Testing and tuning OF for accuracy

Hello everyone. Now that I have my machine put together and the waste board complete, I set out to tune the machine. I checked both X and Y distance traveled and ensured Z was perpendicular to the waste board (though I have not trammed the waste board or flattened it yet).

After much testing, I found X travel to be off by 1/32 over 30" and Y to be off by 1/64 over 30". So I adjusted the steps per rev a little and dialed in the machine. Now there is no perceptible difference between the distance traveled and the distance commanded to travel.

Happy to provide additional details on the process and the formula for correcting if anyone is interested.

Iā€™ve provided some pictures of my setup if anyone wants to check their machine.

X Setup

X @ zero

X before adjusting (no pictures of the dialed in final product)

Y Setup

Y Zero

Y After Adjusting

While checking the Z carriage, I found it to be tilted 0.0254mm forward over 40mm (0.0006125/mm) and tilted right 0.127mm over 40mm (0.003175/mm). I tried for about an hour to correct the left/right tilt but was not able to get rid of it. I seems when I tighten the screws, the z-carriage returned to itā€™s position each time, despite my best efforts. Though the error is perceptible, it is very minor and should not cause a problem. But, we will know for sure when I flatten the waste board and check for ridges.

During my testing, I noticed no perceptible backlash in any direction, which was a welcome surprise (though expected given the screw drive).

All in all, the machine was nearly perfect out of the box, and that is light years ahead of my X-Carve. A little tweaking and now it is dialed in. I wish I could get the left/right tilt out, but Iā€™m going to live with it for now.

I will do some cutting tests next to make sure Iā€™m getting finished results with the same level of accuracy.

Iā€™m curious if anyone else has tested their machine and are seeing similar results?

-Tom

22 Likes

Great breakdown Tom!
I think it would be great if you posted a step by step on how you adjusted it! I know it would be very helpful to many!

6 Likes

Sure thing - Iā€™ll work on some instructions later or tomorrow. I hope get a video of it done, but thatā€™s a much more involved process.

12 Likes

I too would appreciate that. Iā€™m sure text and a few pictures should be adequate to get the process across to us.
Tom

2 Likes

Have you also considered measuring repeatability? You mentioned no noticeable backlash, which is great, so I am curious how the machine performs on repeatability tests without load, and after running some of the cutting tests you plan to do. I did not see any documentation on the OF site (apologies if I missed it) on the expected accuracy and repeatability their CNCs.
Tom

1 Like

EXCELLENT post Tom. Thatā€™s one of the things I love about this OF community is the willingness to share and help each other with the machines to make them better. Looking forward to your write up.

3 Likes

Tom,

To get that left/right a little more accurate, what I did was loosen the screws on the router mount and inserted small wedges of paper at the opposing bottom edge where the router mounts as a shim, that tweaks the tilt just slightly. You can use different thickness of paper to adjust more/less. I was able to get my tilt within 5/1,000th of an inch, which gives me perfectly smooth carves.

Hope that helps,
-Alex

13 Likes

Great tip Alexā€¦ Did you use 1-2-3 blocks to set up your router?

1 Like

Grant,

No I bought a Tramming Gauge to make sure I am perfectly aligned on both front to back as well as left/right.

I discussed that here: Tramming the Router

-Alex

3 Likes

Thanks Alex, much appreciated

1 Like

Excellent suggestion Alex - never occurred to me to shim the router. I was focusing on the router plate.

@TMToronto - I will run more tests for tonight. When I mentioned backlash earlier, I should have noted I was getting highly repeatable results with the machine returning to the same position, but I did not measure it beyond eyeballing the position. And I learned that precision is for measuring systems - the appropriate term for motion systems is repeatability. TMYK :wink:

-Tom

5 Likes

I ran a quick pass in X and Y after getting set up, but to be honest, not to the extent of @cyberreefguru Though now I might. lol. I used a 22 degree engraving bit. Checking for tram on the router I used a small machinist square, and a block of granite. I shimmed the granite to level in X and Y. Inserted an endmill (old one) upside down, and used the square against the endmill and a flashlight. I can just barley see any light in the X, and the Y is as spot on as I can see.
So not the precision ( or probably the proper) way of doing things, but in a pinch. lol.

3 Likes

@TMToronto et al -

I did some more testing over the last 2 days, and unfortunately I donā€™t have anything exciting to post as a result. I ran a number of tests sending the machine various distances at various speeds and returning it to the same known location. I tested the return location using a dial gauge. Each time the machine returned to the same location within a fraction of 0.001 - the resolution of my gauge. Iā€™m not convinced the little ā€˜slipā€™ I saw wasnā€™t due to the gauge moving on the router a little. But was literally like 0.0001 each move, after 5 time back and forth, I could barely tell the dial was moved from zero (and it might of just been my angle, though I tried to view from the same angle each time). So thatā€™s all great news on repeatability as well as accuracy.

Starting Position

Ending Position after 3 Runs of 100mm Left and Right

Iā€™ve also posted a short slow motion video of the machine returning to zero position.

More to come as I get into the machine and build more experience.

-Tom

8 Likes

Thank you very much for doing these tests and posting the results. The fact that your findings show very good repeatability is great news.

4 Likes

I promised I would write the procedure I used to test and adjust the accuracy of the OF, so here it is. Decided to add it to this thread rather than create a new one.

Things you will need:

  1. OF Woodworker (same process for machinist, but adjust sizes)
  2. an accurate 36" ruler (a mm rule would be easier, but I didnā€™t have one)
  3. 1/4" shank, 1/4" 45 degree v-bit (larger bits will work, but makes it more difficult to see)
  4. Blue tape or masking tape
  5. Optional - fine grained 12" machinist ruler

Process:

  1. Put the v-bit into the router collet and tighten down
  2. Home the machine
  3. Jog the machine back (Y-) 100mm, and right 100mm (X+).
  4. Tape the 36" ruler to the waste board in the X direction.
    ** Ensure the ruler is parallel to the Y axis, or collinear with the X axis (your choice ;))
    ** Place the ruler near the front of the waste board, but not so far forward the bit wonā€™t reach the ruler
  5. Jog the machine so the v-bit is centered over the 1" mark on the ruler. This is your starting point.
  6. Zero the X and Y coordinates
  7. In the MDI section of the OF controller, type the following commands, pressing play after each one.
  • G91
  • G20
  • G1 X30 F100
    ** G91 sets the machine to relative positioning mode
    ** G20 sets the machine to imperial (inch) mode
    ** G1 X30 moves the machine right, 30", feed rate of 100 in/min
  1. Check if the v-bit is completely centered on the 30" mark of the ruler.
    ** If it is, congratulations, the machine is well calibrated in the X direction. Jump to step 18
    ** Assuming the machine needs some correction, continue on in the process
  2. Using the machinist ruler, measure the distance from the v-bit to the 30" mark. This is your error or offset. Letā€™s assume the error is +0.125".
  3. In the MDI section of the OF controller, type the following command and press play
  • G1 X-30
    ** this jogs the machine back to the zero position we set in step 6.
  1. Calculate the offset for the error discovered in Step 9.
  • Adjustment Factor = Fa = (Distance Traveled/Distance Commanded)
  • Example, Fa = 30.125/30 = 1.004166666666667 = 1.004167 (distance traveled = 30" + 0.125")
  1. In the OF controller SW, click the menu expansion icon in the top left and click on ā€˜Motor 0ā€™
  2. Scroll down to the option ā€œtravel-per-revā€ under the ā€œMotionā€ section. You need to adjust this value to compensate for the error.
  3. The default value for the travel-per-rev is 10mm. Multiply the value times the offset (Fa).
  • 10*1.004167 = 10.04167.
  1. Enter the new value (ex, 10.04167) into the text box and click Save in the top left.
  2. In the MDI section of the OF controller, type the following commands, pressing play after each one.
  • G1 X30
    ** G1 X30 moves the machine right, 30"
  1. Remeasure the offset, if any. If there is still an offset, restart the process at step 10.
  2. If no offset, then repeat the process for the Y direction with the following changes:
  • G1 Y30 instead of G1 X30 for step 7 & 16
  • G1 Y-30 instead of G1 X-30 for Step 10
  • Motor 1 instead of Motor 0 for Step 16

I hope this is helpful for folks and if anyone has questions, let me know.

-Tom

36 Likes

Thank you for this! I donā€™t have my machine yet (recently ordered it, so have a bit to waitā€¦), and this is my first CNC machine, but Iā€™ve stashed this away for when I can run through this. Appreciate the detail and clarity!

Mike.

Thank you for taking the time to post this. I will try it once I receive my OF.

Thanks for posting this, Tom. Very useful. Reading through the instructions, I think I spotted a small error. If you zero at the 1" mark and then jog over 30" the bit should be at 31". In steps 8. and 9. the measurement should be off the 31" mark.
-Chip

1 Like

Thanks Tom, my machine is scheduled to arrive this month. I will be following your procedure.

#2. An accurate ruler. Is there brands guaranteed to be accurate? I have 3 that were used in my old workā€™s print room, each from a different manufacturer. None of them line up to each other by the time you get to their ends. Some stainless, some aluminum. Iā€™ve never had much trust in rulers after seeing that.

Anyone have a good source for a metric version in the states?