[ftkalcevic]

What's the most accurate switch/sensor when it comes to repeatability?

Background...

I want to build a tool Z height sensor. I don't want to use the electrical contact type (tool completes the circuit to ground). I want it to be either a mechanical switch, or optical or other proximity sensor.

It will be a plunger type design, either hitting a mechanical switch, or moving past an optical sensor (reflective, or slot), or an inductive or capacitive sensor.

(I also plan on putting a limit switch at the bottom of the travel to stop the system - I have broken many expensive tools because of touch off mistakes).


I don't know what repeatability I need to aim for. I'm thinking 0.01mm or better.


Searching through digikey, I can't find much information on repeatability of optical sensors. This one, the omron EE-SX670, (datasheet here http://www.ia.omron.com/data_pdf/dat...eet_csm483.pdf) has a graph on page 4, "Repeated Sensing Position Characteristics" but I don't understand what each of the deltas represents. 0.002mm repeatability looks good, but 0.04mm is a bit much.


Digikey also has a precision limit switch, also an omrom D5A Series, (http://www.ia.omron.com/data_pdf/dat...eet_csm591.pdf) which has a repeatability of 0.001mm, but is $198 - way too much.



I've seen the DIY home/limit switch thread on this forum, but reading about the hysteresis of the magnets makes me question the accuracy.


Any thoughts?

Frank

[underthetire]

Most tool sensors I have seen, BIG, renishaw, metrol, etc, all use a mechanical type, however it's more like a pin that is spring loaded against some ball bearings. When the shaft of the pre setter is pushed down, it breaks the contact (normally closed) giving the skip signal to the control. Also, the break away stylus is a nice addition, it has a small link between the switch body and sensor head that breaks with too much force, but is cheap and easy to fix.

[Al_The_Man]

There are some fairly high precision probes such as heidenhain multi direction measuring probes that use a an optical method, the probe sits in a gimbal which is the fulcrum point so with the probe being around 3" in length the opto on the opposite short side of the gimbal, the measuring distance is magnified.
Other makes follow a similar system.
Al.

[underthetire]

Actually, our most accurate, down to micro inch, is the good old lvdt probes. You think 200 is expensive though, start pricing a lvdt set up! Think the last one with amp was ~6k. Ouch.

[RomanLini]

I designed a Z height tool setter with mechanical break contact, that offers very high repeatability.

It can be easily cut from some 12mm flat stock as a 2D cut.

You can see it in this thread, about half way through;
http://www.cnczone.com/forums/cnc_wo...ll_router.html

It's repeatability is better than 0.01mm and it easily measures the thermal expansion in my spindle shaft and tools once they get warm.

[ftkalcevic]

I designed a Z height tool setter with mechanical break contact, that offers very high repeatability.

It can be easily cut from some 12mm flat stock as a 2D cut.

You can see it in this thread, about half way through;
http://www.cnczone.com/forums/cnc_wo...ll_router.html

It's repeatability is better than 0.01mm and it easily measures the thermal expansion in my spindle shaft and tools once they get warm.

That's a great simple design. The only issue is it wont hand over-travel very well. I didn't really spell out my requirements, but I also want to be able to set fine PCB milling cutters and drills.

I've never seen LVDT sensors before. Googling suggests they are pretty simple (conceptually) - Macro Sensors - LVDT Basics - What Is An LVDT? - How Does An LVDT Work?. I wonder how DIY they are.

[mcarvey]

You should be able to get that level of repeatability with any of those 4 types of sensors, if they are implemented in a well thought out manner. The key is to understand the principles that govern their operation and then design accordingly. For what you're trying to do I think that either mechanical or photo-interrupter would be the easiest to implement since they are likely to use less external circuitry.
I also looked at the EE-SX670 datasheet and agree that the "Repeated Sensing Position Characteristics" seems to be missing some vital information. I was able to find a more detailed databook about photomicrosensors that sheds some light on what the graph means. Page 25 should be helpful, as well as searching for that phrase.

http://www.google.com/url?sa=t&rct=j...EcSEewOOg5Uwuw

I'm not an expert on this, but I'll share my guess as what the deltas could represent. First some terminology. So the photosensors have a light emitter and detector, seperated by some a distance (lets say 4mm) Suppose we had a perfectly opaque, wafer thin interrupter plate. If we place the this interrupting plate right next to the emitter, lets call it d_e+0mm. 1mm from the emitter would be d_e+1, etc. d_e+4mm would be right next to the detector.
The edge of the interrupter plate creates a shadow, and since the light source is not perfectly narrow the shadow will be crisper if the interrupter plate is closer to the detector, and fuzzier if it is closer to the emitter. Since the photodetectors work based upon a light threshold, it would make sense that the interrupter plate's position in the slot has some affect on the repeatability. I could be wrong though.
Then again, the graphs don't really show anything to support this, so it may just be temperature dependent. Also, It's also not clear what impact the no. of repetitions means. Is that the worse case? Average?

Hope this helps,
-Matt

[Karl_T]

I designed a Z height tool setter with mechanical break contact, that offers very high repeatability.

It can be easily cut from some 12mm flat stock as a 2D cut.

You can see it in this thread, about half way through;
http://www.cnczone.com/forums/cnc_wo...ll_router.html

It's repeatability is better than 0.01mm and it easily measures the thermal expansion in my spindle shaft and tools once they get warm.

I went looking through your thread and couldn't find it. I did find a broken link to a Z height on page 7. Was that it?

Sure like to see what you did

Karl

[ftkalcevic]

I went looking through your thread and couldn't find it. I did find a broken link to a Z height on page 7. Was that it?

Sure like to see what you did

Karl

This is a direct link to it.

http://www.cnczone.com/forums/740808-post18.html

[RomanLini]

That's a great simple design. The only issue is it wont hand over-travel very well. I didn't really spell out my requirements, but I also want to be able to set fine PCB milling cutters and drills.
...

It handles overtravel of 1mm or so, which is totally irrelevant as it detects the descending tool to 0.01mm accuracy and always stops the descent.

It actually stops the tool before you can see any visible deflection of the top plate, my friend says it's "magic".

Why do you need overtravel? Do you expect the sensor to fail? As a break sensor it can't fail on a descending tool, it is fail-safe in that direction.

I use it often for very fine tools like PCB milling tools and micro endmills etc. It's given 12 months service with zero defect and zero wear (well no wear I can detect).

Thanks for providing the direct link, the tool height setter is in post #18 of my build thread.

[ftkalcevic]

Why do you need overtravel? Do you expect the sensor to fail? As a break sensor it can't fail on a descending tool, it is fail-safe in that direction.

I have many years of inexperience when it comes to touch off probes. I've broken some expensive cutters and damaged a nice touch off gauge because of operator error, bad programming and faulty hardware. As a hobbiest, I need all the protection I can get.

[RomanLini]

Yep I respect that. I still don't understand what you think would fail?

The tool comes down, and when it kisses the sensor plate it breaks open the contacts, and the machine stops. Unless you think the software will fail? If the wiring fails open circuit (or the contacts are dirty) the tool would not start to descend so fails are in a "safe" manner where the tool does not descend.

[Al_The_Man]

As an observation, maybe it should be noted how industrial probes act in the case of G31 and how the macro's are written.
All industrial type probes I have worked with, Renishaw, Heidenhain, Mitotoyo etc, have a fair degree of over travel allowance in all axis after the trip has taken place.
This is mainly to protect the probe.
In a system such as these, the G31 is triggered by High speed skip input that causes the controller to capture the axis in a set of registers.
In a typical moves sequence the probe is moved initially at high feed rate to the initial measuring point, this includes what is known as a protected move so that if the probe were to collide with a jig or part, the skip input would trigger and the next move would be an instant stop.
This hysteresis of the probe allow for any over-travel that may occur after the skip has taken place and to avoid damage to the probe if an instant stop does not occur.
Many probes such as Renishaw are many thousands of $$ so this safety margin is built in for that reason.
Al.

[RomanLini]

Thank you Al I appreciate your time to explain the situation and now I understand the problem a lot better.

I probably get a bit insulated from the "real world" products and their issues as I write my own PC CNC software, and that communicates with my own microcontroller software that is inside my CNC machine and make sure it's all a custom application that works together instantly and without fault.

To ftkalcevic- Sorry for not understanding the need for a larger amount of overtravel! If you were still interested in my Z height sensor design it would be very easy to increase the amount of overtravel by increasing the clearance under the top moving plate, and just ensuring the spring system was reliable through a larger range of travel. I think 1/2" of travel would be easy enough to attain while still keeping the same benefits of a hard top plate and break style contacts.

[ftkalcevic]

To ftkalcevic- Sorry for not understanding the need for a larger amount of overtravel! If you were still interested in my Z height sensor design it would be very easy to increase the amount of overtravel by increasing the clearance under the top moving plate, and just ensuring the spring system was reliable through a larger range of travel. I think 1/2" of travel would be easy enough to attain while still keeping the same benefits of a hard top plate and break style contacts.

No offence taken. The first cutter I smashed was when converting the Mach3 probe routine from imperial to metric. It worked fine - touching off and setting the Z height to 50mm, but then it tried to rapid travel to a safe height of 5". That was my mistake - it should have moved to z125, not z5 - it tried to go to z=5mm, 45mm into my height sensor.

I was looking for a solution to stop stupidity like that - a limit switch to shut down after 5mm of over travel. I do like your solution - simple and accurate. But I'm also fascinated by the LVDT probes. I want to try and build one of those first.

[ger21]

I was looking for a solution to stop stupidity like that

No offense, but it's awfully hard to stop stupidity.

[underthetire]

The lvdt was actually invented where I work, way back before I was even born. They were originally hand made and hand wound, so yes you could build your own. There are some off the shelf amplification boards available now as well.

[ftkalcevic]

The lvdt was actually invented where I work, way back before I was even born. They were originally hand made and hand wound, so yes you could build your own. There are some off the shelf amplification boards available now as well.

I saw some at Digikey. $110 for a coil (.6mm movement), $140 for the board. They only have one coil in stock. Longer travel coils are available at $500 each in lots of 5. Ouch.

I assume these are precision and very linear. For Z Height touch off, it only needs to be electrically stable so the results are repeatable.

[underthetire]

That's pretty cheap actually. Think the last test indicator type with amp we were quoted 6500. Accuracy is down to .00001 or better though. A little overkill for DIY stuff. We find they are very repeatable, especially the air plunger type.

[ftkalcevic]

$250 is still too much for a hobbiest like me.

There's a good DIY example...

LVDTs

I'm sure I can cobble something together for under $20. Maybe $30, with some LED bling.