Awesome. I've always wondered what chatter looks like. It would be cool to see a macro shot of any of the shop tools that move really fast like an air nibbler or something. I also wondered what it looks like when you accidentally run a lathe spindle in reverse (or mill). Not that I would have ever done anything like that, I'm purely asking for a friend that removed a substantial amount of material wondering why the knobs were so hard to turn...
Yay thanks for stopping by! So funnily enough, I think my first milling video on YT I was running the endmill in reverse and really struggling to turn the handles, not understanding what was happening. Doh :) I just might have to investigate that, yunno, for science.
@@BreakingTaps I'm pretty much convinced that all the channels I like also like watching the same channels I like... P.S. I feel like this totally changes my perspective on chatter..... had no clue a rigid looking set up like that would oscillate so much! I mean I could see the endmill and or tool holder and entire head of the machine deflecting that much during chatter but had no idea the work and entire machine table and vise would... crazy
Heck yeah! BTW, Steve Mould did an interesting video on video motion enhancement algorithms that modify the contrast of videos to accentuate motion using lower speed cameras. You guys could experiment with that for some good high speed footage.
Such an underrated channel. Pushing home-manufacturing to a new level with every video. Your editing has improved massively as well over time. Best of luck on your TH-cam journey
Also appreciate the note about editing... it's so hard! I'm humbled everytime I watch a really good video with editing. Slowly getting better, trying to spend more time on it since it makes such a big difference imo.
As an actual CNC machinst, I can tell you that even the biggest and heaviest machines move around like this. Not as much, and not as fast, but your regular Haas mill is a limp noodle compared to dedicated heavy iron for HSM production machining. "Everything is a spring" is indeed correct, and everything that has clearance, has slop to move around. The second thing I want to mention is that your best stepover for "less than ridged enough" tooling is usually 33% and 66%, where the cutting forces want to pull the cutter parallel with the direction of feed, instead of into or away from the work. This tends to give you a more perpendicular cut, because the tool isn't flexing to one side, and can give you a more stable cut because the direction of the cutting forces is countered by the feeding forces of the machine and should have less backlash to move around in.
🙂 Figured if the goal was to make it chatter.... might as well really give it the juice and chatter :) To be fair, the WOC is still pretty light which makes the IPM look a little less ridiculous... still very high though :)
Wow, I always cringe when I hear chatter and in my head I'm just imagining that super hardened metal against metal, but actually seeing what it's doing is crazy. Really great analysis, too. Haven't seen this channel before but I'm glad I stumbled across it.
This is one of the most unique and informative videos I have ever seen on machining. Clearly, the average machinist does not have the equipment nor ability to do this depth of analysis. Thank you very much for sharing it.
Still watching this video 3 years later. There's a wealth of knowledge to gain from it, both about your machine and the effects it has on machines in general. Once again thank you.
That is amazing. I’ve made my fair share of mistakes and have seen the movement of the machine like that at the end. It’s pretty cool seeing it in slo mo
This is the stuff of universities and institutes. Very educational! I think that the factor , which additionally plays a big role, but is hard to estimate for yourself is the effect of temperature. It isn't constant throughout, but we all know how it could worsen a metals property. Great video!
I would LOVE to see this done with all different types of tool holders to see the difference between them. Do ER collet, hydraulic holder, slim fit, shrink fit, rego fit, etc.
"what NOT to do with an endmill", great high-speed footage! I assumed chatter was just bouncing *off* the part, creating just a lateral vibration - it never occurred to me that it would be a circular oscillation. But considering the bit is spinning, I suppose I should have expected that. Thanks for the vid!
I also wanted to ask, when your spindle bogged down and the VFD eventually shut off, if you had a more powerful spindle (5kw), that cut would of then been possible, no?
Thanks! I think a more powerful spindle might have done the trick, although I suspect there might be a configuration issue with my VFD too. Someone else in the comments ran the numbers and the spindle should have been pulling under the 2.2kW limit... so there might be a threshold on my VFD that isn't configured correctly, or maybe I tripped the thermal cutout. Not sure :)
This is just some crazy video analysis. TH-cam's algorithm sent me your way somehow and I am so glad they did. I've always been looking for someone who knew machining AND All this slow mo videography stuff. The way you can measure certain things like how much the spindle moved is extremely helpful. Subbed.
In a "proper" cnc machine all you do is shift the failure point around to different places. Care and maintenance of tool holders and spindle taper. Pull studs. Draw bar springs. Collets or whatever tool holding. Tool balance/runout. Thrust bearing maintenance. Work holding. That's all right off the top of my head. Good cut parameters are exactly as important. The ragged edge of max productivity is every bit as razor thin, the bar is just set a tiny bit higher.
Absolutely fantastic work. Your play by play is super useful. Regarding cut depth... insight + (slowmo) sight is a 10 outta 10! More of this kind of skoolin for desktop amateurs PLEASE!
BT, very nice. I've been running machines since the1970's and this is the first time I've ever seen this. Thanks for the video and yes I will show this around.
Chatter is your depth of cut, machine stability, material being cut, material and profile of the cutter speed lubrication etc... that produces a harmonic frequency that causes chatter and there are multiple orders of harmonic and sub harmonic frequencys that combine and cause mayhem. What I do is finish the profile with a reverse pass (2 to 5 thou finishing cut) i.e. your not cutting, your spinning the cutter in the oppposite to its cutting direction into the job slowly.. outcome a nice shiny smooth finish and yes i am a trained eng tech Ps. Never used carbide cutters on the mill always HSS and never broke any as i would get my marbles booted lol if i did. Cutters are a capital outlay when blunt into the tool and cutter machine and re sharpen.
Speaking from experience after working for a top carbide and tooling manufacturer, I would suggest you revisit how you hold your carbide cutters. ER collets, if that is what you are using are not very good. I suggest a side lock holder or the more expensive shrink fit. For a cheap experimental option I would try the side lock type. Cheers Ian langley ex Seco Tools New Zealand
So I used to work in industrial automation, and one time we were building a machine to remove coating from pipes in a specific way. The project was a pain in the ass and was taking much, much longer than initially estimated but we were making progress. We had a medium-strength steel blade as a kind of scraper to remove the coating at a high RPM - the whole system rotated around the pipe as fast as possible - and the blades were getting stuck and sometimes breaking. The boss came in and told us to use carbide blades. The main mechanical engineer said "no, that's not the issue, carbide will just break MORE easily". Nope nope. Boss says use carbide. Spent a week reworking it to get the other blades in there. First run, blades just shatter. Get out the spares and slow it down. Blades shatter. Do this a few more times after waiting for more carbide blades to get made up. They shatter. Boss finally goes "huh maybe carbide is too brittle." YOU DON'T SAY!!
Great work Sir, very high quality production. I really enjoy your content. You're so advanced compared to my machining that I don't know what to recommend next. Maybe content like this with speeds and feeds using high speed footage, and showing other machining failures would be helpful. Maybe slotting in aluminum and gumming up the cutter, that would be interesting for noobs like myself.
First off, this is really great content! Subscribed. This is a lot of work, but I'd suggest putting a bunch of dots and recording the video. Then do motion tracking on the dots and then plotting the various motions. I don't know of a reasonable way to do this, but someone must. I've done stuff like this with OpenCV, but it was a big effort. Even if you don't get to tracking, at least take the video and maybe someone else can figure it out. It would be amazing to take the video, then see a motion capture that converts that video to a wireframe using the dots' real positions.
Oh my goodness, this is an AWESOME idea, thanks! I'm definitely going to research how to make this possible (and/or who to recruit to collab with if it's outside my skills). I was considering adding some dots to do rudimentary analysis of the table and frame since I was curious how much they were deflecting...but didn't think about going full motion-capture on it and build out a 3D model. Genius! I wonder how much 3D data can be extracted from a single perspective? Maybe do photogrammetry first to build up a 3D model, then the high speed data provides the motion? I dunno... exciting! So much to start reading :)
@@BreakingTaps Don't know if you have a programming background (still working my way through your videos), but here is a link to a site that got me started on OpenCV www.pyimagesearch.com/2015/09/21/opencv-track-object-movement/ In particular, this link shows tracking a dot. So long as your dot is significantly bigger than a pixel (consistent lighting, etc.), getting sub pixel resolution is totally doable (1/10 pixel is doable, IIRC). If you can find off the shelf SW, all the better. Combine this with a cross-colab with another bigger TH-cam machining channel (your content, if you pull this off, will make you the star of the episode), and I would bet you would get a huge jump in your subscriber count / dopamine fix.
Perfect, thanks! I do indeed have a programming background... not the best at Python but can muddle through. Will take a look and see if I can get something working over the weekend! I didn't realize OpenCV could track so robustly, sub-pixel resolution is super cool.
Kind of makes sense, endmill engagement slows down spindle, when you hit 0 or low number spindle can speed up and bounce when endmill goes back into material.
great images. Had a Fadal at my last job that no one knew how to use so I learned how so I am familiar with feed and depth parameters. I also did some high speed vid of a fluids so I know how hard it is to get good footage. What you did is really good and informative to watch. The oscillation of the material took my by surprise.
When you have a deeper or wider cut you are putting more force and vibration energy into the machine's structure. Double the width, double the force, quadruple the energy. These machines don't have much that will absorb that energy. It would be very interesting to see a super-slow-motion shot of a machine with tracking dots, which could tell us what part of the machine is doing most of the deflection. * I didn't catch which mill you are using, but most of them use cast iron for a fair bit of the framework. Cast iron has about 23x the internal friction of steel (iopscience.iop.org/article/10.1088/1757-899X/147/1/012031/pdf). That said, the mills tend to use thick cast parts which are very rigid. Unless the cast iron is doing the deflection, it can't absorb the energy. * I suspect that most of the vibrational energy usually gets absorbed by the work piece, where the cutter is peeling away a chip. This is the place where metal is deflecting the most, so has the most potential to absorb energy. I believe this is why shallow cuts can sometimes chatter worse -- the kick coupled into the machine from each cut is relatively larger compared to the energy dumped by peeling away the chip. It would be interesting to find the resonant frequencies of the mill, and then see if the chips that are cut at different points in the resonance show more and less plastic distortion.
It could be interesting to see how the endmill moves when only one of the flutes is engaged at a time, ie. cutting deep with a single flute endmill with a relieved back side of the cutting edge, compared to ie. a 4+ flute one in the same situation. I've been taught that single flutes are really good only for very shallow depth and large radial engagement. I'd wager the deflection would be smaller on these high-speed spindles compared to a low-RPM mill, but it could be still fun.
Ooh, I like that idea. I have a number of single flutes on hand. Would be neat to see how it compares to to 2/3/4 flute in a few different cutting scenarios. I bet their chip evacuation is pretty awesome to see too, since the gullet is so large.
@@BreakingTaps Oh, one more thing - not sure how usable it is for this use case, but video motion magnification (see ie. people.csail.mit.edu/mrub/vidmag/ for the academic side, software included) could be interesting too and potentially reveal any smaller amplitude harmonics in the vibrations. Its commercial use is more in monitoring tiny concrete / pipe vibrations for early failure, but it's worth a shot if you get it working. :)
Depends on weather you are climb milling or conventional milling also it depends on path direction. Tool length. Holders collets. And also speeds and feeds
I've seen footage from regular cameras of industrial machinery that is run through software which then amplifies tiny movement and it is amazing to see. Apparently it's used to look for potential wear or weak points. Would love to compare this rather flexible machine with something like Piotr Fox Wysocki's epoxy granite machine here on youtube. His finishes are magical, made even more so by the fact that it's a home built machine.
I've only experienced chatter on a regular column drill and I've gotten the feel for how much force to put on the handle over time. Too little load = chatter, too much torque = spindle stops or stuff goes flying, or cutter gets busted. I haven't operated a CNC mill or lathe but I think that manual experience is quite important for gauging how much load to put on a tool.
The usual fix for chatter is to slow your spindle speed, which increases your chip load, same as increasing the feed rate or similar to taking a wider RDOC.
This is the most wonderful and bonkers machine-y science-y public service. Please tell me you've got a Patreon set up! I'd love to see something similar in aluminum, maybe take a look at variables influencing chip welding?
Thanks! That's a great idea re: welding... I'd love to see what that looks like too (lots of things to play with there too, with/without coolant, etc). No Patreon yet, but perhaps I should start to look into one now that the channel is starting to grow a bit more :) Thanks for the suggestion!
Hi, I have a serious question. Do you think putting a flywheel on a small 500w "desktop" CNC machine would server an advantage in reducing chatter? I can tell from your video your machine clearly has a lot of momentum from the way it kept spinning. Do you think this momentum helps the tool chip better. And do you think the gyroscopic effect would help prevent the tool from doing that oribal motion we can see when chatter is occuring? Please let me know your thoughts!
Nice video! Don't forget, when you reduce your step over you need to up your feed rate to get the same chip thickness. Investigate chip thinning. On your .012 step over your feed rate was too slow to "load" the endmill.
So I'm guessing what this means is that chatter is caused by the bit not gripping anything which then allows it to get throwing off balance when ejecting chips or when it gets the next bite of material, those induce vibrations that then hit resonance frequency and continue to grow until your cut is completely garbage?
What a great video! A couple of observations. Based on your values at 24000 rpm and 1/4 in dia., your end mill is cutting at 1570 SFM. Yikes, that is high even for carbide on a rigid milling machine, let alone a router. Niagra cutter recommends 100-400 SFM for mild steels. niagaracutter.com/speedfeed The chip load per tooth is .0016, which falls in the range for mild steel. The material removal rate is 2.25 cu inches per minute, which is amazing.
Oh yeah, totally. 100% agree! I probably should have stressed that some more... these are awful settings and probably murdering the tools even on the "good" cuts :) I have an older video with (a little) more sane settings for steel... still high SFM but it's a tricky balance to find something that hits the torque/power curve of these spindles without going too fast and burning up the carbide. I recently discovered a "torque boost" setting on my VFD which pushes more voltage at lower RPM, I might give that a shot if to see if I can get the RPM down some more. It starts to get dodgy in steel around say 16k RPM, so if the torque boost could get me in the 12-16k range it'd be soooo much better comparatively speaking. IIRC I still have to use it 16-18k right now which is pretty high SFM on 1/4". Might also just get a better VFD, I hear the "vector" drive VFDs are pretty nice :)
@@BreakingTaps These are rated for higher speeds. www.guhring.com/Documents/Tech/SpeedFeed/6759.pdf?version=082020092010. Can your VFD display motor amperage like th-cam.com/video/h9q6j7POgOI/w-d-xo.html?
Nice! HSMAdvisor shows that the cut that caused the VFD to give up should have required only about 2 of the 3 HP claimed for that spindle. Were you monitoring spindle power/current?
Oh now that's interesting, I hadn't punched in the numbers myself. The VFD can show voltage and current draw, but it's sorta hidden away and not easy to access (and I don't have the Avid machine wired up to grab that data yet). Might setup a secondary camera on the VFD to see what it reaches. Maybe it wasn't overpower... there was a generic error code and I assumed it was power without actually validating it :)
@@BreakingTaps I added your machine to the HSMAdvisor "Cloud". Some VFD's can provide analog and/or digital outputs that might help. Since torque is proportional to motor current, and you know speed, you don't need voltage to calculate cutting power.
You're putting out some phenomenal content, especially compared to the bigger channels out there. How are you liking the benchtop pro 2x3? Thinking about moving my ATC over to a bigger frame than my 6040 and thinking about it. Would you buy it again?
Thanks! I've been pretty happy with the Benchtop Pro so far, especially considering the abuse I've put it through :) Most of the things I dislike are not actually about the machine itself (mostly Mach4-related). The only major complaint/irritation/obstacle is the z-clearance of the gantry... you eat up 8-9 inches of travel pretty quickly once you start using a vise or taller workholding. Even if you raise the spindle higher on it's mounting, you still have the gantry itself to clear. So that can be a real limiting factor sometimes. I also wish I had splashed out for the larger steppers (or servos), but otherwise very happy with the mechanical parts of the machine. I don't really have room for a larger machine, and it still feels like a better deal than a comparably sized Tormach or whatever, so I'd probably buy it again yeah :) I loooove my ATC, but now that Fusion has limited toolchanges on free it puts a bit of a damper on things. Just an additional cost I guess to get a Fusion license, but irritating :(
If you'd like to know the theory of chatter look at "Machining Dynamics: Frequency Response to Improved Productivity" by Tony L. Schmitz and K. Scott Smith. He was my mentor and Dean (respectively) during College and they studied this phenomena extensively. What it really boils down to is the natural frequency of the machine and what impulse the endmill induces into the system. The goal then becomes finding which resonant frequency produces the most efficient machining (volumetric material removal) given the machine constraints. You're definitely on the right track using the frequency response of the system to tell you what's going on take a look at this old demonstration: th-cam.com/video/uv3yUCl27wM/w-d-xo.html But the cutting parameters are not chosen arbitrarily, they are chosen in a "stability lobe" essentially a region where the system is in a stable frequency response region. Here is a video of Dr Scott Smith explaining the history and discovery of this th-cam.com/video/fzqiSzmSvhk/w-d-xo.html He does a great job explaining it in very simple terms.
Oooh, thanks for the book recommendation! Would love to dive into the theory more and really get a good understanding of what's happening. I've seen those "lobe" diagrams before and understand them conceptually, but not really sure how to go about discovering/generating them for a particular machine, or really what part(s) of the machine/parameters really affect it. Cheers for the videos and recommendations! It's such an interesting field of study!
@@BreakingTaps You already know, although it's deceptively simple. spindle speed, depth of cut, axial engagement, feed per tooth, etc. It's literally all the standard cutting parameters you area already familiar with, that are the inputs to your system. Generating the stability lobe plots is straight forward, it's called a "Tap test" in industry. You put a small piezo accelerometer on the tool tip, then you tap it with a small hammer that has an accelerometer on it as well. The accelerometer on the hammer tells you how much impulse force you've applied to the system and the accelerometer on the machine gives you the frequency response of the machine. Here is a demonstration of how it's done th-cam.com/video/XcZmOIBRcUw/w-d-xo.html The accelerometer is simply attached with wax and the software is based on NI LabView to collect the data and pass it through a fast Fourier transform to get the fundamental peaks. The software and hardware is really nothing special. The real magic is in the mathematics that Dr. Smith goes into great (and readable) detail in his book. He and his colleagues created the company "BlueSwarf" to license the complete package technology to company under various trade names, but the underlying fundamentals are the same. The technology is used primarily in the aerospace industry where the big challenge is to make very thin walled parts from monolithic billets. You can get fantastically thin walls which is quite amazing. I could go into more detail, but the book is about $30 and well worth the read.
@@capnthepeafarmer Picked up a copy! I have some time over the holidays, gonna see if I can recreate this tap test... I have some accelerometers in my parts drawer and the test looks surprisingly simple (assuming I can replicate the math :) ). Cheers!
@@BreakingTaps The book should include a CD with matlab m code so you can work on the problems through the book. My suggestion is to read chapter 1, then go straight to chapter 4, then if you're inclined to do so review the theory in the remaining chapters.
@@BreakingTaps and coincidentally, MSCDirect just licensed their technology. Before I think it was KennaMetal, but you can check out their little promo page. www.mscdirect.com/solutions/millmax?intcmp=Millmax_HomePage_Slider_202012_v1
CLIMB MILLING is what you're doing: the bit CLIMBS ahead and magnifies any looseness in the "ways" (its the worst way to cut). YOU DON"T MENTION THIS AT ALL!!! IT MAKES ALL THE DIFFERENCE IN THE WORLD! With that rotation a PROPER cut would go from Left to Right! Were you trying to fail?? Throw us a bone here and let us know, unless of course you don't know what you're doing in the first place........
Are you sure your running at what you said of 24,000 rpm, I would suspect more like 2400 rpm ! At 24,000 you would be burning that endmill up without coolant, very few small mills run at that speed, And by the frames per second divided by the rough revolutions showing on that end mill it speaks more of 2400 rpm. These small mills rarely go over 2400 rpm, And you are climb milling with steel, That is a no no unless you are cutting very small amounts. 1 or 2 thou. Cut into steel and you cut will look much better, climb milling is for finishing not gouging ! And on the you tube link it says TAPS Should be endmills
I've broken plenty of taps too, no worries! Just need to find a way to get it on high speed :) Definitely 24k RPM, it's a 24k ceramic bearing, 2.2kW direct drive spindle. Regarding the footage, there are places where I sped it up so it wasn't quite as slow/boring, but if we look at a section that's untouched like the 18,000 FPS portion in the middle, 24k RPM == 400 rotations per second. And 18k FPS / 400 == 45 frames. The YT video is displayed at 24 FPS, so that's 1.875s of realtime. If you look at the top of the end mill, there's a small marker spot right where the flutes start, and it shows up about every 2s visually. So yeah, definitely running at 24k :) FWIW, I chose parameters specifically designed to chatter and look bad on high speed, but the spindle does run pretty well in steel with decent parameters. Still too fast (on an SFM basis) for steel, but doesn't burn up endmills immediately either. Cheers!
Awesome. I've always wondered what chatter looks like. It would be cool to see a macro shot of any of the shop tools that move really fast like an air nibbler or something. I also wondered what it looks like when you accidentally run a lathe spindle in reverse (or mill). Not that I would have ever done anything like that, I'm purely asking for a friend that removed a substantial amount of material wondering why the knobs were so hard to turn...
Yay thanks for stopping by! So funnily enough, I think my first milling video on YT I was running the endmill in reverse and really struggling to turn the handles, not understanding what was happening. Doh :) I just might have to investigate that, yunno, for science.
@@BreakingTaps I'm pretty much convinced that all the channels I like also like watching the same channels I like...
P.S. I feel like this totally changes my perspective on chatter..... had no clue a rigid looking set up like that would oscillate so much! I mean I could see the endmill and or tool holder and entire head of the machine deflecting that much during chatter but had no idea the work and entire machine table and vise would... crazy
Heck yeah! BTW, Steve Mould did an interesting video on video motion enhancement algorithms that modify the contrast of videos to accentuate motion using lower speed cameras. You guys could experiment with that for some good high speed footage.
@@BreakingTaps I'm so relieved to know I'm not the only one. haha
Why helicopter's self destruct
Such an underrated channel. Pushing home-manufacturing to a new level with every video. Your editing has improved massively as well over time. Best of luck on your TH-cam journey
Also appreciate the note about editing... it's so hard! I'm humbled everytime I watch a really good video with editing. Slowly getting better, trying to spend more time on it since it makes such a big difference imo.
As an actual CNC machinst, I can tell you that even the biggest and heaviest machines move around like this. Not as much, and not as fast, but your regular Haas mill is a limp noodle compared to dedicated heavy iron for HSM production machining. "Everything is a spring" is indeed correct, and everything that has clearance, has slop to move around.
The second thing I want to mention is that your best stepover for "less than ridged enough" tooling is usually 33% and 66%, where the cutting forces want to pull the cutter parallel with the direction of feed, instead of into or away from the work. This tends to give you a more perpendicular cut, because the tool isn't flexing to one side, and can give you a more stable cut because the direction of the cutting forces is countered by the feeding forces of the machine and should have less backlash to move around in.
YES! It's so rare to see good footage this slow of machining, and such a fascinating analysis.
Thanks Winston, really appreciate it!
@@BreakingTaps 200 IPM.......😂😂😂,MILD STEEL
When i Use Max 100 IPM, but thats Aluminum Materials ,KOR 5
Thanks for the Video
🙂 Figured if the goal was to make it chatter.... might as well really give it the juice and chatter :) To be fair, the WOC is still pretty light which makes the IPM look a little less ridiculous... still very high though :)
@@BreakingTaps Looks the Titans CNC Academy,
Thats 1600 IPM
KOR 5
1/2 Endmill
Aluminum Material
Not Break
This fast, you mean 😁
Wow, I always cringe when I hear chatter and in my head I'm just imagining that super hardened metal against metal, but actually seeing what it's doing is crazy. Really great analysis, too. Haven't seen this channel before but I'm glad I stumbled across it.
This is one of the most unique and informative videos I have ever seen on machining. Clearly, the average machinist does not have the equipment nor ability to do this depth of analysis. Thank you very much for sharing it.
Still watching this video 3 years later. There's a wealth of knowledge to gain from it, both about your machine and the effects it has on machines in general. Once again thank you.
That is amazing. I’ve made my fair share of mistakes and have seen the movement of the machine like that at the end. It’s pretty cool seeing it in slo mo
This is the stuff of universities and institutes. Very educational! I think that the factor , which additionally plays a big role, but is hard to estimate for yourself is the effect of temperature. It isn't constant throughout, but we all know how it could worsen a metals property. Great video!
I would LOVE to see this done with all different types of tool holders to see the difference between them. Do ER collet, hydraulic holder, slim fit, shrink fit, rego fit, etc.
"what NOT to do with an endmill", great high-speed footage! I assumed chatter was just bouncing *off* the part, creating just a lateral vibration - it never occurred to me that it would be a circular oscillation. But considering the bit is spinning, I suppose I should have expected that. Thanks for the vid!
Breaking that tool was so worth it for the footage. Absolutely amazing. Thank you
WOW. I think you are the first person I have subscribed to after only 1 video. Fantastic footage and crazy from a physics/material perspective.
This is beyond educational. I hope you realize the service you have done here. It makes the science behind machining visible. Thank you.
I also wanted to ask, when your spindle bogged down and the VFD eventually shut off, if you had a more powerful spindle (5kw), that cut would of then been possible, no?
Thanks! I think a more powerful spindle might have done the trick, although I suspect there might be a configuration issue with my VFD too. Someone else in the comments ran the numbers and the spindle should have been pulling under the 2.2kW limit... so there might be a threshold on my VFD that isn't configured correctly, or maybe I tripped the thermal cutout. Not sure :)
This is just some crazy video analysis. TH-cam's algorithm sent me your way somehow and I am so glad they did. I've always been looking for someone who knew machining AND All this slow mo videography stuff. The way you can measure certain things like how much the spindle moved is extremely helpful. Subbed.
In a "proper" cnc machine all you do is shift the failure point around to different places. Care and maintenance of tool holders and spindle taper. Pull studs. Draw bar springs. Collets or whatever tool holding. Tool balance/runout. Thrust bearing maintenance. Work holding.
That's all right off the top of my head. Good cut parameters are exactly as important. The ragged edge of max productivity is every bit as razor thin, the bar is just set a tiny bit higher.
This is AWESOME! It’s so cool to be able to see what you hear! Thank you for investing in this.
Absolutely fantastic work. Your play by play is super useful. Regarding cut depth... insight + (slowmo) sight is a 10 outta 10!
More of this kind of skoolin for desktop amateurs PLEASE!
This! This is golden content. When you had asked earlier what we wanted to see, this is exactly it. Great content!
Great to hear, thanks! Appreciate the feedback here and in those polls, definitely helps me :)
This was so inCREDIBLY cool to watch... Subscribed!
The fancy editing was awesome
BT, very nice. I've been running machines since the1970's and this is the first time I've ever seen this. Thanks for the video and yes I will show this around.
Great footage!!!
Also, we usually say machine "rigidity" instead of "stiffness".
Great footage! Really gives a greater understanding of what is really going on
Really interesting. Thanks for posting.
Chatter is your depth of cut, machine stability, material being cut, material and profile of the cutter speed lubrication etc... that produces a harmonic frequency that causes chatter and there are multiple orders of harmonic and sub harmonic frequencys that combine and cause mayhem.
What I do is finish the profile with a reverse pass (2 to 5 thou finishing cut) i.e. your not cutting, your spinning the cutter in the oppposite to its cutting direction into the job slowly.. outcome a nice shiny smooth finish and yes i am a trained eng tech
Ps. Never used carbide cutters on the mill always HSS and never broke any as i would get my marbles booted lol if i did. Cutters are a capital outlay when blunt into the tool and cutter machine and re sharpen.
Speaking from experience after working for a top carbide and tooling manufacturer, I would suggest you revisit how you hold your carbide cutters. ER collets, if that is what you are using are not very good. I suggest a side lock holder or the more expensive shrink fit. For a cheap experimental option I would try the side lock type. Cheers Ian langley ex Seco Tools New Zealand
So I used to work in industrial automation, and one time we were building a machine to remove coating from pipes in a specific way. The project was a pain in the ass and was taking much, much longer than initially estimated but we were making progress.
We had a medium-strength steel blade as a kind of scraper to remove the coating at a high RPM - the whole system rotated around the pipe as fast as possible - and the blades were getting stuck and sometimes breaking. The boss came in and told us to use carbide blades. The main mechanical engineer said "no, that's not the issue, carbide will just break MORE easily".
Nope nope. Boss says use carbide. Spent a week reworking it to get the other blades in there. First run, blades just shatter. Get out the spares and slow it down. Blades shatter. Do this a few more times after waiting for more carbide blades to get made up. They shatter. Boss finally goes "huh maybe carbide is too brittle." YOU DON'T SAY!!
Haha, if only someone had known ahead of time and could have warned him! 😂
Great work Sir, very high quality production. I really enjoy your content. You're so advanced compared to my machining that I don't know what to recommend next. Maybe content like this with speeds and feeds using high speed footage, and showing other machining failures would be helpful. Maybe slotting in aluminum and gumming up the cutter, that would be interesting for noobs like myself.
Oooh I like that aluminum welding idea, would be really fun to see that happen in slow motion! Cheers!
First off, this is really great content! Subscribed. This is a lot of work, but I'd suggest putting a bunch of dots and recording the video. Then do motion tracking on the dots and then plotting the various motions. I don't know of a reasonable way to do this, but someone must. I've done stuff like this with OpenCV, but it was a big effort. Even if you don't get to tracking, at least take the video and maybe someone else can figure it out. It would be amazing to take the video, then see a motion capture that converts that video to a wireframe using the dots' real positions.
Oh my goodness, this is an AWESOME idea, thanks! I'm definitely going to research how to make this possible (and/or who to recruit to collab with if it's outside my skills). I was considering adding some dots to do rudimentary analysis of the table and frame since I was curious how much they were deflecting...but didn't think about going full motion-capture on it and build out a 3D model. Genius! I wonder how much 3D data can be extracted from a single perspective? Maybe do photogrammetry first to build up a 3D model, then the high speed data provides the motion? I dunno... exciting! So much to start reading :)
@@BreakingTaps Don't know if you have a programming background (still working my way through your videos), but here is a link to a site that got me started on OpenCV www.pyimagesearch.com/2015/09/21/opencv-track-object-movement/ In particular, this link shows tracking a dot. So long as your dot is significantly bigger than a pixel (consistent lighting, etc.), getting sub pixel resolution is totally doable (1/10 pixel is doable, IIRC). If you can find off the shelf SW, all the better.
Combine this with a cross-colab with another bigger TH-cam machining channel (your content, if you pull this off, will make you the star of the episode), and I would bet you would get a huge jump in your subscriber count / dopamine fix.
Perfect, thanks! I do indeed have a programming background... not the best at Python but can muddle through. Will take a look and see if I can get something working over the weekend! I didn't realize OpenCV could track so robustly, sub-pixel resolution is super cool.
Wow!! Did not expect such a good video!! Great job man!! Loved it
Kind of makes sense, endmill engagement slows down spindle, when you hit 0 or low number spindle can speed up and bounce when endmill goes back into material.
great images. Had a Fadal at my last job that no one knew how to use so I learned how so I am familiar with feed and depth parameters. I also did some high speed vid of a fluids so I know how hard it is to get good footage. What you did is really good and informative to watch. The oscillation of the material took my by surprise.
When you have a deeper or wider cut you are putting more force and vibration energy into the machine's structure. Double the width, double the force, quadruple the energy.
These machines don't have much that will absorb that energy. It would be very interesting to see a super-slow-motion shot of a machine with tracking dots, which could tell us what part of the machine is doing most of the deflection.
* I didn't catch which mill you are using, but most of them use cast iron for a fair bit of the framework. Cast iron has about 23x the internal friction of steel (iopscience.iop.org/article/10.1088/1757-899X/147/1/012031/pdf). That said, the mills tend to use thick cast parts which are very rigid. Unless the cast iron is doing the deflection, it can't absorb the energy.
* I suspect that most of the vibrational energy usually gets absorbed by the work piece, where the cutter is peeling away a chip. This is the place where metal is deflecting the most, so has the most potential to absorb energy. I believe this is why shallow cuts can sometimes chatter worse -- the kick coupled into the machine from each cut is relatively larger compared to the energy dumped by peeling away the chip.
It would be interesting to find the resonant frequencies of the mill, and then see if the chips that are cut at different points in the resonance show more and less plastic distortion.
This video circles around the point. Great job.
It could be interesting to see how the endmill moves when only one of the flutes is engaged at a time, ie. cutting deep with a single flute endmill with a relieved back side of the cutting edge, compared to ie. a 4+ flute one in the same situation. I've been taught that single flutes are really good only for very shallow depth and large radial engagement. I'd wager the deflection would be smaller on these high-speed spindles compared to a low-RPM mill, but it could be still fun.
Ooh, I like that idea. I have a number of single flutes on hand. Would be neat to see how it compares to to 2/3/4 flute in a few different cutting scenarios. I bet their chip evacuation is pretty awesome to see too, since the gullet is so large.
@@BreakingTaps Oh, one more thing - not sure how usable it is for this use case, but video motion magnification (see ie. people.csail.mit.edu/mrub/vidmag/ for the academic side, software included) could be interesting too and potentially reveal any smaller amplitude harmonics in the vibrations. Its commercial use is more in monitoring tiny concrete / pipe vibrations for early failure, but it's worth a shot if you get it working. :)
Woaaah, that's a cool technique! Will dig into this, it looks super useful.
Great quality video and information, subscribed
Really amazing video! Learned something today
Depends on weather you are climb milling or conventional milling also it depends on path direction. Tool length. Holders collets. And also speeds and feeds
So cool to see!
I've seen footage from regular cameras of industrial machinery that is run through software which then amplifies tiny movement and it is amazing to see. Apparently it's used to look for potential wear or weak points. Would love to compare this rather flexible machine with something like Piotr Fox Wysocki's epoxy granite machine here on youtube. His finishes are magical, made even more so by the fact that it's a home built machine.
I've only experienced chatter on a regular column drill and I've gotten the feel for how much force to put on the handle over time. Too little load = chatter, too much torque = spindle stops or stuff goes flying, or cutter gets busted. I haven't operated a CNC mill or lathe but I think that manual experience is quite important for gauging how much load to put on a tool.
Hey, awesome footage, really interesting to see. Thanks for sharing. :)
Interesting that the collet allows for so much movement. It would like to know if tightening the collet holder more impacts the outcome.
Cool vid man, subbed, look forward to more
That's awesome man.
Came here after hearing about you on Ep 111 of Within Tolerance Podcast.
I think I’ll stick around a bit and explore. 😉👍🏻
Oh, neat! I didn't realize it was featured on the podcast, will have a listen myself! Thanks for watching! ❤
The usual fix for chatter is to slow your spindle speed, which increases your chip load, same as increasing the feed rate or similar to taking a wider RDOC.
IDK why this was recommended to me but I enjoyed the content good work
Oh my aching spindle bearing!!
I had a foreman that would say "if you ain't cutting, you're just rubbing". This is the proof.
On a properly rigid machine you can skim tenths at will with zero chatter. I do it all the time in 17-4 H900 and Titanium alloy.
Wow, very eye-opening video! Nice to see what's really going on with chatter.
NEAT footage! Rather wild how much deflection is going on.
Great footage and explanation.
First video, you got me!
Awesome, thanks!
An experienced ear is a very useful tool for adjusting feeds and speeds.
That was fascinating to watch, very cool. =]
This is super fascinating! I wonder if it would be possible to correlate what your audio analysis to the video?
This is extremely relevant information, thanks for doing the research!
Good nerdy stuff👌🏼😀
Wow, thanks!
Great job! 👍
Very interesting, you've got a new subscriber.
This is the most wonderful and bonkers machine-y science-y public service. Please tell me you've got a Patreon set up! I'd love to see something similar in aluminum, maybe take a look at variables influencing chip welding?
Thanks! That's a great idea re: welding... I'd love to see what that looks like too (lots of things to play with there too, with/without coolant, etc). No Patreon yet, but perhaps I should start to look into one now that the channel is starting to grow a bit more :) Thanks for the suggestion!
Great video!!!
Hi, this is really interesting. For me it would have been really helpful to also hear sound samples from the different footages. Thanks 👍
I never thought it'd happen, but watching this video makes me want to draw a free body diagram.
I was taught to avoid climb cutting whenever possible.
Great video! I love the tune! Subbed.
the slo mo blues at the end tho
0:50 Videotaping the endmill... Got you BetaMax recorder out for this? Amazing video; subbed. Thanks buddy.
Excellent video! I wonder if you can use the camera to show the difference that the amount of flutes on a tip does?
Working on it! Setting up some shots for single flutes right now, and might extend it to compare single vs double vs triple flutes in aluminum. :)
0:11 - Chatter? More like endmill _shatter,_ right guys, right?
I'll get my coat.
Use a 3 fluted Endmill. The teeth are 120 degrees apart. 2 and 4 fluted endmills chatter more because they have opposing teeth.
Jesus that was wild.
Awesome video
amazing
Quality content
It's lack of spindle/machine stiffness which appears to be the big problem here judging by the opening frames.
Hi, I have a serious question. Do you think putting a flywheel on a small 500w "desktop" CNC machine would server an advantage in reducing chatter? I can tell from your video your machine clearly has a lot of momentum from the way it kept spinning. Do you think this momentum helps the tool chip better. And do you think the gyroscopic effect would help prevent the tool from doing that oribal motion we can see when chatter is occuring? Please let me know your thoughts!
18 seconds: endmill has left the chat (ter)
Nice video!
Don't forget, when you reduce your step over you need to up your feed rate to get the same chip thickness. Investigate chip thinning. On your .012 step over your feed rate was too slow to "load" the endmill.
And by the same token, reducing the rpm could have had the same effect. The limitations of the machine tell you which to choose.
So I'm guessing what this means is that chatter is caused by the bit not gripping anything which then allows it to get throwing off balance when ejecting chips or when it gets the next bite of material, those induce vibrations that then hit resonance frequency and continue to grow until your cut is completely garbage?
Yep, that's my interpretation as well! And once you hit the resonance, it's pretty much game over until you break out of the resonance zone
What a great video!
A couple of observations. Based on your values at 24000 rpm and 1/4 in dia., your end mill is cutting at 1570 SFM. Yikes, that is high even for carbide on a rigid milling machine, let alone a router. Niagra cutter recommends 100-400 SFM for mild steels. niagaracutter.com/speedfeed
The chip load per tooth is .0016, which falls in the range for mild steel.
The material removal rate is 2.25 cu inches per minute, which is amazing.
Oh yeah, totally. 100% agree! I probably should have stressed that some more... these are awful settings and probably murdering the tools even on the "good" cuts :) I have an older video with (a little) more sane settings for steel... still high SFM but it's a tricky balance to find something that hits the torque/power curve of these spindles without going too fast and burning up the carbide. I recently discovered a "torque boost" setting on my VFD which pushes more voltage at lower RPM, I might give that a shot if to see if I can get the RPM down some more. It starts to get dodgy in steel around say 16k RPM, so if the torque boost could get me in the 12-16k range it'd be soooo much better comparatively speaking. IIRC I still have to use it 16-18k right now which is pretty high SFM on 1/4".
Might also just get a better VFD, I hear the "vector" drive VFDs are pretty nice :)
@@BreakingTaps These are rated for higher speeds. www.guhring.com/Documents/Tech/SpeedFeed/6759.pdf?version=082020092010. Can your VFD display motor amperage like th-cam.com/video/h9q6j7POgOI/w-d-xo.html?
@@BreakingTaps Thanks for taking the time reply. The vector drive may be nice. Love the content, please keep it up.
Nice! HSMAdvisor shows that the cut that caused the VFD to give up should have required only about 2 of the 3 HP claimed for that spindle. Were you monitoring spindle power/current?
Oh now that's interesting, I hadn't punched in the numbers myself. The VFD can show voltage and current draw, but it's sorta hidden away and not easy to access (and I don't have the Avid machine wired up to grab that data yet). Might setup a secondary camera on the VFD to see what it reaches. Maybe it wasn't overpower... there was a generic error code and I assumed it was power without actually validating it :)
@@BreakingTaps I added your machine to the HSMAdvisor "Cloud".
Some VFD's can provide analog and/or digital outputs that might help. Since torque is proportional to motor current, and you know speed, you don't need voltage to calculate cutting power.
You want something fun the cut. Try Inconel 625.
Carbide is brittle until you try to bust apart a broken end mill.
Great vid..!! When the hell are you/ US going metric 😂….
You're putting out some phenomenal content, especially compared to the bigger channels out there. How are you liking the benchtop pro 2x3? Thinking about moving my ATC over to a bigger frame than my 6040 and thinking about it. Would you buy it again?
Thanks! I've been pretty happy with the Benchtop Pro so far, especially considering the abuse I've put it through :) Most of the things I dislike are not actually about the machine itself (mostly Mach4-related). The only major complaint/irritation/obstacle is the z-clearance of the gantry... you eat up 8-9 inches of travel pretty quickly once you start using a vise or taller workholding. Even if you raise the spindle higher on it's mounting, you still have the gantry itself to clear. So that can be a real limiting factor sometimes. I also wish I had splashed out for the larger steppers (or servos), but otherwise very happy with the mechanical parts of the machine. I don't really have room for a larger machine, and it still feels like a better deal than a comparably sized Tormach or whatever, so I'd probably buy it again yeah :)
I loooove my ATC, but now that Fusion has limited toolchanges on free it puts a bit of a damper on things. Just an additional cost I guess to get a Fusion license, but irritating :(
Isn't the mill milling in the wrong direction as per the rotation of the milling bit? I thought you were suppose to mill in the opposite direction?
You can do either. What is shown is called climb milling, the other way is conventional milling
Few yards movement is good enough.?
Nice
If you'd like to know the theory of chatter look at "Machining Dynamics: Frequency Response to Improved Productivity" by Tony L. Schmitz and K. Scott Smith. He was my mentor and Dean (respectively) during College and they studied this phenomena extensively. What it really boils down to is the natural frequency of the machine and what impulse the endmill induces into the system. The goal then becomes finding which resonant frequency produces the most efficient machining (volumetric material removal) given the machine constraints. You're definitely on the right track using the frequency response of the system to tell you what's going on take a look at this old demonstration: th-cam.com/video/uv3yUCl27wM/w-d-xo.html
But the cutting parameters are not chosen arbitrarily, they are chosen in a "stability lobe" essentially a region where the system is in a stable frequency response region.
Here is a video of Dr Scott Smith explaining the history and discovery of this th-cam.com/video/fzqiSzmSvhk/w-d-xo.html He does a great job explaining it in very simple terms.
Oooh, thanks for the book recommendation! Would love to dive into the theory more and really get a good understanding of what's happening. I've seen those "lobe" diagrams before and understand them conceptually, but not really sure how to go about discovering/generating them for a particular machine, or really what part(s) of the machine/parameters really affect it. Cheers for the videos and recommendations! It's such an interesting field of study!
@@BreakingTaps You already know, although it's deceptively simple. spindle speed, depth of cut, axial engagement, feed per tooth, etc. It's literally all the standard cutting parameters you area already familiar with, that are the inputs to your system. Generating the stability lobe plots is straight forward, it's called a "Tap test" in industry. You put a small piezo accelerometer on the tool tip, then you tap it with a small hammer that has an accelerometer on it as well. The accelerometer on the hammer tells you how much impulse force you've applied to the system and the accelerometer on the machine gives you the frequency response of the machine. Here is a demonstration of how it's done th-cam.com/video/XcZmOIBRcUw/w-d-xo.html
The accelerometer is simply attached with wax and the software is based on NI LabView to collect the data and pass it through a fast Fourier transform to get the fundamental peaks. The software and hardware is really nothing special. The real magic is in the mathematics that Dr. Smith goes into great (and readable) detail in his book. He and his colleagues created the company "BlueSwarf" to license the complete package technology to company under various trade names, but the underlying fundamentals are the same.
The technology is used primarily in the aerospace industry where the big challenge is to make very thin walled parts from monolithic billets. You can get fantastically thin walls which is quite amazing.
I could go into more detail, but the book is about $30 and well worth the read.
@@capnthepeafarmer Picked up a copy! I have some time over the holidays, gonna see if I can recreate this tap test... I have some accelerometers in my parts drawer and the test looks surprisingly simple (assuming I can replicate the math :) ). Cheers!
@@BreakingTaps The book should include a CD with matlab m code so you can work on the problems through the book. My suggestion is to read chapter 1, then go straight to chapter 4, then if you're inclined to do so review the theory in the remaining chapters.
@@BreakingTaps and coincidentally, MSCDirect just licensed their technology. Before I think it was KennaMetal, but you can check out their little promo page. www.mscdirect.com/solutions/millmax?intcmp=Millmax_HomePage_Slider_202012_v1
CLIMB MILLING is what you're doing: the bit CLIMBS ahead and magnifies any looseness in the "ways" (its the worst way to cut). YOU DON"T MENTION THIS AT ALL!!! IT MAKES ALL THE DIFFERENCE IN THE WORLD! With that rotation a PROPER cut would go from Left to Right! Were you trying to fail?? Throw us a bone here and let us know, unless of course you don't know what you're doing in the first place........
Hooked !
Are you sure your running at what you said of 24,000 rpm, I would suspect more like 2400 rpm !
At 24,000 you would be burning that endmill up without coolant, very few small mills run at that speed,
And by the frames per second divided by the rough revolutions showing on that end mill it speaks more of 2400 rpm.
These small mills rarely go over 2400 rpm,
And you are climb milling with steel, That is a no no unless you are cutting very small amounts. 1 or 2 thou.
Cut into steel and you cut will look much better, climb milling is for finishing not gouging !
And on the you tube link it says TAPS
Should be endmills
I've broken plenty of taps too, no worries! Just need to find a way to get it on high speed :)
Definitely 24k RPM, it's a 24k ceramic bearing, 2.2kW direct drive spindle. Regarding the footage, there are places where I sped it up so it wasn't quite as slow/boring, but if we look at a section that's untouched like the 18,000 FPS portion in the middle, 24k RPM == 400 rotations per second. And 18k FPS / 400 == 45 frames. The YT video is displayed at 24 FPS, so that's 1.875s of realtime. If you look at the top of the end mill, there's a small marker spot right where the flutes start, and it shows up about every 2s visually. So yeah, definitely running at 24k :)
FWIW, I chose parameters specifically designed to chatter and look bad on high speed, but the spindle does run pretty well in steel with decent parameters. Still too fast (on an SFM basis) for steel, but doesn't burn up endmills immediately either.
Cheers!
ImageJ, the Fiji version?
Yep!
Which camera is this? Phantom?
What song is at the end?