a N/O and a N/C contact on the switch or the relay, but it is a simple circuit.

However, my feeling is that if I have to add a circuit, I'll just use a proximity sensor with an opto-isolator to handle the two different voltages. By doing that, no debouncing is required. Twelve to twenty-four volts is supplied to the proximity sensor and five volts is fed to the computer via the opto-isolator.

[javeria]

Originally Posted by Richards

Twelve to twenty-four volts is supplied to the proximity sensor and five volts is fed to the computer via the opto-isolator.

That's how it is set up on both of my machines and works flawlessly.

[xkadet]

Thanks for the positive input and all the suggestions. Completely altered my sensing strategy based on the inputs and reading this thread. I'm really just casting about trying to figure out what's a reliable approach that won't make things miserable down the road or take forever to implement.

I think most everything anyone would need to implement a good limit sensing strategy is here, now. I've attached a fairly lousy pict of my router as of a week ago. Now all I've got to do is put in the work table and wire the electrical, and then the fun starts. Going from 6 sensors to 3 will be a treat.

regards,

[isladelobos]

HI all.

Reading about This Pdf, I think not is possible only one Relay?

[martin77pl]

I am reading this thread as well. I received my proxies yesterday. Them interface relays shown in post 50 by Gerald are made in Poland. Do I need four of them as there are four proxies?

[Gerald D]

Yes, you need a relay for each proximity switch - 4 relays in total

[gixi]

I found some time ago this item on eBay and I'm wondering if is good for our project. On eBay the item number is :330322787424. And the seller is carolbrent. He also sell limit switches. All at good price. But again: this is good for MM project ?
He sell a kit : CNC stepper motor Limit BOARD COMBO WITH 6 SWITCHES kit.
It looks OK from my point of view, and most important is easy to get conected al switches.

[Gerald D]

That board is for mechanical limit switches, not for proximity switches. Mechanical limit switches have mounting and wood dust problems.

[cncb]

I apologize if this has been answered, I read this thread in its entirety! Is the reasoning for 4 proxy switches so that there is a second proxy switch on the gantry for reasons of autosquaring? Thanks.