Hmm ... a magnetic brake / clutch certainly would take the 'load' off the stepper. That's getting into some more 'industrial' setup, no ?

Sorry for the 2nd part, it was late night and the thinking cap was not on ... nor any lights.

Someone needs to make a map of compatible/custom parts providers instead of just the 'OBPS distributor map' which was a bit harder to find then in the past (maybe just the late hour at work here too).

 

It's not dropping when unpowered, it is sliding under the weight and slight vibration of the router when it is powered. Just sitting there, all power off, the Z is nice and steady BUT not firm... Place a bit of pressure downward, and it will slide down as if sinking in quick sand (they say if you don't move you won't sink). It will not slide up (again, quick sand sucks you down not letting go ...)

Keeping the Z stepper powered prevents the 'slide', whether the router is on or off. However, the stepper is sweating ... like anyone of us would be if our arm was stretched out with a heavy weight in hand, we all would fail at some point (minutes for some, hours for others). The NEMA 17 might be good enough for a spindle. But it can't handle a 2+ HP router.

Maybe using some form of counterweight would allow the NEMA 17 to handle the router. For me, it was "easier" to upgrade to a NEMA 23 (capable of 269 oz/in at that).

As for the eStop, or even limit switches, it certainly would be nice to just 'freeze' the machine without any loss of position, maybe turning the spindle off (just to be safe).

Let the user figure out if he/she needs to abort or just tweak something to resolve issue.

Nicer still would be the ability to skip through code : restart from any given position without moving tool through the skipped code ... maybe even a 'step' mode to find or skip the trouble spot in the G-code while having the machine working (not a simulation, more of 'move tool up in air and then step forward or back through code without doing Z' until user finds a good spot to continue the milling/routing).

 

That's an idea ... could just align through holes on each layered pieces for placing dowel at stacking time. Just have to get an easy way to evenly slice the model within SketchUp to create each layer. There must be a plugin/add-on for that, eh ?

 

I guess that was way too obvious for me to find, eh ?

Thanks for the link. I will look into it shortly ... myOx is back in business !

 

YES ! myOx is now in business ...

1) Changed the little NEMA 17 (Z axis) for a NEMA 23 a bit earlier today and what a difference it makes ! Even with all power off the 'sliding' under the weight of the router is gone, bye bye. Before, the slightest pressure downward and router would slide down. Now, I need two hands and considerate pressure to get it to move. If you are going to use an heavy router or spindle, don't bother with a NEMA17 for your Z axis. Lesson learned at this end ...

2) finally added the fans to cool down the CNC xPro properly. After going to the NEMA 23 above (total of 4 now), just heatsinks on the driver chips was not enough. About an hour into a test job there was a small wandering of the router. Now 2+ hours into a test job and all is beautiful in myOx's world. It looks ready to work 24x7 !

3) tried the latest version of SketchUcam, along with a free web based raster to vector service, and that is working just fantastic ... for 2d and 2.5d jobs. I need to check why a complex curve has generated G-code doing 'needling'. It keeps going up and down on that one trace instead of following the curve like all the others done to date. A little detail which makes the job run for nearly ever and ever and ever. It is a good test of endurance, right?

I will have to clean up the quick install of the 2 fans - testing a small 80x80 as well as a bigger 140x140 ... In fact, I will need to finish the shields.

As I removed the NEMA 17 I had a bit of a servicing issue with myOx - I will need to redo the wiring and use some quick connects to make maintenance, upgrades and the likes much easier as myOx as now reach a stage where it can transform

First major change will probably be to replace the 750mm lengths of v-slots used for the Y axis to 1000mm or even more. At 750mm, I only get about 20" of work for the Y axis. I keep finding myself planning things requiring at least 24". X has 50", but Y is so small compared to it. So, yes, a 24" x 24" machine would of been a nice configuration to start with. I will need to figure something for the existing table ... I don't want to redo it. Maybe some brackets to hold the Y rails ?

Looks like I won't really care if it rains all summer did I just say that ?

 

Thanks for hint. I did scale the design before and, I am almost certain, a tad AFTER selecting the cut lines. So it is most likely as you say. More recently, I also rotated (90 degrees) the entire design, trace lines, tabs et al.

The problem is I did erase the cut line, using the PhlatBoyz Erase Tool - a few times in fact, then I did the cut line and it still generates tens of thousands of gcode segments (I think it's like 250,000 moves !) In fact, at this time, it still shows the orange trace BUT the gcode no longer generates anything for that (out)line. The other lines, the tabs, etc. are still there though and they have nice gcode. It's just that one 'fancy' line/trace being affected.

I guess I'll have to start over from the design before applying in SketchUcam tools ... I must have some extras in the file at this point... It won't take too long, now that I learned SketchUcam prefers a solid surface with NO cut outs. I need to let SketchUcam do the cutouts on its own ...

 

Not certain what you mean or how it applies to the sliding downward of the Z axis. However, the NEMA23 has fixed the 'sliding' issue, with and without power being applied, so I am all smiles now ...

 

I understand quickly, you just have to explain slowly

Got it ... my message had too many points, eh ? I'll have to come back to your suggestion and the need to 'freeze' instead of killing a run after an eStop/limit switch trigger.

For now, I have a hand close to power outlet. One quick pull and it definitely stops everything in an extreme emergency ... There's no coming back from that drastic a stop !

 

Upon further inspection, turns out the 'curve' I was trying to follow was actually thousands (really!) of tiny line segments itself. The vectorizing process traced the outline, but didn't create curves for all outlines ...

So I did a quick search, Sketchucation came through with TIG-Weld plug-in. If there was a plug-in to slice, we had to have one to weld. I'm starting to think web searches are like reading documentation : last resort option !

So I deleted one of the larger tiny line segments, did a select connected ... with one of the others. SketchUp promptly selected the rest of the contour. Did a TIG-Weld to them. Drew a line to replace the deleted one. Selected it and the already welded contour. Welded the two and ... tada ! ... got meself one nice CURVE.

But wait, SketchUcam still creates an outline in multiple segments AND no g-code for that curve ... I'll have to start from 'stracth' (before first SketchUcam step) just to make certain I'm working with a clean SKP. But me thinks I'm unto something here ...

 

Worked around the problem ... I did an outward offset of half the bit diameter of that complex curve, after a weld so it be a single curve. Then I did a 'cut on line' 90% deep (in 3/4" G1S plywood) and when I generated the g-code it was only about 8000 lines instead of the 200k+ earlier just for that curve ! A big improvement indeed... The only catch is I can't place tabs along those lines, thus the 90% deep cut. I'll have to finish with knife and file ... or maybe flip the board, as I have to engrave something there, and do a cleanup pass.

I guess as good as SketchUcam is, there is some slight of hand to be done in some cases ... I saw some video showing how to generate the g-code in sections and then join (and clean up) the files into a one file to some times get around a few issues, especially with 2.5d and some cases of 3d done/doable with SketchUcam.

 

It's been on my wish list since the very beginning of my build along with their controller (up to 9 axis), drivers, etc. At the time, PlanetCNC seemed expensive as I was not certain of making myOX work... I have never been known to be 'good' with mechanical stuff.

As I get more confortable with myOX, now that it seems all under control and properly working for hours non stop, a next generation myOX is coming back on my radar ... but the PlanetCNC remains pricey for my budget although its capabilities are becoming more and more interesting (and useful) to me.

 

Did a near final run of a sign for my niece's new business. It's about 20" x 30" cut in 3/4" plywood, with a bit of engraving but mostly cut right through thinking of doing a spinning element. The later proves a bit tricker - inserting long enough pins for rotation not so obvious given the various cutouts in the design ...

Job took about one hour to complete. I can live with that given the dimensions and cutouts done ... It certainly was way better than any of the previous tests where after 3 hours it was still far from half done - same design !! Better skills at SketchUcam is where I shrunk the time line.

I might need to make SketchUcam generate some features in separate Gcode files to later join as one to avoid some tricks that are proving not as convenient. The interlacing of cutouts doesn't seem to always work if done in one shot. I traced a few lines for these cutouts, but they don't allow for tabs (!) Cutting them to 90% or so deep makes for some difficult manual clean up ... For now, I'll just split these lines which will also get rid of the 'spinning' feature of the finish projects. So that works out for me.

The design calls for the engraving to be done on both front and back. I CNCed reference holes to use dowels as pins for perfect alignment when flipping the work piece. Problem seems myOX is too precise (or more likely the dowel not quite to spec). The holes are exactly the dowel's diameter - too tight of a fit for the need ! There's not even room for glue if I wanted to stack layers ...

Dummy me, wanting to shave a few minutes off the run time, I minimized the safe height. Turns out my work piece was slightly higher at the far Y. Also due to dummy me - I didn't properly clean my work table ! End result - some gouging into the surface. Man it's hard to learn by our mistakes ... Lesson learned: SketchUcam is very efficient with air time of the tool end, the spoiler board is also there for a reason and now I understand why most go beyond 100% when doing full depth cuts. Raise that safe Z height well above your work piece, you won't regret it.

... a few pictures and a movie soon to follow as I adjust the design, add a few elements and cut a new sign, both sides.

Now I'm looking for things to engrave / cut with myOX The wife is keeping her car out of the garage, so I can do lots of wood chips and dust.

 

The problem was that the work piece was more off level than the spoiler board due to some 'dirt' between the two. My bad

Never thought of the otherwise obvious : if spoiler surface is Z zero, it should keep longer, unless I do like some and purposely do inside/outside overcuts. I have seen a few in various forums suggest as much as 140% ! How thick is their spoiler ?!?

I forgot about the little detail of allowing some offset for fitting pieces together. After all, we're not working with spandex here. Also my bad

At least I'm learning, right ?

 

Right now my imagination is bringing me everywhere I don't need to be SketchUcam is not being consistant probably due to my fault. However, it is difficult to figure out what I'm doing wrong in SketchUp Make to cause SketchUcam to choke.

For example, one 0f the digits in the sign mentioned earlier was being routed over and over - just wasting time as it traced it 5 or more times before going to the next element. Especially when I had traced it exactly like the other 20 or so characters. Why was that digit/character causing SketchUcam to generate gcode in a loop ?? I had to zap and redo all SketchUcam elements - then the characters were all generating proper g-code. Just plain weird ...

Then, there is that complex curve. I had to TIG-WELD it for SkecthUcam to be able to generate efficient g-code instead of the pecking taking 6 hours when the job can be done in about 15 minutes. Now, when I split that curve into two, closed each half and TIG-WELD each half, why is SketchUcam simply refusing to generating efficient code forcing the multiple hour pecking ? The process worked for a larger and more complex curve (both half as one). Why not for the 'simpler' halves ? The difference seen is the halves contained nothing but scrap material being removed. As a whole it contailed several other elements ...

Problem is I can't let SketchUcam do an inside or outside cut, required to use tabs to keep the element attached to the rest of the design while being cut. I seem forced to do a trace cut which does not allow for tabs. This was reason I split into halves - manually did equivalent to a tab in spaces separating the halves ...

Grrrr.....