Electronic gearing for gear hobber

[Garwood]

I've picked up the David Brown gear hobber they used when they built the Mayflower, atleast it looks like it could be. This is in a package of running hobbers, so I'm not crushed if it can't be saved. It has suffered a lack of lube in it's nether regions and has major internal bearing/gear/shaft detritus. I don't think it's worth the effort/expense to rebuild the geartrain. The hob head and work table and ways are all fine.

I'm pretty anti-retrofit, but gear hobbers are not really commodity machines these days. Atleast not in the USA it seems.

I have googled this some. You got the Linux CNC guys and you got Kflop guys and then you got a few Motion Guru posts on this site talking about how simple this is with the right hardware while sharing dead links to products.

My goals would be- Control 3 or 4 axis (hob spindle, work table, hob up/down feed. Hob infeed can be manual) Hardware cost under a few grand, new or used. I can hunt Ebay. Avoiding steppers. OK with Chinese servos/drives if they work. Some kind of user interface that I can teach other people how to use- Could be anything with a screen if it works. I can't do software, but I do have some good resources for Python and C coding if needed.

What hardware would work? Vector drive and Servos? All servos?

This is a 12" gear hobber weighing around 3 tons. Probably needs 5-7.5HP for the hob I would guess. Work table and axis feeds could probably be 1HP or less.

 

[implmex]

Hi Garwood:
Zahnrad Kopf posted in your other recent thread about the realities of electronic gearing and trying to cut accurate gears with it.
He's not a fan!

There was another very long winded thread about gearmaking (9 pages) originally posted by drcoelho in which the topic of electronic gearing was broached and motion guru as well as carbide bob and others commented on the topic there.
Here is a link:

Gear Hobbing: Servo or Stepper motor

I'm early in developing a gear hobbing add-on to my Deckel manual mill along the lines of what Andy Pugh (YouTube and YouTube) and John Stevenson (Electronic hobber - test run) did but with more professional components. The approach is to have an encoder on the spindle, a PLC controller, a...

www.practicalmachinist.com

It's worth a read.

Cheers

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

 

[Garwood]

Well how do all the CNC hobbers made in the past 40 years do it then?

Where is the hardware threshold where you can actually do this?

 

[implmex]

Hi again Garwood:
I don't know.

But from this post I gather that some relationships are still mechanically geared:
** - One of the contributors to at least some minor error in what ultimately relates to accuracy is the fact that they are simply employing "electronic gearing", AKA - "servo slaving". This has some limitations with regard to encoders, computers, and the (literal) flexibility of belts on pulleys, allowing one point of some minor error to creep in, eventually. This, as opposed to even some "cheap" CNC Hobbers that still utilize Worm Wheels to accomplish some of the reductions, and maintain (at the least) some rigidity and accuracy of timing.

This is from post #10 in your other thread.

Cheers

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

 

[EmGo]

Hi again Garwood:
I don't know.

But from this post I gather that some relationships are still mechanically geared:

No. Nobody with a reputation does this. CNC gear hobbers and shapers are common and work better than mechanical ones. This has been true since about 1980. People who don't know what they are talking about should not speak.

Now, the thing with retrofits ... one of the main reasons that cnc hobbers/shapers are better is that they get rid of all the monkey-motion and related structural drawbacks of a mechanical train driving the moving parts. With a retro, most of the time you're still going to be stuck with shafts and beevils and holes in the castings for the moving components to go. You'd have to redesign the machine to put the motors where you want. Why bother ? If you want to do a retro, just get an early or wornout older nc machine and do the work to that. Otherwise you're doing a bunch of work to get something that is design-wise second-rate. It's not worth the effort (usually. Maybe you have a 72" machine that would cost $50,000 just to move, in that case probably makes sense. But not for a 12"-er).

I wouldn't bother, Gar. You can buy a better starting point for cheap, and the cost of retro-ing is going to be much more than you anticipate. We love old junk, and free is major motivator, but sometimes it's better to just scrap stuff. You could set up a lemonade stand and do better with your time than retroing an old David Brown. If it's worn out, everything will be worn out, so you're still going to have to do major rework. Not worth it.

 

[Garwood]

No. Nobody with a reputation does this. CNC gear hobbers and shapers are common and work better than mechanical ones. This has been true since about 1980. People who don't know what they are talking about should not speak.

Now, the thing with retrofits ... one of the main reasons that cnc hobbers/shapers are better is that they get rid of all the monkey-motion and related structural drawbacks of a mechanical train driving the moving parts. With a retro, most of the time you're still going to be stuck with shafts and beevils and holes in the castings for the moving components to go. You'd have to redesign the machine to put the motors where you want. Why bother ? If you want to do a retro, just get an early or wornout older nc machine and do the work to that. Otherwise you're doing a bunch of work to get something that is design-wise second-rate. It's not worth the effort (usually. Maybe you have a 72" machine that would cost $50,000 just to move, in that case probably makes sense. But not for a 12"-er).

I wouldn't bother, Gar. You can buy a better starting point for cheap, and the cost of retro-ing is going to be much more than you anticipate. We love old junk, and free is major motivator, but sometimes it's better to just scrap stuff. You could set up a lemonade stand and do better with your time than retroing an old David Brown. If it's worn out, everything will be worn out, so you're still going to have to do major rework. Not worth it.

I completely follow you 100%.

Find me a CNC hobber with a Fanuc or mits control that I could have working on my floor for under $15k. I don't think a guy can do that right now. Maybe when the next crash happens and fuel gets down to $2/gallon so shipping from the midwest is possible.

However, it does kinda sorta seem like a crafty fellow could get rid of, oh, 98% of the monkey motion gearing shit with 3 servo motors- One driving the hob spindle through a toothed reduction belt (yes, I said belt, I know, but real CNC's have been using good belts to do serious work for a long time, we could even get crazy and put the encoder on the hob spindle), one servo direct coupled to the work table worm and one servo driving the original acme hob vertical feed screw.

I can set the whole hobber on a HBM and machine servo mounting flats if needed. I have a huge stockpile of 1980's Japanese CNC hard parts like servo couplings, limit switches and way too many NOS cable chains, and rolls nipponflex conduit. Hell, I even have a complete Fanuc 10M operator console.

If I could get the right hardware, get it mounted and turn one of my nerdy friends loose on it for a couple weeks (they'll say 2 weeks and the GUI software will take them a month, I know how it goes) in trade for making parts for their projects do you think I could have something that cuts splines and gears as good as a mechanical hobber, but had the benefit of fast part changeovers, maybe even load monitoring to throw an alarm if something went wrong (I can't imagine I would hear a 200RPM hob crashing in my shop over the other sounds)?

Am I just way too naïve here?

 

[CarbideBob]

If I could get the right hardware, get it mounted and turn one of my nerdy friends loose on it for a couple weeks (they'll say 2 weeks and the GUI software will take them a month, I know how it goes) in

Dream on.
Yes it will move at this point and cut metal. .....It will not make good parts.
25-50 more weeks and it might make a passable part.
So many real world details here. Servo lags, axis sync, sample times, gain loops, stability across a wide range, the digital domain world and its hate of analog.
It seems straightforward but it is not.

On the other hand the only way to learn is to try it. Prepare for some frustration.
Bob

 

[EmGo]

I have a huge stockpile of 1980's Japanese CNC hard parts like servo couplings, limit switches and way too many NOS cable chains, and rolls nipponflex conduit. Hell, I even have a complete Fanuc 10M operator console.

Oh, there's another little hurdle. Most of the builders use their own custom controls. The ones who do use Fanuc hardware have their own software. I don't think they will be handing it over any time soon

It's possible to write your own, Fanuc will give you what you need, but you have get on their distributor or manufacturer or custom builder list ? Our sponsors have done that, it wasn't low cost and couldn't be justified for just one machine, I imagine.

Am I just way too naïve here?

Nah, nobody starts out knowing all about a subject. But the reality is, it's so simple to set up and run a conventional gear machine that there's really no point to retro-ing one. We did do retros but they were on 3 meter machines and for a specific purpose : for a small hobber it could be a fun project but for just making gears, you're way ahead to pick up a nice hobber to start with.

You could find a cnc carcass cheap if you were patient and hunted, they do exist, so maybe keep an eye out but in general, mostly not worth the effort except as a "look what I did !" kinda thing.

Honestly, you'd be way better off working on your product line. That'll bring you so much better returns on your time ...

 

[Garwood]

Fanuc has some docs suggesting they made a special version of the 11M for hobbing. Seems like many cnc hobbers were made with the 15MA control (that'd be my choice for a control).

Fanuc has some special docs about setting up a 16iT/18iT for hobbing by slaving the live tool to the C axis.

As there are thousands of I series T controls out there and they apparently support it out of the box that would probably be the smartest choice.

I also happen to have an 18iT lathe with live tools.

 

[memphisjed]

A good drive for the "spindle" with indexers (not encoder) tied to hob. This lowers your counts down to 2 per cut (up-down=1 rpm=2 counts). why go thru having 240000000 counts when you only need 2?
Why servos for motion? You are not going fast, You are not needing torque feedback. Servos are jittery, weak, and old tech for stable slow motion. Steppers are taking over for good reason. (not chinese {no offense EG} drives and 24 volts at 2 amps). ABB, seimens, GE, and even the newer geckos are impressive. They are cheap, easy, and you can likely remove gear reductions on your positioning axis. They holding power is better than servos constantly compensating for your application.
cat scans, mri, atms - all steppers now. Linear motors are steppers.
Controls, depends on how much interface you can fake. kflop, acorn, planet (stable as wago plc machines) all have index/encoder for electric gearing built in- moving on to automation direct, mits, seimens, heidemen, benkenoff (highly do not recommend), wago, controlled automation gets you any interface you want for the low cost of designer + coder + revisions + tax and shipping. These are better if you can stomach the cost, which is literally add 2+ zeros to the prosumer controls.

You might even get away with a stepper for the spindle running a cv drive (up down limits are program signals), then you can use a single industrial ardinio for the motion.
more torque than a 5hp motor: https://www.automationdirect.com/ad...s/single_shaft_stepper_motors/stp-mtrac-42202
makes your head spin numbers of torque: https://www.kollmorgen.com/en-us/pr...rque/m1xx-series/mh172-series/#Specifications

 

[EmGo]

A good drive for the "spindle" with indexers (not encoder) tied to hob. This lowers your counts down to 2 per cut (up-down=1 rpm=2 counts). why go thru having 240000000 counts when you only need 2?
Why servos for motion? You are not going fast, You are not needing torque feedback. Servos are jittery, weak, and old tech for stable slow motion. Steppers are taking over for good reason. (not chinese {no offense EG} drives and 24 volts at 2 amps). ABB, seimens, GE, and even the newer geckos are impressive. They are cheap, easy, and you can likely remove gear reductions on your positioning axis. They holding power is better than servos constantly compensating for your application.
cat scans, mri, atms - all steppers now. Linear motors are steppers.

Umm, there is no positioning or holding involved with hobbing. Both axes are rotating in a timed relationship. Also, depending on your tooth count and hob material, they can rotate at a pretty good clip. For example, even with hss hobs, with a 3" hob doing about 100 fpm you'll be spinning the hob at about 150 rpm and for a six-spline the table will be going 25. With a double-start hob it'll be running 50. That's not fast by lathe standards but for a 12" diameter work table, that's cooking right along.

Get into carbide and the numbers go up.

 

[memphisjed]

Umm, there is no positioning or holding involved with hobbing. Both axes are rotating in a timed relationship. Also, depending on your tooth count and hob material, they can rotate at a pretty good clip. For example, even with hss hobs, with a 3" hob doing about 100 fpm you'll be spinning the hob at about 150 rpm and for a six-spline the table will be going 25. With a double-start hob it'll be running 50. That's not fast by lathe standards but for a 12" diameter work table, that's cooking right along.

Get into carbide and the numbers go up.

150 rpm is not fast. 25 rpm is really not fast. If not a shaper hobber then it is perfect for steppers. Less than 300 rpm and no reason for reduction (error points) you need with servos. 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.
You can spend money on good servos, really good money on precision gear reduction head unit, and then drives, control that can chain it all- and gain nothing. You could place a load meter on each axis for fun I guess?

 

[EmGo]

150 rpm is not fast. 25 rpm is really not fast. If not a shaper hobber then it is perfect for steppers. Less than 300 rpm and no reason for reduction (error points) you need with servos. 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.

This is totally wrong, but not worth going into. Steppers flat-ass will not work on a real hobber. Maybe a hobby shop toy cutting 32 dp but definitely not the real thing. Sorry.

edit: here's a little 16" liebherr with a video that gives some idea of how they work.

Liebherr Universal L 401 Gear Hobber - Revelation Machinery

Made In Germany Maximum Diameter Gear Cut: 16” Maximum Rated D.P. (Module) In Steel: 3.18 Dp Maximum Axial Travel: 12” Minimum Number Of Teeth: 6 Diameter

revelationmachinery.com

It's got 20 hp. This is not even today's technology, where they run faster and harder and carbide hobs dry. I'll see if I can find something newer ... this isn't stepper-motor territory.

btw, shouldn't touch this thing with a fifty foot pole. Only an effing idjut puts water-based coolant into a gear machine. It ruins them.

 

[memphisjed]

This is totally wrong, but not worth going into. Steppers flat-ass will not work on a real hobber. Maybe a hobby shop toy cutting 32 dp but definitely not the real thing. Sorry.

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.
You are the big proponent of linears, which is why I think the jump in stepper control/drives happened.
I have seen in two years car paint lines go from servos to steppers, and the latest rfi I have seen has the robot arms running steppers. These are for the sea colored squished circle factory, so doubt they are hobby grade.

 

[EmGo]

and why not?

What's 56 amps at 460 volts ? 35,000 lbs for a 12" diameter machine .... modern hobbers are not toys. Do you have a 30 hp stepper motor handy ?

Liebherr Model LC-282 CNC Gear Hobber w/ Auto Loading dry or wet cutting LH 90 controls 2003 | Piselli Enterprises

Liebherr model LC-282 CNC Gear Hobber, with Auto loading, Dry cutting and / or wet, LH-90 Cnc Controls, Chip Conveyor, inspect under power, excellent condition NEW 2003.

pisellient.com

And here's a brochure which can give a better idea of what's involved

https://www.liebherr.com/shared/media/gear-technology-and-automation/broschures/gear-cutting-machines/liebherr-hobbing-machines-lc200-500.pdf

I don't want to discourage Garwood but ... honestly, a control from a lathe is not going to give him what he wants. Better to just buy an L401 or L301 or be happy with the Overton he has. CNCing a trashed David Brown is going to be a giant time sink for no advantage. A decent used L300 will run rings around it even if he gets it working.

 

[memphisjed]

He stated 5hp spindle. You are saying 30. Then you run a motor, not servo, not stepper. E-gearing that to steppers that will easily handle those movements, speeds, and loads.
You have not stated why movement can not be stepper, or even why a servo is better in this case. 5 years ago sure, now the almost equal has been crossed to where steppers are superior in many cases.

 

[EmGo]

He stated 5hp spindle. You are saying 30.

Why would you recreate the abilities of 1946 ? If he retros, then there's no point in making it "almost as good as it was way back when" except a lot more expensive and fragile.

Not that Mr Brown would be able to handle those powers, so here we go doing all this work for something that isn't going to perform Even a manual L400 of 1970 has 20 hp. This is like so much work for nothing, while a lathe control that has five axes is a little unusual, and it certainly isn't going to transfer straight across to a machine that is so different, and Fanuc isn't going to hand over the keys to modify the software for nothing but ...

Then you run a motor, not servo, not stepper.

The drives on modern cnc hobbers are servomotors. They've tried linear motors on the table, they still have a problem with cogging. Most likely your steppers would at low speeds as well ?

All in all, doing this would be a nightmare for no gain whatsoever. The basic structure just isn't up to the job. He'd end up with a more expensive and fragile version of a machine that was good in 1950.

If he wants to, that's fine but ..... a lemonade stand would be a better investment.

p.s., Gar's mistaken about 3 or 4 axes being enough. There's no point if you can't do two-cut, retract away from the part when it's got a shoulder, crowning and taper are both valuable, hob shift is a necessity, it's like why do it if it's not even as capable as a manual machine ?

 

[Cole2534]

If he wants to, that's fine but ..... a lemonade stand would be a better investment.

I thought we preferred hot dog carts?

Anyway, post up some good videos about gear hobbing please.

 

[EmGo]

I thought we preferred hot dog carts?

Been a while snce I had a good hot dog

Anyway, post up some good videos about gear hobbing please.

It's not that exciting after the first few thousand parts ... but a two-tool bevel generator, that's fun

 

[Peter from Holland]

Fun video of gear hobbing in the big

Peter