Gear Hobbing

Blimey! When the 'early articles' that I've read about hobbing were written, the stepper motor hadn't even been conceived :)

Andrew

The stepper motor is actually an AC synchronous motor that has been adapted.

Gareth google John Stevenson and gear hobbing and you'll find he did it years ago :)

Andrew

Actually, I've been in touch with the Three Amigos, John, Giles Parkes and John Florentin for some time. It was for Giles that I produced the Hobnail spreadsheet a few years ago, and I have a 60DP hob here on loan from Giles.

I did produce a hob attachment for my mill in order to generate some

200 tooth gears for my retro radio dial, but it was too flimsy in the end, so am imminently about to try the same on a 3-D printer.

But, thanks anyway.

Using a PLL to synchronize a (divided down, higher frequency) oscillator to the hob's rotation will get the average speed of the hob quite well, but the angular synchronization will only be as good as the number of impulses per hob-rev you're synchronizing to, methinks.

belts, gears etc. are only necessary when synchronizing the gear to the hob if one is not using any digital intelligence between the sensor and the stepper.

If one is using a home brewed optical disc, then one hundred pulses per rev is easily achievable (use a spare changewheel, spray the teeth and gullets black, then skim the paint off the tops of the teeth). If one is using a commercial encoder, 512-1024 pulses/rev are bog standard. to get the right number of output pulses, use a PIC, Arduino, Banana Pi, to add:- hob_ppr *desired_gear_teeth/stepper_ppr into a counter as a floating point number.

Decrement it by one every pulse from the encoder.

Then output a stepper pulse and re-add the number to the counter every time the counter becomes negative.

If I remember the logic correctly, that's what I used for a PIC based counter that started off as a commercial rev limiter for racing motor bikes, developed on to LED gearchange speed indicators for bikes and racing cars and was even modified to measure the barring speed of steam turbines to ludicrous accuracy. In the end, it got used between 0.01Hz and 250kHz, all on a 4MHz PIC!

I'm thinking Tony Jeffree's Division Master, as sold by Lester Caine can cope with A/B type division by proper use of the settings and it can be actuated from a buffered optical encoder. John S. used one, but I don't know if it was in that mode.

Digital is simpler than analogue for this, to my mind ;-)

regards

I will shortly be hobbing a set of gears for a friends "Minnie" Traction engine. I use Linux CNC on an X2 mill. I fitted the spindle with a 100 hole disk with 2 through hole sensors giving me A and B Quadrature. I designed and built a 72 to 1 indexer driven by a stepper. Worm Wheel hob and Worm made on the lathe. Worm Wheel hobbed as it spun freely. The shaft of the indexer I made in ER32 form so components mounted in a collet. I can obviously mount a faceplate on the outer thread to give me a rotary table.

20DP Hob is currently on the way from China. Main cost has been the Stepper motors and drivers, time is free these days!

Works for me

Richard

Sorry, misled you, for _TWO_ division chains are needed, one for the workpiece and one to do the phase comparison with the hob optical disk. When I posted, I was really only thinking about the workpiece division down to a stepper motor,

Are you thinking of a traditional machanical hobber when you say that?

I fell into that trap, for there is a difference between driving a workpiece continuously and dividing the circle because, when dividing the circle, you can output a differing number of pulses to counteract the rounding errors of your calculation.

When I did my (all-too-flimsy) set up to try to hob a couple of anti-backlash

200-tooth gears for my retro radio dial project, I had a 48 slot optical encoder on the mill spindle (salvaged from an old 8" disk drive, wish now that I'd kept the 8" disk drive :-( )

and a 200-step per rev stepper motor driving a 1:48 worm and wheel salvaged from a defunct dividing head, thus giving me the 200 teeth that I wanted with no division being needed.

Unfortunately, despite setting up so that the axis of the hob was SBO to the middle of the two clamped-together gear blanks, it cut the teeth on the side, but the tooth spacing around the rim was beautiful :-)

Sorry, I don't follow that despite holding back my commentary on your original post!

Pause for further thought ...

Actually, yes I do!

A very clever way of circumventing the rounding error, except that you'd need to be aware of how floating point was implemented, especially on a PIC, lest you run out of time between pulses.

In my case, because I'm a professional hardware / software engineer I try to keep away from a busman's holiday, (Raspberry Pi still in its packing

15 months after impulse purchase :-( ) and in any case, I've a pile of SN74L presettable divide counters in my junk box which are exceeding their use-by-date.

COMMENT RETRACTED!

(See later post !!!!)

Shirley, the hob should have been lying at the lead angle of the hob, relative to the gear face? I am assuming a helical hob of the type I have bought.

Richard

In the case of hobbing a spur gear, yes, indeed, but in this case it was an attempt at a worm-wheel with the cut plunging downwards and not transversely as for gears.

Ah now I understand "cut the teeth on the side". The two clamped together was the start of a method of backlash elimination? I did this on my wheel in the indexer, time will tell if I need it, all put together in the original cut state at the moment. Richard

Yes, that is right, for after cutting in tandem, then one would be spring loaded against the other to pinch both sided of the worm thread.

Anti-backlash gearing was a common feature of tuning dials in old radios, otherwise after carefully tuning in an interesting signal, when you took your hand away from the tuning knob, the radio would jump off frequency.

In my case, to alleviate the difficulty of machining a worm to match the circular pitch of whatever DP or Module hob you had to hand, I resolved to use off-the-shelf M10 rod as the worm, and the hob was an M10 tap.

A linked idea was to have an over-long section of M10 rod with a nut to travel along it to carry the dial pointer, the nut also to have a spring-loaded brother to prevent backlash there.

However, because of the flimsiness of my set-up, I have moved away from that approach, but I suspect that the real problem was my eagerness to evaluate the technique resulting in a rushed job that was insufficiently braced structurally.

That is what I did on my indexer, springs and all. However the springs were only to help move the gears against the teeth faces to take up the backlash, once there I have two M4 screws that lock the two gears together until the next time.

Richard

replying to gareth, paul mayhead wrote: Hi Gareth How can I get a copy of Hobnail please?

Blimey, that's a blast from the past!

My Email address (when you demunge it) is valid.

Also, out of interest, what forum are you posting into?

Gareth. I am giving a talk on the Jacobs Hibbing machine to SMEE next year. I would like to mention your hobnail spreadsheet. May I do that?, giving you due credit of course. Regards, Chris Robinson

Blimey, That's a blast from the past, and you're lucky that I just happened to dip my toes into the Usenet news group uk.rec.models.engineering tonight to see your request!

I did buy a Jacobs hobber 4 years ago from Pennyfarthing Tools, but in the end sold it on without doing anything with it because my interest in hobbing really only arose through dealing with Giles Parkes of the SMEE and ceased after the death of Giles.

Since then, I acquired a Corry wheel cutting engine and a set of involute cutters that satisfied my need (and sets of cycloid cutters for clock making of course)

By all means talk about Hobnail, but its 10 years at least since I developed the spreadsheet and it's still on this laptop and you are welcome to a copy of it. Up to date email address is correct on this posting.