GME's New 80/20 CNC Build - My Design

[ger21]

You can't look at it as an absolute distance.
You have to use the angle deviation, and use trig (or CAD) to calculate the distances.
And you also can't use an example of 45° angle, which makes everything the same.
Say the gantry is tilted 1°.

At the tip of a 1/2" bit, tilted 1°, the difference between the front and back is .0087".
At the tip of a 60mm bit, tilted 1°, the difference between the front and back is .0412".

If you were to make a cut .002" deep, about 1/4 of the 1/2" bit would be cutting (roughly 1/8")
With the 60mm bit, the same amount of the bit would be cutting, only about 1/8".
When surfacing the spoilboard, you'll be cutting deep enough so that the entire bit is cutting.
Assuming 100% stepover, the 60mm bit will leave ridges nearly 5x taller than a 1/2" bit, when both are tilted at a 1° angle.

So it's not distance you are correcting, but angle.

As you reduce the stepover, the ridges get smaller, but only because subsequent passes are removing the high spots left from the previous pass.


The only thing I'm saying about bit diameter, is that if you're spoilboard has .001" ridges with a 60mm bit, you're not going to notice any error during normal cutting.

[GME]

You can't look at it as an absolute distance.
You have to use the angle deviation, and use trig (or CAD) to calculate the distances.
And you also can't use an example of 45° angle, which makes everything the same.
Say the gantry is tilted 1°.

At the tip of a 1/2" bit, tilted 1°, the difference between the front and back is .0087".
At the tip of a 60mm bit, tilted 1°, the difference between the front and back is .0412".

If you were to make a cut .002" deep, about 1/4 of the 1/2" bit would be cutting (roughly 1/8")
With the 60mm bit, the same amount of the bit would be cutting, only about 1/8".
When surfacing the spoilboard, you'll be cutting deep enough so that the entire bit is cutting.
Assuming 100% stepover, the 60mm bit will leave ridges nearly 5x taller than a 1/2" bit, when both are tilted at a 1° angle.

So it's not distance you are correcting, but angle.

As you reduce the stepover, the ridges get smaller, but only because subsequent passes are removing the high spots left from the previous pass.


The only thing I'm saying about bit diameter, is that if you're spoilboard has .001" ridges with a 60mm bit, you're not going to notice any error during normal cutting.

Thank you, Gerry. Cloud lifted. Your statement, "[a]s you reduce the stepover, the ridges get smaller, but only because subsequent passes are removing the high spots left from the previous pass" really brought it home for me. As a threshold matter, my measurement of the amount of deviation was flawed. I failed to account for stepover and the fact that subsequent passes would be removing high spots. So, at the bottom line, I really don't know the amount of deviation, and without having that number, I can't calculate the deviation angle. That also means I have no idea how much I need to shim.

The biggest problem is trying to calculate the deviation angle. I suppose I could make one pass at a depth sufficient to ensure full bit contact, measure the difference in the DOC at each end with a dial indicator, and use trig to get to the deviation angle. From that, I can use trig to calculate how much I need to shim to correct the deviation at the gantry level. Another option would be to level my glass plate on the spoilboard, and use my dual indicator tramming gauge to measure the height deviation. I'm thinking the tramming gauge should yield more accurate measurement. Trig does the rest. I'm not comfortable just laying the glass on the spoilboard, because there is twist involved and it will probably result in an inaccurate measurement without leveling.

You may be correct about not seeing the error in normal cutting. I'll have to crunch the numbers and see how it works out with the right ones. I am now confident that the deviation is more than .001", since I was reducing the true deviation (cutting off the high spots) with each successive surfacing pass. I'm pretty sure the deviation is at or less than .005", since I had full contact at .005" DOC. but if it's as much as .005, that's 5X more than I originally thought it was.

Does the methodology outlined above make sense, or am I still missing something?

Thank you so much for getting me on the right track.

Gary

[wmgeorge]

It would be simple just to do a test surface say area with your spoil board bit and another area the same size with common 1/2 end mill. A test plot in both directions.

[GME]

It would be simple just to do a test surface say area with your spoil board bit and another area the same size with common 1/2 end mill. A test plot in both directions.

Thanks, Bill. Certainly a possibility. To make it work, I'd have to use 100% (or maybe 90%) stepover. Otherwise, I would risk cutting down some of the angle with each succeeding pass. As I noted earlier, a single pass would probably do the trick as well. The 3rd option was to use my tramming gauge. My issue is with the ridges I get when surfacing side-to-side parallel with the X axis. I was able to handle running parallel with the Y axis with side-to-side tramming.

I decided to use a leveled piece of 3/8" thick piece of plate glass and my tram gauge to get a more accurate handle on the deviation and angle. Turns out the degree of deviation changed a bit, partly because this time I took my measurements at the extreme ends. I didn't do that when I measured before, but I was within a few inches. Turns out the deviation roughly doubled on the left side, and decreased by a little more than half on the right. Percentage wise, big differences. However, in raw numbers, we are talking about very small changes.

In my next post, I'll write more about methodology and get into the weeds on the numbers. I'll also discuss the pros and cons of what to do next. My intent is to document solutions for myself, and to provide others with some troubleshooting tools. I'm just happy that Gerry helped me see where I was going wrong. He changed my direction, and I found a decent methodology to for getting to the relevant numbers. A short, but very productive collaboration. I might have figured it all out on my own, were it not for it being some 50 years since I studied trigonometry in school, and my not having much use for it during my working years. I used integral calculus now and then, but not trig.

I should add that there are some very nice trig calculators online. They make make number crunching a breeze. Here's the one I used: Right-Angled Triangle Calculator

Gary

[GME]

The saga continues . . .

To get to the numbers I needed, I started by leveling a 3/4 glass plate on the spoilboard. I used 2 setups, one on the extreme left, and another on the extreme right. This was important, because the gantry beam has some twist. I can tell before measuring, because I have ridges in one direction on the left, and in the opposite direction on the right. At the midpoint, there are no ridges. That means that as the Z axis moves from left to right, it is tilting one way at one end, has a neutral position at the center , and then tilts the other way as it continues it's travel.

I used my Edge Technology Pro Tram System to take measurements. Here's a picture of it:



I oriented it so the dials faced the sides of the CNC. Using only one of the dials, I got the glass plate leveled. I zeroed the dials according to the instructions that came with the Pro Tram using the supplied magnate, and lowered the gauge until the plungers were depressed about .050". Then, it was a simple matter of reading the differences between the 2 indicators.

Here's what I found (fractions rounded):




Correction to the right side drawing. The .015" measure is wrong. It should read .0015".

The angle errors (.054 and .017 degrees) seem fairly small. However, when projected out over distance, t#he amount of error in thousands of an inch can increase rapidly over distance. My interface plates are 9.25". Assuming I shim between the horizontal lower interface plate and the gantry mounting plate, on the left side, it will take about .009" to correct the error. The right is much better. It will only take about .003" on the right over 9.25".

As noted in post #384 above, I used an online calculator to crunch the numbers. It saved a huge amount of time, and I could select how many decimal places up to 9. I highly recommend it.

Okay, let's look at what it all means in day-to-day operation. Aside from my 60mm surfacing bit, I limit my cutter to 1/2" diameter and under. I use 1/8" and 1/4" most often. On the left side, .054 degrees translates to a deviation of only .000471" for a 1/2" cutter, .000236" for 1/4" and .000118" for 1/8". Not much deviation. The right is much better: .000148" for 1/2", .0000742" for 1/4" and .0000371" for 1/8" - a minuscule amount. All of this is at 100% stepover. With a reduced stepover, the error should be even less.

Unfortunately, while the calcs provide the theoretical amount of shimming to correct for the deviation, the practical may not comport with theory. That's because the aluminum gantry beam not behave like the theory suggests it should. There will also be issues when it comes to re-squaring the gantry. Making adjustments means loosing mounting screws. Once loosened, the gantry will predictably try to return to its original shape. That will throw off measurements when trying to make adjustments. What should be a quick and easy setup process may turn into an exercise in trial and error. Seems like a long and frustrating process.

ger21 suggested in his post #381 that I might not notice the error during normal cutting operations. I tend to agree. If start a project requiring a higher level of precision, I can always make the adjustments at that time. I'll just hold onto the data I collected, so I don't have to reinvent the wheel.

Conclusion: For now, I believe I will leave things as they are. If I find the need for higher precision, I will make the changes at that time. In some cases, I can also mitigate the issue somewhat by cutting parts using a raster cut running with the Y axis.

If anyone would like additional detail, just ask.

Gary

[wmgeorge]

A wise decision, Now get started making things!

[GME]

Overdue for an update.

Call it a touch of OCD, or whatever, but I just couldn't let go of the gantry issue. So, I fixed it. I first trammed the gantry (front to back) with the Z axis to the extreme right. Then, I moved the gantry to the extreme left and recalculated the amount of error. The gantry is held in place by 5 pairs of bolts. So, I calculated the amount of shimming required to attain the correct angle at each bolt - maximum shimming at one end and tapering to zero at the other. I guess I did the calculations correctly, because I got it right on the first try. Tramming looks good in both directions all the way across. I ran my 60mm surfacing cutter, taking off .010" and the surface smooth - no ridges.

For my first project, I decided to cut some new hard maple hold downs. The t-track spacing is closer together than my previous machine and the hold downs from back when weren't optimal. Here's a Vectric Aspire rendering:



Cutting the hold downs revealed a new problem to address. I was cutting them with the length of the wood oriented from front to back, so that the slot was cut parallel to the Y axis. I was cutting 7 pieces at a time from one stick. The first 3 slots cut as expected, but the last 4 were off. The X axis went negative about 1/8" and held that position for the remaining pieces. The second stick was fine. The 3rd stick went off for the last 2 pieces.

My first thought was noise. Mechanical didn't make sense, because I used roughly the same positioning for each stick I cut. I also knew I hadn't made a positioning error, like at an angle to Y, because all 7 pieces would have had their slots cut at an angle. Not the case.

Previously, I tried using an AM radio tuned off station to try to get a read on noise. There was definitely noise there, but unless it was giving me problems, I was going to live with it. Turns out living with it wasn't an option after all.

mactech54 suggested specific line filter (a TDK Lambda) in another thread. I ordered and installed one. Cleaned the noise right up, or at least as far as I could tell with an AM radio. I could detect no change in the AM static, which was very different than before, without a filter. Without the filter, there was no question about noise. As soon as spun up the router, the noise was obvious. With the filter, no noise at all. Can't beat that. Still need to cut some parts to ensure the problem is solved, but I'm confident.

More to follow. . . .

Gary

[wmgeorge]

Good news Gary. It seems like SOP to add a line filter when using those Chinese drives.

[Sterob]

Good work compensating for the twist Gary.
It sounded like a daunting task to me....lol

I missed Mactecs comment about the noise filter. Which one did you end up buying?

Steve

[GME]

Good work compensating for the twist Gary.
It sounded like a daunting task to me....lol

I missed Mactecs comment about the noise filter. Which one did you end up buying?

Steve

Thanks, Steve. It was a bit scary taking on the twist. I worked on it in my head for days, before putting my hands on. Lot's of preparation paid off.

The noise filter is a TDK-Lambda (manufacturer), model: RSEN-230L. It's good to 250VAC; 30 amps. I have no idea where you would get one in Australia. I got mine from Mouser Electronics here in the U.S. They sell globally. I took a quick look at their global locations. The closest to you appears to be Singapore. Here's the address I found:

3 Changi North Street 2
LogisTech Building #03-01B
Singapore 498827

Here is a link to the Mouser page containing the filter I bought: https://www.mouser.com/ProductDetail...rrencycode=USD

Here's a link to the thread where mactec made the recommendation. It's on the 2nd page, post #21. https://www.cnczone.com/forums/diy-c...metimes-2.html

Gary

[GME]

Good news Gary. It seems like SOP to add a line filter when using those Chinese drives.

Thanks, Bill. Funny, but I didn't have problems with my first machine, which ran on the same electronics. I'm guessing that it had to do with physical separation. With my first machine, the VFD and control boxes were separated by more distance, and the cable to my spindle was separated from all of the other control cables. With this build, the boxes are very close together, and the Spindle cable is in the same cable tracks as all the other cabling.

I'm with you. A line filter is a must for anyone using HY VFDs. Fortunately, they are a relatively inexpensive solution.

Gary

[mactec54]

Good news Gary. It seems like SOP to add a line filter when using those Chinese drives.

There is no difference between any VFD Drives it does not matter where they are manufactured or by who they all put out the same EMI, if they don't then they are not
power up

How cables are placed shields are Grounded play a big part of a noise free system

Placement of the VFD Drives in a separate cabinet can sometimes give a false impression that you don't have a noise problem

[Sterob]

Thanks, Steve. It was a bit scary taking on the twist. I worked on it in my head for days, before putting my hands on. Lot's of preparation paid off.

The noise filter is a TDK-Lambda (manufacturer), model: RSEN-230L. It's good to 250VAC; 30 amps. I have no idea where you would get one in Australia. I got mine from Mouser Electronics here in the U.S. They sell globally. I took a quick look at their global locations. The closest to you appears to be Singapore. Here's the address I found:

3 Changi North Street 2
LogisTech Building #03-01B
Singapore 498827

Here is a link to the Mouser page containing the filter I bought: https://www.mouser.com/ProductDetail...US¤cycode=USD

Here's a link to the thread where mactec made the recommendation. It's on the 2nd page, post #21. https://www.cnczone.com/forums/diy-c...metimes-2.html

Gary

Thanks Gary,
I think TDK have a presence in Australia. If not, Mouser is a possibility.

[GME]

There is no difference between any VFD Drives it does not matter where they are manufactured or by who they all put out the same EMI, if they don't then they are not
power up

How cables are placed shields are Grounded play a big part of a noise free system

Placement of the VFD Drives in a separate cabinet can sometimes give a false impression that you don't have a noise problem

I know what you mean about putting the VFD in a separate cabinet can give a false sense of security re EMI. I put mine in a separate cabinet mistakenly believing it would make a huge difference. It didn't. What made the difference was proper grounding, and a good line filter. On my first machine, where the spindle cable was not properly grounded (grounded at only one end), noise didn't interfere with operation, probably because I put the spindle cable in a separate cable chain running in the opposite direction and physically separated from the other cabling. I'm guessing that the physical separation and distance kept me from suffering from noise induced issues, despite the fact there was still a lot of noise.

On my current machine, I have the spindle cable running in the same cable chain as stepper and proximity switch cables, which seems to be the more conventional setup. Proper grounding and filtering lets me get away with it.

Thank you so much for your willingness to help us get our systems working properly. I got my VFD settings from your various posts (2.2 kw HY), and learned the proper grounding and filtering methods from the information you have so often provided. Your specific product suggestions were also very helpful. You are a valuable resource and much appreciated.

Gary

[GME]

Here's a brief update. I made some modifications for my dust collection. I have a Kent CNC dust shoe with the 4" fitting. Using it alone caused my dust collection hose to be in my way when setting up stock for cutting. The attached photos show my mods:



I used 1/2" Baltic Birch (BB) to make the part that holds the hose fittings. They were attached to z with some straight pieces of 1/4" thick aluminum bar and 1.5x1.5x.25" angle on each side of the cable chain. It made a very rigid support. The fitting just below the BB is just a standard 4" inline hose connector. The fittings on top consist of a Rockler rotating fitting capturing the inline connector from the top, and a Dust Right fitting attached to the rotating connector. The rotating fitting allows the hose to swivel was the Z-axis moves around the table.



This is a closer shot. In it you can see the copper wire on the right side. Although hard to see, it's attached to an alligator clip clipped to the hose clamp above the BB support. The wire attached to a corresponding grounding post on the dust Dust Right fitting that connects to the hose attachment.



It's a stretch attaching the dust collector hose. I'm 5'-7" and can just reach it without having to use some sort of step. In the extreme left, you can see the PVC pipe with grounding wire wound around the outside. There is also a wire on the inside of the PVC as well as a ground wire inside the hose. For even more grounding effect, I have ground wires attached to the steel spiral wire in the hose. You can see the orange wire nut in the first photo as well as this one. The ground wires go to an earth ground. I went from hair standing on end static and tripping limit sensors to essentially no static.

NEXT UP: A rundown on the removable dovetail/mortise accessory I designed and built that allows cutting joinery off the front of the CNC.

Gary

[GME]

Here is my joinery jig designed to cut standard joinery off the front edge of my CNC. Although it's my design, I borrowed heavily from the Leigh D4 series commercial jigs. In fact, I did more than borrow from the jig design, I also borrowed the clamps and springs from the D4 jig I've owned for many years. The clamps are the black ones at the top and front corners.



The base for the jig composed of 2 pieces of 80/20 smooth extrusion, 1.5" x 4.5" x 42". If you look closely at the front, you will see 5 holes drilled in the top slot. The holes are for access to button head cap screws used to attach the two extrusions together. When connected, I have 4.5" of height at the front and 6" of depth across the top.

The stock is clamped using a 1.25 x 1.25 x 42" x .125" wall square aluminum tube. 5/16" studs cut from threaded rod are attached to 80/20 fasteners captured in the extrusion slots and held rigid with a 5/16" nut.

You can also see 3 holes in the aluminum tube, 2 with studs and the one in the center open. The studs are readily removable. They allow me to use any width of stock and be able to tightly hold it in place. I have some 5/16" knobs on order, which I'll use to tighten down the tube.



This is a closer shot to the right side. You can see vertical and horizontal pieces of 1" x 1/8" thick aluminum bar stock used as a material stop. The same arrangement is on the left side. Cut the vertical side a little long, so the top extended about 1/8" above the horizontal face of the extrusions. I used a square to align the vertical face, and the 1/8" projection gave me a hard surface to use to align the horizontal stop perpendicular to the vertical.



This is a closeup shot of the right side from a different angle. You can see the stops and how they attach. You can see the compressed screen behind the vertical tubing clamp




This photos show the side of the aluminum tube that contacts the stock. The red material is 220 grit glued on sandpaper. I learned that stock can slip on the Leigh jig and sandpaper cured the problem.

You will also note that the large holes in the tubing. I cut them on my mill. They are large enough to allow the springs to easily go inside the tube, and are large enough to go over the nuts holding the studs captive. That means I can clamp all the way down to the surface of the stops, which allow me to use stock that is thicker than the 1/8" thickness of the stops. I doubt I'll be cutting much on stock in the 1/8"-1/4" range, but I will have the capacity.

There is big difference between the Leigh jig and mine. The Leigh jig uses fingers to position stock vertically and horizontally, and has a large void in the body to allow the cutters to enter the stock without contacting the jig body or parts. That cannot be easily replicated using 80/20. My solution follows:



My solution is a 1" thick piece of acrylic with tapped holes and a 1/4" thick piece of aluminum bar stock screwed to it.



Here is another shot of it in place. I made it large enough to use the clamp to hold it in place with the top bar clamp. I just clamp the stop in place slide the stock up until it makes contact. This stop will work with 3/4" thick and thinner stock. I'll need to make a different stop of thicker stock. I probably won't be making many dovetails in thicker stock, but I will definitely be making tenons in thicker material.



Finally, there is the question of how to attach the jig to my spoilboard so I have a rigid connection that won't interfere with stock mounted horizontally, as when making single pass half-blind dovetails. My solution was 3 pieces of 1.5" x 1.5" aluminum angle. As is the usual case, the two faces of the aluminum angle were not perpendicular. They were close, but no cigar. I use my mill to square the faces, and drill the holes. Using 80/20 fasteners, I was able to slide the angle until it aligned with my t-slots and used 1/4" studs and slide-in t-nuts to hold the jig in place. I ended up with a rock solid connection.

Oh, one other thing. I believe I mentioned it in a previous post, but I ran a cutter across the front edge of the spoilboard to establish a straight edge perpendicular to the Y-axis. So, I just but the jig up against the front edge, tighten it down, and I'm good to go. No other alignment necessary.

I am happy to answer questions. No secrets here.

Gerry, (ger21), if you happen across this post, feel free to share it wherever you want in conjunction with Joint Cam. If you have suggestions where I might post this elsewhere, I'm open to suggestions.

Gary

[nlancaster]

I am stealing your joint fixture design.....

Looks great!!

[Ntl]

Hey Gary are you going to post pictures of the finished machine? I wanted to see how all your hard work turned out. How did it wind up budget wise,were you close to your target?

[GME]

I am stealing your joint fixture design.....

Looks great!!

Thank you. I'm glad you like it. Steal away! Let me know if you have any questions.

Gary

[GME]

Hey Gary are you going to post pictures of the finished machine? I wanted to see how all your hard work turned out. How did it wind up budget wise,were you close to your target?

Good evening, Dan.

Sure, I can post some pictures when I'm finally done.

Right now, I'm rethinking my support structure. My first machine had two torsion box cabinets as the support base. Each cabinet was 60" long x 24" deep. there was a torsion box at each end and in the middle. I then added a torsion box top that the CNC set on. I used hockey puck adjustable feet that screwed into coupling nuts epoxied into the torsion boxes. So, I had 5 feet per cabinet. Very rigid; no vibration issues. Solid as a rock.

My plan was to add some aluminum gussets to the legs and support braces to increase rigidity. I even have the gussets rough cut and ready for final cutting on the CNC. I was about ready to cut them, when the idea of making new cabinets for this machine occurred to me. I repurposed my old ones, although I could resurrect them, if they are tall enough. I need to measure them. Also, I don't want a total of 4' of depth. I don't need it, and I prefer to have my electronics underneath the spoilboard, rather than on the side, where they take up too much space. I want a final bed height of of something around 38.5" to give me a final height of around 40" for joinery purposes. I really wish I could have way more height for joinery, but I'd need a ladder to use the machine.

I'm still chewing on which way to go, although I know cabinets would provide the more solid substructure. I suppose I could weld a frame, but I want something perfectly flat. I couldn't get that by welding, but can get there with torsion boxes.

Although I have far from unlimited resources (I'm retired after all) I really didn't have a rigid budget. The greatest costs were in purchasing the aluminum and hardware from 80/20, so I had those costs up front. On top of that, I replaced some cabling, and bought some aluminum plate and angle. I also bought Hiwin linear rail components and the Avid CNC Z axis. The extra costs probably totaled around $2,500 or so. I'm at about $5,500, all in, excluding motors, electronics, spindle, etc, which I already had and were functioning properly. A few hundred more than the Avid Pro 4x4 kit + leg kit, but my machine is nearly twice as heavy (the extrusions are twice the size of Avid's, except for the gantry beam) and I have much higher quality (and more expensive) linear rails. Cost wise, it worked out at about what I expected. No major surprises.

Gary