Holding brake needed on a 4/5th axis when using a harmonic drive?

I've been designing a small-ish 5axis based around a bt30 spindle driven by a 2.5hp servo for a little while now. I have already acquired some fairly beefy used Harmonic drives for the 4th and 5th axis and I am having trouble figuring out if it is necessary to incorporate a holding brake.

Since they are supposed to be zero backlash I don't have the need for a holding brake to reduce chatter. If the cutting forces are high enough to require a brake, I would probably already be at the point of damaging the Harmonic drive. At least, that is what I am thinking. Am I completely off?

Here are some specs of the stuff I've already picked up for reference:
10k BT30 spindle
1.8kW Servo for spindle
CSF-40-50 for the 4th axis
CSG-32-50 for the 5th axis

I have zero practical knowledge when it comes to these Harmonic drives.

I have a DIY harmonic drive 4th axis and haven't found it to be the limiting factor in overall rigidity of my admittedly small-ish machine & setups. I don't have a brake on mine, but I'm using a NEMA34 stepper which has very good holding torque.

Having said that...

I've read - no first hand knowledge - that Haas and some other MFGR's have generally chosen other reduction mechanisms because the harmonic drive can have torsional flex in the flex spline. True, they have effectively zero backlash, but the flex spline is not as stiff as a cycloidal or double-worm drive. I suppose that the flex can contribute to chatter or broken tools under extremly high loads - imagine drilling with a spade drill way off center.

And unlike worm drives, harmonic drives can be back-driven, especially the lower ratio versions. I don't know if cycloidals can be back-driven.

This is why I chose to use a drive size which is significantly larger than is required for my size 4th axis. I probably could have used a 17mm size, but I went with a 32mm - it's about 10x stronger than I need to whip my steel trunnion table around. But it's also stiffer, torsionally, than the smaller drives.

The one reason I can think of where a brake would be seriously beneficial is when a servo is used to drive the harmonic (or other) gearbox. It's possible for servos to 'dither' or hunt in between positions, or if the encoder position tolerance is very high, then the drive could be back-driven a little bit by cutting forces before the servo tries to recover the position. Here's a case where an axis brake would put a positive stop on everything. In contrast, a servo-mounted brake would stop dithering/hunting and back-driving, but wouldn't help with the flex spine being twisted.

In my situation I will eventually add a brake - but to the tailstock. I want the ability to add some drag or be fully locked - because even though the harmonic drive isn't moving or flexing, the trunnion table is still free to flex and rotate around the tailstock.

If you're using a big stepper then those two harmonic drives you've selected will be massive overkill from a pure strength standpoint. If you're using a servo, then only experience will dictate if your particular servos, gear ratios, and system tuning will hold the axis in position well enough without a brake for your setups. Just try to keep heavy cutting forces as close to the center of rotation as possible - the B-platter should be lower than your A-rotation center so the expected part center is in line with A once fixtured.

That's a good point, I wasn't thinking about flex in the drive itself.

I'll also be going with DMM AC servos for everything.

Thanks so much for your input, I'm definitely using brakes now. At the very least there's no reason to put stress on the HDs when not in motion so the brake should help with that and any simultaneous machining will be limited to lower cutting force finishing operations for the most part.

It's only a 2.5 hp spindle so I won't be using any huge tooling but I'd say it's better to be safe.

Thanks again.

You aren't going to stress the HD with those servos. Based on the spindle motor choice I'm guessing you have a G0704 to Tormach 1100 sized machine. If this is the case, you aren't putting giant multi-kw servos on the axis drives - your axis servos will stall and error out before you damage the HD. I wasn't kidding earlier about your choice of HD's - look up the load ratings on the 40 and 32-series drives. Those are used on robot arm joints capable of hundreds of kilos at huge extension lengths.

I'd suggest designing a brake for the A-axis spindle, but don't bother spending time & money building it until after you see if there's an issue. Assuming you're going with the DMM 750W Nema 34's (appropriate size for 6"-8" 4th axis), you can get them with built-in 24VDC brakes. Buy one of those and see if it's good enough to stop any servo hunting you may experience.

And if you don't need it, use that one for the Z-axis to stop the head sliding down.

The reason I'm appearing to contradict my earlier statement is that unless designed very carefully, a brake will influence the position of the axis when applied. A disc brake is significantly less likely to move the axis than a drum brake in this regard.

A gentlemen on this forum has recently designed & built a rather nice 4th-axis + mill-turn (lathe) spindle combo. He NEEDS a brake, but that's because the only thing holding the spindle is the servo and a comparitively low pulley ratio - something like 2:1. You will have a 50:1 torque increase - a 25x increase over his arrangement. I really don't think you need anything like his sweet disc brake - just (maybe) a servo brake easily controlled by the servo drive. No extra parts, no extra air lines.

Do a search for "mill turn spindle" here and have a look at his setup. He needs a brake.

Then have a look at my setup. I really don't need a brake, except maybe a servo brake when I get around to switching from the stepper.

The difference is that he can get lathe RPM and I cant - but I can get away with 10x less motor.

https://www.cnczone.com/forums/attac...d=425602&stc=1https://www.cnczone.com/forums/attac...d=425604&stc=1https://www.cnczone.com/forums/attac...d=425606&stc=1https://www.cnczone.com/forums/attac...d=425608&stc=1

I've been brainstorming for a while and I really appreciate the suggestions, it's easy to get stuck on things.

I really like the idea of leaving a spot for brakes and add them if I happen to need it. I still don't have a concrete design yet but I guessed my trunnion would end up in the 200-250lb range with a center of mass somewhere between 3-5" off center. I figured worst case scenario I was around 125ft-lbs so that's why I went with a 40 series HD.

You are also dead on in that I will going with a 750w for the A and a 400w for the C axis.

This is a very rough design but this is where I am at for the moment, just for reference.

https://www.cnczone.com/forums/attac...d=425616&stc=1

18" X - 24" Y - 15" Z travels
Each piece will be welded and machined and then bolted together. None of our machines at work are anywhere near big enough so I need to work with smaller pieces, which has been that hardest part to design around.

Another thing I'm really hoping to incorporate and being able to mount an optional ~18 x 24" table over the trunnion and then I can run it as a 3 axis with larger vises and parts.

I like the wine rack ATC. Especially because you could put an umbrella or chain-drive in the same spot and not have to deal with moving it in to position - just rotate to the tool slot.

Some thoughts - and I understand you're in the spit-ball phase:

1. Optimize it for 4/5 axis work and don't compromise it by trying have a modular 3-axis table in there. Lifting a multi-hundred pound 'table' in and out just to run some bigger pieces - yuck. Get a Bridgeport for the 'dumb' stuff.
2. The head is way larger than it needs to be. Looks like you're going after the DMG-style 'gantry' type of thing. That's good, but consider building the spindle cartidge holder out of heavy-wall tube/pipe with a flange and just welding it to a plate; trucks and ballscrews go on the back. Cuts way down on the big 'box' thing that may interfere with tool/holder clearance during 5-axis moves.
3. Gantry. Think about the cutting forces and maybe go with more of an A-frame looking thing rather than a big box tube. Use smaller size tube framing and bolt plates to it to stiffen things up along the expected load directions (up/down, front/back, twisting). Also think about how much the head is going to be extended and unsupported when a short tool is down near the platter. Tha'ts a lot of leverage on the head and gantry.
4. You have a ton of X & Y compared to your platter size and stated intended part size. Side forces on the part away from the platter center will be pretty high. This may be a case where instead of a brake, you install an air-driven locking pin in the platter & trunnion. A tapered pin driven up in to the platter bottom near the edge would be used to lock it down hard during anything but continuous platter moves. You'd have a limited number of indexing points, but depending on diameter you should be able to use a 1/2" pin and get 22.5 degree (or 10 or whatever) increments. Similar idea to a collet indexer used on a mill, and a double-acting pancake cylinder should be able to yank a short pin out of the hole pretty positively in case there's a bit of binding due to rotational loads. I'd definitely use some hard-as-rock drill bushing for the pin guide, and see if anyone makes hardened conical inserts - check the usual fixturing suppliers.
5. Frame. I don't see any point in having the travels extend more than an inch or so past the platter center of rotation. I realize you can envision tilting the A-axis forward with a tall part and need to get out to the end of it, but if you think about the tool and approach strategies you can probably rotate the platter for access. And I think that most of your Z travel is for tool clearance - not because you're realistically going to be clamping a 12" tall part on the platter and tilting it forward and back 90 degrees.

If you cut down on the frame forward of the trunnion you'll have better loading/unloading access, easier reinforcement of the trunnion end mounting points, and you can triangulate the ends down. Look at it this way: you need the Z to get all the way down to the platter surface, minus some probe or tool length. That's your Z-bottom. Then add typical fixture height, and then part stock height. Then tool clearance for the longest thing (drill or long narrow access end mill holder), then tool taper & pull stud clearance. You'll find you can probably get away with 8" stock in the fixture at best.

But you only need the head/slides/gantry/mount really robust at positions between the bottom of travel and the point where the longest tool contacts the tallest stock ('tallest" also includes wide stuff when the trunnion and platter are rotated). This may change how your head is mounted to the gantry, and where you put the most bracing or whatever.

Looking forward to developments.

1) Definitely agree here, 5 axis is the priority. I'm the only one at work who's allowed to touch our Brother MX140 and while it's a really nice machine, not being able to do simultaneous 5 axis is what motivated me to play around with something like this. If I did do a table it would just be a couple rails with a fixture plate sat on top of it, nothing super heavy.
2) I'm with you again, the head is just a carry over from a previous idea and have yet to really refine it. I'm mostly dealing with the arrangement of the spindle servo and ATC pneumatics. The servo maxes at 3500rpm so I need to do some gearing with pulleys to get that up higher which is why I may need a little more room otherwise. But, I do like the idea of finding some 6" tube and just using that.
3) You're right again, I really shouldn't have shown you where I was at since it is so unfinished. I've really only had time to figure out the base framing and general arrangement/geometry. I'm to the point where I've started thinking about the gantry a little, figuring out the tool change and making sure I have places/clearance for some bellows. I'm also hoping to minimize the use of tubing at least in the gantry and head to prevent any resonance.
4) The larger than strictly necessary travels are another issue I have with our Brother. There's been a few jobs where it would have been so much easier to have like another inch of travel. I'd hate to be in the same predicament here and at this scale another few inches of travel doesn't really cost any more. You have some interesting idea there for indexing and I'll definitely mull that over.
5) Well I finally found something to slightly disagree with. Not that it wouldn't work, but it would really bug me to not be able to reach my entire work envelope from the top just to save a few inches of Y travel. It would add a lot more repositioning and just generally complicate programming. I am still working on being a little more realistic with work envelope and sliming down the travels to increase rigidity, but it's a struggle. The 15" of Z travel, again probably isn't necessary. I doubt I will have the rigidity for super long tool stickouts but I do think it would be fun to play around with larger sculptures and what not out of lighter weight materials. Again, I might need to get more realistic.

I'm at around 70" overall height with the head up right now and I'm worried the trunnion itself will be a little too low to the ground. I can go a little taller but I'm probably looking at some combination of reducing Z travel, raising the trunnion up, and making the gantry a bit taller. I've limited myself to any one part being no longer than 36" for machining purposes which is where I am at for the height of the side walls.

I still have so much work and it doesn't help that I keep changing my mind. You have given me a ton of great suggestions.

It doesn't help that I've also been designing and building a small envelope, sub micron resolution lathe for work to turn down Xylan coating so my brain is a little burnt out.

Your Z-travels probably are necessary, maybe even more is needed. I guess what was inartfully trying to express is that you need plenty of travel to clear parts, fixtures and still have room for tools & holders. But that when you start refining the design it doesn't need to be super-rigid through the upper half of the travel - and the revese is true for the lower half of the travel.

I see your point about the Y-travels. Why fuss over a 5-axis program when all you need is a 3-axis, right?

Mori DMU 40 or something. Gantry appears to be just above the work envelope, as is the lowest point where the Z-axis is supported. Nothing says you can't extend the Z-axis up and make the top trucks higher on a triangulated/cantilevered thing. Gets you travels up high and rigidity down low in the cut.

And your gantry can be made from welded plate, as can the X-slide.

https://proxy.duckduckgo.com/iu/?u=h...Vo_VX8.bmp&f=1

Oh ok, I totally missed your point about increasing Z travel. The picture I posted doesn't give a good sense of that so here's one from the front.

https://www.cnczone.com/forums/attac...d=425638&stc=1

I've been planning around an ambitious/worst case scenario 12in cube work envelope so with 15in of travel and with the longest bt30 holder I can pretty much machine the trunnion and clear all but the biggest workpieces. But then I can probably add 2-3" of Z travel, not introduce too many rigidity issues, and still be within my 36" max machining envelope so maybe I'll do that.

I did have a design at one point where the gantry just barely cleared the work envelope, not sure why that got shelved. It's a good idea. I can raise my trunnion a little further off the floor leaving more room for chip pans and coolant trays and beef up the gantry a little to compensate for the increased height.

I've bounced back in forth a few times between a Haas UMC-750 style machine and a gantry type. I think I'm going to settle for this because for one, I'm having an easier time designing it out of smaller pieces and two, I think it's just more rigid.

This has been so helpful, I think I have some tweaks to make. Hopefully I can come back with something more impressive here in a few weeks.

Hi Shin - A comment on the framework. Having so many members will mean having many vibration modes. Its usual to divide"spaces" up evenly like you have BUT this means you also create a shape that has a natural strong vibration made. So break spaces down with triangles (triangles vibrate less then squares and rectangles) or use an uneven spacing so each member has many modes vs a preferred mode. Can I ask why the corner posts are made from an open section then filled in with shear webs? Peter

Ok, I hadn't really considered optimizing the tubing configuration. I'm kind of stuck with tubing for the base due to practical reasons, I just kind of assumed it wouldn't matter too much. I didn't really like how it was anyways.

As for the corner pieces, I was trying to avoid using a bunch of flanges to assemble everything so the idea was to cap each piece of tube and have some tapped holes. I could then have through holes on the corners and just bolt them together. Those weren't really any weaker than 1/4" wall tube in my simulations. It seems a little easier for final machining (not sure why) and I like how it looks. I admit it's probably overly complicated, but I need practice welding. With the ways things are going I'll change my mind soon anyway.

Stop criticizing my awful design, I just wanted to know about brakes. :D

Quote:

---

Originally Posted by shinji2k

Stop criticizing my awful design, I just wanted to know about brakes. :D

---

You steped in this one, buddy, when you hinted at something interesting to people with nothing better to do than Cletus McDingDong FEA's.:rolleyes:

Your design isn't awful, just needs refining. But you've got about 10.2 jillion steps to go before a brake or indexing shot pin comes in to play...

As to the framing, just use square tube and weld it all up - don't bolt it. Grind the welds flush and have it stress-relieved. Then bolt on side panels and grout them in with epoxy - do the same for all the other precision surfaces. You'll basically make a stiff but out of parallel/square/whatever frame, and all the precision surfaces will be screwed to that and grouted with epoxy. You do one surface at a time and nudge it in before the epoxy sets. Once the epoxy sets you torque all the surfaces/plates down and then bolt on the moving elements.

Easy, cheap, just takes some care during assembly. And if you have access to a decent sized surface plate it becomes significantly easier - the frame Y rail surfaces are done upside down as a pair so the rails are completely parallel in Z. Same-same for the X-rails mounted on the gantry.

And if you want to dive down another rabit hole, look in to filling the voids in the frame (between stiffeners) with epoxy/polymer granite.

Hmmm, I do like the idea of starting with the gantry. We have some pretty large surface plates at work. I could make the gantry, get it all flat and level and square and use that and my machine level to shim up and align the Y axis rails. Getting the trunnion mounted would be the only tricky part, but I could use the same process to shim up the mounting plates and just indicate it all in with the existing axes.

Quote:

---

Originally Posted by spumco

You steped in this one, buddy, when you hinted at something interesting to people with nothing better to do than Cletus McDingDong FEA's.:rolleyes:

Your design isn't awful, just needs refining. But you've got about 10.2 jillion steps to go before a brake or indexing shot pin comes in to play...

As to the framing, just use square tube and weld it all up - don't bolt it. Grind the welds flush and have it stress-relieved. Then bolt on side panels and grout them in with epoxy - do the same for all the other precision surfaces. You'll basically make a stiff but out of parallel/square/whatever frame, and all the precision surfaces will be screwed to that and grouted with epoxy. You do one surface at a time and nudge it in before the epoxy sets. Once the epoxy sets you torque all the surfaces/plates down and then bolt on the moving elements.

Easy, cheap, just takes some care during assembly. And if you have access to a decent sized surface plate it becomes significantly easier - the frame Y rail surfaces are done upside down as a pair so the rails are completely parallel in Z. Same-same for the X-rails mounted on the gantry.

And if you want to dive down another rabit hole, look in to filling the voids in the frame (between stiffeners) with epoxy/polymer granite.

---

After tightening the bolts, will the precision get worse? Because epoxy which now acts as the underneath material is not so hard as steel. When it sustains the localized strong forces at those bolt/screw areas, it may deform itself downward a little(especially when the parts are linear rails or linear actuators which are not a so big strong parts)?!

Some real-world info.
You problem has nothing to do with strength and everything to do with torsional rigidity on 5 axis effects.

Simple milling or drilling, at the edges of a 2-axis trunnion workpiece, say 250 mm cube == 10", == 5" from trunnion center, can easily create torsional forces of several hundred Nm, on both axis.
While only using the power of a basic drill == 1/2 hp power output, give or take, aka 8 mm end mill.

Look at the torsional strength aka rigidity of a Haas Trunnion, 2 axis, the smallest one.
Look at the strength of the locking in Nm.

Your trunnion will twist like a pretzel on both axis 4 and axis 5.

Quote:

---

Originally Posted by hanermo

Some real-world info.
You problem has nothing to do with strength and everything to do with torsional rigidity on 5 axis effects.

Simple milling or drilling, at the edges of a 2-axis trunnion workpiece, say 250 mm cube == 10", == 5" from trunnion center, can easily create torsional forces of several hundred Nm, on both axis.
While only using the power of a basic drill == 1/2 hp power output, give or take, aka 8 mm end mill.

Look at the torsional strength aka rigidity of a Haas Trunnion, 2 axis, the smallest one.
Look at the strength of the locking in Nm.

Your trunnion will twist like a pretzel on both axis 4 and axis 5.

---

Yeah that trunnion was pretty crap and was more of a stand in. Here's where I am at now. I'm still tweaking as I go but I'm much happier with it.

https://www.cnczone.com/forums/attac...d=428718&stc=1
https://www.cnczone.com/forums/attac...d=428720&stc=1
https://www.cnczone.com/forums/attac...d=428722&stc=1
https://www.cnczone.com/forums/attac...d=428724&stc=1

Quote:

---

Originally Posted by raychar1234

After tightening the bolts, will the precision get worse? Because epoxy which now acts as the underneath material is not so hard as steel. When it sustains the localized strong forces at those bolt/screw areas, it may deform itself downward a little(especially when the parts are linear rails or linear actuators which are not a so big strong parts)?!

---

Yep. But if you keep the epoxy very thin - i.e. file/grind/mill the underlying surfaces fairly flat - epoxy deflection will be lost in the noise. With a concentrated point load the epoxy isn't going to hold up, but when the forces on the mill head are transferred through the gantry and then to the trucks and finally to the rails, the forces are spread out quite a bit. The load on the epoxy interface is broader and epoxy is quite strong/stiff in compression.

Then the limiting variable is the frame rigidity. The frame will be strong - but very much a tuning fork. I would guess that with unfilled tubes this thing will sing and howl like a cat with a firecracker in its bum.

Which is why filling the tubes with something stiff would be beneficial. It will help - not cure - vibrations and help the frame settle and reduce flexing. But the primary objective is to keep the frame from flexing so much that the trucks don't bind up.

A filled-tube frame is far from ideal, but this is not a professional machine tool builder. This is a hobby-type project and helping somebody optimize their design given thier limitations on tools, materials, and skills means not getting all OCD on perfection.

Will this be perfect? No, of course not.

If he wanted it significantly better he'd just get some surplus granite surface plates or countertops and have them cut up. Then glue in some inserts and bond it all together in to a ferociously rigid frome. But if you don't have a decent headstone/monument shop nearby or the rigging equipment to deal with multiple hundreds of pound of granite lumps - what's the point of spitballing a completely different design?

Will it work? Yep. And the OP, if this thing ever gets built, will work out what the actual limitations of using it are.

Oh there will definitely be resonance issues. But then I have had good luck in the past dampening with a loose fill like sand or the crushed oyster poultry feed stuff if too much weight is an issue. The gantry and trunnion will definitely get some crushed oyster and the base will get sand as needed.

Does anyone use epoxy as substrate for the linear rail to sit on? .. like for 15mm/20mm rails.If it is of 1mm thickness, it is difficult for so thick of it to fill all well. So, the method will like ...after adding the shims underneath for precision leveling adjustment of the rail position, boundary pieces added on its four sides for trapping the being poured resin, screwing out and rail is taken out, fill a layer of epoxy resin, put the rail on and screw the bolts again.. But, will the shims slightly dislocated during these operations and also the two times of screwing operations will all make the precision can't be kept?

1 mm is thick.
The point of the epoxy is to 'fill in any gaps' and prevent flexing. Start with a good flat surface before you add the rails. Then an epoxy thickness of 0.1 - 0.2 mm is enough.

Cheers
Roger