Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: DonMarkey on March 30, 2019, 08:55:18 AM
-
I have never done any probing. How user friendly is it? I need to probe and cut a mating part. Basically the matching convex to a tapered radius. Just for multiple one offs of different parts, I doubt I will ever reproduce the same part. I am trying to decide which way to go. Probing or build a tracer attachment.
-
Hi,
I assume from your description that you are using a lathe?
You need to turn a part to mate with a convex mirror?
Probing you may be surprised is remarkably easy and you can make an adequate probe yourself without spending
anything. It occurs to me you could try it without a great deal of expense or time.
Can you explain a bit more about what you are doing?
Can the convex surface be mounted in your machine to probe it?
Craig
-
Sorry a little more info. This is a hurco knee mill that I retrofitted, basically an oversized Bridgeport. This is for making rifle quarter ribs. I made a tracer for my old manual mill but I sold it with the mill. With that I made a casting out of epoxy by pressing the barrel into it, then copied the contour. I was thinking with probing I could set the barrel up and probe the area I want to fit the rib to and then mirror it somehow. So if you aren’t familiar with quarter ribs, it is basically a piece or steel mated to the top of the round barrel and soldered on. This is where the sight is then dovetailed into.
Thanks
Don
-
Hi Don,
OK that makes sense.
I see a couple of challenges that might occur if you probe the barrel.
First I imagine for a solder joint to be effective accuracy is important, say less than 0.001".
If you were to probe in the Z direction accuracy of the probe data would be good at and around the centerline
of the surface but accuracy would degrade as you approached the flanks of the surface. If that loss of
accuracy is a problem then the probe axis would have to be rotated such that it was approximately at
right angles to the surface.
That would not be insurmountable but adds considerable complexity.
How 'far around' the barrel does the quarter rib have to extend?
Craig
-
I could set it up at 90 to use x y ? Here is an example
(http://i27.photobucket.com/albums/c192/donmarkey/1917/BE1B3256-A583-44D3-9319-A4957CA15BF0_zpsytheujrd.jpeg) (http://s27.photobucket.com/user/donmarkey/media/1917/BE1B3256-A583-44D3-9319-A4957CA15BF0_zpsytheujrd.jpeg.html)
-
Hi,
that is quite a complex shape and so I would guess that simply probing in the Z axis direction would
not be good enough.
Yes if you rotate the barrel OR alternately the probe then you would get better results. The data collected would
form a 3D point cloud. That gets away from the rather simple G31 probe command.
I'm not saying it cant be done but that extra complexity will give you cause for thought.
My strength is Mach4 and with Mach4 you can have up to four probes. What I would do is have one probe
operate normally, ie vertically and two others operating at 900 to the first ie horizontally,
one in the positive Y direction and one in the negative Y direction.
To my knowledge all three probe point data would still end up in the one probe file and would constitute the
raw data for a 3D point cloud.
How you would go about that in Mach3 is a little more problematic. I suspect it could be done but I would
defer to Mach3 gurus.
Craig
-
I wouldnt be opposed to Mach 4 if that is something that could be done in the hobby version. I would just need to work a couple things out for an upgrade. Good food for thought.
Thanks
Don
-
Hi,
yes, up to four probes is common to both Mach4Industrial and Mach4Hobby.
I might point out that I feel sure something could be cobbled together in Mach3 just as well, its just I'm not sure
how. Changing to Mach4 on this basis alone might be dubious.
Craig
-
I Googled "quarter rib dimensions" and see the side touching the barrel appears to be defined by 2 dimensions.
http://www.scoutrifle.org/index.php?topic=2580.0
Dimensions are as follows: A = barrel diameter under the center of the front pair of screws
B = barrel diameter at the chamber
It should be a relatively easy job to make a CAD file for this.
Are the dimensions of the top details specific for different brand scopes? Are these published?
-
Problem is unless you contour the barrel yourself which is rare there is no way of knowing the barrel dimensions short of measuring it yourself. And many are hand polished so that changes everything. In the past you would make a dam to hold epoxy and make a reverse cast of the barrel then just copy with a tracer mill or tracer attachment. This is one of the instances where the dimensions are irrelevant as the part is custom fit and the next one will be different.
Thanks
Don
-
That should be a simple 4th axis probing function. Mount teh barrel in teh 4th center teh probe in Y axis. Then probe teh X axis for the length of the rib. Then rotate 4th say .25deg then probe back along teh X. Repeat until you have probed the area of the rib. That will create a point cloud for you to process for cutting.
Just a thought, (;-) TP
-
Hi Don,
Terry's (BR549, good to see Terry) is the classic method, namely rotate the object you are trying to probe
or measure.
That may represent a challenge, rotating the rifle accurately (between centers?) given that they are so long.
The same result could be achieved if the probe were fitted in a rotating head. It would require making
a probe head assembly that fitted in your spindle and yet could rotate in the Y plane.
Depending on how often you have to do this work it may prove simpler/cheaper in the long run to go back to
your epoxy mold.
Craig
-
Hi Don,
I have a variation of the idea of either rotating the rifle under the probe or rotating the probe head.
My proposal is to have three probes (plunger micro switches, you know the cheap and cheerful ones!) mounted
on a plate which in turn mounts in the spindle.
To probe at or near the center line of the rifle you would monitor the output of the center (vertical probe)
and move it with Gcodes lke:
G31 Z-1 F4
If you are not familiar with probing G31 is the probe code. The move (in absence of X and Y coordinate data) is
downwards to Z=-1 inch. When a probe event is detected the movement stops and the motion controller reports
the XYZ position of the event.
This is very much the common use of the G31 probe command. It is not however the only way to use it.
for instance if you coded:
G31 Y-1 Z-1 F4
Then the motion would be in the YZ plane at 450 to vertical. If you monitored the probe parallel to
the direction of motion it would record a probe event as normal.
You can see where this is going. What I'm trying to do is arrive at a compromise that does not require either
rotating the rifle OR rotating the Probe head, because such rotation device would require extensive 'making'. This
alternative is an approximation to the rotating Probe head idea.
I will admit also that because Mach4 has the possibility of four separate probes the idea occurred to me to utilize that
ability. This could still be done in Mach3 but would require all three probe heads to be wired in parallel so that if any
one of them triggered a probe event and its associated data point would be recorded. That would work fine provided
that if for instance you are probing with a vertical movement (-Z direction) and either the right or left hand probe were
to strike some part of the rifle that would cause a trigger event erroneously.
Does the idea appeal?
Craig
-
Hi Don,
I had these switches in mind:
https://nz.element14.com/honeywell/bzc-2rq1-a2/switch-basic-top-pin-plunger/dp/1525197 (https://nz.element14.com/honeywell/bzc-2rq1-a2/switch-basic-top-pin-plunger/dp/1525197)
Note this is the New Zealand site of Element14 so the price is in NZD. Works out to about $13 USD. Element14
I think might be called Newark in the US.
I thought that if you could drill a small hole in the plunger and glue in a 1/8 bit of wire as a probe extension.
All three switches could be attached to a flat plate of some description which would have a shank that fitted
in your spindle.
These switches are pretty good quality and 'guess' you could get 0.001 repeatability with them. One of the features of
these switches is about 5mm of overrun. That is the travel that the plunger can make AFTER the switch event without
bottoming out and damaging itself.
As you can see I'm thinking that you could make a thee axis probe for under $50. I imagine that there would be some
coding required, but that is time and brain power intensive, no dollars involved!
Craig
-
Simply mount the barrel in the 4th and use a steady rest to support the rest of the barrel. :) Yo can then probe in X/A in either of 2 modes long runs up and down X indexing A or A with short moves of X.
That would seem to be the easy way using just 1 probe.
(;-) TP
-
If the barrels are round, not oval shaped, I think there might be an easier method which would use only one probe.
The centerline of the barrel would be trammed in to the X axis of the mill before measuring and used as the zero reference.
A single set of measurements would be taken down the centerline. They would be mirrored and adjusted for the tool diameter to create a tool path. That path would then be mathematically rotated a few degrees at a time for successive passes.
What I am suggesting is in effect measuring the radius at many points and converting it to a tool path. Rather than writing a custom program to do the conversion the values could be transferred to a CAD program and it would do the heavy lifting.
If you already have a method of converting probing data to G code the set of readings could be mirrored and the rotated passes added before entering into that program.
-
Assuming you'd machine the final part with a ball endmill, make a stylus of solid steel the same dims as the ball endmill. I made one with a .375" bearing ball silver soldered to the end of a .250" drill blank mounted in a piece of delrin. Simply isolate it electrically and do a "contact" probe grid in the size necessary, find and replace all Z-'s with Z+ to mirror. Just like a duplicator, it should come out "dead nuts". Wouldn't it ?
Lemme know if I'm way off, just thinking.
Russ
EDIT: might replace the + and - the other way round.
-
Bluing on the barrel may interfere with electrical contact.
If the part is not symmetrical in the X axis inverting both Z and Y would rotate the part 180 degrees.
If you use a spherical tip you're really measuring points that are the radius of the sphere away from the part. The surface being measured is not flat so you can't just subtract the probe radius, especially at the edges. The probe radius would need to be subtracted from the measured position of the center of the probe at the angle from the measured position to the center of the barrel. This means both Y and Z would be adjusted. Then the tool radius would be subtracted in the same manner to calculate the center of the tool path. If the probe and the tool have the same diameter, that diameter could be subtracted instead of doing the radius twice.
Some additional adjustment might be required if the angle of the tapered section is too steep. A quick calculation says that error from a small taper angle might remove an extra thousandth or two of an inch.
-
Good point. Yes, I see that now, thanks.
Hmmmmmm .... plan b. ::)
-
The required compensation problem is one of the reasons why I suggested the approach in the earlier post. The measuring could be done in one pass and then expanded by computation to define the rest of the part. A probe with a bar shaped tip with a V edge aligned with the Y axis should work well.
There are other potential problems with the 3 probe method too.
In the measuring phase how do you measure the offset between the 3 probe tips in all 3 dimensions and then adjust and combine the readings into one file?
The plunger of the microswitch doesn't slide in a precision bearing. Will it be pushed to the side when measuring against an angled surface creating an error?
-
Hi MN300,
There are other potential problems with the 3 probe method too
Yes you are correct. I have some ideas about how the 'translation' would occur. I think in terms of Lua and have
some success in manipulating probe data files to achieve a specific goal. It is none-the-less a programming
problem, ie it costs nothing but time and effort.
I'm just wondering how it is you propose to mount a rifle (4 or 5 or 6 feet long?) and rotate it? Its not
a regular shape and would also require a big machine, may be bigger, even much bigger than Don has got.
Craig
-
I am not proposing to physically rotate the rifle. Once you have the measurements down the center line the rest of the measurements would be calculated for different angles. It's just simple trigonometry. It would take a fair amount of custom programming to make G code or a file that could be used to make G code.
A person experienced with CAD could use the initial set of measurements to create a cross section and rotate it into the full profile. The CAD program would do the rest of the work and generate the G code.
-
Hi,
the problem occurs at the flanks of the barrel where the surface is vertical, you lose precision, badly if
probing vertically. If you had an omnidirectional probe yes that would work, but as you know they cost a bomb.
Craig
-
Hi,
this is an example of a cheap probe:
https://www.ebay.com/itm/Econo-Probe-3D-Digitizing-Probe-Touch-Probe-Mach3-CNC-Router-MIll-Tormach/254148224305?hash=item3b2c6a2531:g:GlEAAOSw~8JcQ0h2 (https://www.ebay.com/itm/Econo-Probe-3D-Digitizing-Probe-Touch-Probe-Mach3-CNC-Router-MIll-Tormach/254148224305?hash=item3b2c6a2531:g:GlEAAOSw~8JcQ0h2)
A second hand (looks to be in great order) Reinshaw:
https://www.ebay.com/itm/RENISHAW-MIP-MANUAL-INDEXING-PROBE-HEAD-STYLI-ACCESSORIES/302900075114?hash=item4686405a6a:g:19EAAOSwD-dbrHpH:sc:USPSPriorityFlatRateBox!97203!US!-1 (https://www.ebay.com/itm/RENISHAW-MIP-MANUAL-INDEXING-PROBE-HEAD-STYLI-ACCESSORIES/302900075114?hash=item4686405a6a:g:19EAAOSwD-dbrHpH:sc:USPSPriorityFlatRateBox!97203!US!-1)
Craig
-
Hi Craig,
I think you may be misunderstanding what I suggested. One line of measurements taken along the X axis would give you the radius of the barrel at many points. To do this you would setup the center line of the rifle to be in line with the X axis.
Regardless of how you get the measurements you would eventually need to know where the center line is, either by measurement or calculation, to make the G code. This is because you need to consider the radius of the probe and the tool and the angle to the center line. If you measure with a thin probe in line with the center line of the barrel the probe radius could be considered to be zero. This allows the use of a simple probe like you have described earlier.
John
-
Hi,
but Don has told us that the barrel is not round, or at least perfectly so.
https://www.machsupport.com/forum/index.php?topic=39769.msg266789#msg266789 (https://www.machsupport.com/forum/index.php?topic=39769.msg266789#msg266789)
Craig
-
I interpreted Don's comment to mean the diameter is not consistent. If the barrel is not round then my suggestion doesn't apply. The geometry problems involved in converting the measurements still apply.
I'm curious to know how you would generate G code.
-
Hi,
The geometry problems involved in converting the measurements still apply.
I'm curious to know how you would generate G code.
All good questions......we'll have to wait until Don replies and can give us a bit more to go on.
Craig
-
Hi,
but Don has told us that the barrel is not round, or at least perfectly so.
https://www.machsupport.com/forum/index.php?topic=39769.msg266789#msg266789 (https://www.machsupport.com/forum/index.php?topic=39769.msg266789#msg266789)
Craig
I think what Don meant is that the "contour" can end up anywhere.
The way the barrel blank is turned, lathe contoured and then spun sanded an polished all results in any cross section of the barrel perpendicular to the bore is a perfect circle. (near enough for practical purposes, anyway) I've done a few barrels and this was the case in my experience. I would think that all you would need to do is probe in the X down each side at centerline Z height in the area needed to determine the dia. at whatever interval you desire. The concentric circles could be easily drawn and connected in 3d, then sliced to leave the desired segment of the contoured cylinder. Cam the rest.
If it is truly round, would you need anything more than the dia. at intervals ?
-
Probing function is easy and accurate. The point accuracy will only be as accurate as your set-up
and machines capability. Only you can define just how acurate the probing needs to be and how
automated you want the actual probe process to be.
I took a practical approach to the matter in that i wanted to be able to convinently and quickly
probe a piece to duplicate it or find dimensional information. If it is not easy and convient to do
you will not use it. Complex becomes a PITA also, so KISS applies, be pracatical for your intended
use.
I created a probing page / wizard for both the lathe and mill. A detachable index ring on the lathe
chucks allow for accurate indexing. As Russ replied, the barrel, for practical purposes, is symetrical
about the center line. If mounted in a lathe,you could probably take 30 probes on the top at 0 degrees,
15 probes at 90, 180, and 270.
The probed point data is is used by CAD to actualy draw the probe profile or just just show points.
In cad you could easily compare the probing at 90, 180, and 270 for any shape variation.
I must say thanks to Terry for the point conversion into CAD. In the end I still had to read a CAD manual
to get it to work in two different programs.
In summary here is what I do and works for me:
1. Set up work on lathe, mill, what ever.
2. Acquire point data
3. Have CAD draw the point data
4. Save as a dxf file
5. Use file in CAM or whatever program to get the GCODE.
No fancy probes and only one which allows side and point contact required ( make you owne for $1).
Make a visual one for realy fast and easy point collection, look into camera / optical techiniques.
A lot of options for doing things and YOU DON'T HAVE TO SPEND MUCH MONEY!
Note that electronic probing doesn't work on non- metallic material so you need options.
FWIW,
RICH
-
Giving this a bit more thought here is how I would do this project.
Mount the barrel in teh 4th and center the probe over teh barrel in Y axis.
Move teh X point to teh start point of teh probing routine. Probe tehesurface in Z and set teh Z to the radius of the barrel at that point.
Next probe in X for the distance of the rib to be duplicated. This gives you a single line contour of teh barrel.
Save out teh points and import them into CAD.
From CAD create a angular array from teh end view. The total angle would be teh wide of teh rib section.
Rotate teh points 180 degrees to create a negative of teh barrel form.
That would give you an easy set of points to create a 3d surface from. ( Point Cloud)
Export teh points out or convert the points into a surface file.
Then from a 3d CAM create a 3d toolpath to recreate the barrel contour.
Just a thought , (;-) TP
-
Wow. Lots of food for thought. I am thinking a casting and then straight one to one duplication is going to be my easiest approach. Then the probing with a stylus matched to ballmill diameter would work?
Don
-
Hi,
yes that would work but you can omit the casting step.
If you probe the barrel with a matching probe, save the data, then the ball mill can follow the path from the previously
stored data.
Craig
-
So I am overthinking this? If I probe the centerline, then probe again .020” off center and so on I can use that as my step over and convert the z+ z-? What am I missing?
-
Hi Don,
If I probe the centerline, then probe again .020” off center and so on I can use that as my step over and convert the z+ z-? What am I missing?
That is the essential idea....you are not missing anything.
Most CAD/CAM programs have a function that converts a set of points, or point cloud, to a Gcode toolpath. It is still
the same idea but is automated to produce the Gcode. You could for instance probe at 0.02 stepovers but to get the
accuracy you might decide to mill at 0.005 stepovers, the CAD/CAM program allows that.
Carig
-
Now you are assuming I have or even know how to use cad/cam. Haven’t used that since the late 80s. I use conversational cnc on one mill or write g code manually for the mill I converted to Mach3 right now. But I’m learning.
Thanks
Don
-
Hi,
CAD/CAM is not absolutely essential to making a probe point cloud work it certainly helps. It may well be that the effort
required to master one of the programs out there is a greater effort than casting the part as you have done before.
Fusion360 is a very capable CAD/CAM program available free from Autodesk. If you have a 64 bit PC and OS download it
and try it out. I suspect it will do as you require. The only question is whether you are prepared to stump up the time
to learn it?
Craig
-
Hi Don,
thinking a little more about your situation. I suspect that learning to use a CAD/CAM may be a hurdle.
Evidently you are quite capable of conversational programming and manual programming.
I myself use MillWizard from NFS ($75), it adds what amounts to conversational programming. There is one feature of
that program that may appeal to you. Its the 'Polyline' feature. You program a series of points with either lines or
arcs between each point and MillWizard will generate the Gcode.
I have attached a couple of pics that might explain it a little better. The first pic is where I have programmed a number (5)
points (X, Y coordinates) with a straight line in between each. The second pic is the toolpath that MillWizard generates.
There are other options as well, tool offsets, direction of cut etc. While it is tedious to program in the points its perfectly
doable and in fact very simple.
Such a method would not be practical if you had a point cloud of a thousand points but probably useful up to 100 points.
Download a copy and try it. As a demo you can't save the Gcode but is in all other ways fully functional. The advantage of
MillWizard is that you can be producing Mach compliant code within 1/2 an hour.....its so straight forward.
Craig
-
Thanks I’ll look into it.
Don