[nateman_doo]

Starting a new build since my 18x18 Tormach Build worked out so well. Looking for something with less travel, but more rigidity.

The first part is the base. A 0.625" slab of precision ground Aluminum. Was debating on Steel for all three plates, but this plate would be too cumbersome to deal with as steel. It will be buried in concrete anyway... so whats done is done.

I had 2 notches milled out to butt the rails against to have as proper alignment as I can with used rails:



Rail up against the side. used a through punch to mark the aluminum for drilling.


The Tormach build took me over TWO YEARS to make. This one should take FAR less time.

[nateman_doo]

Now, my first question to the general public...

I already have 420oz Stepper motors, the power supply, the Bob.. and everything on my X2 build.
Would you scrap it all (or sell) the old stuff, and buy Servo motor kits for each axis... or leave the old electronics in place?

So far, the biggest advantage I see in Servo motors are the encoders. I have had my steppers stall, and Mach doesn't know this and it throws the entire coordinate system off and cuts the part completely off where it should (generally resulting in a wasted part). Since now I cut more then 1 part at a time, that feedback would be very helpful. But at 600 per axis... its a HUGE investment in a machine that is already going to become a HUGE investment.

Since everyone here knows much more then I do... I would be interested in others thoughts.

[Zach_G]

$600/axis sounds pricey, automation tech sells a 3 axis kit for $320/axis plus the belts and pulleys for gear reduction. If you're willing to wade through broken English manuals and take some risk you can get an AC servo set by ordering direct from China via AliExpress. Ballscrews are also a lot more affordable direct from China, that's how I bought mine. I'd want some large diameter (20-25mm) precision ground ballscrews with preloaded double nuts for the most rigidity.

Got some drawings or models of the final design or do we have to wait till the build is complete!?

[nateman_doo]

Sorry, your going to have to watch it unfold. I tend to design on the fly, or design as you go. Lots depend on me getting a hold of things.

[fixridermtl]

nateman_doo, empty some of your pm folders please. The system won't let me reply.

[nateman_doo]

Ordered the saddle plate of steel. 12x24x1 Generic low carbon steel. 80 something lbs. Probably as much as all 3 of the plates in my Tormach build!

[nateman_doo]

Installed the master rail:

See the clamps bowing? I have it as tight up against that lip as i can.

Now here is how I tested the height:


Dragged the indicator stuck on the 2x4x6 block, and slide the block up and down. Absolutely ZERO movement on the indicator... as I will show here:

the indicator isn't really in focus (damn cell cams) BUT, you can see the indicator not moving at all, until I slide the block out from under the indicator.

[nateman_doo]

I am working on the second rail. I have bolts in all the holes, and I am pushing, and pulling the rail into place while sliding the dial indicator along. SUUPPEER tedious, and downright annoying. However, you can't put a price on precision.

[nateman_doo]

Would you guys accept this as straight?

[nateman_doo]

So after spending the afternoon on it..I got it to under 0.0005. I threw on the 0.0001 gauge (analog just seemed to give more feedback in this case), pushed and pulled the rail as best as i think a pair of used e-bay rails can be.

[nateman_doo]

I tried a new technique. Instead of squaring the vise before I put a part it, I put a part in and squared the vise with the piece. ordinarily, that wouldn't be a bright idea, but the part was a precision ground squared block:

(I squared it at 12")




So these blocks are the "saddle joints" since I don't have a better word for them:





They create space for larger ballscrews, and they give something for the large saddle plate to bolt into (instead of the linear guides themselves. Not to mention, keeps the spacing of the blocks. I will use a few bolts, and pins to hold the solid plate to the joints (if i ever have to take it apart in the future). Somehow, I have to fit this already 100 lbs beast on my Tormach build to drill the holes through the saddle plate into the joints.

[nateman_doo]

So.. back to this picture:


I was thinking of a way to secure the 80/20 to the lower saddle.
Attachment 224484
I will drill out the threads in the lower plate, and get bolts long enough to go through the lower plate, and pull the extrude up, and push the rail down, sandwiching the plate.

[nateman_doo]

Now, back to this photo:


I figured, why not re-calibrate the steppers. So after a few minutes messing around, I got this:


after screwing around a bit more, I got it to around +/-0.0001 using an edge finder

[nateman_doo]

The edge finder i used was an electronic ball type. Mine happens to be 0.1964 from the edge, so I took it to the left edge and went to -0.1964, and went 12" and dropped the Z down. It read 12.1965, so i moved the edge finder back 0.0001, and it lit up. I figured that was good enough for the work I do.

[nateman_doo]

So, I am looking at this kit:
3-Axis NEMA23 350oz in 60VDC/20A PSU G320X Gecko Driver | Automation Technology Inc

Has anyone mounted the encoders on the shaft? I would think that would be better then the motor, in case the coupler breaks (just had that happen last night on my other machine). What happens when the motor gets the command to move, but it sees the shaft isn't moving?

How does a 350 oz servo compare to a 540 oz stepper?

[Bubba]

[QUOTE

Has anyone mounted the encoders on the shaft? I would think that would be better then the motor, in case the coupler breaks (just had that happen last night on my other machine). What happens when the motor gets the command to move, but it sees the shaft isn't moving?[/QUOTE]

I have my X encoder on the mill mounted on the shaft and has been working well for about 14 years! I had a belt break one time and because the motor turned and the shaft didn't, so it faulted out on an error!

You just have to make sure there is no backlash between the motor and shaft.

[nateman_doo]

In reality, wouldn't that help get rid of any backlash between the coupler and the shaft? Or is it that sensitive where if it moves a hair, and if it doesn't see movement it faults? Oh...and on belts, you think they are better then direct drive? I know its nice to be able to tuck the motor under the table and use gears, but I always feared backlash with belts and gears. I would love to hear more of your thoughts on that since I am still in the design stage.

Ideally, I would think some type of linear encoder on the actual ballnut/table would probably be the most accurate. Either way I really think I would like it on the shaft, I just have never touched an encoder so I have no idea how to accomplish it until I have it in my hand. I am still deciding on ballscrews/motors at the moment. I want something thicker then the RM1605's that I used before.

[Bubba]

Nate,
Bear with me as my "design" work was back in 2000!
In a simplistic way, the servo "dithers" between two adjacent pulses when at rest and while this is minor in a tight system, it can be drastic if there is a time differential between when it sends the pulses and when it receives them. I didn't believe it when I did my system as I had a dro on the mill and used that initially thinking the same thing as you. In about 3 seconds, I discovered the system was going into a violent oscillation that kept getting worse by the millisecond! You need tight coupling between the servo and the encoder/shaft.
Also remember, most servo motors are rated for a lot higher rpm than steppers so you need reduction to get them to the torque level you need. This is a combination depending on the max rapid you want vs the power needed to do the job. this is a system that needs design and not just throw whatever at it..
For reduction, I tend to stay away from gears as they have considerable backlash unless you throw many $$$$ at them. On the other hand there are several different types of "timing" belts that are reasonably priced and have minimal or no backlash. The current flavor of the month seems to be the HTD type of belt as it is claimed to have no backlash. On my mill/drill and hopefully soon to be done lathe, I am using the trapezoidal type of belt (specifically the XL 037 style).

Design your system to have no backlash and you will be a lot happier. Backlash compensation in the control program is no substitute.

For the absolute best accuracy, you need a control program that will accept both a "shaft" encoder and linear scales on the axis.

HTH

[Zach_G]

Wish I had a torque-speed curve to illustrate that those servos will far exceed steppers performance-wise but you absolutely must gear them down. 4:1 ratio would allow over 200IPM rapids on 5mm lead ballscrews and provide 200/1400oz-in continuous/peak torque over the majority of that speed range. By comparison, a direct drive 370oz-in stepper would give comparable rapid speed but only maybe 50oz-in torque availabe at higher speed range.

As for belts, trapezoidal shapes such as M, L or XL designations are prone to backlash because the trapezoidal shape requires extra clearance to allow the belt tooth to enter the pulley slot when rotating. Curvilinear belts such as GT or HTD are shape optimized to allow meshing with little to no clearance required. Last time I priced pulleys Timing Belt Pulleys & Timing Belts - Power Transmission Components - Synchronous Drive Parts | B&B Manufacturing had pretty good pricing and selection.

There's a surprising amount of flex in those RM series "anti-backlash" ballscrews - I got the 20mm versions and still crave more rigidity. Two ballnuts preloaded against eachother would probably do it and the nuts are so cheap it's not worth the time to experiment with oversized balls.

[nateman_doo]

Thank you both for the input. having zero experience with gears, or servo's, I crave the input. I plan to use the RM2505 series ballscrews, with 2 nuts compressed with wave springs. Thats the easy part. As for the gears, and timing type belts, That is something i will DEFINITELY look here for help.

I like the gears and timing pulleys because I will tuck the motor under the table, and I will use a sliding mount to keep tension on the belt. Also I think I will encorporate some type of additional rollers to keep as much surface area contact with the gears as possible.

I am browsing through McMaster for pulleys, but they don't have matching sets for 4:1