Linear rail alignment for better precision

Hi all!

This is a video on how I attempted to get more precision out of my build.

I used a precision level to level the base, which took a long time. I then assembled the rails and gantry per the Avid instructions. I was a bit surprised when I tested my linear rails for level-ness; they were not anywhere near level, and actually had a bend in the middle. This is due to the instructions aligning the rails with two alignment jigs. I measured them with calipers and discovered they were about 0.005" different. This made it impossible to use them to get everything level, so I re-did it based off the machinist level.

I made a video of the process in hopes that it might help someone else make their build more accurate!


Those who want the most accuracy out of their Avid PROs should check their rail alignment. The procedure and jig provided by Avid does not guarantee precise alignment of the rails. My PRO4848 rails, both X and Y, were off by several thousandths before I discovered and corrected it.

1 Like

How did you correct it? Did you use a machinist level, like I did above?

I didn’t use a level , I just made corrections until it was close to parallel to the bed using a dial indicator in the spindle chuck and a 1 meter long straight edge on the bed. Not as good as your process and not as accurate as I would like but I didn’t want remove the spoil board and start over again. When the spoil board wares down and I have to remove it I will probably use a version of your process to get it where it should be.

1 Like

I’m all in favor of assembling the machine as precisely as possible and took many similar steps when I was building mine. But… they are ultimately still bolted to 80/20 in the other dimension and 80/20 only specifies:

Flatness: 0.004" /in width
Straightness: 0.0125" /ft length not to exceed 0.118 over 20’
Twist: 0.25 degree/ft

I suppose you could shim it straight, but I think realistically understand that there will always be error and know how to measure and plan for it.

1 Like

That steel tubing that @corbin used for his base has just slightly better specs than the 8020 but not by much. But with the steel he could grind it back to spec after welding it. The 8020 is a lost cause. Everyone who futz with it is wasting their time.

My advice, get it close and compensate in software.

Doesn’t that not matter? If you align the rails with a level, then that is what the machine rides on. The spoilboard will be flattened with respect to the level machine, and also be level with no twist.

Your method will ensure that the spoilboard ends up level, but if you try to cut a straight line in a piece it won’t necessarily be straight. Just imagine that your gantry extrusion is slightly curved in the plane of the spoilboard. It could still be level using your measurements, but whenever you cut a line along the x-axis it will make a curved cut.

The x-axis gantry could be bent like a semi-circle. As long as the rails on the gantry are perfectly level it won’t make a difference.

I must be misunderstanding what you are saying…can you draw a picture?

Actually that reminds me that one of the things that irked me is that the rail rests on the end caps. The extrusions have a 2 degree “drop lock” that gets flattened if the thing being bolted to it covers the entire profile. The rails don’t cover the whole profile so they are actually partly recessed… until they get to the end caps. So all Avid machines will have a slight curve at the ends of the X axis. This bend cannot be measured with a level (it can with a straight edge, I’ve done it).

1 Like

Here’s a picture of what I mean:

The rails could still be level, but imperfections in the other dimension won’t show up on the level.

1 Like

I am sure you all have heard the saying, “lipstick on a pig.”

In this case, a hand welded and tuned steel frame with 8020 sat on top would essentially be “a pig on lipstick.”

1 Like

Ah yes! I hadn’t even considered that issue. I wonder how far off mine is over the 8’ table…I’ll have to figure out how to do some measurements, and possible corrections.

I don’t know how far down the rabbit hole you want to go, but with a $20 webcam, a laser, and some software, you can measure significant distances down to a few microns:

Measuring flatness

A good touch probe and a long precision straight edge would work too. You can make the corrections in software instead of trying to make the machine physically perfect.

However, if your primary working material is wood, be mindful of your time and diminishing returns. A piece of wood will move more than your current precision if you just stare at it for a few minutes or take it into a different room.


Awesome…thanks for the link - I’ll look into it. I am doing wood, so it simply may not be worth my time to even go down the rabbit hole.

When you guys say “fix it in software”, do you mean set something in Mach 4, or in my CAD/CAM designs by just accounting for issues by designing it in? (Fusion 360 and Vectric VCarve). If you set something in Mach 4…what are you setting?

I haven’t checked my Alignment blocks yet but was wondering in the video your putting the Gauge on the top of your spoilboard. Wouldn’t that be better to be on the 8020?

That’s a good question!

If you check your blocks, check them from the machined area to the bottom; in the video I did it incorrectly. Turns out mine are 0.010" off in that direction, so it was even worse.

I didn’t explain really well on how I used the dial indicator; I put some notes in the description to help clarify: I use the level to see if the rail needs to go up or down. I use the dial indicator to see how much I’m moving it up or down, so I can only move it the smallest amount (I moved it about .0002-.0003" at at time, eyeballing between the thousand line markers). It doesn’t matter what the dial indicator is sitting on, since it is just relative movement.

You don’t want to use your dial indicator as a reference down the entire rail; it would only make it as level as your 8020.

I hope that makes sense!

Yes it does :wink: thanks!!

1 Like

I went down the rabbit hole of sanding flat 40x80 extrusions (1 for each y-axis rail) on granite surface plates, and then epoxying+bolting them to main 160x80 extrusion once I had aligned them flat using the same precision level from your first picture. This moves the y-axis rails 160mm further apart and required a new z-axis mounting plate.

Epoxy is currently curing, but I can share the process if anyone interested.

Really like the precision level from Amazon. Started using a method where moving it down the length of something and recording the slope change in excel, then graphing it to get a “topography”, and using that to fine tune sanding flat. I think was able to get to +/-0.0001” down length of 1000mm 40x80 extrusion.

1 Like

Awesome! I’d be interested to see pictures of the process. If you have a second…post a few !

I think if I do it again, I think I’ll use tinkeringtechie’s video suggestion using some cheap webcams.