Hi Peter, I have come across this once before. I believe its called a Double enveloping worm. It was on a pod on a North sea tug that had run into a submerged obstacle and had ruined the gear. The gear was bronze and was sacrificial as the worm was so complicated. I managed to model the gear and cut the teeth using a custom ground fly cutter on a horizontal borer with a mega complicated set up. Later on I found out that the pod had been made by a Japanese company that had a lead time of 24 months if it was still in business. Tug is still chugging away last I heard . Challenging job but the best ones always are. ( We didn't have any 5 axis machines at the time, in fact at that time there were no 5 axis machines in the UK).
@@fredrezfield1629 Hi Fred I wish I could, it was about 20 years ago. I do remember the gear blank was set up on a rotary table and the head was offset about 14 degs, I cant remember the dia of the fly cutter, but I do remember it was a pain cutting the teeth. It was just plunge cutting and the deeper I got the bigger the cut got, cutting both flanks at the same time. I did rough it out with a straight grooving tool first. I had to engineer some backlash into the thing as the ship was loosing megabucks a day in drydock and I only had one chance to get it right ( no test fitting). The ship was called the MV Norman. Sorry I couldn't be more specific, but it worked fine and I got a letter of commendation from the ship owners. The blank was roughly 1.1m in dia with around 65 teeth, each tooth was about 20mm deep, I cut involute teeth although the original had straight teeth ( this gave me a bit extra clearance).
As a fellow machinist walking knee deep in post processor customization in Edgecam for Fanuc and Okuma lathes and 4 axis machining VMC-s I raise my hat! this job is some real test of skill and understanding of circumstances!
@@johnhansen1684 It's possible, although the user interface isn't the most user friendly. I'm working on a few post processors right now and it takes some trial and error.
@@johnhansen1684 I think the limit is the sky when we talk about edgecam post customization. all of the post macro is contained and readable in the .ctx and .mxt files. those files are compiled from the .cgt and .cgd files, which are opened in the codewizzard which give you a pretty decent customization interface. the Codewizzard files are compiled from a bounch of txt files, and by using .cgt files with .cge files on can even replace whole sections of the base macro. so basically one can rewrite the whole thing if he wants. with that said i am pretty sure that requires an awful lot of knowledge in math and macro porgramming. macro programming is never an easy task, so yeah it is pretty hard given that a 5 axis CGD file is compiled from 8-10 txt files which contain ~15k lines of macro coded by the edgecam team. the basic stuff tho is not that hard to figure out tho. also it has a pretty decent help integrated into the codewizzard.
@@ferencungvari3089 Interesting, i have played around with WorkNC but not looked into edgecam that deep. we got some edgecam license where I work would be fun to look into it. do you have any sources for information other then just trail and error?
@@johnhansen1684 codewizzard help in the menu is a goldmine. I got the luxury to watch our distributors guys doing the initial customization and built a friendly relationship with them so in case of desperation i can always have a chat with them while they show me the principles through teamviewer. I think it would be wise to chek it out if some dude put up something to skillshare or similar sites. That said the customization is always a custom problem for the given environment so if you can master the principles, you can get by. Also turning on the NC debugging functions in codewizard helps a lot. There is also a customization whiteboard iside where one can implement his own procedures than call those in the cases where neccesary alas keeping the thing professional.
OK, Tis is so far over my head but to listen to your brain working and the machine working is most enjoyable. Heads like yours take machining to the nest level and beyond. Thanks for sharing.
Very cleaver design, and I like it. Consider the small amount of surface footage difference across the cam followers flat length of engagement. I have come across this situation before, but in a different way. You are very clever, and you will more than likely create your own outer race to compensate for the difference. Thank you for letting me enjoy another wonderful video of brilliance.
Peter, had I read down the comments a little further, Cory Keck's comment would have sufficed.Thank you Peter, and Cory. I'm only 73 and grateful to have my same youthful excitement about this field of work. Bill P.
@@thedillestpickle i have broken xometry with parts like this before. it would give me a ridiculously low quote and i would hit go just for laughs and get a phonecall back saying they couldn't do it for that
That's awesome. I remember reading somewhere that the thread pitch on the worm is not constant and the slot is wider than rollers. This allows the roller to roll with clearance on the opposite side and the exiting and entering rollers preload against each other while the ones in the very center are actually clearance on both sides.
One way it could work is if the screw was pinched between each adjacent roller and the groove was wider. Imagen a straight track pinched between two roller bearings. Now twist that around in the worms shape. That would do it.
The techinique for producing these dates all the way back to the 1920s. It is very similar to single point screw cutting except that the tool is driven along an arc spanning the length of a workpiece which was been concave turned along the same arc. This is generally done using a rotary table which is synchronised to the spindle in which the workpiece is mounted.
Happy to see someone with your resources thinking about a DIY version of this beautiful thing. I hasten to add that on a number of occasions watching a video of a rotary table running using this roller cam method, something was wrong in all of them. The rollers looked like a snug fit between flanks but they cannot be. Opposite surface direction. So after seeing your video , got all enthused again and laid out a 2d sketch to the following conclusion: The distance between outside edges of two rollers can be pitch X 2. Making sure , of course the crest or flank between them doesnt touch anything. Ill be watching. Thanks Peter. Bill Phillips, BC, Canada
Look at this video. A close up of the cam roller drive. Look closely you will see the rollers reverse direction as they pass thru the worm. th-cam.com/video/2uHlH3H-1k0/w-d-xo.html
I know a fellow colleague who keeps telling me that you need a straight up model and always trust the post for that model… Using Mastercam. Your process is a great example of having to work around the software limitations.
If I had an accurate model of the worm it would be better to program from in Esprit. Then I could use real 5 axis cuts on the surfaces. As it is I'm really only cutting to the wire curves, of the ruled surface. I can sort of make finish cuts by offsetting the cut from center-line. But its not the most ideal. Also I could put proper chamfers and finish the thread ends. If they were also modeled.
it is unbelievable the Mazak CNC machine's five-axis lathes can be programmed and perform curves thread this brilliant machinist man can program and make all kinds of parts out of a CNC machine,!
The only way would be a machine dedicated to doing exactly this. Or maybe a horizontal boring mill with a rotary table on top of its normal rotary B axis table positioned such that it's center of rotation was at the worms radius. And then all that geared to operate at the gear ratio needed. from the B axis drive handwheel. Then to cut it you would only have to set the proper depth with the quill and rotate the B axis thru the 90 degrees while the worm rotates at the correct ratio. Or maybe rotate the smaller rotary table's handwheel to drive the large B axis thru the gearing to cut it. Simple don't you think?
Brother uses Roller Cam gears for the tool turret and their 4th axis. As does Sankyo. The trick is, as you suspected, tapered cam rollers in tapered grooves.
Hmmm.. so are the Brother rollers tapered larger toward the OD, which would seem to give line rolling contact? Maybe not significant at this radius, especially if they have a little crown. Also, if you split the worm in two, could shim axially to take out clearance, or maybe could rotate one half slightly to take out the clearance, which might preserve the continuity of the arc? Very impressive! Thx!
Instead of splitting the cam for preload, you could model two rule surfaces with slightly mismatched pitches. I believe this is how the machine tool makers do it. Incidentally, Mori has a machine (NT) with a massive direct drive servo controlling the head’s rotation. That concept hasn’t come into widespread use, probably because it makes the machine a lot bigger and more expensive. But I’ve always thought that might be the best approach on a small custom machine.
Are you able to sweep cut with a solid in esprit? I've done something similar in Solidworks before by creating the spiral like you have and then driving a cylinder (endmill) around the spiral wireframe as a sweep cut. Not sure if you could do it with two rails though to control the "tool axis" of the cylinder.
So it's just like a worm gear, but with several more teeth engaging at once. So a very strong construction. This would be super useful in adding even more strength to something like a "Torsen differential"- for heavy duty applications. Very interesting. I like it.
It’s exactly like a worm gear but with rolling contact between the elements rather than sliding contact. Big reduction in wear and friction and it allows you to eliminate any backlash by preloading the rolling elements.
Ive been thinking about making one of these since your B axis video. I dont think you can split it in two to achieve preload. The join would create a lump that the roller has to climb over. Tapered wont work either. The rollers must contact on one side only. Otherwise they cant rotate. The only thing I can think of is every second roller is offset to get preload. Thanks Peter for getting the thought process going on this one!
There are a couple of ways. One is two worms on one gear. The other is this. Visualize a straight track with a ball bearing on both sides loaded into each other. Now twist that track into the shape of this worm. It’s the same thing. The slot is wider than the rollers. They ride on the spiral track in between them. If that makes sense. Not on both sides. I favor the two worm idea. It is easer to preload by just rotating the worms against each other.
@@EdgePrecision that is also a great idea. Kind of like a ballscrew setup running in a worm arc. Deflection will give some backlash though I think. I hope a roller worm company sees your video and chimes in with some insight.
Could you grind something like this on your tool / cutter grinder? Also, an idea for filming: I think if you used through spindle air blast with through tool coolant tools you could get the best of both worlds in terms of clear footage, and tool loading / wear. Of course you wouldn't get the lubricity that coolant provides, so maybe it would be more of an intermediate step between no coolant and flood coolant. Amazing work as always. Thank you for taking the time to 'play' with things like this, and then share the results with us. Things like this are what advance the art in my opinion.
It's not really a shape that the tool grinder is set up to do. It almost looks like a very steep fluted tool, and you could just grind it like that. But look carefully. The flute (Groove) changes angle from sloped almost 45 degrees at the ends to vertical in the middle (Remember me putting the 1/2' dowell pin in it in the video). So you can't do it like grinding a flute. But the machine does have the same capability of motion as the Mazak. So it maybe could be done with a small diameter grinding wheel (Like the endmill) like on the Mazak. But unfortunately the machine does not have the travel range in the B axis perpendicular to the works axis. It only can travel to a positive 10 degrees in the B axis. This requires 45 degrees. This is not a problem for tool grinding but it can't do this.
@@EdgePrecision I agree. The edge of the helical slot sides is not a straight line anywhere. A grinding disc could not do it because the slot wall on the side towards the middle of the worm length is an overhang and an inside curve. It can be cut with a straight tool but not a disc. I did a lot of cutting of helix paths and the cross section of the wall was not rectangular due to the cutting being done with a disc. The slots were all below .004 inches and the tool selection was very limited. Rough with an endmill and finish grind?
We had a similar worm gear driving a cam that rotated our active sonar hydrophone receivers on our Burke destroyers. This almost looks identical, actually.
@@EdgePrecision yeah lines up time wise. The Burkes came into service in the 80s really, so it's been about 40 years. I haven't been to sea in 25 years or so, I assume they're not using mechanical actuators for the arrays anymore anyways. Probably more a static phased array, makes maintenance simpler.
Surely the rollers only roll on one side if the slot width is the same as the roller? Struggling to see the benefit over a conventional worm, can’t see how this design would help reduce backlash, but probably just my lack of understanding. Amazing machining though. Do you have a plan for this, or is it a purely intellectual exercise?
This is the drive system used in Mazak variaxis c and a axis drive. A company based in Japan makes them for Mazak. They also manufacture many toolchanger gearboxes for companies like Mori Seiki and Mazak. If we break a cam follower in one of them we have to wait months to get the correct cam follower, which are ground to fit precisely in the worm. I'd imagine you may have a similar drive to rotate your spindle.
Yes the B axis has this kind of drive. My guy that I have work on my machine (Not a Mazak person. But very good.) says that of all these type of machines in this area (there are many). That this drive is very reliable. They never have any trouble with them.
@@EdgePrecision They are very reliable. We have I think 7 vrx machines with them, 9 including our other facility. I have replaced cam followers on two, but these are machines with the 640 control that have been in full production for 16 years, 3 shifts, six days a week. Still, the reason for the replacement was a seal failure/oil leak and subsequent bearing failure. Mazak will require you to send them the machine, not field serviceable. However, we were able to talk them into selling us the parts and we did it ourselves. It was quite a job, but the machine is currently making us money. I really would love to play with an integrex. My company just doesn't do millturn, not sure why. We have the parts, but not enough smart guys like yourself.
Rapids tend to be slower because of “in position checks” that the CNC will perform. So when the rapid move is complete, the CNC will verify its position is in tolerance of where is was commanded
Manually screwing in each of the 24 rollers through a hole in the side of the case or body of this drive system.... While also incorporating lash adjustment in the worm in two axis or on a cam'd bushings? This is gonna be very interesting. I am sure of all people, you will pull this off just fine.
👍 I would think you'd have to feed the cutter in both directions, reversing the worm so to maintain a climb cut on both sides of the groove. Wouldn't the cam follower need to run in an oversized groove or one side would be counter rotating and opposing the rotation of the worm? I'm trying to understand the preload idea. Thanks Peter! You're the best!🤔
you could generate a 3d model with a swept cut feature in Solidworks. You would probably need 4 guide curves for a square cut profile with one on each corner.
So I found a bit of success not by using guide curves but by applying a prescribed twist to the profile in the sweep cut. Though I'm just using a tapered helix path to approximate his much fancier curves.
@@makingitwithnick This is totally a surface modeling part. Was able to get the bottom face of the follower channel, don't really have a need to finish the part. Modeled a helical surface (boundary surface with helixes) and the horn-shaped surface (revolve surface). Intersection curve, did this for both corners then boundary surface between the two and you're done.
Thats a really cool idea ! It is possible that the cam followers have a curved surface (I think they call it crowning), which would help. Otherwise there might be a differential in rotational speed between the top and bottom of the roller outer surface. There will be some tolerance on the angular location of the cam followers, so you probably dont want the slot to be too good a fit, probably as you say, better to cut the slot oversize and preload axially. I dont know if the manufacturers will guarantee the concentricity of the outer bearing surface to the 'bolt' on the cam followers - perhaps they aren't usually used for such high precision requirements. I am stretching my knowledge of worm design (!! not very deep), but I would guess that your design would be alot lower friction than a standard worm, but the load capacity might be smaller (especially the double envelope type)
@@EdgePrecision I am trying to think whether a 'ball worm' might be possible.. a bit like a ball screw, but keep the balls on the same plane as the wheel (in some sort of carrier) and both the worm and the wheel have the same groove that would fit standard ball bearings...
@@EdgePrecision I was thinking that you could make two identical halves and preload the bearings by rotating them relative to eachother. would that work in this design? it's stretching my ability to model the movement in my head.
This is a good idea. A crowned follower would indeed eliminate the speed mismatch but by having a single contact point instead of line contact, this would decrease stiffness, torque capacity and wear resistance. I believe a tapered follower could be used to eliminate the surface speed mismatch through the whole contact line between the follower and worm.
This was just a test. As I said in the video. I would rough it with a smaller tool. Then do finish cuts on the walls. In that case I would climb mill them. But the real part would be steel and heat treated and ground after that.
Hi Peter, Not sure what CAD software you're using but fusion 360 has a "sweep" operation that you can pick a profile and a path. You'd use your helix as the path and the cross section of the cut for the profile. Hope this makes sense! Excellent job machining that especially without the full model!
Yes and I tried that in Fusion. But it doesn't work for this. I even used both curves, it still doesn't work. It for some reason distorts the surface when it wraps around this path. I even tried to just fill between the two curves by drawing lines across their ends. This will fill in SpaceClaim and Fusion but I doesn't remain a flat curve surface. So as I thought I could just give it thickness. But that doesn't work either. This shape really messes with cad softwares. So the sweep fill or extrude dont work. I think if a software could do this. I could define a gear ratio between the gear and the worm solids. Now if I had a solid projection representing the cam roller. Then I could say do a cut thru the worm with that as I rotate the worm with that rear ratio. basically it would be following my curves as you saw in the video, but the two parts would be constrained with the gear ratio. If that makes sense. Basically exactly the way the machine cut it.
@@EdgePrecision CAD tools that can model this are able to use 3D shapes (here a cylinder) as "profile". The reason why a "flat" profile doesn't work is because of the expansive nature of the cylinder that the cutter/bearing represents when this "worm" rotates by.. parts of the cylindrical profile simply create a sort of under/overcut that a flat 2D profile is unable to produce when being dragged along a helical path.. And yeah, I know this because I tried to model something like this in Inventor with a 2D profile and run into those under/overcuts.. later versions (inventor 2022 or so I think) have 3D profile operations who can do what you originally wanted to do by using for example a cylinder as profile, but I don't have that. There also was/is a guy who programmed this into/with FreeCAD some years ago and made YT videos of it: channel is called 'after CNC' Video is titled: 'FreeCAD.凸輪 Video 3：Roller Cam_Dynamic Simulation'
This was just a test to see if it could be done. No attempt was made for proper finish or dimensional accuracy. Or even the proper material and heat treat that would be required. Now that I have a idea what to do. I have different ideas on the design to improve on how it would work. Thanks.
It's called a Globoid Worm Gear, if someone hasn't already said that. I spent hours trying to model one of these in Solidworks but finally gave up. I'll have to look at your approach of creating a ruled surface and see if I can get it to work.
In the cam software I was able to create a proper ruled surface. I could not make a proper one in the cad software. For some reason having the rotations messes it up.
that's a cool worm. I'll bet if you could find a CBN grinding wheel (so it would not get smaller while grinding) you could probably finish grinding on the mazak. I'll bet that super light grinds would do it. Of course that's coming from a hobbiest, so what do I know.
I think if you really wanted precision the Mazak is too far away with its rotational centers. It would really need a special fixture setup for the grinding. I have a idea what could be done. It would have to be made but it’s not to complex.
A purpose built machine with a large rotary table, slide and spindle on it bolted down to a base plate and geared to a dividing head/tail stock might be able to manually do it to some degree.
Well that is certainly an interesting Chinese Puzzle piece! Could you grind it on your tool & cutter grinder? It is similar to a drill or endmill flutes. (Yea, I know you wouldn't want to do that will aluminum.)
Drill and endmill flutes follow a cylinder shape. This shape is following a totally different shape. The edge of a sphere. Also the lead has to vary based on the rollers location projected perpendicular to that sphere/circule.
Shouldn’t the endmill length from the B axis center match the minor diameter of the root? Seems like too much movement on the Z, but final assembly with a drive operating in torque mode and a hydraulic brake when necessary makes sense. Really cool!
Because the machines B axis center of rotation isn’t in the same place as the cam roller center. The Z has to move with the B axis to rotate at the cam roller center.
Almost easier to program that by hand. Would alternate the roller position on the gear so they roll on different sides of the worm. Could use 12mm rollers for this, could starfish the gear for mechanical preload
I wonder how to grind those surfaces... maybe a gear wheel like the one you show, but instead of roller a grinder wheel in one of the roller spots. But that will make the grinding super special, complex and expensive; but very doable. Maybe rollers are not needed if the worm and gear are driven by two motors fully controlled, like a CNC, but just to axis.
The grinder only has to sweep the length of the screw. So if it pivots on the wheel center it would only have to swing about 90 degrees. So the worm wold need to rotate six revolutions. So how to do that with the most precision is the question. I’m thinking a Precision lead screw that moves a linkage 90 degrees for 6 revolutions. I can think of a way. That could be very precise. So the grinder could be a tool post grinder with a small ID wheel. Maybe.
As far as the CAD goes, can you make a loft cut? In Alibre, you can define a 3d path for your spiral, and then position sketches along it of the groove's cross-section. If the CAD goes find you worthy, it should just interpolate from sketch to sketch, virtual cutting out the grove as it proceeds. Mind you, Alibre frequently squicks out and adds weird twists if the sketches aren't to its liking.
I my software SpaceClaim I tried everything I could think of. I also tried Fusion with no success. I tried a loft. I tried a extrude. I tried to use the two curves as guide curves and a section translate. It all doesn’t work. There must be a way. I just don’t know how.
@@EdgePrecisionHere is a possibility, though I'm afraid it may be more work than it's worth. In OpenSCAD you could define the roller wheel, then intersect that with your worm model to leave the shape you desire. It would be a fair bit of coding (OpenSCAD is billed as the programmers CAD) but I believe it could do it.
@@ahfreebird This is what is necessary (I think). If first a gear ratio between the solid of the worm and the disk with a solid projection like the roller. Then be able to rotate them thru each other so the roller projection cuts thru the worms solid at that ratio. almost the same as the machine cuts it. But I don't know if a software could do that.
Some thoughts: Cam rollers are crazy strong. Maybe you only need two or three to engage at once...? Here's the idea: Maybe you could add a second wheel and then clock it ahead (or behind) so that you have rollers located in between where you currently show them. Then widen the cam groove and have two rollers run in the same groove and by clocking the wheels you can change the effective pitch take up the backlash or even pre-load. ONe wheel (and rollers contact one wall of the groove while the other is turned back to contact the other wall. I have no idea what your space constraints are but a wider groove and move half of the cam rollers to a second wheel might get you somewhere. Cool problem. I'm glad it's not mine. Cheers and good luck. Edit: Maybe you make the cam in the reverse profile so that the tooth protrudes. Then you have your wheel as you have shown, and you then mount a second roller to pinch the tooth that protrudes. You can make a grind spacer or some belleville washer stack to preload the cams to the tooth profile. I think that's how they make cam style index tables.
Two worms with wider grooves then the rollers is the way to go. I think. Each loaded against each other. Quite easy to do by rotating the worms at their belt flanges.
@@EdgePrecision You mean a second worm on the opposite side of the wheel you have shown? Then rotate one wom slightly farther advanced until you remove all clearance? If so, it certainly could work. The worm is the harder part to make.
Would you want to share a simple drawing so we can maybe try and model this? Maybe at least just the overall shape of it which you showed in the video? I love challenges like these. Also, the bottom of the groove must be slightly convex because that's just how endmills are made right? Isn't that an issue?
No not really because there is clearance between the ends of the cam rollers and the groove bottom. Email me at Peter.edge.pre@gmail.com I will send you the model in this video. Put roller cam in the description so I don’t delete the email.
I think you could generate preload by adjusting the rollers (every other). Look at it like pairs of rollers squeezing a rib. One non eccentric and one eccentric roller so one adjusts the preload. This also allows a small gap so there is no rubbing on the non drive side of the roller.
@@EdgePrecision Two worms would definitely allow you to preload the gear. The only problem I see with that is that both worms would have to be driven in sync ( additional motor, coupling, chain, belt, something) as you can't back drive a worm with a gear.
I think offsetting every other cam would be work, but i dont think having eccentric adjustment would be a good idea as it would be very hard to get consistency for all roller pairs and so you would get a motion error as it rotates. There will be a certain amount of flex of the follower bolt to take up the hole position tolerance. If they are bored around the perimeter you should be able to get a good position.
@@mikebrowne5152 I don't think this would be a problem using a timing belt. Like many servo motors are connected to the machines. Making the couplings adjustable so as to load the worms against each other to eliminate the backlash. This would also increase the load capacity of the drive.
@@lawmate In thinking more on this. I think there are two ways to go. Make two worms with clearance on the rollers. Then load one against the other to control backlash. This make the wheel with the cam rollers easy. Just a even roller spacing all around. The other way. If you visualize a straight track with a bearing on each side preloaded into each other. Now take that track and wind it into the worm shape. That's what the worm is. Now position the rollers in pairs on the wheel so that the worms spiral track rides between them. They don't necessary have to be in align with each other each pair could actually be angled to take advantage of the lead better. Now in fact each pair could be mounted in a rod that is fitted very closely in a bore on the OD of the gear plate. Now if they could be rotated by some means as a pair to adjust for the back lash. So there would be twelve pairs on the gear running on the worms track this way all the bearings would be always rotating in the same direction. With the first idea with two worms the bearings will have to reverse directions going thru each worm. Probably not a big problem. But I think it would be better if each bearing always rotated in it's same direction for high speed operation. I like this concept and think I will make a design based on this idea.
This is why modern computer simulations have made this kind of machining easier to to. In the past I would have to sort of guess where everything needed to be. Run it and carefully watch. If there was interference redo the setup until everything cleared in the machine. Much more stressful.
Export the solid and curves as an IGES and I would be happy to take a shot at it for you. This mechanism has always been of interest to me. I think the geometry looses fidelity going back and fourth but it’s very minor. 5 decimal vs 7 kind of deal.
@@chrispoirier384 send me an email at peter.edge.pre@gmail.com I will send you the model. Put in the description roller cam drive. So I don’t delete it.
Are you able to add a thread/helix feature around another helix feature in your software because that would be the easiest way. You'd set the 2nd helix based on the the gearing ratio and do. Set the normal of the cross section to follow the first helix and 1 corner point of the cross section to follow the secondary helix. so as the cross section follows the first helix the second helix will cause the cross section to rotate according to the gear ratio.
The problem is it isn’t a helix. You would think you could do a simple sweep with the two guide curves. But the sweep won’t work. I tried it in two different softwares. If a software could do a subtract of one of the rollers rotating around on the gear at the ratio of in this case 24-1. Just like the endmill cutting it. That would work.
I can model that in solidworks by getting creative with swept features. programming that in mastercam even with just the wireframe would be really easy with just the two guide curves. don't even need the second guide curve, just the bottom surface curve and a point at the center of the radius
Then model it and program it. Machine it and video record it. Then send me the video and I will post it for you. I would like to see what you come up with.
Peter, I can see the difficulties that the rollers add. If the point of this is accurate movement instead of brute force thru mechanical advantage, how about replacing the rollers with a precise delrin dowel? Or a steel stud with a Delrin lining on the outside? CNC ballscrews live years while interfacing with Delrin.
The B axis on this machine uses this exact same system. Just do a search for roller cam drive on TH-cam. You will see this is a way many precision rotary tables work.
I think it would be easer and less complicated, to use two of the worms on one roller gear. Loaded against each other. Often times the simplest solution is the best.
What happens if you hit the motion hold button while it's doing a pass? Does everything stop but the spindle? Big red button in the case of an emergency? cool video!
@@EdgePrecision Just pull them apart. The same way they are "stuck together" when you think of the necessary order of operations when installing the rollers, increasing the distance between center-lines loads the outtermost rollers against the groove wall. The rollers are "captured" when they are all installed and in the worm groove, pulling them apart effectively loads the rollers.
You don't have to split it, if you increase the distance between center-lines the outtermost rollers would be loaded as they have the highest angle of incidence. I hope that makes sense, a roller in the center (with the groove perpendicular to the worm axis) would pull right out without binding, but a roller at a significant angle to the worm axis would be pressed against the groove wall if you tried to pull it away perpendicular to the worm axis.
th-cam.com/video/2uHlH3H-1k0/w-d-xo.html Here is a link to a commercially made drive of this type. Look carefully at the rotation of the rollers. Look at the top of each roller. Rou can see the rotation easer there. On the middle of the worm drive the rollers are reversing direction. After thinking about this some more. I favor the idea of using two worms on one roller gear. Then loading each worm against each other by rotating them in the drive (Timing belt maybe). But the cam bearings still have to reverse direction. Or make the the screw on the worm narrower and position the rollers in pairs. Loading them into each other as if on a track. This way the bearings always rotate the same direction (Opposite for each side but always the same for each bearing). This idea I also like. But you are correct Splitting the worm does not load the bearings evenly around the gear's circumference (If that makes sense). But this almost appears in the video what is happening.
It’s not impossible but the tool would have to have the correct clarece angles. But I don’t think the machine would be able to move quick enough to make it practicable.
That is absolute cnc porn. Interestingly, (speaking of making a cnc) I went round the Mazak factory here in the UK and they use entirely Mazaks to build Mazaks 👌. A small army of integrexs and horizontals. So you have the right kit.
There isn't really as I said in the video a solid model of what I'm doing here. I can send you a .stp file of the curves I used to make my program. email me and put in the description Worm Curves so I don't delete it. # peter.edge.pre@gmail.com
Hi Peter, I have come across this once before. I believe its called a Double enveloping worm. It was on a pod on a North sea tug that had run into a submerged obstacle and had ruined the gear. The gear was bronze and was sacrificial as the worm was so complicated. I managed to model the gear and cut the teeth using a custom ground fly cutter on a horizontal borer with a mega complicated set up. Later on I found out that the pod had been made by a Japanese company that had a lead time of 24 months if it was still in business. Tug is still chugging away last I heard . Challenging job but the best ones always are. ( We didn't have any 5 axis machines at the time, in fact at that time there were no 5 axis machines in the UK).
so you cut it in a boring mill? can you be more specific?
@@fredrezfield1629 Hi Fred I wish I could, it was about 20 years ago. I do remember the gear blank was set up on a rotary table and the head was offset about 14 degs, I cant remember the dia of the fly cutter, but I do remember it was a pain cutting the teeth. It was just plunge cutting and the deeper I got the bigger the cut got, cutting both flanks at the same time. I did rough it out with a straight grooving tool first. I had to engineer some backlash into the thing as the ship was loosing megabucks a day in drydock and I only had one chance to get it right ( no test fitting). The ship was called the MV Norman. Sorry I couldn't be more specific, but it worked fine and I got a letter of commendation from the ship owners. The blank was roughly 1.1m in dia with around 65 teeth, each tooth was about 20mm deep, I cut involute teeth although the original had straight teeth ( this gave me a bit extra clearance).
Double-enveloping worm gear developed by Samuel L. Cone. Sometimes called Cone gearing.
Awesome work Peter!
Amazing work as usual Peter.
I have learned so much from you over the years. Thank you.
outstanding work Peter! absolutely fantastic.
As I always say, your the best Peter. Please keep doing it.
Love your videos, aspired me to taking machining seriously. Super cool stuff and awesome skill.
As a fellow machinist walking knee deep in post processor customization in Edgecam for Fanuc and Okuma lathes and 4 axis machining VMC-s I raise my hat! this job is some real test of skill and understanding of circumstances!
Is it possible to make your own Edge Cam pp? Heard it is a big pain to do that ore do changes in a Edge Cam PP
@@johnhansen1684 It's possible, although the user interface isn't the most user friendly. I'm working on a few post processors right now and it takes some trial and error.
@@johnhansen1684 I think the limit is the sky when we talk about edgecam post customization. all of the post macro is contained and readable in the .ctx and .mxt files. those files are compiled from the .cgt and .cgd files, which are opened in the codewizzard which give you a pretty decent customization interface. the Codewizzard files are compiled from a bounch of txt files, and by using .cgt files with .cge files on can even replace whole sections of the base macro. so basically one can rewrite the whole thing if he wants. with that said i am pretty sure that requires an awful lot of knowledge in math and macro porgramming. macro programming is never an easy task, so yeah it is pretty hard given that a 5 axis CGD file is compiled from 8-10 txt files which contain ~15k lines of macro coded by the edgecam team. the basic stuff tho is not that hard to figure out tho. also it has a pretty decent help integrated into the codewizzard.
@@ferencungvari3089 Interesting, i have played around with WorkNC but not looked into edgecam that deep. we got some edgecam license where I work would be fun to look into it. do you have any sources for information other then just trail and error?
@@johnhansen1684 codewizzard help in the menu is a goldmine. I got the luxury to watch our distributors guys doing the initial customization and built a friendly relationship with them so in case of desperation i can always have a chat with them while they show me the principles through teamviewer. I think it would be wise to chek it out if some dude put up something to skillshare or similar sites. That said the customization is always a custom problem for the given environment so if you can master the principles, you can get by. Also turning on the NC debugging functions in codewizard helps a lot. There is also a customization whiteboard iside where one can implement his own procedures than call those in the cases where neccesary alas keeping the thing professional.
Awesome work. I have thought about building a machine do make these, but i sure wish i had a mill turn like yours. Awesome work!
OK, Tis is so far over my head but to listen to your brain working and the machine working is most enjoyable. Heads like yours take machining to the nest level and beyond. Thanks for sharing.
Nice piece of work and explanation of Geometry for this worm
Well done Peter! 👍🏻
Impressive. Serious math going on with this tool path. Thanks Peter
I really like this idea! Awesome concept! 👍👍
Very cleaver design, and I like it. Consider the small amount of surface footage difference across the cam followers flat length of engagement. I have come across this situation before, but in a different way. You are very clever, and you will more than likely create your own outer race to compensate for the difference. Thank you for letting me enjoy another wonderful video of brilliance.
Peter, had I read down the comments a little further, Cory Keck's comment would have sufficed.Thank you Peter, and Cory. I'm only 73 and grateful to have my same youthful excitement about this field of work.
Bill P.
That is impressive on so many levels. I'm going to try modelling it on my CAD software just for fun. Thanks for sharing.
Send it to a local machine shop and ask for a quote just for fun lol
@@thedillestpickle i have broken xometry with parts like this before. it would give me a ridiculously low quote and i would hit go just for laughs and get a phonecall back saying they couldn't do it for that
I just modeled it. I'll see what xometry says about it.
$1900 USD
That's awesome.
I remember reading somewhere that the thread pitch on the worm is not constant and the slot is wider than rollers. This allows the roller to roll with clearance on the opposite side and the exiting and entering rollers preload against each other while the ones in the very center are actually clearance on both sides.
One way it could work is if the screw was pinched between each adjacent roller and the groove was wider. Imagen a straight track pinched between two roller bearings. Now twist that around in the worms shape. That would do it.
This would explain why the rollers seem to be barrel shaped, facinating.
very good video Peter,,thanks for your time
The techinique for producing these dates all the way back to the 1920s. It is very similar to single point screw cutting except that the tool is driven along an arc spanning the length of a workpiece which was been concave turned along the same arc. This is generally done using a rotary table which is synchronised to the spindle in which the workpiece is mounted.
congrats for your example!!
Happy to see someone with your resources thinking about a DIY version of this beautiful thing. I hasten to add that on a number of occasions watching a video of a rotary table running using this roller cam method, something was wrong in all of them. The rollers looked like a snug fit between flanks but they cannot be. Opposite surface direction. So after seeing your video , got all enthused again and laid out a 2d sketch to the following conclusion: The distance between outside edges of two rollers can be pitch X 2. Making sure , of course the crest or flank between them doesnt touch anything. Ill be watching.
Thanks Peter.
Bill Phillips, BC, Canada
Look at this video. A close up of the cam roller drive. Look closely you will see the rollers reverse direction as they pass thru the worm. th-cam.com/video/2uHlH3H-1k0/w-d-xo.html
Looks like a part for Tim Burton. Nice work Peter.
beautiful as always.thanks.😍🙏
I know a fellow colleague who keeps telling me that you need a straight up model and always trust the post for that model… Using Mastercam. Your process is a great example of having to work around the software limitations.
If I had an accurate model of the worm it would be better to program from in Esprit. Then I could use real 5 axis cuts on the surfaces. As it is I'm really only cutting to the wire curves, of the ruled surface. I can sort of make finish cuts by offsetting the cut from center-line. But its not the most ideal. Also I could put proper chamfers and finish the thread ends. If they were also modeled.
it is unbelievable the Mazak CNC machine's five-axis lathes can be programmed and perform curves thread this brilliant machinist man can program and make all kinds of parts out of a CNC machine,!
Awesome video. Thanks.
Interesting concept and certainly something that would be super challenging to make on manual machines!
The only way would be a machine dedicated to doing exactly this. Or maybe a horizontal boring mill with a rotary table on top of its normal rotary B axis table positioned such that it's center of rotation was at the worms radius. And then all that geared to operate at the gear ratio needed. from the B axis drive handwheel. Then to cut it you would only have to set the proper depth with the quill and rotate the B axis thru the 90 degrees while the worm rotates at the correct ratio. Or maybe rotate the smaller rotary table's handwheel to drive the large B axis thru the gearing to cut it. Simple don't you think?
@@EdgePrecision Yep 2 properly geared rotary tables should do the trick.
I've using CAD for 40 years, and how you would go about modelling that is messing with my head big time.
Thank you so much for sharing
Surface modeling :)
S Lunin of Spiral Bevel has mastered methodology of modeling for this family of gearing
Brother uses Roller Cam gears for the tool turret and their 4th axis. As does Sankyo. The trick is, as you suspected, tapered cam rollers in tapered grooves.
Hmmm.. so are the Brother rollers tapered larger toward the OD, which would seem to give line rolling contact? Maybe not significant at this radius, especially if they have a little crown.
Also, if you split the worm in two, could shim axially to take out clearance, or maybe could rotate one half slightly to take out the clearance, which might preserve the continuity of the arc?
Very impressive! Thx!
Instead of splitting the cam for preload, you could model two rule surfaces with slightly mismatched pitches. I believe this is how the machine tool makers do it. Incidentally, Mori has a machine (NT) with a massive direct drive servo controlling the head’s rotation. That concept hasn’t come into widespread use, probably because it makes the machine a lot bigger and more expensive. But I’ve always thought that might be the best approach on a small custom machine.
This is exactly right! - th-cam.com/video/mh9Q8p6CHx0/w-d-xo.html
Are you able to sweep cut with a solid in esprit? I've done something similar in Solidworks before by creating the spiral like you have and then driving a cylinder (endmill) around the spiral wireframe as a sweep cut. Not sure if you could do it with two rails though to control the "tool axis" of the cylinder.
So it's just like a worm gear, but with several more teeth engaging at once. So a very strong construction. This would be super useful in adding even more strength to something like a "Torsen differential"- for heavy duty applications. Very interesting. I like it.
It’s exactly like a worm gear but with rolling contact between the elements rather than sliding contact. Big reduction in wear and friction and it allows you to eliminate any backlash by preloading the rolling elements.
Very cool. Thanks for sharing.
Ive been thinking about making one of these since your B axis video. I dont think you can split it in two to achieve preload. The join would create a lump that the roller has to climb over. Tapered wont work either. The rollers must contact on one side only. Otherwise they cant rotate. The only thing I can think of is every second roller is offset to get preload. Thanks Peter for getting the thought process going on this one!
There are a couple of ways. One is two worms on one gear. The other is this. Visualize a straight track with a ball bearing on both sides loaded into each other. Now twist that track into the shape of this worm. It’s the same thing. The slot is wider than the rollers. They ride on the spiral track in between them. If that makes sense. Not on both sides. I favor the two worm idea. It is easer to preload by just rotating the worms against each other.
@@EdgePrecision that is also a great idea. Kind of like a ballscrew setup running in a worm arc. Deflection will give some backlash though I think. I hope a roller worm company sees your video and chimes in with some insight.
Some CAD packages use a feature called a tool body sweep to model features like that.
Could you grind something like this on your tool / cutter grinder?
Also, an idea for filming: I think if you used through spindle air blast with through tool coolant tools you could get the best of both worlds in terms of clear footage, and tool loading / wear. Of course you wouldn't get the lubricity that coolant provides, so maybe it would be more of an intermediate step between no coolant and flood coolant.
Amazing work as always. Thank you for taking the time to 'play' with things like this, and then share the results with us. Things like this are what advance the art in my opinion.
It's not really a shape that the tool grinder is set up to do. It almost looks like a very steep fluted tool, and you could just grind it like that. But look carefully. The flute (Groove) changes angle from sloped almost 45 degrees at the ends to vertical in the middle (Remember me putting the 1/2' dowell pin in it in the video). So you can't do it like grinding a flute. But the machine does have the same capability of motion as the Mazak. So it maybe could be done with a small diameter grinding wheel (Like the endmill) like on the Mazak. But unfortunately the machine does not have the travel range in the B axis perpendicular to the works axis. It only can travel to a positive 10 degrees in the B axis. This requires 45 degrees. This is not a problem for tool grinding but it can't do this.
@@EdgePrecision I agree. The edge of the helical slot sides is not a straight line anywhere. A grinding disc could not do it because the slot wall on the side towards the middle of the worm length is an overhang and an inside curve. It can be cut with a straight tool but not a disc. I did a lot of cutting of helix paths and the cross section of the wall was not rectangular due to the cutting being done with a disc. The slots were all below .004 inches and the tool selection was very limited. Rough with an endmill and finish grind?
We had a similar worm gear driving a cam that rotated our active sonar hydrophone receivers on our Burke destroyers. This almost looks identical, actually.
This is a old idea. The B axis on this machine is driven this way. And this machine is almost 18 years old.
@@EdgePrecision yeah lines up time wise. The Burkes came into service in the 80s really, so it's been about 40 years. I haven't been to sea in 25 years or so, I assume they're not using mechanical actuators for the arrays anymore anyways. Probably more a static phased array, makes maintenance simpler.
Surely the rollers only roll on one side if the slot width is the same as the roller? Struggling to see the benefit over a conventional worm, can’t see how this design would help reduce backlash, but probably just my lack of understanding. Amazing machining though. Do you have a plan for this, or is it a purely intellectual exercise?
Nice. Now let's see a mini 5 axis machine 🙂
This is the drive system used in Mazak variaxis c and a axis drive. A company based in Japan makes them for Mazak. They also manufacture many toolchanger gearboxes for companies like Mori Seiki and Mazak. If we break a cam follower in one of them we have to wait months to get the correct cam follower, which are ground to fit precisely in the worm.
I'd imagine you may have a similar drive to rotate your spindle.
Yes the B axis has this kind of drive. My guy that I have work on my machine (Not a Mazak person. But very good.) says that of all these type of machines in this area (there are many). That this drive is very reliable. They never have any trouble with them.
@@EdgePrecision They are very reliable. We have I think 7 vrx machines with them, 9 including our other facility. I have replaced cam followers on two, but these are machines with the 640 control that have been in full production for 16 years, 3 shifts, six days a week. Still, the reason for the replacement was a seal failure/oil leak and subsequent bearing failure. Mazak will require you to send them the machine, not field serviceable. However, we were able to talk them into selling us the parts and we did it ourselves.
It was quite a job, but the machine is currently making us money. I really would love to play with an integrex. My company just doesn't do millturn, not sure why. We have the parts, but not enough smart guys like yourself.
That is a COOL piece
Very nice Peter, I think the preload comes from the rollers, as I remember every other roller is offset so they roll on each side of the worm.
You can visualize the spiral as a straight track with bearings on each side. Now wind that into the spiral shape of the worm. That is what this is.
Rapids tend to be slower because of “in position checks” that the CNC will perform. So when the rapid move is complete, the CNC will verify its position is in tolerance of where is was commanded
Manually screwing in each of the 24 rollers through a hole in the side of the case or body of this drive system.... While also incorporating lash adjustment in the worm in two axis or on a cam'd bushings? This is gonna be very interesting. I am sure of all people, you will pull this off just fine.
I always enjoyed watching your interesting videos
What program did you use to design the Roller Cam Drive?
The cad software is SpaceClaim and the cam software is Esprit TNG.
@@EdgePrecision Thank you for answer.
I don't know because I don't use programs like SpaceClaim or Espirit, but it looks really good.
Yes, two worms with differential adjustment would be the best way to allow the clearance AND eliminate backlash.
Did you put a frilly skirt on when doing the Cam CAM?
👍 I would think you'd have to feed the cutter in both directions, reversing the worm so to maintain a climb cut on both sides of the groove.
Wouldn't the cam follower need to run in an oversized groove or one side would be counter rotating and opposing the rotation of the worm? I'm trying to understand the preload idea. Thanks Peter! You're the best!🤔
This was just a test. Yes what you are saying would be true for a real part. As well as heat treating and grinding.
you could generate a 3d model with a swept cut feature in Solidworks. You would probably need 4 guide curves for a square cut profile with one on each corner.
I was thinking the same thing and I'm playing around with that, not much success so far.
So I found a bit of success not by using guide curves but by applying a prescribed twist to the profile in the sweep cut. Though I'm just using a tapered helix path to approximate his much fancier curves.
@@makingitwithnick This is totally a surface modeling part. Was able to get the bottom face of the follower channel, don't really have a need to finish the part. Modeled a helical surface (boundary surface with helixes) and the horn-shaped surface (revolve surface). Intersection curve, did this for both corners then boundary surface between the two and you're done.
Thats a really cool idea ! It is possible that the cam followers have a curved surface (I think they call it crowning), which would help. Otherwise there might be a differential in rotational speed between the top and bottom of the roller outer surface. There will be some tolerance on the angular location of the cam followers, so you probably dont want the slot to be too good a fit, probably as you say, better to cut the slot oversize and preload axially. I dont know if the manufacturers will guarantee the concentricity of the outer bearing surface to the 'bolt' on the cam followers - perhaps they aren't usually used for such high precision requirements. I am stretching my knowledge of worm design (!! not very deep), but I would guess that your design would be alot lower friction than a standard worm, but the load capacity might be smaller (especially the double envelope type)
Another way to preload and increase load capacity is to put two worms on one gear. And pre load them to each other.
@@EdgePrecision I am trying to think whether a 'ball worm' might be possible.. a bit like a ball screw, but keep the balls on the same plane as the wheel (in some sort of carrier) and both the worm and the wheel have the same groove that would fit standard ball bearings...
@@EdgePrecision I was thinking that you could make two identical halves and preload the bearings by rotating them relative to eachother. would that work in this design? it's stretching my ability to model the movement in my head.
This is a good idea.
A crowned follower would indeed eliminate the speed mismatch but by having a single contact point instead of line contact, this would decrease stiffness, torque capacity and wear resistance.
I believe a tapered follower could be used to eliminate the surface speed mismatch through the whole contact line between the follower and worm.
Very neat. If you turned the horizontal axis in the opposite direction to climb mill, would that give a better surface finish?
This was just a test. As I said in the video. I would rough it with a smaller tool. Then do finish cuts on the walls. In that case I would climb mill them. But the real part would be steel and heat treated and ground after that.
Way above my limited knowledge. What can't you make??? "amazing"
Going to try and model this when I have some free time at work
There are some videos on TH-cam of people doing it. But it’s not a simple as it seems. Good luck.
Hi Peter,
Not sure what CAD software you're using but fusion 360 has a "sweep" operation that you can pick a profile and a path. You'd use your helix as the path and the cross section of the cut for the profile. Hope this makes sense!
Excellent job machining that especially without the full model!
Yes and I tried that in Fusion. But it doesn't work for this. I even used both curves, it still doesn't work. It for some reason distorts the surface when it wraps around this path. I even tried to just fill between the two curves by drawing lines across their ends. This will fill in SpaceClaim and Fusion but I doesn't remain a flat curve surface. So as I thought I could just give it thickness. But that doesn't work either. This shape really messes with cad softwares. So the sweep fill or extrude dont work. I think if a software could do this. I could define a gear ratio between the gear and the worm solids. Now if I had a solid projection representing the cam roller. Then I could say do a cut thru the worm with that as I rotate the worm with that rear ratio. basically it would be following my curves as you saw in the video, but the two parts would be constrained with the gear ratio. If that makes sense. Basically exactly the way the machine cut it.
@@EdgePrecision CAD tools that can model this are able to use 3D shapes (here a cylinder) as "profile".
The reason why a "flat" profile doesn't work is because of the expansive nature of the cylinder that the cutter/bearing represents when this "worm" rotates by.. parts of the cylindrical profile simply create a sort of under/overcut that a flat 2D profile is unable to produce when being dragged along a helical path..
And yeah, I know this because I tried to model something like this in Inventor with a 2D profile and run into those under/overcuts.. later versions (inventor 2022 or so I think) have 3D profile operations who can do what you originally wanted to do by using for example a cylinder as profile, but I don't have that.
There also was/is a guy who programmed this into/with FreeCAD some years ago and made YT videos of it: channel is called 'after CNC'
Video is titled: 'FreeCAD.凸輪 Video 3：Roller Cam_Dynamic Simulation'
Great video 👍 Maybe do a single lobe and test out different machining ops to get a better finish.
This was just a test to see if it could be done. No attempt was made for proper finish or dimensional accuracy. Or even the proper material and heat treat that would be required. Now that I have a idea what to do. I have different ideas on the design to improve on how it would work. Thanks.
@@EdgePrecision nice to get a design that has to be thought about and process to get a finished piece thank you!
It's called a Globoid Worm Gear, if someone hasn't already said that. I spent hours trying to model one of these in Solidworks but finally gave up. I'll have to look at your approach of creating a ruled surface and see if I can get it to work.
In the cam software I was able to create a proper ruled surface. I could not make a proper one in the cad software. For some reason having the rotations messes it up.
that's a cool worm. I'll bet if you could find a CBN grinding wheel (so it would not get smaller while grinding) you could probably finish grinding on the mazak. I'll bet that super light grinds would do it. Of course that's coming from a hobbiest, so what do I know.
I think if you really wanted precision the Mazak is too far away with its rotational centers. It would really need a special fixture setup for the grinding. I have a idea what could be done. It would have to be made but it’s not to complex.
A purpose built machine with a large rotary table, slide and spindle on it bolted down to a base plate and geared to a dividing head/tail stock might be able to manually do it to some degree.
I saw a video on TH-cam doing just what you are describing. But I could not find it.
Man oh man, that toolpath is slicker than frog shit on wet concrete! Helluva job!
Well that is certainly an interesting Chinese Puzzle piece!
Could you grind it on your tool & cutter grinder? It is similar to a drill or endmill flutes.
(Yea, I know you wouldn't want to do that will aluminum.)
Drill and endmill flutes follow a cylinder shape. This shape is following a totally different shape. The edge of a sphere. Also the lead has to vary based on the rollers location projected perpendicular to that sphere/circule.
Keeping the machine out of G0 is faster since it's not waiting for 0 velocity at the end of each block! Clever little cheat right there 👍
Shouldn’t the endmill length from the B axis center match the minor diameter of the root? Seems like too much movement on the Z, but final assembly with a drive operating in torque mode and a hydraulic brake when necessary makes sense. Really cool!
Because the machines B axis center of rotation isn’t in the same place as the cam roller center. The Z has to move with the B axis to rotate at the cam roller center.
Almost easier to program that by hand.
Would alternate the roller position on the gear so they roll on different sides of the worm. Could use 12mm rollers for this, could starfish the gear for mechanical preload
I wonder how to grind those surfaces... maybe a gear wheel like the one you show, but instead of roller a grinder wheel in one of the roller spots. But that will make the grinding super special, complex and expensive; but very doable. Maybe rollers are not needed if the worm and gear are driven by two motors fully controlled, like a CNC, but just to axis.
The grinder only has to sweep the length of the screw. So if it pivots on the wheel center it would only have to swing about 90 degrees. So the worm wold need to rotate six revolutions. So how to do that with the most precision is the question. I’m thinking a Precision lead screw that moves a linkage 90 degrees for 6 revolutions. I can think of a way. That could be very precise. So the grinder could be a tool post grinder with a small ID wheel. Maybe.
As far as the CAD goes, can you make a loft cut? In Alibre, you can define a 3d path for your spiral, and then position sketches along it of the groove's cross-section. If the CAD goes find you worthy, it should just interpolate from sketch to sketch, virtual cutting out the grove as it proceeds. Mind you, Alibre frequently squicks out and adds weird twists if the sketches aren't to its liking.
I my software SpaceClaim I tried everything I could think of. I also tried Fusion with no success. I tried a loft. I tried a extrude. I tried to use the two curves as guide curves and a section translate. It all doesn’t work. There must be a way. I just don’t know how.
@@EdgePrecisionHere is a possibility, though I'm afraid it may be more work than it's worth. In OpenSCAD you could define the roller wheel, then intersect that with your worm model to leave the shape you desire. It would be a fair bit of coding (OpenSCAD is billed as the programmers CAD) but I believe it could do it.
@@ahfreebird This is what is necessary (I think). If first a gear ratio between the solid of the worm and the disk with a solid projection like the roller. Then be able to rotate them thru each other so the roller projection cuts thru the worms solid at that ratio. almost the same as the machine cuts it. But I don't know if a software could do that.
Some thoughts: Cam rollers are crazy strong. Maybe you only need two or three to engage at once...? Here's the idea: Maybe you could add a second wheel and then clock it ahead (or behind) so that you have rollers located in between where you currently show them. Then widen the cam groove and have two rollers run in the same groove and by clocking the wheels you can change the effective pitch take up the backlash or even pre-load. ONe wheel (and rollers contact one wall of the groove while the other is turned back to contact the other wall. I have no idea what your space constraints are but a wider groove and move half of the cam rollers to a second wheel might get you somewhere. Cool problem. I'm glad it's not mine. Cheers and good luck.
Edit: Maybe you make the cam in the reverse profile so that the tooth protrudes. Then you have your wheel as you have shown, and you then mount a second roller to pinch the tooth that protrudes. You can make a grind spacer or some belleville washer stack to preload the cams to the tooth profile. I think that's how they make cam style index tables.
Two worms with wider grooves then the rollers is the way to go. I think. Each loaded against each other. Quite easy to do by rotating the worms at their belt flanges.
@@EdgePrecision You mean a second worm on the opposite side of the wheel you have shown? Then rotate one wom slightly farther advanced until you remove all clearance? If so, it certainly could work. The worm is the harder part to make.
Would you want to share a simple drawing so we can maybe try and model this? Maybe at least just the overall shape of it which you showed in the video? I love challenges like these. Also, the bottom of the groove must be slightly convex because that's just how endmills are made right? Isn't that an issue?
No not really because there is clearance between the ends of the cam rollers and the groove bottom. Email me at Peter.edge.pre@gmail.com I will send you the model in this video. Put roller cam in the description so I don’t delete the email.
@@EdgePrecision I see, thank you for answering, that is a pretty interesting piece! Email sent!
AAPSOLUTLY BRILLIANT
I think you could generate preload by adjusting the rollers (every other). Look at it like pairs of rollers squeezing a rib. One non eccentric and one eccentric roller so one adjusts the preload. This also allows a small gap so there is no rubbing on the non drive side of the roller.
Two worms loaded against each other is easer and does the same thing.
@@EdgePrecision Two worms would definitely allow you to preload the gear. The only problem I see with that is that both worms would have to be driven in sync ( additional motor, coupling, chain, belt, something) as you can't back drive a worm with a gear.
I think offsetting every other cam would be work, but i dont think having eccentric adjustment would be a good idea as it would be very hard to get consistency for all roller pairs and so you would get a motion error as it rotates. There will be a certain amount of flex of the follower bolt to take up the hole position tolerance. If they are bored around the perimeter you should be able to get a good position.
@@mikebrowne5152 I don't think this would be a problem using a timing belt. Like many servo motors are connected to the machines. Making the couplings adjustable so as to load the worms against each other to eliminate the backlash. This would also increase the load capacity of the drive.
@@lawmate In thinking more on this. I think there are two ways to go. Make two worms with clearance on the rollers. Then load one against the other to control backlash. This make the wheel with the cam rollers easy. Just a even roller spacing all around. The other way. If you visualize a straight track with a bearing on each side preloaded into each other. Now take that track and wind it into the worm shape. That's what the worm is. Now position the rollers in pairs on the wheel so that the worms spiral track rides between them. They don't necessary have to be in align with each other each pair could actually be angled to take advantage of the lead better. Now in fact each pair could be mounted in a rod that is fitted very closely in a bore on the OD of the gear plate. Now if they could be rotated by some means as a pair to adjust for the back lash. So there would be twelve pairs on the gear running on the worms track this way all the bearings would be always rotating in the same direction. With the first idea with two worms the bearings will have to reverse directions going thru each worm. Probably not a big problem. But I think it would be better if each bearing always rotated in it's same direction for high speed operation. I like this concept and think I will make a design based on this idea.
Nice, tenx, ✌️
No matter how much checking and double checking, that must catch your breath when the "Y" head swings toward the cabinet at speed.
This is why modern computer simulations have made this kind of machining easier to to. In the past I would have to sort of guess where everything needed to be. Run it and carefully watch. If there was interference redo the setup until everything cleared in the machine. Much more stressful.
шикарно )
wow, very nice
In Creo I think you could model this easily once you have the curves using variable section sweep.
If the software could do a subtract with a roller/cylinder following the curves I have already drawn. My software wont do that.
Export the solid and curves as an IGES and I would be happy to take a shot at it for you. This mechanism has always been of interest to me. I think the geometry looses fidelity going back and fourth but it’s very minor. 5 decimal vs 7 kind of deal.
@@chrispoirier384 send me an email at peter.edge.pre@gmail.com I will send you the model. Put in the description roller cam drive. So I don’t delete it.
Is it possible to request a small sample production of roller cam gear?
I just did this for a demonstration in a TH-cam video. To show the toolpath required to machine such a part. There is no production of this.
Dang, youtube has scrambled your video for me. I love your channel but I guess I'm going to miss this one.
I don’t understand how that could happen. No one else has complained. It must be something on your end.
@@EdgePrecision I'm sure it is, they don't like me on youtube. Looks like an interesting project too.
@@NikColyerMachineWorks Just press refresh
@@calholli I wish. I pushed refresh 5 times before I gave up. Thanks anyhow.
Are you able to add a thread/helix feature around another helix feature in your software because that would be the easiest way. You'd set the 2nd helix based on the the gearing ratio and do. Set the normal of the cross section to follow the first helix and 1 corner point of the cross section to follow the secondary helix. so as the cross section follows the first helix the second helix will cause the cross section to rotate according to the gear ratio.
The problem is it isn’t a helix. You would think you could do a simple sweep with the two guide curves. But the sweep won’t work. I tried it in two different softwares. If a software could do a subtract of one of the rollers rotating around on the gear at the ratio of in this case 24-1. Just like the endmill cutting it. That would work.
I can model that in solidworks by getting creative with swept features. programming that in mastercam even with just the wireframe would be really easy with just the two guide curves. don't even need the second guide curve, just the bottom surface curve and a point at the center of the radius
Then model it and program it. Machine it and video record it. Then send me the video and I will post it for you. I would like to see what you come up with.
@@EdgePrecision haha i will see if i can find the time
Peter, I can see the difficulties that the rollers add. If the point of this is accurate movement instead of brute force thru mechanical advantage, how about replacing the rollers with a precise delrin dowel? Or a steel stud with a Delrin lining on the outside? CNC ballscrews live years while interfacing with Delrin.
The B axis on this machine uses this exact same system. Just do a search for roller cam drive on TH-cam. You will see this is a way many precision rotary tables work.
Any idea how does the grinding process works after this ?
The grinding for this would take a special setup. But I now have a different idea for this. But that also would need a special grinding setup.
Were there is a will there is a way. Fantastic job well done.
There's a _lot_ going on here; I'll have to watch this again at least once, and have a real good read through the comments too.
Could you offset 2 rollers on each pin for no backlash?
I think it would be easer and less complicated, to use two of the worms on one roller gear. Loaded against each other. Often times the simplest solution is the best.
What happens if you hit the motion hold button while it's doing a pass? Does everything stop but the spindle? Big red button in the case of an emergency? cool video!
Yes everything stops. Remember this is using the C axis to rotate the turning spindle. So it’s not a treading cycle.
At 4:09 is there a way to capture the 'model' the simulation produced and put that back into CAD? I'd love to print one in plastic to play with. 🙂
That may be possible. I think I can save a stl file of the simulation result. I will have to check into that.
Do you use inverse time on your rotary feed?
I don’t even know what that means? The cam software applies the feed in relation to the radius that the tools tip is from the center of rotation.
@Edge Precision is the code using G93 or G94 with the rotary axis?
Which cam software use?
Esprit TNG.
What CAM system are you used?
Esprit TNG.
It would seem that one way to obtain zero backlash would be to have another worm which could be used to just rotate enough to to take up any slack.
I think this would be good and also increase the load capacity.
@@EdgePrecision Just pull them apart. The same way they are "stuck together" when you think of the necessary order of operations when installing the rollers, increasing the distance between center-lines loads the outtermost rollers against the groove wall. The rollers are "captured" when they are all installed and in the worm groove, pulling them apart effectively loads the rollers.
You don't have to split it, if you increase the distance between center-lines the outtermost rollers would be loaded as they have the highest angle of incidence. I hope that makes sense, a roller in the center (with the groove perpendicular to the worm axis) would pull right out without binding, but a roller at a significant angle to the worm axis would be pressed against the groove wall if you tried to pull it away perpendicular to the worm axis.
th-cam.com/video/2uHlH3H-1k0/w-d-xo.html
Here is a link to a commercially made drive of this type. Look carefully at the rotation of the rollers. Look at the top of each roller. Rou can see the rotation easer there. On the middle of the worm drive the rollers are reversing direction. After thinking about this some more. I favor the idea of using two worms on one roller gear. Then loading each worm against each other by rotating them in the drive (Timing belt maybe). But the cam bearings still have to reverse direction. Or make the the screw on the worm narrower and position the rollers in pairs. Loading them into each other as if on a track. This way the bearings always rotate the same direction (Opposite for each side but always the same for each bearing). This idea I also like. But you are correct Splitting the worm does not load the bearings evenly around the gear's circumference (If that makes sense). But this almost appears in the video what is happening.
What software is that
The cad is SpaceClaim and the cam is Esprit TNG.
I wonder how it would turn out using a single point cutter turning it? Is it possible?
It’s not impossible but the tool would have to have the correct clarece angles. But I don’t think the machine would be able to move quick enough to make it practicable.
@@EdgePrecision Thanks for the reply. Love your videos.
That is absolute cnc porn. Interestingly, (speaking of making a cnc) I went round the Mazak factory here in the UK and they use entirely Mazaks to build Mazaks 👌. A small army of integrexs and horizontals. So you have the right kit.
Your tool and cutter grinder wouldn’t work to grind it with it?
That might be possible. But I would have to figure a way to program it. That would be the problem.
@@EdgePrecisionConversational machines are wonderful until you need to do something they weren't intended to, aren't they?
@@ahfreebird In thinking about this. The machine doesn't have the travel range in the B axis to do it.
What is the machine?
Mazak Integrex e650H
The Brother T200 rotary table roller gear:
th-cam.com/video/8K8dytXUGR0/w-d-xo.html
exactly, brother have been using cam rollers on their ATC turrets for nearly 30 years.
Как правильно называется этот редуктор?
I wouldn't have come up with the idea to combine roller cams and worm drive in a thousand years. Crazy...
Плохо, что нет перевода.Как убежали, по технологии обработки, далеко от нас. А я до сих пор работал на станке 75 года прошлого века.
Any chance You can share 3d model or TNG project file? I'd love to make the same part on my 5ax mill!
There isn't really as I said in the video a solid model of what I'm doing here. I can send you a .stp file of the curves I used to make my program. email me and put in the description Worm Curves so I don't delete it. # peter.edge.pre@gmail.com