Electronic gearing for gear hobber

[memphisjed]

Plenty of controls with built in e gearing (canned ladder logic) have 5 and more axis abilities. These are not dynamic motions when you need 5 axis simultaneously so you can get away from not needing heidemans. The 840 can not do that without moving to Modbus (Fieldbus in Siemens speak).
If it is slow enough for cog issues then a reduction is needed, less than a servo.

Garwood has old machine, this is not a question if new bigger production machine can do more faster- it is can you retro a old machine to work good enough at reasonable price. Is it worth it is different question. I shaved ten thousand +- off his price; so hotdogs with slaw on him.

 

[Zahnrad Kopf]

Belt drive the spinning hob is smart, you introduce a vibration dampening unit close to the cutter.
These speeds, honestly, servos are a waste of money and gain in headaches.

That would be a huge error. Maybe one can "get away" with it on some small Fine Pitch gear in a mild material, but one would never want to do it with anything even *mildly* medium to large Pitch. Much less, with a decent sized Hob. Good grief. No.

EDIT - (caveats of practical belt size relatively appropriate to machine capacity apply - I'm not even remotely interested in theoretical consideration of trying to shoehorn a 3" wide 8mm Pitch belt driving a Hob spindle on a BC 6 class machine) Anything can be done. We've all done something that proves a point or was done just to satisfy desire or curiosity. No one is arguing against that. But the *practical* reward versus effort in the OP's question is what's being considered.
/EDIT

 

[Vancbiker]

.......
now the almost equal has been crossed to where steppers are superior in many cases.

When I see builders like Makino, Grob, Yasda, or Hermle switch to steppers for their drives, I'll start to believe this^.

 

[Garwood]

I would embrace steppers with open arms if they could do the job well.

I would believe it if I saw it. Any examples of steppers doing real work? Like several hp worth of work?

And no, I don't need or want 30hp on this little hobber. 5hp for the hob would be fine I think. This is an ancient, lighter built machine.

Belts-

Most cnc lathes have belt driven spindles and belt driven spindle encoders.

All old haas 4th axis' have a cheezdick belt inside (with encoder on the motor even!).

And on that note, the work table on this David Brown turd is pretty much the same thing as a Haas hrt310.

 

[EmanuelGoldstein]

and why not? No way around that they run a more constant velocity than a servo, have much faster accel/decel, and can take a harsher environment.

This sort of ignores the dynamics of gear hobbing tho. How do you control speed with a stepper ? The process of hobbing teeth is like a very interrupted cut at slow speeds. Maybe at 32DP this won't be a problem but with bigger teeth, the hob motor has to be actively controlled. Mechanical hobbers often have big flywheels on the hob spindle for this reason. With a servo, as it slows down you pour the coals to 'er, when it speeds up you tighten the nosebag. Servos are very good this way, how do you actively get a stable speed out of a stepper with a very non-constant load ?

Fun video of gear hobbing in the big

Hard to tell from the glitzy video but doesn't look like hobbing at all ...

This is an ancient, lighter built machine.

You're trying to build a Swissmak

 

[Zahnrad Kopf]

All old haas 4th axis' have a cheezdick belt inside (with encoder on the motor even!).

And.. more importantly, you will note that they also have a Worm and Wheel. Case closed.

 

[memphisjed]

That would be a huge error. Maybe one can "get away" with it on some small Fine Pitch gear in a mild material, but one would never want to do it with anything even *mildly* medium to large Pitch. Much less, with a decent sized Hob. Good grief. No.

EDIT - (caveats of practical belt size relatively appropriate to machine capacity apply - I'm not even remotely interested in theoretical consideration of trying to shoehorn a 3" wide 8mm Pitch belt driving a Hob spindle on a BC 6 class machine) Anything can be done. We've all done something that proves a point or was done just to satisfy desire or curiosity. No one is arguing against that. But the *practical* reward versus effort in the OP's question is what's being considered.
/EDIT

Your edit beat me to breakfast dinner. Belts work if small length, fat width. Big boned and curvy. Our 25 hp drills are belt driven. Not the same as a v or flat belt on a black and decker drill press.
You have very nice gear reduction spindles that each gear is custom matched by the likes of yourself. These almost have negative noise value. Not practical for this case- awesome, just not practical.

 

[Garwood]

And.. more importantly, you will note that they also have a Worm and Wheel. Case closed.

Hold on ok, I don't think you should close this case just quite yet! I know you know your shit about gears! I'm not looking to argue with you, but I fail to see the difference between a new 2022 Okuma lathe (or any other top mtb) using a belt to drive the spindle (which you can use to generate extremely precise threaded things) and using a belt to drive a hob spindle which is basically a precision thread with teeth on it.

My hrt310 comparison is to the work table. The gearbox in the work table would be retained.

How does a real cnc hobber do it? Direct drive? Gears? What if there was a belt in there because belts work pretty ok?

 

[memphisjed]

how do you actively get a stable speed out of a stepper with a very non-constant load ?

Steppers do not vary speed with load. You run slaved to spindle speed, controller then hums out speed to stepper drive (same as servo). They run all they got all the time. Good for when you need constant speed, bad if need load feed back from them.

 

[EmanuelGoldstein]

Steppers do not vary speed with load. You run slaved to spindle speed, controller then hums out speed to stepper drive (same as servo). They run all they got all the time. Good for when you need constant speed, bad if need load feed back from them.

In this case they ARE the spindle. The table drive takes a fraction of the power that the hobhead takes. And both of them have to rotate at a constant speed, timed to each other under radically varying loads. Even with your description it does not sound like what steppers are suited for. It certainly is totally different than what you were promoting in the beginning.

 

[implmex]

Hi Garwood:
You wrote:
but I fail to see the difference between a new 2022 Okuma lathe (or any other top mtb) using a belt to drive the spindle (which you can use to generate extremely precise threaded things) and using a belt to drive a hob spindle

A lathe spindle sees mostly a load that is continuous, and predictable, even when threading.
Do you believe that a workpiece spindle on a hobber sees the same sorts of loads?
I'm not so sure it does.
Emanuel and Zahnrad Kopf have both pointed out the same...the loads vary a lot, so the system needs to be exquisitely sensitive and robust as hell in a way a lathe (even a very good stout lathe) does not.

Also the hob and the workpiece need to maintain excellent synchronicity in real time when both see those fluctuating loads.
I can't think of a turning case where this is so.
Even threadmilling an interrupted thread does not require the cutter speed to be synched to the workpiece speed.
It does requre synchronicity between spindle and Z axis ballscrew, of course, but to my mind that is very different
Nobody really cares if the threadmill slows down a bit, but if a hob slows down a bit and the workpiece motor cannot compensate instantly to maintain the synch, bad things happen.

Cheers

Marcus
www.implant-mechanix.com
www.vancouverwireedm.com

 

[Zahnrad Kopf]

Hold on ok, I don't think you should close this case just quite yet! I know you know your shit about gears!

The longer I live, the more resolutely convinced I become of the opposite of that statement. That's not even mildly a joke. I've been enthralled about the generation of repeating forms and Involutes for 45+ years now. In retrospect, I feel like I really don't know nearly enough.

I'm not looking to argue with you, but I fail to see the difference between a new 2022 Okuma lathe (or any other top mtb) using a belt to drive the spindle (which you can use to generate extremely precise threaded things) and using a belt to drive a hob spindle which is basically a precision thread with teeth on it.

It is actually much more complex than that. But, since you mention it, they're not really that good at it. At all.
-> Recent comments in a thread, related to this topic.

Now, full disclosure - I've a handful of small Hobbers that are *perfect* for this kind of thing, just sitting in a corner, waiting for something to do. And, in point of fact, I've even taken the time to discuss it with some people very well versed in controls and familiar enough with the application to be able to pull it off nicely enough for my own satisfaction, while also having experience in making/retrofitting a CNC machine or four, from scratch. ( both using readily available parts kits *AND* before such things were available ) < I.E., back when no such things existed and hadn't even been considered > I went so far as to create a spreadsheet of parts, plan the mounts and CAD them up.

In the end, I never went so far as to do it because there simply isn't really a large enough gain to justify the cost and efforts. As far as I can tell from my own personal point of view, "CNC" isn't enough of a reason to spend the amount of time, efforts, or money required to do it *well*. The real value is in the automation of things like multi pass/multi depth cutting ( roughing & finishing ), Crowning, Hob Shift, pesky reduction combinations, and loading/unloading of blanks. That's where the real payoffs are. And if one isn't making enough of a specific part to justify *those* concerns, then the whole effort is pissing into the fan, in my humble opinion.

At the end of the day, we already know that the proposal is *possible*. So you owe it to yourself to ask yourself just why *so many* antiquated, manual, gear driven Gear Making machines exist and (largely) still DOMINATE the world.

The answer ( I think that you will find ) is that they far outpace the alternatives in economy, linked with quality.

 

[Garwood]

The longer I live, the more resolutely convinced I become of the opposite of that statement. That's not even mildly a joke. I've been enthralled about the generation of repeating forms and Involutes for 45+ years now. In retrospect, I feel like I really don't know nearly enough.



It is actually much more complex than that. But, since you mention it, they're not really that good at it. At all.
-> Recent comments in a thread, related to this topic.

Now, full disclosure - I've a handful of small Hobbers that are *perfect* for this kind of thing, just sitting in a corner, waiting for something to do. And, in point of fact, I've even taken the time to discuss it with some people very well versed in controls and familiar enough with the application to be able to pull it off nicely enough for my own satisfaction, while also having experience in making/retrofitting a CNC machine or four, from scratch. ( both using readily available parts kits *AND* before such things were available ) < I.E., back when no such things existed and hadn't even been considered > I went so far as to create a spreadsheet of parts, plan the mounts and CAD them up.

In the end, I never went so far as to do it because there simply isn't really a large enough gain to justify the cost and efforts. As far as I can tell from my own personal point of view, "CNC" isn't enough of a reason to spend the amount of time, efforts, or money required to do it *well*. The real value is in the automation of things like multi pass/multi depth cutting ( roughing & finishing ), Crowning, Hob Shift, pesky reduction combinations, and loading/unloading of blanks. That's where the real payoffs are. And if one isn't making enough of a specific part to justify *those* concerns, then the whole effort is pissing into the fan, in my humble opinion.

At the end of the day, we already know that the proposal is *possible*. So you owe it to yourself to ask yourself just why *so many* antiquated, manual, gear driven Gear Making machines exist and (largely) still DOMINATE the world.

The answer ( I think that you will find ) is that they far outpace the alternatives in economy, linked with quality.

Thanks for that insight! That makes really good sense. Manual hobber will never do those things without an unjustifiable amount of work. And having my only gear experience being milling a few hundred splined shafts on CNC mills over the years, I don't know a huge majority of the things I will need to know to make my own gears on the hobbers I just bought.

I thought the "electronic gearing" might be a decent get by to patch up an old hobber, but sounds like I would be far ahead to focus my energy on applying nice hobbers to making my parts.

I would still be interested in furthering the discussion of exactly what hardware it would take to do the job and real curious about how a genuine CNC hobber does it, both the hardware and the software aspect.

 

[memphisjed]

In this case they ARE the spindle. The table drive takes a fraction of the power that the hobhead takes. And both of them have to rotate at a constant speed, timed to each other under radically varying loads. Even with your description it does not sound like what steppers are suited for. It certainly is totally different than what you were promoting in the beginning.

Changed spindle maybe to spindle no. The spindle can bog down (it will by nature of ac motor) and the rotary slows with it thru syncing. Of all things this is a good spot for stepper. Highly controllable, exact speed to command, high torque, slow rpm. The spindle is what you read variations from, everything else matches that, near instantly.

Steppers do not do Hp of work (speed and torque), they just do torque. You wouldn’t want a real plasma/laser/mill table with them because they are slow, really slow. Rapids at 300 or 500 ipm wouldn’t sell many production machines.
Robot surgeries (no jitters), the giant ass spinning magnets of mri, these are steppers. MachineDesign has Pr piece of Lin motion designing steppers for satellites because reliability and stability (vibration/noise) are better with them.
They are not better, just better at a some things.

I do not see how Garwood lathe sync cycle wouldn’t work. The machine will not be Zach level. He is starting with an old machine that already wrecked a few to many times- and lost its teeth.

Practical is questionable for a a few valid reasons.
Doing it for reasonable price? Yes.
I vote just do it. Practical is for suits.

 

[CarbideBob]

Hold on ok, I don't think you should close this case just quite yet! I know you know your shit about gears! I'm not looking to argue with you, but I fail to see the difference between a new 2022 Okuma lathe (or any other top mtb) using a belt to drive the spindle (which you can use to generate extremely precise threaded things) and using a belt to drive a hob spindle which is basically a precision thread with teeth on it.

My hrt310 comparison is to the work table. The gearbox in the work table would be retained.

How does a real cnc hobber do it? Direct drive? Gears? What if there was a belt in there because belts work pretty ok?

Is there an encoder on the spindle or just the motor?
Then we add that a thread is continuous in one direction, hobs bang and release. Belt drives don't like that so much.

Unsure if someone is mixing steppers with step and direction servos.
While there are linear steppers linear motors are AC servos not steppers.

 

[johansen]

So they make servos that can take step direction, let's assume for the moment that they are as stiff as a stepper of equal power rating. Suppose you have a 1 or 2kw servo driving the hob through a 10:1 gear reduction.

Keep in mind that the stepper with micro stepping is good but at it's limit of torque it is one step behind where it should be. The bigger the stepper the stiffer it is. For the math I'm using 1.8 degrees per step.

suppose you have a 10:1 gear ratio driving the hob from a stepper. One step is 1.825 degrees behind the commanded position. If the inertia of the servo is low, the hob will be + 0.X degrees ahead and .182 degrees behind of where it should be, with every single tooth that engages the cut.

How stiff is the gear train on a "real" gear cutting machine? And secondly does the gear train stiffness actually matter to your needs.
Now this is at it's torque limit, which should not be reached. So maybe half that 0.05 degrees wobble on the hob, neglecting the backlash of the 10:1 reduction.

Suppose you have a 40:1 worm gear in the base of the machine to rotate the gear blank. 1 step behind is 0.045 degrees behind the commanded position. Forces are low, so this should not be a problem.

I just bought a house so been busy but when I get my shop setup I will be making an open loop stepper driven gear hob, for the random gears I need to make occasionally.

 

[Garwood]

Is there an encoder on the spindle or just the motor?

Lathe spindle motor has an encoder inside it. Spindle motor is belted to spindle (cog or multiple V-belts). Spindle has additional encoder that is cog belted to it. I believe encoder inside the motor is used for spindle drive feedback and the spindle driven encoder is used for threading and C-axis stuff by the control. I say this because I have a couple lathes I can unplug the spindle encoder and all it does it alarm out when you go to thread.

All 8 of the CNC lathes I own/have owned have been built this way from 1979 through 2008 with the exception of a few that had multi-speed gearboxes gear driving the spindle, but those still had the spindle encoder driven with a little cog belt.

 

[Garwood]

Changed spindle maybe to spindle no. The spindle can bog down (it will by nature of ac motor) and the rotary slows with it thru syncing. Of all things this is a good spot for stepper. Highly controllable, exact speed to command, high torque, slow rpm. The spindle is what you read variations from, everything else matches that, near instantly.

Steppers do not do Hp of work (speed and torque), they just do torque. You wouldn’t want a real plasma/laser/mill table with them because they are slow, really slow. Rapids at 300 or 500 ipm wouldn’t sell many production machines.
Robot surgeries (no jitters), the giant ass spinning magnets of mri, these are steppers. MachineDesign has Pr piece of Lin motion designing steppers for satellites because reliability and stability (vibration/noise) are better with them.
They are not better, just better at a some things.

I do not see how Garwood lathe sync cycle wouldn’t work. The machine will not be Zach level. He is starting with an old machine that already wrecked a few to many times- and lost its teeth.

Practical is questionable for a a few valid reasons.
Doing it for reasonable price? Yes.
I vote just do it. Practical is for suits.

OK. So what you are proposing is that steppers would work fine for the work table feed, hob vertical feed and table in-out feed, but driving the hob is not stepper territory.

Say the hob is driven with a 5HP 3 phase motor with a VFD and an easy accel rate. Encoder is applied to the end of the hob spindle for feedback. Stepper for work rotation is slaved to hob spindle encoder.

What seems to get really complicated in my mind is when you start thinking about this slaving stuff, the work table rotation is a function of hob rotational position and vertical feed position. Helicals are not cut with the helix inline with the vertical feed. With a developed "CNC" control this is all accounted for, but slaving an axis to another? How do you make slave gearing work for helicals on a gear hobber?

 

[johansen]

How do you make slave gearing work for helicals on a gear hobber?

Seems to me that is rather easy.

Here is an example:

Gear rotates 500 times, hob spins 4999 times, hob travels x distance across the gear face.

For a spur gear the hob would rotate 5000 times for a 10 tooth gear.

Helix angle is set by the ratio of the distance x of the gear to the pitch of the gear. At the end of the cut, the hob is 1 rotation behind where it should be for a spur gear.

For the opposing hand, the hob would need to turn 5001 times.

 

[EmanuelGoldstein]

Changed spindle maybe to spindle no. The spindle can bog down (it will by nature of ac motor) and the rotary slows with it thru syncing.

Ah. So you don't plan to make it work conventionally, just keep everything synched as the main motor changes speed. That sounds more feasible. Maybe not ideal but at least has a chance of working. Just put a honkin' big motor n the hob spindle.

I still wouldn't bother messing with it tho, there's decent running machines out there for a lot less than what he'll have into it and the best it could ever do is match the performance of a 1950 David Brown. That's assuming the table worm is not worn out.

And the software is not going to be a picnic to write ...