CNC - calibration issues for inlay work

[Keith Outten]

According to the American Society of Non-Destructive Testing the unaided human eye cannot discern any measurement less then 1/64". That is the national standard for inspection using a scale. Any measurement below a 1/64" graduation has to be performed using a precision instrument.

I use a quality 48" scale as my shop standard. Every non-precision measuring device is compared to the 48" standard and as long as all of my scales visually agree with my standard that is sufficient for my shop. The 48" standard is never used for production work.

CNC work is unique to itself and I can't remember a time when non-precision work was co-mingled with precision measurements. In other words I never check my CNC work with a scale which I think is what everyone here is saying. Tooling is inspected to determine the actual size and when it is not accurate I make up the difference in my software as many here have suggested. Consideration has to be made to allow for tooling that sees a temperature increase when it is used but that is only in very rare situations, I try my best not to even consider working to machine shop tolerances.

I am well aware of the temperature variation in my shop and its effect on jobs such as inlays particularly when an acrylic male is an inlay into a wooden base or the reverse situation. An acceptable very close tolerance fit in the evening when my shop has warmed up for many hours normally will not fit the next morning when the temperature decrease is just a few degrees. When material types are identical this is less of a problem but its still a concern.

[Jim Becker]

Btw, I have a 24" Starrett rule but I may look into getting one of these 48" models.

I actually ordered the 48" rule I linked to and should have it tomorrow I believe. I'll try to remember to let you know how I feel about it. I've been REALLY happy with the steel rules I have now and they are all I use for measuring when I have the choice. My 8m tape and 25' tape are reasonably accurate as far as I can tell, but even an accurate tape can have some little bit of variability due to the hook end maybe being sticky or over time, getting worn. For CNC work...the more accurate you can measure something, the better you can zero in on what you need for adjusting your toolpaths for critical work. This is no place for a "cheap" caliper, either...DAMHIKT! And a caliper is my preferred measuring tool around the CNC when it can fit on or into what needs measured.

[David Falkner]

Good, I'm eager to hear how you like the 48". Btw, I rarely use the hook if I need an 'accurate' tape measurement. And I realize using 'accurate' and 'tape' in the same sentence shouldn't happen unless 'not' is also in the mix but you know what I mean. Like on my 48" measurement on the CNC, I started it at 2" so the hook wasn't involved.

My dial caliper is a Mitutoyo 6" I bought in the mid 70's and it seems like a pretty good one. There's probably not a day that goes by when I don't use it at least one time.

David

[Jim Becker]

I try really hard to avoid the "start on the 1 or 2 inch mark" thing simply because the old brain subscribes heavily to Mr Murphy and his antics. Heck, even starting at the end I sometimes, um....well...you know...have "D'oh!" moments. I'll let you know how the long rule turns out. The nice folks from Amazon say it's arriving sometime tomorrow.

[Mark Bolton]

I try really hard to avoid the "start on the 1 or 2 inch mark" thing simply because the old brain subscribes heavily to Mr Murphy and his antics. Heck, even starting at the end I sometimes, um....well...you know...have "D'oh!" moments. I'll let you know how the long rule turns out. The nice folks from Amazon say it's arriving sometime tomorrow.

Everyone flubs "buring the inch" on occasion. A good way to avoid that mistake (if your scale/rule is long enough) is to burn 10". Start on the 10, and run out from there. It seems to trigger your brain better than having to subtract 1.


Burning the inch is way more accurate even with a scale. You often see machinists choking up on their scale and measuring from the 1 as well.

[David Falkner]

Interesting how the mind works; I like starting on 2", 3", 7" etc. because the number at the end makes me double check everything instead of taking an ending number for granted. Besides, I love math and do most of it in my head.

David

[Ted Reischl]

I am curious about needing to be dead on accurate at 48"???

A machine shop achieves close tolerance by using equipment that was manufactured to close tolerances or the use of glass scales, DRO's etc, not with a scale. When I talk tolerances at that distance think less than .005. A 48 inch scale will grow and shrink that much depending on temperature.

So, where am I going with this? IF you know the accuracy of your motion method, for instance a precision ball screw or a well made rack and pinion then what we are doing when we check the travel with a dial indicator is ensuring that we have the steps set correctly. The precision is in the screw or rack. If you use a precision dial indicator, one that reads tenths then moving an inch will tell you a lot about the accuracy of your machine. Let's say it is two tenths off in one inch. Over 48 inches that would be about .010 total. So adjust the steps a tad.

This is sort of like old vertical milling machines with worn out lead screws. Pretty tough to work with using the dials on the cranks, but and a DRO? Precision has returned.

Here is an idea for measuring how accurate your machine is: Buy or borrow a DRO and attach it to your machine, it will tell you far it has actually traveled. Good ones are quite accurate, cheap ones are .002 per 6 inches of travel. Here is a cheap one that would help you out:

https://www.amazon.com/Digital-Large...8T6M3BV644060R

After using it on your cnc you could hook it up to some other machine in your shop, I use one on the RAS.

[David Falkner]

I assume by 'tenths' you're referring to ten thousandths, Ted - correct? That's the way the math works out, anyway.

It's been a long time since I've been around tool & die folks who talked in 'tenths' and meant ten thousandths, but to most woodworkers I know a tenth is just something close to an eighth and both are too small to be of any use.

I like that DRO and the price isn't bad at all. That may have to be my first tool purchase for 2020!

David

[David Buchhauser]

I actually ordered the 48" rule I linked to and should have it tomorrow I believe. I'll try to remember to let you know how I feel about it. I've been REALLY happy with the steel rules I have now and they are all I use for measuring when I have the choice. My 8m tape and 25' tape are reasonably accurate as far as I can tell, but even an accurate tape can have some little bit of variability due to the hook end maybe being sticky or over time, getting worn. For CNC work...the more accurate you can measure something, the better you can zero in on what you need for adjusting your toolpaths for critical work. This is no place for a "cheap" caliper, either...DAMHIKT! And a caliper is my preferred measuring tool around the CNC when it can fit on or into what needs measured.

Hi Jim,
I am curious as to your evaluation criteria for the accuracy of your new rule (or any or your other rules). I own an assortment of US made steel rules acquired over the years. Some 36", 48", and 96" lengths. The better quality ones, and those which seem to agree with each other the closest, have etched graduation marks as opposed to some of the others with painted or silk screened markings. These are all what would be considered as "inexpensive" general purpose rules. I consider any of these sufficient for general carpentry, metal fabrication, and so on - but certainly not sufficient for accurate calibration of a cnc machine or any other critical measurements.

Keith has mentioned: "I use a quality 48" scale as my shop standard. Every non-precision measuring device is compared to the 48" standard and as long as all of my scales visually agree with my standard that is sufficient for my shop. The 48" standard is never used for production work."

I am interested to know the particulars of Keith's shop standard 48" scale - manufacturer, model number, and exactly how accurate is it?

Since I do not own a high quality 48" scale that I can use as a standard to check my existing shop rules, I really have no idea how accurate they are. I only know that they do not all agree with each other.

It would seem that what is required is a rule/scale with a know accuracy, preferably legitimately traceable to NIST with certification to verify this (see below).

David

https://www.mcmaster.com/2056a21

Certified High Accuracy Ruler.jpg

[David Buchhauser]

So, where am I going with this? IF you know the accuracy of your motion method, for instance a precision ball screw or a well made rack and pinion then what we are doing when we check the travel with a dial indicator is ensuring that we have the steps set correctly. The precision is in the screw or rack. If you use a precision dial indicator, one that reads tenths then moving an inch will tell you a lot about the accuracy of your machine. Let's say it is two tenths off in one inch. Over 48 inches that would be about .010 total. So adjust the steps a tad.

Hi Ted,

I used the procedure you describe to adjust the "turns ratio" for a new cnc I am setting up. This is an Avid Pro4824 with Centroid Acorn controller and 4HP atc spindle. Using a dial indicator with 0.0001" resolution and by adjusting the "turns ratio" in the software (this was a repetitive process) I was able to achieve 0.0001" per inch for the Z axis, 0.0003" per inch for the X axis, and 0.00015" per inch for the Y axis. This equates to 0.0008" over 8" for the Z travel, 0.015" over 48 1/2" of X travel, and 0.0035" over 24" of Y travel.

This particular machine is intended for aluminum machining, and most of my parts will most likely be smaller than 12" x 12" - so this will be more than sufficient for my needs. As mentioned in a previous post, small adjustments can be made in the cutter offset to achieve the desired dimensional results.

With the combination of precision rack and pinion/stepper motor drive/electronics this is effectively a "mechanical" DRO system. It had occurred to me that I could use this particular cnc machine to check (within reason) the accuracy of my long steel rules. Using a DRO with precision etched glass optical scales (as you have suggested) would be another option for checking the accuracy of my long steel rules. I had really never considered this before, since I do not use these for any sort of "precision" work. But I do have DROs on several of my vertical milling machines - so perhaps this is something I will do just for fun.

David


X axis calibration
20191123_234049_resized.jpg

Y axis calibration
20191123_235346_resized.jpg 20191123_235352_resized.jpg

Z axis calibration
20191124_000833_resized.jpg

Turns Ratios
20191124_001405_resized.jpg

[David Falkner]

Nice looking setup, David, very clean. Do you not use any backlash compensation in your software or is it just not on this screen? Just curious...

David

[Mark Bolton]

I am curious about needing to be dead on accurate at 48"???.

To me at least the issues isnt necessarily being dead on at 48"(though in the woodworking world thermal issues are not going to effect you at that scale) but using that measurement as a cross reference for accuracy after calibrating on a small scale clearly shows multiplied error. If you calibrate your machine to .001 in 4" that is going to equate to .012 in 48 or .025 in 100, etc. Even techs setting up 100K commercial wood cnc's do a longest dimension or diagonals check on a machine with a tape.

I can run a diagonal check on my machine (60" x 100") at 116.625 and see very accurately where Im at with a tape as well as checking the machine travel accurately with a dial indicator and a mag base.

[Jim Becker]

I am curious as to your evaluation criteria for the accuracy of your new rule (

Honestly, I'll compare it with my existing which all seem to match up. Consistency is more important to me here than knowing if the rules are "dead nuts" accurate, at least for woodworking. While this conversation is in the CNC area, I honestly don't use my rules for "critical" measurements for CNC purposes. That task is given to my digital caliper within its range and if I have to check something larger, I'm going to acknowledge that my eyes are going to come into play if I have to use a rule. I often make adjustments in CNC tool paths measured in hundredths and sometimes in thousandths and that's just not going to happen with a rule no matter how accurate it is.

[Keith Outten]

David,

I purchased a Woodpeckers 48.5" Paolini Bench Scale to use for my shop standard. I don't consider the shop standard a calibration method its just to check to verify a reasonable accuracy for measuring tapes and other shop scales.

I checked it for accuracy using a 24" veneer caliper using the bench scale sliding stop at 24" and then I made a second measurement from 24 to 48" verified using a 10X magnifier lupe.

It is as accurate as the eye can see given the way I verified the accuracy, but the worst case scenario is that it is probably within 1/64" over the length of the 48.5" scale. I never work to a 1/64" tolerance using shop scales, my eyesight is not what it used to be even when wearing glasses and temperature variance in my shop prohibits working to that level of tolerance. Precision instruments are calibrated at 68 degrees F and I know that my shop varies more then 5 degrees in any hour of the day based on whether the dust collector is running or not.

99% of the work I do is ADA signs, its pretty rare that I have the need to do precision work. The exception to this rule is that every one of my ADA signs have precision inlays for the tactile text but that work is done on my Trotec Laser Engraver. It engraves the plaque and cuts out the letters, numbers and icons and the width of the laser beam provides a perfect fit every time without the need to measure or verify any dimension other then the depth which I check with a digital dial caliper when I set the speed and power.

[Brad Shipton]

Good job sourcing your problem. I always sigh a great relief when I finally find a problem. So many times I forget to write down the changes made or the problem I was having...

Mark's squareness test is great. For me, when I have butt joints between panels, I hate to see gaps. I use Cabineo connectors frequently for boxes, and re-cutting parts annoys me.