I'm currently deciding whether to do a design in #OpenSCAD or #FreeCAD or maybe some of both?
I may convert an old 12-chord autoharp from my wife's family into a 21-chord unit by making narrow chord bars. If I do that, I'd like to try #3DPrinting the chord bars with integral TPU dampers, programmatically generated.
Button rank: Which of three rows should the button go on?
Notes: Which notes are in the chord?
Strings: Note names of every string, along with their relative station.
Name: Chord name printed into the bar.
Then I could remove from the model sections of the TPU damper part of the bar for every string corresponding with a note named in the list of notes for a particular damper, so that it doesn't damp the notes in that cord.
This feels programmatic, and OpenSCAD would be a rational choice. I know that one can program macros for FreeCAD, but it's harder to make FreeCAD macros part of a project instead of part of an installation as far as I know. On the other hand, I'd love to generate STEP files that express more semantics, rather than STL files, which argues again for FreeCAD. Maybe I could find a way to drive this from a spreadsheet in FreeCAD instead of with code. 🤔
Or maybe we should stick to the 12-chord version for now until we decide this is too limiting. 😀
@smellsofbikes Six of the chord bar return springs were missing. I happened to have the right gauge of spring wire, so my #ElectronicLeadScrew came in handy for winding new screws with the right form. I filed a nick in a piece of scrap in a tool holder to guide the wire, made one throwaway spring with a mandrel at the final ID, turned the mandrel down by the difference, then wound the springs with a couple tight turns, then about 17mm of the main pitch, then finished with the fine pitch again. Much easier than trying to switch the gearbox mid-turn. Wound the springs turning the spindle by hand. The ELS meant that turning the spindle was easier than if I were driving the whole transmission and carriage along with the spindle by hand.
But then, when I characterized my own lathe's gearbox and tried to use that for my #electronicLeadScrew, I spent over an hour of actual debugging time (plus head-scratching away from the lathe and resting and coming back to the problem) before I realized that a simple math error happened to lead to what looked like a complete misunderstanding of the gears, through sheer coincidence.
When I bought my SK-Tank #3DPrinter a couple years ago, I hadn't learned how to use shape binders in #FreeCAD to design parts around a collection of other parts in particular positions. I didn't like the hotend mount SecKit provided, so I designed my own, but it was hard to make changes to my models.
Of that set, the nozzles I printed in PETG are starting to melt, and before the last time I touched them up had actually caught blobs and caused layer shifting. Didn't happen much, so it took me a while to figure out what was happening.
Today I took my old models, started over from scratch on each body, but used shape binders to the STEP files for the hardware, and between the parts, to design similar but more stable parts, taking advantage of what I've learned in the meantime, like moving the part cooling fan slightly for better airflow, a larger nozzle aperture, . That was substantially faster, and was easier to tweak the models in the middle as I worked through the design.
I left the volcano block in the model even though I changed to a CHC heater block; it is strictly smaller and so will fit, but this leaves my options open if I hate the CHC heater block.
I didn't bother creating an assembly and adding hardware to my design; it's just a bunch of M3 screws and inserts.
I'm now printing the set in ABS on my Qidi X-Max 3 at least to test clearances, and when my glass-filled fiberglass arrives, I might re-print it for additional stiffness after I'm confident in the quality of the design on cheaper normal ABS.
Almost every screw will be fixed with heat-set inserts, which means that I also need to make the pinecil-heatset driver adapter I keep forgetting to get around to making, now that my lathe is back in operational status.☺
I decided that since I use my pinecil now for soldering (because I can use USB-C power), I should make the press fit insert adapter for the TS-100 instead.
60mm of 5/8" (~8mm) brass rod, turned down to 2.5mm to fit inside an M3 thread for a few mm, then turned to a 4mm shoulder. On the inside, drilled out to 5.5mm (5.55mm shaft on TS-100), then slotted with a 1/64" slitting saw so that it is a clamp fit. It holds tight, but if I push a putty knife into the slot on one side, it slides right off. Just right.
I have the parts around the house to make one of those jigs that holds the soldering iron vertical, but I'll try without that first. I think I can probably insert them straight enough without that.🤞
The weird surface finish on the 4mm thick part is because I was playing with my #electronicLeadScrew adjusting the feed rate up and down while cutting. I love that feature!
Back to projects now that some kids are settled in college, and today I finished my #electronicLeadScrew project. A few weeks ago my wife pointed out the arithmetic error that led to wrong pitch. Got that fixed, and now threading is accurate within the precision of my instrument (within .001" over 1").
This all started when I wanted to thread a ridiculously small screw to a shoulder. Today, I cut a thread to a shoulder easily. I won't even need to cut a thread relief any more! The thread tool will cut the relief naturally when the tool reaches the shoulder.
Also, the ability to adjust feed rate on the fly is 🧑🍳💋 — I can watch the finish on one pass and adjust it until the finish is what I want on the next pass, and then nail it on the second pass.
I'll never have to change the change gears again. Unless this one breaks, I guess. Or I'm using the mini lathe. Hmm, maybe I need to do another one of these on the mini lathe! 😀
I thought that I was basically wrapping up the #ElectronicLeadScrew project tonight, but the servo goes into overload while the lathe isn't even turning.
It works fine, I go off to look things up on the computer and check calculations, and a while later the servo is just flashing 4 red blinks in its overload code. This makes no sense. But I'm sure not trusting it to cut a thread until I have reason to expect it not to suddenly overload. ☹
OK, if I turn the power on but don't turn the lathe spindle, the #electronicLeadScrew servo doesn't go into overload. So it must be that I was driving it too hard and it just took a while for the heat to trip a sensor, so it just didn't trip while it was actually running.
I guess I'll switch gears to a lower gear to give it more mechanical advantage.
Re-calculating to put it (back!) into the lower gear (that I started with), I found on a sheet of paper that I had referenced an arithmetic error that happened by chance to exactly cancel out moving it into the higher gear (a factor of 2.5). So now at least some reason to think that I can get it working right in the lower gear by math instead of magic.
I have a servo:output ratio of 275:24. Nice round numbers those! 🙄
It's a good thing I'm using a servo, not a stepper, here. In order to get an integer number of steps per turn of the lead screw, I have to search for a number of servo input steps/rev that gives me an integer number of steps per rotation that is 2000 or less, so that the control board can keep up sending steps. (I check for "close to an integer" here because I might end up using a different even lower gear, and some of those configurations have no actual integer solutions, so I allow myself 10 parts per thousand error in this code.) Here's my quick one-liner:
$ python -c 'print(str([z for z in [(i, i*275/24) for i in range(100, 500)] if z[1]%1 < .01 and z[1] < 2001]))'<br></br>[(120, 1375.0), (144, 1650.0), (168, 1925.0)]<br></br>
I can program my servo to take 168 steps/rev and it watches the rate and interpolates for smooth movement.
If I were using a stepper I would have to approximate or completely change how I'm driving my #electronicLeadScrew physically.
My lathe supports metric and imperial module pitches. These are, as far as I know, only useful for cutting worms.
I'm trying to decide between implementing module pitches for the #electronicLeadScrew or just putting the change gears back on if I ever decide to cut a worm...
@chaseadam helped me diagnose why my #ElectronicLeadScrew wasn't working. The terminal I was using for the contactor to energize the servo's DC power supply has power when the lathe is energized but not while it is running.
Silly me, I'd originally intended to use the net numbered 10 to energize that contactor, then thought I'd made a mistake and changed to the net numbered 4. But that drops when I actually engage the motor. Turns out 10 was right in the first place. It really helped to have someone behind the lathe poking with a multimeter while I engaged the controls at the front!
Now I just need to work out my new ratio and reverse the direction now that I'm feeding through a gear, and I'll be ready to test.
As I'm putting my lathe's quick change gearbox back together to start getting the #electronicLeadScrew going, I realized that I don't have a proper bent funnel to get the oil in. The fill port is on the side (the top isn't accessible). But I have a #3DPrinter#FreeCAD and #PrusaSlicer handy.
The angled top is intended to let me print it upside down without supports. The angled bottom faces help the oil go where it is supposed to.
We'll see how that goes.
Hopefully in a bit, some lovely fresh ISO68 oil will fill the sump in the gearbox and I'll be back in business. 🤞
I think I have finally worked out all the exact arrangements of every setting for the gearbox on my lathe.
It has four generally independent settings: A/B/C, R/S/T, X/Y/Z, and 1-8. Not every combination is meaningful. Some gears are driving gears in some configurations and driven in other configurations! There are a total of 27 gears if you count the three compound gears as two gears (I do; they set up different ratios).
I'm going to write it all up, with pictures; the irony is that assuming my #electronicLeadScrew works I'll use exactly one of those settings for everything from now on. But it somehow seemed a waste to have the gearbox open and not work it all out.
I think this documentation is actually complete. I spent some quality time sitting in front of my G0709 lathe with pencil and paper, sliding things around, taking pictures, checking the parts diagrams in the manual, and writing things down. I'm happy that I got to what I think is a completely correct characterization without actually having to disassemble anything further. 🎉
Quite ironic that I won't actually use much of this information with my #electronicLeadScrew but hopefully it will help someone else someday. In any case it's intellectually satisfying to have worked it all out.
It's a good thing that I opened my quick change gear box to clean it while getting the #electronicLeadScrew in place. The bottom was full of filth which stuck to magnets. That's cleaned out and magnets in place as chip catchers for the future.
But also now I have worked out how the gear train works. Not all settings drive both the lead screw and the feed bar, but the gear train in the ATW1 setting is 20:50 19:38 24:33 to the feed bar, and 36:35 from the feed bar to the lead screw. This looks like the highest reduction ratio I can manage with any gear combination in the box. I don't think I could have understood this without taking the box apart.
Time to button this up. The new gasket should arrive Saturday.
The only question is whether I want to document the exact gear ratios for all the gearbox settings before I close it up. I probably won't use them, but someone might, and I don't really want to pull it apart again if I can help it.
After discovering that my gearbox is too stiff to run my #ElectronicLeadScrew servo directly coupled, I opened the gearbox to clean it, and ponder replacing it with an ELS box instead of feeding through it, and spent some time thinking this through.
I have a loop of HTD-5M timing belt, a 30 tooth pulley, and a 10 tooth pulley, which would give me 3:1. I'd have to bore out the 30-tooth pulley and cut a keyway to fit the shaft, though. And I'm not sure that 3:1 is enough mechanical advantage.
But then I realized that I have a 66-tooth gear from the change gear set with the keyway cut, and if I had an 11-tooth gear to drive it, that would give me 6:1.
The 66-tooth gear measures 102mm across, which is 1.5mm module. (One module is added to the radius past the pitch diameter to the tips of the teeth, so measuring the diameter in mm and dividing by two more than the tooth count gives the module.)
#FreeCAD has a great involute gear packages, so I started by #3DPrinting an 11-tooth 1.5mm module gear for the servo.
I have also ordered a set of 1.5mm module 20°PA cutters. It's about time I start putting that BS-1's index plates to good use. 11 divisions is 3 21/33 turns per div. Fortunately I have an index plate with 33 holes.
Maybe I'll start with delrin before advancing to aluminum or steel. I've never cut gears before, and I'd like to start with something with which I'm less likely to come to grief with one wrong turn of a handle.
The servo for my #ElectronicLeadScrew is mounted again, this time with a 6:1 gear reduction.
The 66-tooth gear needs to be held in place by some sort of spacer. The most convenient spacer was another gear. I measured with the 42-tooth gear and made the servo mount a fork. Then I realized I had a 33-tooth gear and didn't actually need to do that. It will be fine.
I ended up making it more compact by mounting the gear in the other direction, with the teeth toward the motor.
I'm glad I made 15mm of teeth on that printed gear; it gave me some wiggle room installing this.
I got the servo to quit singing when turned on by setting the secondary notch filter to 250Hz, which seemed close to the frequency I heard. So far so good. Bug...
When I power everything up and turn the spindle, the servo starts to turn, and then trips its alarm and stops. It does this even if it is not physically connected to the lathe gearbox, so it doesn't seem like it could reasonably be complaining about it taking too much torque to turn.
I guess I should go back to factory default settings and try again, but as far as I can tell I didn't change much. Steps/rev and the notch filter is I think all I changed.
As far as I can tell, the servo doesn't say anything more than "an error occurred" by lighting a red LED and asserting a signal, so I have no idea what error it thinks it ran into.
When I fired up my #ElectronicLeadScrew yesterday, I noticed that the RPM it displayed was a bit higher than what was written on my lathe. I confirmed with an optical tachometer; it just runs a little fast.
I suppose someday I'll convert to a three-phase motor and a VFD and get both adjustable speeds (still using the gearbox for torque) and better surface finish (maybe). But not today.
I installed it and powered up the lathe, and heard a humming noise.
"Ugh, something is rubbing on the fan blades."
Nope.
"Oh no, sounds more like it's coming from the original electrical box for the lathe. What on earth did I do wrong in there?"
Nope.
It was the servo singing, and it stopped singing when I turned the spindle. It needs some tuning now. But to tune it, I need to figure out gear ratios such that I can turn both the lead screw and the feed bar at the same time. And I really need to do that before I tune the servo, I think.
I've got a lot going on right now, so #ElectronicLeadScrew progress will be sporadic over the next few weeks, I expect.
I made a cable for the console when I was first testing. I used an ethernet wire. I bonded all four striped wires into a ground, and the four solid colors were the four remaining contacts. It worked.
Then when I put it in a cabinet, I cut that cable, connected each conductor to one of the 8 pins in a male "aviation" plug on each end, and then made a female-female cable in between using the same pin numbers for the colors that I used at the ends.
Somewhere in that set of three cables, STB got shorted to GND, but only when I connect them all together. Each individual section tests fine. 🤔
I really didn't like using the slide-on dupont connectors for this anyway, so I ordered some 5-pin 2.54mm screw terminal blocks which I hope will fit in the project box, and I'm just going to run ethernet wire through glands instead. One simple cable. Back to what worked, but less likely to vibrate loose while the lathe is going. I hope.
Tonight, I got the #ElectronicLeadScrew cabinet screwed to the back of the lathe, and the supply wires connected to the primary electrical box.
The power sequencing works. When I turn on the lathe e-stop, it powers up the control board and the fan; when I arm the lathe to run, the contactor powers up the power supplies for the servo and the (future) toolpost grinder spindle.
The lead screw console lights up but doesn't display, so I think I have an electrical problem somewhere in between. Probably a bad connection. If I'd been smart I would have buzzed it all out on the bench, but I got impatient.
Power to the servo means that I can try auto-tuning servo parameters with it driving the carriage, regardless of when I fix the console.
The easiest solution for the console is probably to temporarily give up on the aviation connectors and just make a new cable and run it through glands without disconnects. That would be 32 solder joints fewer to fail. Then I can put the cable set with the aviation connectors on the bench and figure out what is wrong with it when I feel like it. (Never?)
Anyway, feels like I'm getting close. Might be next week before I have enough time to finish it, but despite the console problem I'm feeling OK with where this is now.
Very little time tonight to work on the #ElectronicLeadScrew. Just wired up the 4-pin aviation connector for the toolpost grinder motor. Right now using only two conductors for DC, but this lets me upgrade to 3-phase ESC/VFD later if I want to. Also modeled spacers to hold the M6 mounting screws captive to make installation easy and avoid scratching the lathe with the mostly-flush screws.
Spent a little bit of time on the #ElectronicLeadScrew tonight. I got various sizes of label-printer heat shrink tubing today and WHERE HAS THIS BEEN ALL MY LIFE‽ and now I'm seriously pondering disassembling three dozen connections to add more labels to wires in the box.
I created a toolpost grinder control cable with an aviation plug on one end, and inside wired the matching plug up to the associated power supply. With heat-shrinked labels on both ends of every wire. I just need to add a few more wires for the toolpost grinder motor (I'll use a 4-pin aviation plug even though I need only two for the DC motor I'll be using now). I don't know when I'll get around to the physical build for the toolpost grinder, but at least when I do I'll have the power ready. I don't really want the TPG project to get in the way of the ELS project but the rubber blanks in place of aviation plugs ate at me.
While #KiCAD is my tool of choice for schematics, it's probably the wrong tool for the schematic that I'll put on the inside of the ELS cabinet. I seem to remember #Dia having some basic electrical symbols, and that might be the right tool for the job... Yeah, I see it has several sheets of the right symbols for this. Any other tools I should consider for simplified schematics?
The #ElectronicLeadScrew cabinet is fully wired up inside except for adding wires for the future toolpost grinder. Right now I have rubber plugs blocking those holes, but I might install aviation plugs and wire them up before installing it on the lathe, since it will be rather less convenient later.
I can't wire it up to the primary electronics box on the lathe until I physically install the lead screw electronics enclosure on the lathe, because the cable connecting them just goes through glands, and is not terminated at an external connection. That will let me get the length right.
There was barely enough room in the control panel box to install an aviation plug, which I realized after putting together the male ends on a cord to attach to it. But it worked.
I'm powering my new #ElectronicLeadScrew from the lathe. When I engage the e-stop, the control board will be powered on, and when I arm the lathe, the power supply for the servo (and, ultimately the toolpost grinder) will be energized through a contactor. Fortunately the lathe came with a clear electrical diagram, so it's pretty easy to see how to do that.
I had to unscrew the electronics box from the lathe to drill the hole for the power connections from the electronics box to my new project box. In doing that, I discovered that the power cord for the lathe went directly through a hole without a gland, though it did have some strain relief. It was a thicker cord than the original from the manufacturer, so I had to drill its hole out bigger as well in order to accommodate a gland big enough to install it correctly. Fortunately the assortment I bought for the smaller glands also had one large enough for the purpose.
I also drilled all (I hope!) the holes in the project box, and can finally install all the parts and start wiring them together.
Last night, I drilled and tapped holes in the #ElectronicLeadScrew enclosure for mounting din rail, cable raceway, and the toolpost grinder power supply. But I want the screws to be flush with the back surface (because I'm going to mount it flush), so today I'm #3DPrinting custom spacers for the M3x5 screws so that they will fully engage with the threads I tapped in the enclosure, but not protrude out the back.
I still haven't decided how to run the wires for the control box, encoder, and servo, so that's probably another four holes to cut, since they are very unlikely to be using the knockouts.
Then I have to decide how to cut the holes for the fan and the air outlet. James Clough used an angle grinder. I'd rather not, but also I don't have a drill press, and the case is too tall to use the mill as a drill press for at least the bottom hole for the fan. I might get away with it for the outlet hole in the door though. 🤔