When digitizing with a probe mounted on a vertical mill do most mills move through a grid and then come down at each point for a z height measurement or do they actually work around the object measuring the from the side of the object?

All that I have seen mesure in the z dirrection. Moving in a grid like pattern

Renishaw and Heidenhain and others make a X,Y,Z, probe's, usually on a CNC mill macros are written using the G31 function (skip input) and will capture whatever you want, typical macros are for center of hole/boss, outside of boss, edge of part, tool length etc.
3 axis probe's are not usually cheap,Up to $10k's but I picked up one on ebay for just over $100, I am just in the process of implementing it on a mill I have.
Al

Al...
Could you post a picture of that probe that cost only a $100?
Wow! what deal!

It's a Heidenhain and it apears to be new or never used.
Al

OOooo! nice find Al! I've been interested in digitizing but never came across something on E-bay like you have....

Maxnc makes a probe that work ok for most things. +-.002 or so, not a cmm but ok for most things.

Has anyone used the digitizing probe available for DeskCNC software?
I could not even find a picture of it, but I figured for $225.00 what the heck, so I ordered one, I have not recieved it yet, but will post some pics and my two cents worth when I get it.

:idea:

There is a picture with the product announcement.

http://www.cnczone.com/forums/showthread.php?s=&threadid=1511

Fred Smith - IMService
http://www.cadcamcadcam.com

I bought one last week for use with mach2. I set it up wrong and broke the stylus. Then the second time I used it the stylus came unscrewed. Gotta make sure its tight. I learned to do small things. A 0.03 stepover x and 0.06 y at 45ipm on a 18.25x18.25 square will take about 180 hours to digitize. So I just did a 1/4 of it with a .125" stepover and it only took 5 hours. Now i'm trying to figure out how to get the point cloud into a solid model. Wish I would have run it a 0.06" stepover. could have been done in 11 hours.

Donny

Whiteriver,
A couple of questions if you don't mind. You will have to excuse me, I ordered mine with DeskCNC, it was not too expensive, so I figured I would check it out. I understand that it is just a 5 way probe switch, but what are you doing to import your image, do you run the machine in a grid formation and the probe reads only on the z-down, or does it read on the x-y plane as well?
What type of file is generated by your point cloud after digitizing? Can this file be opend in Rhino3D and "tweeked" before re-machining?

:idea:

I guess if I would have read the whole thread I could have answered my own question.
So can you use this probe in the x-y planes for edge detection with DeskCNC?

:idea:

The DeskCNC probe with DeskCNC will autoscan a defined cubic volume. The setup process for 3D involves jogging across the corners of the extents, then setting scanning rates and stepover size for X and Y. This automatically calculates the number of points and estimates the resulting .STL file size. When the scan is completed the data is automatically back compensated for the radius of the probe.

It is not designed for edge finding, but if calibrated against a known standard, could probably be used for this purpose.

Fred Smith - IMService
http://www.cadcamcadcam.com

So as it is traveling horizontal it contacts an object it registers position, or it horizontals only at the uppermost z position programmed into the cubic volume you are scanning, and only measures in z down position at each vector point determined by stepover grid?

The probe should never move to the Max Z height except at start and stop. If it does, the probe may have been disconnected. During scanning, it appears that the probe tip is in constant contact with and rolls over the surface.

Just one of the many benefits of a serial port CNC controller. ;-)

Fred Smith - IMService
http://www.cadcamcadcam.com

I have one of these probes coming this week so have been reading this with interest, would defining a narrow max/min band work to use the probe as an edge finder for 2D?

I thank you very much for the info, I can't wait until mine arrives!

We wrote macros for Fanuc MACRO-B many years ago that would not only do a simple grid (touching in -Z), but we could also do:

1) Profiling, where the Z is held constant and the initial probing direction is either +X, -X, +Y or -Y. Let's say the probe starts touching in +X. As the probe works its way along the part, if the angle of the surface exceeds 45 degrees, it changes direction to touch in +Y. If the angle keeps changing past the next 45 degree angle, it switches to -X, etc. This lets you digitize all the way around a cam or any 2-D contour. If you also program a Z step increment, the probe moves up/down by that amount and it digitizes are around the part again. This routine works great for steep-walled pockets, cams, etc.

2) Multi-directional probing in X-Z, where you probe a row of points along the Y axis. The direction of the probe touching automatically changes from -Z to -X to -Z to +X to -Z as the surface gets steeper or shallower than 45 degrees from vertical. When a row of points is collected, the probe retracts and goes to the first point in the next row in Y. This cycle is great for digitizing along a horizontal surface, down into a cavity (or up a steep wall) then along a horizontal surface again.

2) Multi-directional probing in Y-Z, where you probe a row of points along the Xaxis. The direction of the probe touching automatically changes from -Z to -Y to -Z to +Y to -Z as the surface gets steeper or shallower than 45 degrees from vertical. When a row of points is collected, the probe retracts and goes to the first point in the next row in X.

We also used a sub-macro to actually touch the surface in each direction. That sub-macro could be set to take a single touch, or a double-touch for better accuracy.

We found that simple -Z grid digitizing almost never worked well in the real world. Steep walls and corners need to be digitized too, and these cycles were developed to accommodate all those surfaces.

I am allowed one dumb question per week.
Some years ago I used a Gorton Panto Mill for copy milling.
Would it be possible to set up a piece of material and an original on the table an perform the machining using direct instruction to the cnc controller from the probe input. in other words a cnc copy mill without the need to generate G-Code in the normal way.

wfiles

The more I learn The Less I seem To Know.

If you want to copy mill, I would recommend you use a copy mill. They are dedicated to this task and a CNC will not be as productive.

With DeskCNC, you scan directly to the CAD file, and make the g-code. It's not the "normal" way because it's all included in one package. You never need to even see the G-code, and can make as many exact/ modified/ or scaled copies as you want.

Fred Smith - IMService
http://www.cadcamcadcam.com/hobby

I am well aware of the attributes of DeskCNC. Maybe I posted in the wrong forum.
My question was not an attack on conventional CNC methods ,more to investigate the technical possibility of using current technolgy and components to simulate what was a conventional mill with dedicated and unique motion control.
I am still curious to know if the digitizing probe output can be used as direct input to control steppers at least on x,y axis.
Maybe my thoughts are too far outside the square?

Bfiles

Do you plan on using 2 seperate machines? One to cut while the other digitizes? The probes output is a 5V signal. The software knows where it's at when it gets that 5V signal, and would somehow need to send that info to the other machine to follow along?

One problem is that digitizing is a slow process, and it usually consists of vertical plunge moves. Not the best way to duplicate a surface. I'm guessing that running the g-code generated from the digitized data would be faster by a factor of at least 10 times perhaps even 100 times faster depending on material.

Speed is not the problem, nor is there necessarily a need for 2 machines. The problems are scaling and compensation. Cutter compensation can be ignored if the contact probe and the cutter are the same size, expecially in 2D. However, when you get to 3D, the probe stylus and the cutter have to be identical size and shape, or the resulting part will be distorted. Cutter compensation requires lookahead to be processed, and this is not possible when doing live scanning. This brings us back to preprocessing a scanned surface and then using Cad-CAM methods to calculate cutter compensation.

Any automated probing routine can easily do tracer milling. Just mount a spindle contact sensor on the same machine frame and keep the chip cleared as the scanning/cutting proceeds. If the probe stylus and the cutter have the same shape and size, the part will be identical to the original.


Granularity of the digitizing may also affect the resulting part contours.Since the machine is digital, you need to define a step over distance. If the step over distance is larger than the tolerance of the surface, there will be stepwise marks in the cutting. Again seperating the scanning and cutting processes can allow processing of the data to provide smooth surface cutting motion for both 2D and 3D applications.

A taper attachment on a manual lathe is one of the most common "copy" techniques. The lathe slide is forced to follow the angle set by the taper mechanism. Imagine that this taper attachment was replaced by a powerful mill cutter spindle on the tool post and a rigid follower mechanism with a pattern shaped like a wooden shoe or a model boat hull. Now use the lathe lead screw to force the contact stylus to follow the shape of the pattern(as it rotates). You have a copy mill. It is fairly accurate and very productive.

Fred Smith _ IMService
http://www.cadcamcadcam.com/hobby

I was thinking in 2D rather tha 3D . With the copy mill the cutter and the probe had to have the same dimension.
If the steppers had and encoder built in as they do in printers and scanners, would this make the concept more feasible.
I am assuning that some encoders indicate position rather than speed.

It really has nothing to do with an encoder. Any CNC machine that can follow a contour with a probe, 2D or 3D, can also simultaneously, drag a spindle along the same path. If it is done in real time, the probe stylus and the cutter have to be the same size to make a duplicate part. This assumes a single set of 2 axis slides for 2D, ie a single machine. A pantograph type of fixture could also be used for scaling, but NOT for cutter size compensation.

If the probing and machining are decoupled ( ie through a cad model and g-code) the duplicated part can be scaled, mirrored, modified, smoothed,etc.

There are usually limitations on the response time of a probe, that may make it possible to machine the part faster than it can be probed. Parallel port controllers have definite bandwidth issues with probing speed. That will probably be the limiting factor. For a serial port, dedicated controller like DeskCNC ( 125,000 sps), the physical switch opening and closing becomes the limiting variable. The DeskCNC probe can physically detect around 20,000 to 30,000 points per hour.

Fred Smith - IMService

Just one of the many benefits of a serial port CNC controller. ;-)[/url]
Can someone explain this statement?