Circle problems. Thoughts?

I have been struggling for months to determine what is happening. I understand that perfect circles are hard to obtain. However, I know I was able to do them better earlier on. I have sightly oval circles no matter what I try and I know I’m square and trammed.
I get sightly better results testing at extremely low speeds. 10ipm through plywood. The smaller they are the wise they look. .5 inch circle will be .5 on one direction and .48 on the other. Same for later ones but less pronounced.
I’ve had it since 2017 and I’m not sure how much time I’ve really used it. I decided to purchase the spare parts kit to replace the belts and the pulleys. Do you think that could fix my issue?
I created dog holes in my spoil board and every one wobbles when dogs are inserted. So I purchased a 20mm boring but from Amana to have it work. However, I watched several videos of people getting great results and I’m at my wits end on this. I bought mach 4 this morning as well as the spare parts. $400 just to see if I can get circles back.

Is it only in one direction? Which direction?

Seemingly in the x axis direction. However, the measurement that equals what I asked to create is 45 degrees looking down the y. Measuring 45 from the left to right like this /
The smallest is always left to right and largest is front to back.

I went through this too…

Some things to try:

  • Re-tighten all the bolts in your gantry
  • add a ramp to your toolpath, use a slow plunge rate
  • slow down acceleration settings in Mach 3
  • add weight to your base if you are using the keg kit. Hundreds of pounds if possible.
  • re-tighten all the leg bolts :slight_smile:
  • check the tension on all drive belts, tighten per assembly instructions

Good luck!

Did you actually fix your issue and was it that it vibrated things loose?

I changed acceleration to 25 and just recently returned it without seeing a change.
I always try to use a ramp unless I can do a helical into the material. I tried showing to 10 ipm and it seemed slightly better but, who knows.

Once I get the spare parts in, I’m going to have to go through and remove parts. Might as well work through everything and tighten down again.

I have the leg kit and I use it to store a bunch of wood slabs. I can’t say how much weight though.
I thought about, at one point, casting concrete blocks to fill that whole area. I saw someone add sheet steel to surround their base and make it more rigid. I’ve always thought there was a little too much moment of the base while changing direction. It could use a super beefy steel tube frame that’s welded.

There are so many bolts in this beast. I remember putting that together and spending half a day getting that done. I did find at one point that a bolt in a linear bearing rail had come completely loose. It is probably a good thing to check every so often. I never put the covers over those anyway. Too much work.

Random thoughts from someone who made a CNC router out of plywood and knows all about sloppy cuts…

If you try to move the spindle by hand, how much will it move? Is it about as much as your circles are out by? Putting a dial indicator on the bit and seeing where you can push against to get different reactions might tell you something.

I wonder if it depends on the toolpath used? I mean, if you always start your circle holes in the same relative location (say, the X- edge), have you tried a different one (say, the Y- edge) in case the changing load is pushing the bit around by different amounts at different locations within the circle?

Are you testing with a finish pass, or just roughing passes? If you rough a row of 0.470" holes, then go back and do finish passes at varying speeds, do you still see the same problem?

Another thing to try is to cut the circles in the opposite direction (climb vs conventional, or cw vs ccw). If this moves the egg shape from / to \ that might tell us something.

I think the CNC world needs the equivalent of a 3DBenchy :wink:


I will try running circles from different points. I have used both climb and conventional and will double check what happens again tonight.
When I push on the spindle, when motors are on, I get no noticeable movement. I can put an indicator on to see what the actual movement is.
Slowing the ipm to 10 have the same output as running at 120. So, I suspected that it’s hardware slop and not deflection from force.

There’s this, maybe it has some tips

1 Like

I’m still figuring that same problem with my diy cnc try to measure your travel with a ruler 1000mm for example and see if both y and x are traveling exactly the same i used to just mount a dial indicator and it would give me numbers within tolerance but when I used a ruler I saw the difference if you have the same problem you can either re calibrate your axis or just tweak the number in the motors tuning tab on mach3 good luck.

1 Like

Your machine is a bit older like mine so you might also be experiencing a problem with the tension post catching on the tension bracket.

The tension post is an all threaded M8 screw that runs through a bracket and keeps tension on the R&P. You will find it described here; 4. Rack and Pinion Drive - PRO CNC Machine Assembly Instructions

That bracket that the post (M8 screw) runs through will spin slightly and inhibit the tension post from freely moving. Essentially the thread will catch in the bracket and it will no longer move.

When it freezes on one side (usually the side that has the larger error, but do check both sides) the pinion gear will get pushed further and further off the rack and it will start having a swinging motion or oblong cuts in the product.

So go to the drive assembly and try to push down on it. It should freely move down and compress that spring and then return when you let off. If it doesn’t move or it makes a popping sounds as the threads slip in that bracket then that is possibly the problem.

Also, take two fingers and just push on each side of the gantry with the machine on and motors engaged. Push back and forth. There should be zero play.

I solved it by buying a partially threaded M8 and put some grease on it.

CORRECTION: The tension post is the part with the threading in which the M8 screw is threaded.

1 Like

I checked over what you said and you’re right about the full thread bolt getting caught. I may have had the bolt too tight though. I could barely push it down. I’m not sure how tight it should actually be. Three turns after hand tightening is on the instructions. How much should it might when pressing?

While looking at that I felt the pinion and it’s definitely worn down smaller than the part that doesn’t interface with the rack. So the parts coming in may be the trick. If they wore unevenly that might be the issue. If even, steps per changes could fix it.

I’m also going to replace the bolts with partially threaded ones.

We will see in a couple days. After installing mach 4, the machine sounds different and moved more smoothly. I don’t know if tuning is better or if the software itself is the reason.

1 Like

Yay! Another happy customer :wink:

I could be wrong but I think its about 0.25 inch or 6.35 mm in travel. But it should move freely.

Getting a partially threaded M8 that will fit is actually pretty hard. The problem is the length of threading for partially threaded screws that are common are too short to fit. I couldn’t find a 90mm long partially threaded screw that would screw far enough into the swivel to actually add tension!

I finally found a 85mm that has just barely long enough threads to fit. Being an aircraft guy the number of threads sticking out of an assembly is like 2 or 3 and this is just about that; McMaster-Carr

Please double check the length and the thread length before buying though. I am often wrong :frowning:

1 Like

I ordered some this morning. I was wondering if the length of threads would be long enough. I could always add a couple washers if a few more threads need to engage. Currently the bolt sticks out 5 or so mm out of the bottom. We will see.

I couldn’t reply yesterday after so many people have good input because I signed up yesterday and got my first day posting limit.

I really appreciate the insights though. I have been sifting through posts anywhere between 5 and 15 years old trying to figure this out. It’s my fault that I spent so much time tuning and dealing with software instead of going to the most simplistic solve from the beginning. I tend to go to harder things first for some reason. I troubleshoot systems for a living and somehow never go to the easy part first. My heuristics are out of whack.

I wonder if the tightness against the rack caused more wear or if I should just expect a certain amount of it over a set amount of time and always have the parts available for replacement every so often. Avid, please keep making these parts.

1 Like


I find its a function of time. 6-12 months is the factory recommendation. Me, about 8 months and I really, really need to change it out.

You will also burn through bearings because the design of a CNC gantry weighing that much doesn’t allow an elegant solution for a floating bearing.

So like on my retracting carousel you will find that one linear rail allows movement horizontally towards the other rail. What this does is take small differences in the alignment between the two rails and releases the pressure. This will increase bearing life and it will remove oscillation induced by the friction of the misaligned rails.

On our machines there is no floating bearing for the X or Y axis, probably in an attempt to increase rigidity. This adds vibration and wears the bearings quicker. That isn’t my opinion. That is just industry and manufacturer recommendations.

So that kit AvidCNC sells, you will use all of those parts. Consider them “consumables” when billing customers and calculating part cost.

PS. Please do mark my response as a solution so others can find it. This is a function of the machine aging and that pinion gear wearing that causes the bracket to catch the threads (ok, that last part is pure speculation :stuck_out_tongue: )

1 Like

So, have you replaced every bearing on the machine over time?
I will start calculating the cost over time to deal with it as a consumable. Man, I’m going to have some time in upkeep coming up and should purchase more bearings as well. Two come with the spare parts package and I think I’ll need 8 to deal with the x and y. I wonder how often the z needs replacement. That thing is a real pain to access and set up properly again. I need a clone, without wants of it’s own.


You will find problem bearing installs. Only on one side of the Y axis for me.

I just laser cut 0.25 inch (6.35mm) steel plate “dimensional lumber” as it is referred to in the assembly instructions and I am going to go back and straighten out my cross members. The laser cutter is accurate to about +/-.005 in (0.127mm) or better so this time around it should be correct sized and square.

My original “dimensional lumber” was cut by a dyslexic baboon :rofl:

I think that is what is causing my single sided wear. Doing that this week in fact.

I have had this issue before and it turned out to be a loose tensioning spring on my Y drive belt.

For my build I ditched the lumber. Here’s my notes to Avid on it: - using dimensional lumber to assist in squaring the
crossmembers is good, but using it as a spacing guide means
accumulating error. This means the rear crossmembers could be far off
from their ideal position. Instead, I used a pencil to mark multiples
of 400 mm from the front edge, onto the side rails (i.e. 400, 800,
1200, etc) so that all crossmembers are referenced from the front

1 Like

I remember measuring and marking with a carbide marking pen at each location when I had not put it together. Marking each line using a machinist square across the two rails. Then I made a jig to hold in place while using the square to make sure it was square while attaching. I spent a lot of time on that step because I inherently distrust myself with squareness.

I should go back and check over everything though. It has been since late 2017/early 18 since it has been in its home. I am sure there are other things that could be gone over if I have to get to bearings easily.

I really do not want to have to break it down and attempt to support things while changing them out. I can imagine messing something else up if I try to support the gantry while removing bearings. That cantilever action as it hangs there could possibly screw something else up on the attached side. I suppose only the bearings really. I have access to a mititoyo caliper that is 6 foot. I think I will be grabbing that today to see if it helps in determining the squareness of cross members. I can also use it by mounting to a jig and running longer distances to know my tuning is correct.

I swear. I bought a CNC to make sure that things come out accurate to my design. Having to trust myself to measure well enough still goes into every design because of me measuring during the build. It is a vicious circle. I need an AI controlled extra set of hands attached to me that has real time sub mm accurate vision capabilities for this purpose. Then have it make a copy as accurate for when the bearings on that thing wear. Real bummer. I could not afford that anyway.

OK, now you get my “Perfection is the enemy of Good Enough” speech!

You already know it :rofl: But, yes, maintenance is king!

It’s easier than you think replacing the bearings.

Moving the cross members is also easy. The trick is to loosen the one you want to move, loosen the next one over in the opposite direction your need to move the first. Put in the “dimensional lumber” as a stop and use long clamps between the one you want to move and the cross member you DID NOT loosen (aka, direction of travel). Slowly tighten the clamps as evenly as your can until it is up against the stops.