Building B-Carve

EliasPolitis · March 19, 2018, 8:11pm

Hello puppies. Ok, I had to post this picture just for the padding. My girls love that it’s pink. Dad loves that it’s modular.

Hello 75 Volts.

So, I got two SDSK-2321S (for the X and Y) and one SDSK-2311S (for the Z). SD for step & direction, SK for stepper killer (that’s some ballsy marketing). Each model comes in 3 flavors: D, P, S (in descending order of max RPM and ascending order of max torque). Since I will be direct driving them below 2000RPM, I went with the S variety that packs the most toque.

Here’s the curve for the SDSK-2311S:

And for the SDSK-2321S:

Here’s the bigger one (SDSK-2321S) next to a Stepperonline 269ozin. This is super compact considering it also carries its driver on its back. The default configuration is a 3/8" shaft, but it can be ordered with a 1/4" too. I went with the former and also switched to 3/8"-to-10mm couplers.

The autotuning process with the final load (i.e. the spindle) took about 10mins per motor. In case something changes in the future (e.g. a heavier spindle) a new profile needs to be generated by the tuning process.

JeremiahRandolph · March 19, 2018, 9:57pm

Those sure are real purdy!!

NeilFerreri1 · March 19, 2018, 10:03pm

C’mon… Video!

EliasPolitis · March 19, 2018, 11:06pm

Almost there. The Arduino needs to get wired next, then settle on speed and acceleration per axis. Then I’ll post some footage of the first cut with the new setup.

patience-young-grasshopper

EliasPolitis · March 20, 2018, 9:19am

Wiring. Cheat sheets of each side to avoid back and forths.

Quick and dirty. Don’t tell anyone. We’ll hide it under the carpet and make it nicer at some point later.

A couple of notes.

E-stop on the enable signal. This should be the fastest way to kill the motors in case something goes wrong.

Grbl recompiled with

#define ENABLE_M7

uncommented in order to have two options for controlling external devices (M7 on pin A4 and M8 on pin A3). In the latest release at the time of writing (v1.1f.20170801) it’s commented by default.

Oops, you can still see the bird nest.

Now you can’t

I think we’ve reached the Hello World status.

EliasPolitis · March 20, 2018, 11:19pm

So, why these electronics? Is it because John Saunders thinks they are cool? Well, is there something that he thinks is not cool ? Is it because of the motor and driver all-in-one body? Well, that’s nice but it’s not a big deal. Wiring happens once and you don’t mess with it again.

However, there is one thing that’s not advertised anywhere: they come with a free soft oscilloscope. You don’t have to feel the machine, you can see what is actually happening. No more guesswork. And this is a pretty big deal.

Here’s an example of what happens in a mere 10mm move. Velocity 1,600RPM (8,000mm/min), acceleration 120,000RPM/s (10,000mm/s^2).

Here’s the torque. Max torque reached 19%.

Voltage, steady at 75V.

This one is very critical. Positioning. Not the commanded positioning, it’s the actual positioning. The output. If something was jerky, if there was binding, high friction, it would demonstrate here …as spikes and dents followed by abrupt corrections. The motion bellow is silky smooth.

Here’s a comparison of 120,000 vs 240,000 vs 480,000 RPM/s acceleration. Didn’t break a sweat. Max torque 39%.

There’s also a great debug window below the scope that monitors all signals and alerts in several occasions (e.g. torque saturation, voltage saturation). I was not able to push it that far. It’s funny how I started with my electronics being a bottleneck for my machine and ended up with my machine being a bottleneck for my electronics. Now I have to tear everything down and build a VMC. Just kidding

Conclusion…

So here are the acceleration values the machine can now safely take (without trying to take off).
X: 10,000mm/s^2 (66x higher)
Y: 10,000mm/s^2 (66x higher)
Z: 20,000mm/s^2 (133x higher)

Excuse the excitement, it’s not the numbers only. I just returned from the first cut. Without my jaw . And yes, @NeilFerreri1, I took it on video.

Gotta get some sleep first. And I have the Route 66 song in my head now on repeat. Great.

NeilFerreri1 · March 20, 2018, 11:39pm

The numbers and that data visualization is pretty sweet too, but I’ll be looking for the video soon.

BrandonR_Parker · March 21, 2018, 2:47am

You guys should check out the NYC CNC website. I enjoy watching his videos and there are a few he has posted using ClearPath Servos that you might enjoy. Sorry if I hijacked here a little, but the ClearPath Servos are awesome and I wanted to share the link.

{:0)

Brandon Parker

EliasPolitis · March 21, 2018, 8:03am

@BrandonR_Parker, no problem. I do watch his channel (and Lars Christensen’s and Rob Lockwood’s among others) and I like how he’s self-taught …like most of us.

Ok, so here’s the first cut. (Volume down please)

Here are the cut parameters.

bit: 1/4" 2 flute upcut
speed: 9800RPM
feed: 4489mm/min
DoC: 8mm
WoC: 0.2mm
chipload: 0.229mm
chip thickness: 0.08mm
deflection: 0.001mm
MRR: 7.18cm^3/min

MkII now cuts aluminum as easy as MkI was cutting acetal. And sounds beautifully, zero chatter.

0.08mm is the thickest chip I’ve cut so far. This is what it looks like. Nice and consistent and light colored.

I’m confident this can be fine tuned to 0.1mm and MRR can reach 10cm^3/min too. Also, for first try, not bad finish at all.

One flip and two chamfers later… btw, this little plate is going to be the Z carrier plate for the X-Carve which is switching to a dedicated laser engraver.

HaldorLonningdal · March 21, 2018, 3:35pm

At 5cm^3/min mine doesnt sound half as sound as yours Very well done, and to you usual and appreciated standard!

Really looking forward to the result of fine tuning of this system

HaldorLonningdal · March 21, 2018, 10:31pm

When will you try 30k rpm and 13,5m/minute ?

EliasPolitis · March 22, 2018, 10:13am

<goofmode>

I ordered a truckload of epoxy granite. I thought to use the whole room as a mold and start filling from the floor all the way up to the machine base level. Then half a million RPM/s of acceleration or more would be no problem.

My ooooonly problem would be that with the machine being on the new “floor” my back might start hurting when using it. And I hope my head won’t be hitting the ceiling.

</goofmode>

If you’re blood thirsty for high MRR’s we could satisfy this with wood cutting though.

HaldorLonningdal · March 23, 2018, 7:59am

Hey, I am living out my upgrade fantasy through your build
Thank you for sharing, also the reasons behind the decisions.

IanDudman · March 24, 2018, 10:24am

What a fantastic build am very envious. Wanted to ask where your got the mount for the router, with the coming fins.

EliasPolitis · March 24, 2018, 3:27pm

Thanks. The spindle mount is this one:

There’s also a 80mm version.

EliasPolitis · March 28, 2018, 9:34pm

When life gives you left-over extrusions, make something.

Frame.

8mm acrylic. Had a couple of sheets from office renovations at work.

Measure and mark.

I want the circular saw to be able to easily come off and back on. I wouldn’t trust threads on the acrylic to last. I had once ordered some helicoils but ended up not using them. Now is a good opportunity to try them. They come with the appropriate drill and tap size, a fitting tool and a break tool.

Fitted.

Saw in place. Riving knife off (it would interfere with the cut otherwise).

After the cut is made, saw is taken off, top is slid backwards, secured, frame flipped, saw spaced and placed in position, cut extended for the knife to fit.

Top slid back in place, riving knife on, saw fixed.

Let’s make a sled. 4mm acrylic this time. This is a small saw, saving a few mm in thickness here results in more cut height.

Extrusion fixed.

Bought a couple of long T-nuts and machined the top face and two grooves on a couple of bars.

This is how they fit together.

They slide in the outter slots. That was the original aha moment for making this. Having 3060 extrusions (i.e. two slots when lying flat) allows for one slot to be used for fastening and the other one for sliding.

And a couple of toggle clamps.

Done.

EliasPolitis · March 29, 2018, 5:46am

It’s not typical, but several models have started incorporating them.

Kickback is not very likely to happen with a circular saw (or would that be be kickfront? ). But in the very unlikely event that it happens (i.e. no riving knife and the split pieces push inwards), considering that the work piece is clamped down and cannot jump forward, the saw might try to jump backwards. Too scary to even think about it.

I guess it’s kind of like the airbag. It’s nice to have it for protection in the (hopefully) rare event that it will actually help protecting you.

EliasPolitis · March 29, 2018, 2:53pm

Sometimes I don’t. Then if it tries to run away, I shoot it with a crossbow.

HaldorLonningdal · April 9, 2018, 7:06am

@EliasPolitis

The Clearpath are closed loop, do they communicate between the drivers so they stay in sync relative to eachother?
If for instance one axis would lock up, do the other motors delay untill that unlock?
Or does it go into alarm state halting all axis?

EliasPolitis · April 10, 2018, 4:40pm

As we were discussing at the closed loop stepper thread, in a closed loop system (servo or stepper) the driver knows both the commanded and the actual position and tries to compensate when they are found different. The drivers will also typically have some sort of output signal. In step & direction mode the output signal of a Clearpath’s embedded driver can be configured to report shutdown, torque or speed.

Without using this output, each driver only cares about its own axis. Having switched to single Y transmission (and having each axis properly aligned) I have no particular use of the output signal. In case I was still running dual Y though, I would be more than interested. Since GRBL has no native way to deal with such a signal, I would probably hook up the two Y output signals in a logical OR circuit in order to cut down the common Enable signal (i.e. shutdown all motors) should something went wrong with either Y drive.

Or if I wanted something fancier, send them to the GPIO of the Raspberry Pi and in case something went wrong, shutdown the motors, play an anthem, send a tweet and pay the rent.

Which reminds me that at some point I need to research bCNC’s top killer (and least documented) feature: python scripting. Which should compare to gcode macros as a scooter compares to a Ferrari. (macros as parametric and conditional, not UGS snippet “macros”).