[jbvries]

Hi All,

Let's say I have a log in its natural form and I want to mill an 1/8" groove in a straight line along one axis.

Is there a sensor device or encoder technology that can tell my CNC what the current and maybe forward or surrounding Z height is so it can essentially interpret the contour of the log and keep the groove approximately 1/8" deep along the straight cut?

I would imagine it's possible with a series of laser encoders that send distance information about the toolpath, but it may be one of those, "yeah, this 2 Million dollar CNC does that" haha.

Thank you in advance for any information or knowledge you can share on this subject.

Cheers!
JB

[gfacer]

A UHMW ring or sled and a spindle mount with a spring would be all you need....depends on how sudden the contour could change I suppose.

Sent from my ASUS_Z00TD using Tapatalk

[jbvries]

Would you please elaborate? I'm not sure how you are implying to implement.

[awerby]

He's talking about using a spring-loaded tool, with something attached to it (the sled) that would ride along the surface of your log, raising the tool when it encountered a rise, or lowering it when it hit a valley. Here's a spring-loaded tool holder: Spring Loaded Engraving Tool Engraves Flat, Curved, and Uneven Surfaces with a CNC Machine

[UA_Iron]

What kind of accuracy would you need? An ultrasonic sensor would work well because it can see anything - it does not care about surface characteristics. It would have an analogue output in Voltage or Current to which you would have to interpret into distances. Ultrasonic sensors have deadzones typically - so that first 30-50mm of sense range is not usable. If you use multiple ultrasonic sensors you have to tune them so they don't interfere with each other.

The tough part is probably taking that data and converting it into steps then into your CNC program.

[Old_Phart]

Equally tough would be focusing the ultrasonic to a narrow beam. Especially for the surface of a log. Thats a job for a vision system.
How bout a simple roller attached to the gantry? Neutralize the weight of the spindle and gantry with a gas spring and then just let it roll over the surface?

[Jon.N.CNC]

I've always wondered if simply using a proximity sensor where open a control board sends z down and sensor closed circuit brings z up so it will cycle on/off along a flat surface and deviation would be just a step or two.

Have no idea if it would work tho.

[gfacer]

What I meant was basically a pressure foot type setup (I'm sure a YouTube search will show something on that, axyz makes them for there commercial routers too), but the whole axis is ether sprung or held down with air pressure. Rollers would work too, but maybe less good at edges too? As far as measuring goes, probably much easier to do a measuring pass, convert to a points cloud that a second pass uses to guide a tool.

Sent from my ASUS_Z00TD using Tapatalk

[UA_Iron]

I get emails from Vendors all the time about their technologies. Keyence sent me this just today and made me think of this thread:
https://www.keyence.com/mykeyence/?ptn=001

Or in the attached PDF.

Knowing Keyence, this solution is probably $10k/piece and they won't let you even borrow an evaluation unit to test and see if it will work for the application. And they won't offer you any other value added services like terminating/labeling cables or even minor component assemblies... I've dealt with them before


RE: Old_Phart
Yes, you are right that the ultrasonic is probably not the best solution for this considering the beam dispersion. They make super precise pinpoint type (SICK), but they tend be very expensive and have extremely limited range. I like your idea about the vision system though.

RE: jbvries
You should look up how people use the XBOX kinect system to 3D image items and reverse engineer them. There's a hugely popular thread on here where a guy hacked the xbox kinect, used it to image an 18" tall Iron Man figurine, scaled the model to life-sized and then milled it out on his mill out of foam. He then painted it and put it in a display case. The thing was awesome. I imagine the kinect would be accurate enough for what you intend to do.

[elfrench]

I think everyone is overthinking this

Make an FSR probe and do a search on this board for PCB mapping techniques.

[jbvries]

Circling back around to this to thank everyone for their input. Still on the path to trying this. Cheers!

[jbvries]

I get emails from Vendors all the time about their technologies. Keyence sent me this just today and made me think of this thread:
https://www.keyence.com/mykeyence/?ptn=001

Or in the attached PDF.

Knowing Keyence, this solution is probably $10k/piece and they won't let you even borrow an evaluation unit to test and see if it will work for the application. And they won't offer you any other value added services like terminating/labeling cables or even minor component assemblies... I've dealt with them before


RE: Old_Phart
Yes, you are right that the ultrasonic is probably not the best solution for this considering the beam dispersion. They make super precise pinpoint type (SICK), but they tend be very expensive and have extremely limited range. I like your idea about the vision system though.

RE: jbvries
You should look up how people use the XBOX kinect system to 3D image items and reverse engineer them. There's a hugely popular thread on here where a guy hacked the xbox kinect, used it to image an 18" tall Iron Man figurine, scaled the model to life-sized and then milled it out on his mill out of foam. He then painted it and put it in a display case. The thing was awesome. I imagine the kinect would be accurate enough for what you intend to do.

@UA_iron or anyone with input...

It looks like I can actually get the Keyence sensor for a few hundred bucks on eBay. Do you have any idea on what that would look like interface wise if I wanted to use the output voltage to Mach3 to maintain continuous Z height as X and Y move?

Cheers!

[bobw63]

@UA_iron or anyone with input...

It looks like I can actually get the Keyence sensor for a few hundred bucks on eBay. Do you have any idea on what that would look like interface wise if I wanted to use the output voltage to Mach3 to maintain continuous Z height as X and Y move?

Cheers!

Mach3 does not need to control the z axis in this application as long as the log surface is the only thing driving the depth of the 1/8" cut.
I would set up the program to cut depth to desired level.
Switch the input of the z axis from Mach3 to a linear encoder with a roller riding on the log near the tool.
You may need to change the encoder output from quadrature to step and direction.
As the encoder responds to surface changes in the log, the encoder output would cause step and direction signals to move the z axis in that direction.
Stepper setups will not like rapid surface variations and extreme accelerations.
If I were building a custom z axis head, I would select a servo unit with a step and direction input and then open up following error to prevent faulting out the drive.

[Old_Phart]

Just an idea... How about making a pass over the surface with an indexer and some sort of roller mechanism that merely creates g-code for depth @ degree of rotation. Then a second pass using that code to cut. It would certainly take the response issue out of the equation.
I don't know how practical that would be but it seems viable.

[jbvries]

Just an idea... How about making a pass over the surface with an indexer and some sort of roller mechanism that merely creates g-code for depth @ degree of rotation. Then a second pass using that code to cut. It would certainly take the response issue out of the equation.
I don't know how practical that would be but it seems viable.

Thank you Bob and Old_phart for the input, The sensor will also be used to keep a laser at the ideal distance above the workpiece as to not go out of focus. The sensors available are able to handle the rapid sampling (over 1000 times per second) of a variable surface depth work-piece. My task now is figuring out how to translate it into an actively tracked parameter for Mach 3 (which I have found a lead to here.... https://www.machsupport.com/forum/in...?topic=26673.0 ) .