Building a Gear Shaper (Part 2)
ฝัง
- เผยแพร่เมื่อ 12 มิ.ย. 2024
- 00:00 - Quick recap
00:40 - Operating principle
02:55 - Cutting first test piece
03:38 - Results and improvements
04:32 - A better way?
07:00 - Custom controller
08:06 - Can I vary the stroke length?
08:40 - Cutting a steel internal gear
10:09 - Result and comparison
11:07 - Cutting a compound gear - วิทยาศาสตร์และเทคโนโลยี
"It's just an 8 bit controller, programmed in assembly language" I love how you make that sound so easy 😂
Heaps of PIC/Atmel/STM/etc.. hobby boards out there with mountains of documentation and tutorials. All you need are the fundamentals - you get a few basic examples going like making an LED flash, then perform an actions when a pushbutton is pressed, then displaying characters on an LCD, and you can basically do anything because it's more or less the same thing....
* Making an LED flash is the same as sending pulses to a stepper controller
* Pushbutton presses and LCD projects map to the interface he built.
That's right. Anything that looks complicated, is really just a whole lot of simple steps put together. Assembly language is really the most basic low-level type of programming you can do.
Yes I did some asm and it's completely feasible, I just meant that's some hours of work and you resumed it in two phrases as if it's nothing ! Good job man 👍
Assembly language is surprisingly simple. It's the ancillaries such as hardware interfacing that makes it difficult.
everyone is clowning suggesting learning assembly language. You can simply just use an arduino and the arduino sketch language which is high level easy with million tutorials takes no study at all
You programmed a microcontroller for several axes and added a GUI… in assembly! Andy, your machines and your videos are the cream of the crop on TH-cam. Thank you so much for sharing with us. I appreciate the multi part videos because it means I can more frequently learn from your adventures. Cheers! 🤘
This actually really wasn't that difficult, the screen was just a 4x20 LCD and has it's own character set, I've done more complicated things with a single microcontroller in the past, eg. my EDM machine. I'm glad you appreciate the videos!
Beautiful work. For a couple of years now, I've fantasized about projects requiring ring gears. You've shown us how it's done, so all I have to do is get to work.
Some more information on how your cutters are made could be very helpful.
Thank you Andy for sharing your excellent video, and description. You are very clever and kind for passing on your knowledge to others.
A fantastic project as ever. I love how it evolved and the thought processes behind each change.
You have got to be the most talented versatile person on the Internet. I enjoy watching your videos. How did you learn all these different disciplines? Computers, electrical engineering, programming, machining, mathematics and so on?
Thanks, but I'm really no expert at all of those things! I've learnt just enough from each to be able to do the things I want to do, mostly self taught apart from a couple of college courses. It used to be called 'a Jack of all trades' but I believe the modern term is 'agile'.
Whilst a little knowledge can sometimes be a dangerous thing, I often see projects built by people with far better (say) computing/mathematical skills than me, but then the whole thing is let down because they've used some crappy 3D printed part, whereas if they could just do a little welding or machining, even at a very basic level, it would have been much better.
These animations are incredible. Many thanks for your time and the information.
And it also looks like a pogo stick.
I've been thinking about doing EXACTLY this for a long time. I will DEFINITELY BE BUILDING THIS.
Thank you for reading my mind.
Make sure you check out part one as well, this has more information on the actual construction.
Thank you for your effort I truly enjoyed watching your skills at work please keep sharing and God Bless
I'm glad you know about backing off the cutter... I was told the gear shapers I used backed off 0.050". We also had the gears cut at least 2 passes (rotations of the blank), the second pass mostly being a "spring pass" that improved tolerances and surface finish.
Fantastic,. My only thought with the flexing of the rotary table is maybe don't cut at full depth for the first complete rotation. Take most of the material off then, do a secondary cut and final tooth depth. This will double the process time, but may give a cleaner finished product. Very excited to see this in action, as very envious of your achievements. Keep up your great work work and Hello from little New Zealand
Great project. We shared this video on our homemade tools forum this week 😎
Well done. It’s amazing the number of skills it takes to do these operations e.g. machinist, electrician, programmer, etc.
Congratulations! You are fantastic! Please make more videos. It is like the best school for me. Thank you.
Wow that is amazing. I can’t wait to see it used for real in another project.
Brilliant, Andy. Just brilliant. I am fascinated by your ingenuity.
Amazing work 👍
This is very clever, nicely done. I learned to run manual gear shapers a few years ago-Drummond maxicut and Lorenz. Purely mechanical, quite complex to set up but fascinating. The lorenz's were particularly good, dated from the mid 40's and still ran perfectly. They had a pull back mechanism for the return stroke, it is quite important to stop the cutter dulling. I would highly recommend adding the pull out in the y axis. Also means you don't need to worry about stopping the rotation. Perhaps you could get easier programming if you left the spindles turning constantly, and had a separate pull out system running independently? Sorry to be an internet know all, but maybe it'll be useful
Great show! Sure you will figure out the bugs. Thanks for the look and your time.
You are a genius man
Fascinating
Thanks for sharing
Cheers
Awesome AndysMachines, thanks for sharing!
Fascinating 👍👍😎👍👍 Thanks for sharing
Very impressive, Thank you.
Lee
bravo anzi bravissimo sia nella teoria e nella pratica e nelle spiegazioni.Ti seguo sempre nei tuoi video e cerco di produrre le tue apparecchiature grazie e un caloroso saluto.
Outstanding and inspiring.
It's just an 8bit uC programmed in assembly 😅
More like black wizardry
The Atmel instruction set makes it pretty trivial to do 16 or 32 bit math using 8 bit instructions.
Fantastic work, at a level I could only dream of. Very well explained in the way you described each step logically.
It would be incredibly interesting if you could do a video on how you built and programmed the controller.
Amazing work once again, definitely my favourite engineering channel 👍
Impressive, as always!! Thanks for sharing!
The man's a wizard!
As always Andy, very impressive 👍
Beautiful work.
Great work! Love the idea and your skills…
Wow great video
really very impressive!
So cool! Wow. Inspiring.
very interesting
I greatly appreciate your videos
Fantástico!!!! Como sempre você trás os melhores vídeos em
Mecânica
good job andy
Brilliant!
nice work
He's a wizard Harry
Programmed in Assembly Language 😅 Mad respect!
That is freaking awesome
Impressive! Wow!
I’m in awe. You have inspired me to try something similar. Maybe with a lathe? Replace the compound slide…
Yes, you could certainly do that. A lathe may even be better, there would be less play in the spindle bearings compared to a small rotary table.
Very well done indeed.
Excellent work!I agree on the stroke aswell. The longer stroke on the quil eventually leads to less rigidity anyways.Its common to see wear on gear shaper quil and a mill quil technicall not engineered to do what you do, needs to be taken care of. Im addited to this guys content.Thank you for such valuable working experiments!
10 lines of gcode is too much, let's do 2k lines of assembly instead 😂 You're our hero
2537 lines actually, but I only had to write them once, compared to writing 10 lines of gcode maybe 3 or 4 times so far!
@@AndysMachines what's your position towards sharing the code? Open source and all that. I understand that your machines are rather specialized but surely there's enough people willing to replicate that.
I'm not against open sourcing the code, but it is so specific to this machine that it would need major modifications to work on even anything slightly different, and I couldn't provide support for it. Plus it's still a work in progress. I might put it up on Patreon one day when it's a bit more polished for anybody who's interested.
You are a genius!!!
Fantastic!!!!!
Incrível! Obrigado por compartilhar seu conhecimento, muito bom mesmo!👍
love your content thanks
An nice to see you back again in Part2D2. :)
you are my master !!!
That's amazing ❤❤❤
impressive!
Master!!!
Grandios! 👍👍👍
Impressive..
I think that your assumption (at 10:34) that the profiles should match is incorrect. This would help to explain the tight fit of the gears.
However, this is an impressive project. Thanks for taking us along.
I think they should match throughout the working part of the tooth, since all gears are generated from the same basic rack profile*, but I'm open to anybody explaining to me why they shouldn't? The reason it was a tight fit was because I cut the outer (internal toothed) gear too small due to the deflection in the machine.
*I believe that sometimes internal gear tooth profile has to be modified to avoid interference if the gear driving it is close to the size of the internal gear, but since it is usually much smaller (eg. in a planetary reduction) it's not a problem.
My intro to gear cutting was on This Old Tony's channel, and boy have you taken it to a whole other dimension. Really glad YT suggested your vids to me. Different approaches are based on a needs must basis I'm guessing. Now that you have the gear shaper, are you likely to retire the hobbing approach or are there scenarios where that is more efficient?
I'll probably still use hobbing as it's quicker and more accurate than my shaper set-up. I built the shaper mainly for things the hobbing attachment can't do, such as internal gears.
Doesn't the linkage for the vertical stroke only have one length ?
An arm with multiple attachment points to change the vertical stroke length ?
Enjoying the build, btw.
really nice work! I wonder if you can also lock the rotary axiis during the cutting stroke, and the other axxis as well.
Thanks! I locked the X axis which I'm not using and the mill head to the column. The rotary table locks with two small clamps that unfortunately I can't manually lock and unlock 2.5 times a second! The real problem is that these rotary tables have plain bearings which are adjustable but always seem to have some slop in them.
Time to make one that is kept tensioned to the table :D
Folks this was not done by just any skilled individual - it had to be done by an extraterrestrial life form.
Thanks for the video. Why does the cutter need to rotate?
i like the gear videos; is it possible to make a bevel gear with the slitting saw method?
This is something I haven't tried but yes it should be possible, needs a bit more thought behind it though.
You can make the Gcode more generic by adding parameters. #50=outer ring and #51 = cutter tooth count #52=ratio #53 = radial move. So in this case you would add the four lines #50=31 (outer ring tooth count), #51=8 (cutter tooth count), #52=[#50/#51] (ratio) and then #53 = 9.25 (radial move). Put these lines before G0 Y0.0 . Then in the line where it has the comment (cut full circumference) you replace x1395 with x[360 * #52] and the Y move becomes G01 Y#54 F3.....
This saves you calculating the ratio each time and so is less error prone and makes it easier to copy the code and just make a simple modification to the two main parameters (three if you change the cutter) for different tooth counts. You can also add a pause into the program that asks if the tooth count is correct if wanted. This parameterisation is a useful technique for simple home written code that is required to do the same job but with slightly different numbers and where you don't want to work through the code trying to figure what needs changing each time it changes. You can also add feed rate parameters to suit different materials if required. Something you could consider is adding a parameter calculation that calculates the radial move that brings the cutter up to the workpiece for the first actual cut and put a G0 Y move in before the G01 Y move starts the cutting process. Probably not important for small tooth counts but maybe for large ones it would help and because it is a calculation done in the code it only needs to be worked out once. What about calculating the radial move in the program as well? You could come up with generic code for inside gears, outside gears, module or DP gears and different cutters. The only problem I have with the Gcode in Mach3 is that is does not allow the use of IF branching. What this means is that you need a generic program for inside module, then one for outside module, one for inside DP and one for outside DP. Still only 4 generic programs for all you need, unless you start doing helical cutting as well but that would need more synchronisation.
.
.
G71 G91.1
#50 = 31 (outer ring tooth count)
#51 = 8 (cutter tooth count)
#52 = [#50/#51]
#53 = 9.25 (cutter radial move)
G0 Y0.0
.
.
.
.
G01 Y#53 F3 (feed cutter outwards radially, chip load 0.02)
G01 A360.0 X[360*#52] F130 (cut full circumference)
.
.
Thanks for the detailed explanation, I didn't actually think of doing it that way. I was aware you can use parameters in gcode, but in the past when I've wanted to do something like this and generate gcode from a set of parameters I've written an excel spreadsheet to generate usually just the relevant lines of code that I can then cut and paste. Not that that's any better, but I suppose it gives more flexibility such as being able to include 'IF' statements and things like trigonometric functions (I don't even know if you can do that in Mach 3 or not).
@@AndysMachines No If function but I have used trig to cut elliptical curves.
That steel internal looked like you just need a new cutter and then a stack of bank deposit slips. ;)
Maybe a look inside the rotary table.
Yes, the cutter was a home-made one that I'd previously used in another project and not the sharpest. I also think it would work better if it had more then 8 teeth as the profile would be closer to that of the gear it's cutting.
That's a hell of a build! I was going to make some arguments for going to LinuxCNC, but I can't think of a kinematics style that would work with your setup. Maybe issuing a feed hold anytime the quill is not in its "safe" state. Did you write all the motion control yourself or are you using some libraries? For the ATMegas, AccelStepper is really handy.
Linux CNC can do it by "electronically gearing" the workpiece and the cutter either just using software position or integrate an encoder. The coordiantion of the up-down stroke can be done in lots of different ways. I guess Andy does what he knows because the Microcontroller for him is easier, because he already knows how to do it.
The slotted disc from the video could just be used for simple jogging of the workpiece by a fixed amount.
You make it look simple, while dropping casual bombs "it's just an 8 bit microcontroller written assembly N.B.D."
With something like this could you make a strain wave gearbox?
Yes, it could make the circular spine with internal teeth. That's something I may do with it one day...
👍👍👍Beautiful (electronic and mecanic)👍👍👍
On Atmega 328p, you use Arduino programing?
Could you use Fourth for us mere mortals?
He said assembler in the video.
Очень хорошо😃👍
From my point of view, gear shaping is the best application for flexible home gear production. Personally, I wouldn't make a external gear to gauge the tooth thickness . I would attach the desired balls to a telescopic/snap gauges to measure the OBD. For the correction of the OBD an adjustment block would be beneficial. (just my humble opnionion)
That would be the correct way to measure, unfortunately I don't have a snap gauge like that. It can be done with pins and an inside micrometer (some even have pin jaws). But both of those only work on a gear with an even number of teeth, this gear has 31.
@@AndysMachines For an odd number of teeth it is: Mdk = dk * cos(PI()/(2*z) )+/- Dm (+external). Teleskop Interne Gauge Sets are quite nice, i personally use the for all my internal measurements, good enough for an IT 7.
Would this machine work for bevelled gears, I think it would if you could mount the rotary table at an angle? Also helical gears would be possible as well I think but the code would be a bit more complicated for that i think
You can certainly cut bevel gears on a regular shaper, but I don't remember seeing bevel gears cut on a gear shaper, I'm sure it can be done though. Helical gears are definitely possible, the cutter has to be helical and it spirals down into the workpiece, I think it might be possible to make the advance of the cutter passive with a spring to return it to the start position, which would mean no actual modifications to the machine, just a purpose made tool. I might have to experiment with this.
Honestly I find this so intresting but I'm like a old black Barry phone or erlyer model trying to be a new IPhone my brain just isent smart enough .you sure are way to smart to play with gears your so intelligent I hope you make millions .I'm just amazed even never even working with macheanist wow .I'm like way back in iron age welding and using torches and plasma cutter macheanist are so dam smart
The cutter for internal and external gears is the same? Same tooth shape?
Yes, I used the same cutter. The geometry of internal gears is reversed (ie. addendum is on the inside) so that it will mesh with a regular gear inside it.
great !!!, when will you be taking orders for 65-T 32-DP 20-PA internal ring gears for nine-cylinder radial model engines ? :-) !!!
Are those in demand?
@@AndysMachines the best gear combination for a 9-cyl radial's cam is 20:32 -- 13:65, this allow a relatively large 20-T on the crankshaft (the small 13-T is on the compound planet where it doesn't matter) and the 65-T is mounted with the cam-ring. the other option is 16:32 -- 16:64 which means a much smaller (weaker) crankshaft. since imperial gears are generally only avail in even numbers, and metric gears are generally only avail in multiples of 5, most model engine builders go for the inferior 16:32--16:64 imperial solution. the Bruce Satra, Bob Roach P & W 9 follows full size and uses 20:32--13:65 and the plans call for broaching your own 65-T internal ring gear, and is the main reason why my 9-cyl radial is sitting unfinished on a book shelf while I work on easier(?!) things.
so cool to watch! wouldn't it be easiest to mount an encoder to the up-down linkage? then you'd know the position of the cutter at all times.
I must admit that I have no idea on cost for 'hobby' equipment nor any clue on the processing capabilities of such a microcontroller.
I guess the disc with a slot is a very basic encoder, with only a single index pulse. I could add more slots and another optical switch for more resolution, but I'd always need the single index slot as a reference for the timing.
@@AndysMachines You could always do what car motors do for their crank angle sensor, and instead of having a single slot for the reference you have a *missing* slot instead. This gives you the benefit of having multiple slots for resolution, and by running it through a few revolutions and waiting for the long gap you can find your start/high position. Then you can know when its coming by counting the pulses.
That's a really cool idea! That way you have resolution and indexing with only one channel! I didn't know that's how crank angle sensors work, thanks for sharing, I'll use that idea one day!
If you would mount the encoder to the drive shaft of the quill so the encoder 'sees' alternating clockwise and ccw motion instead of the continuous rotation you wouldn't need a reference point as the top and bottom dead centers provide reference points. But don't let my encoder comments detract from the admiration I have for your build:)
That's a good suggestion and would allow for more flexibility in the programming. Instead of setting the timing by manually positioning the disc I could just enter the 'safe height' into the controller. It's harder to mount an encoder there though, I would probably have to extend the pinion shaft out the other side of the machine and put it there. It would also have to be a quadrature encoder using 2 channels, and I'm running out of I/O on the microcontroller, I think I used all the pins, but might be able to swap something else.
If i was doing this, I'd need a black box to record the data for the catastrophic crash that would inevitably happen.. love your work but this is over my head
Does derilin plastic smell fishy when it's cut? I have an odd piece at work that seems the same that's a bit stinky when drilled or sawn.
It doesn't normally make any smell but if it gets too hot (blunt cutting tool) then it does produce a smell I suppose you could describe as fishy. Drilling or sawing might produce more friction and heat than other cutting methods, the 'fishy' smell could be formaldehyde which POM can release if burnt.
I mean, it was impressive, but any code you wrote for the controller to generate gcode you could have written on your connected PC to generate it too. Do you find going to the effort of making dedicated hardware control boxes like this worth it?
edit: I should say, "because I preferred it" is of course absolutely a very valid answer if that is the answer!
Yes, I've done things like this before by using Excel to generate Gcode from a few input variables, but then I still have to save or copy/paste the code and import it, and the PC connected to the machine is mounted to the wall and not very ergonomic to use. With this controller I literally enter the numbers with a few buttons and off it goes, plus I like it being all self-contained since most of the time this shaping head will be removed and the CNC mill reverted to it's original state. And yes, I often prefer to do things the hard way just to achieve a very marginal improvement.
Could you not store those feed rates in macro variables and call those in the middle of your code, then just change the macro variable to update the whole program at once?
Yes, you're probably right, but I'm not very proficient at writing macros and I think my self-contained solution is more elegant and much easier to set up for each gear. Probably a lot more work to build initially, but that tends to be the way I approach things.
@@AndysMachines Feels to me if you can write anything in assembly language, you could certainly do a macro, even I can write simple G code macros 😂
Ok ok ok great! I actually just asked this quesiton over at the Terminator reddit, could theoretically, the T600 or T800 actually work. Or was it more of them just placing things on a skeleton that looked like it would work. I guess we get to test that with your work!?! I was thinking one would want to test the theories on CGI first, as there might be major reworks needed, and or considering we have come a long way in most these technologies, a redesign would be in order to make things work more smoothly.
I think they actually did a pretty good job with the original T800, the upper half of it at least, the endoskeleton design makes it able to reproduce most of the movements of a human. Remember though that it was built as a movie prop/puppet and needed to be controlled remotely (no CGI). The legs however are not so well designed and with the layout of the actuators I doubt it would actually be able to walk or sit down/stand up. In the first movie the legs were only animated with stop-motion using a miniature model.
@@AndysMachines Upon further discussion on the subreddit, I think i was expecting too much of what it could do, Arnold was able to sell the robot like effects like thata if it had that limited range.
T800 was better, but eventuaally the went to T1000.
I learned about Ameca and realized whata I wasa thinking of was ultimately not possible with that endoskeleton but it really wasn't meant to.
That is pretty awesome that they accomplished what they did!
Therein, looking forwarad to seeing what all you share on your channel, T800 or not, you do really cool work and capture it well!
Thanks!
Seeing all the cutting oil leads me to think that this could also do gear EDM if you made the stroke (or something) current controlled...
Yes, it could be done that way (generated). But then with EDM you could make an electrode to sink the whole gear at once.
yet another machine "I need"....., 🙂need to fix the slop of my x-axis which causes similar flex on the rotary table.
do a"rough" cut first leaving .010 then the finish cut....
why assembly?
Is that windows xp, on the Mach 3 pc?
I mean, if it works, it works.
But, damn that's pretty old.
Yes, it is! And on a 15+ year old PC, it works just fine for this. If it ain't broke, don't fix it. I even used to have my CNC lathe running with Mach 2 on an even older PC. (It broke, so I had to fix it.)
@@AndysMachines Latency of new computers, eg. Ryzen, is terrible. I had to use a garbage board with an ancient CPU to get low enough latency to output stable parallel port signals for stepper motor control with LinuxCNC.
Not sure if later versions of windows support the parallel port required with Mach 3 - Good old XP - reliable and basic to use !
I'm actually using a USB stepper controller board on this PC as it doesn't have a parallel port. Though in the past I've done it directly from the parallel port (maybe with windows 95 even?) This PC only runs Mach 3 and nothing else, XP boots very quickly and works great for this.
To use Mach3 I have an old laptop running XP on my mill and a used cheap touchscreen laptop (Ebay) running W10 on my lathe. Both run through USB controller cards. The lathe had an old XP laptop for years but the keyboard died and so was replaced. Neither laptop connects to the Internet and are used as dedicated controllers and nothing else. Call me cheap if you want, I will take it as a compliment 😉
I wish I was a fraction of this intelligent...
Mr. Pete (TubalCain) latest build is an Atlas Lathe gear cutting attachment. It's less sophisticated and works entirely on a lathe. It might inspire your design.
th-cam.com/video/ZXdk_MSTp04/w-d-xo.html
While Tubal Cain is undoubtedly one of the best machining teachers there is (on TH-cam and in print). I don't really like this method of making gears on a lathe. A better way is to hold the blank in the lathe spindle and use a cutter on a powered tool post spindle (the way Clickspring does it). Though both methods usually rely on having another gear with the same number (or multiple of) teeth to do the indexing, though it doesn't have to be the same size. Or if you don't have another gear you can make a disc with slots around the edge, but as Mr Pete says, if you can do that then you probably already have a mill and dividing head, so you might as well use those to make the gear.
If you add electronic indexing and a spindle lock to the lathe-spindle method then it becomes quite a fast and reliable way to make gears with a lathe.
@@AndysMachines I like the simplicity and lower part count of the lathe design, but its clearly incapable of making more complex gears. As an alternative to slotting around a disc's edge, drill holes around the circumference and use pin for indexing. The drill hole locations can be easily created using a cad program and then printing a paper template and gluing it to a disc blank. Given the cost of a stepper motor and the low torque requirements-- just enough to turn the blank, it's probably worth the little bit of effort to incorporate an electric motor. I would add a automatic spindle lock to reduce the load on the motor's holding torque.
👏👏👏👏👏👏👏😍