Gear Hobber : Design ideas please …

[Michael GilliganParticipant @michaelgilligan61133]

Compact and Simple …

[Michael GilliganParticipant @michaelgilligan61133]

I have a need [desire] to make MOD_0.8 gears, probably from aluminium alloy, in a wide range of tooth-counts.

A small self-contained machine would be ideal, but I want to avoid the complexity of a 'versatile' device like the Jacobs. This will be using one hob [which has already been ordered] and only 'straight cut' spur gears are required.

… Stepper Motor controlled would be of particular interest.

Does anyone know of a suitable design, or have any ideas, please ?

MichaelG.

.

>>> note: I have already decided against 3D printing them

.

Edit: this looked like a promising start, but got nowhere

http://www.modelenginemaker.com/index.php?topic=1159.0

 

 

Edited By Michael Gilligan on 07/08/2018 15:49:29

[Joseph Noci 1Participant @josephnoci1]

Michael, see if Paul White may respond – I have done a design for such a device( the electronics part at least..) for a hobber that Paul has – not sure if it is the Jacobs – I used a Nucleo processor module ( STM micro processor) – the Nucleo is an Arduino work-alike, but a lot faster and easier for me to use! Drives a stepper to turn the gear blank, with an encoder on the hob shaft. 7 segment digital display to set things up – tooth count, etc – very simple stuff – Paul managed to get all the bits email-order in quick time.

I would think the mechanics should not be too difficult – basically a rigid mount to take the rotating blank, driven by a geared down stepper, and on a slide to feed the blank into the hob. If you do not need a fully stand alone device, you could make the blank axle and stepper drive a component that fits in place of the tool post on your lathe, and fit an encoder to the lathe spindle. Happy to pass the design on and assist if you wish to try my electronics

Regards

Joe

edit – syntax…

Edited By Joseph Noci 1 on 07/08/2018 16:04:18

[Martin ConnellyParticipant @martinconnelly55370]

Search for cnc gear cutting on youtube. There are examples of people driving spin indexers with stepper motors driven by arduino type controllers.

Martin C

[Mike CrossfieldParticipant @mikecrossfield92481]

Joe

I would also be very interested in details of your design.

Mike

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Martin Connelly on 07/08/2018 16:11:26:

Search for cnc gear cutting on youtube. There are examples of people driving spin indexers with stepper motors driven by arduino type controllers.

Martin C

.

I evidently haven't looked hard enough, Martin

The only videos I have found so far are for using standard cutters, not hobs

… except, of course, for things like this https://m.youtube.com/watch?v=3j4oqRU9p08 which are rather outside my 'design envelope'

MichaelG.

Edited By Michael Gilligan on 07/08/2018 16:21:00

[Neil WyattModerator @neilwyatt]

The 'obvious' solution to me is something that senses cutter rotation and synchronises to a stepper motor driven spindle carrying the blank.

[John HaineParticipant @johnhaine32865]

In MEW a while back there was a design for exactly this. It needs a high count encoder on the mill spindle to generate pulses that are divided down to drive the stepper that rotates the gear blank. I seem to recall that Sir John also had some ideas on this.

This is something that I've been contemplating to use with my digital dividing head conversion. Failing a high count encoder (which is expensive and tricky to add to the spindle), a better approach IMHO would be a one pulse per rev encoder and a phase-locked loop to multiply the spindle speed up so you set the division ratios using two dividers, one in the PLL and the other dividing down the PLL's VCO to drive the stepper. I've even got as far as buying the program switches! I think it should be doable with a few CMOS dividers and a 4046 PLL chip.

 

Edited By John Haine on 07/08/2018 16:41:40

[Michael GilliganParticipant @michaelgilligan61133]

Posted by John Haine on 07/08/2018 16:39:31:

In MEW a while back there was a design for exactly this. It needs a high count encoder on the mill spindle to generate pulses that are divided down to drive the stepper that rotates the gear blank. I seem to recall that Sir John also had some ideas on this. …

.

Thanks, John

Was it the article that is discussed here ? : **LINK**

https://www.model-engineer.co.uk/forums/postings.asp?th=75580

Must dig that one out.

MichaelG.

.

P.S. … I didn't mention earlier, but I do have a 'DivisionMaster' [from Tony Jeffree's own fair hand] which might get put to use.

 

Edited By Michael Gilligan on 07/08/2018 16:56:40

[Paul White 3Participant @paulwhite3]

Joe having put my activity on the board I must now comment- I had built a Jacobs machine some 20 years ago but found the changing of gears for each wanted gear a bit of a fag, just the same as setting up for screw cutting on the lathe. The lathe problem was solved by Joe giving me the detail and lots of encouragement for an ELS (electronic leadscrew).

This was such a success that I asked Joe for an opinion on removing all the gears from the hobber and doing the same electronically, almost by return he had produced a program and circuit.

It was then down to me to make a new gear blank head and set up an encoder on the hob spindle. This is still work in progress (work being impossible for the last 2 months, too hot). Without doubt if I were starting on a hobber today it would be a stripped down Jacobs with Joes electronics.

It is a fun project in my opinion.

paul.

[Clive FosterParticipant @clivefoster55965]

I can confirm that the CMOS phase lock loop, CMOS divider chain and digital encoder approach to synchronising two stepper driven spindles is practical and relatively easy to implement. Main gotcha is sorting the frequency response components on the phase lock loop right for stable operation. If possible switched ranges set-up is better than trying to do it all in one range.

Stable operation at better than a tenth of a step error is easily achieved, you will need to gear down. Its arguable whether two steppers and all digital control is better than one stepper slaving to one free running servo motor. I've used both but this was back in the late 1970's when CMOS logic was still the latest, greatest thing. No nice microprocessors then. Pretty sure I ended up with a dual CMOS PLL and a voltage control oscillator or two.

Clive.

Edited By Clive Foster on 07/08/2018 16:58:34

[Michael GilliganParticipant @michaelgilligan61133]

Thanks, Paul

Thanks, Clive

Very useful input … much appreciated !

MichaelG.

[BazyleParticipant @bazyle]

Posted by Neil Wyatt on 07/08/2018 16:22:23:

The 'obvious' solution to me is something that senses cutter rotation and synchronises to a stepper motor driven spindle carrying the blank.

All was detailed by John Stevenson on one of the forums (HSM probably) and on his own website when it operated for the system he used on his milling machine for commercial gear production (for the Myford metric approximation gears). Initially using hard divider chips and later a micro when someone else wrote the software.

In essence a 20:1 dividing head plus 200 step motor requires 4000 steps per revolution. The spindle needs to generate 4000 pulses per rev. then it can be divided down to drive the stepper. As a 4k encoder is rare he used I think an 800 pulse one and a 5:1toothed belt drive speed increaser. Other combinations can be used and digital doubling or using falling/rising edges helps reduce the gearing.
If you have quadrature outputs to give a direction and the circuit can follow it you can reverse and keep in step to start a second pass, otherwise you have to have a zeroing procedure or do single pass.

TTL makes it fairly easy to do the dividing if you can get the chips. A software solution has to do integer division not just slap it into the floating point algorithm in the compiler.
I'm not sure if a digital PLL keeps accurate count as the swallow counter means it is only correct on average.

If you are using the lathe for it then once you have the encoder you can do an electronic leadscrew.
Don't forget the blank axis is not perpendicular to the hob axis.

[Pete RimmerParticipant @peterimmer30576]

I've got a hobber Michael. If you want to knock out the blanks I could cut the teeth, or if you're near North Kent you could come and cut them yourself.

Have to check that I have a .8mod cutter first though…

[Clive FosterParticipant @clivefoster55965]

Bazyle an ordinary PLL doesn't keep a accurate count. It just acts to minimise the error between two pulse strings. So long as the PLL frequency response is fast enough and the mechanical system stability is sufficient it won't loose steps. But if it does drop a step out it can't recover. Actually if it does drop out the system design is wrong in some way or other. Something I had major league issues with as that particular system had to three way sync over a wide speed range.

Digital PLL can be made to track lost steps and institute a recovery mode. I'm sure that microprocessors make it far easier today but I'd probably use a set of offset grey code counters to pick up the slip.

In retrospect if I'd had the sense of a midwife toad I'd have re-worked that system into an electronic leadscrew as it had everything needed short of a set of thumb wheels to set TPI and was cheap. Been kicking myself for years over leaving the design behind when I changed jobs.

Clive.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Bazyle on 07/08/2018 17:19:54:

[ … ]

Don't forget the blank axis is not perpendicular to the hob axis.

.

Not forgotten, Bazyle … In fact that was one of the triggers for me to decide to build a 'single-purpose' machine rather than a more versatile one.

The magic number, for the specific hob that I have ordered is 1° 3' 58"

I think we've made good progress on the electronics side. [thanks everyone]

I'm also interested in suggestions for the mechanical design of a stand-alone machine, that preferably fits within [say] a 12" cube.

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Pete Rimmer on 07/08/2018 17:30:09:

I've got a hobber Michael. If you want to knock out the blanks I could cut the teeth, or if you're near North Kent you could come and cut them yourself.

 

Have to check that I have a .8mod cutter first though…

.

That's very decent of you, Pete !

But I think I've got a taste for this project now.

MichaelG.

.

Edited By Michael Gilligan on 07/08/2018 18:42:36

[JasonBModerator @jasonb]

JS's set up on this page and a bit more on p2

Edited By JasonB on 07/08/2018 18:53:33

[Michael GilliganParticipant @michaelgilligan61133]

Posted by JasonB on 07/08/2018 18:52:09:

JS's set up on this page and a bit more on p2

.

Cheers, Jason

… I might need to scale it down a bit

MichaelG.

[I.M. OUTAHEREParticipant @i-m-outahere]

Doesn't give any plans but does show his general set up and software he is using .

Edited By JasonB on 07/08/2018 19:08:47

[I.M. OUTAHEREParticipant @i-m-outahere]

Thanks Jason!

[Neil WyattModerator @neilwyatt]

I have to be honest, I wouldn't use a PLL, I'd just time the rotations accurately with a hall sensor and do the rest in software. a 20MHz timer is going to give you plenty of resolution.

Neil

[Joseph Noci 1Participant @josephnoci1]

A hall sensor and a 20MHz timer just won't cut it I fear…

The easiest and most optimal method is to use the DDA algorithm – an extension of Bresenham's line draw method.

It's simple, works for ANY combination of encoder count value ( sensible encoder values, that is – use at least a 1000PPS encoder) and any number of gear teeth on the blank. Very easy to implement in software, but a 16MHz Arduino will not cut it either! Hence my use of the Nucleo with a small STM processor @ 120MHz – The Nucleo module cost around US$15.00. I can give you all that software if you wish Michael.

A PLL implementation 'does' work, but it's a nightmare working out all the diveders, etc, if done in discrete logic. Also, PLL's are a pain to make stable – phase lead and lag can kill it. A digital PLL is better, easier to stabilise, but needs a really fast processor to up the sample rate.

The DDA method is used in all modern CNC threading and gear cutting implementations.

Joe

Edited By Joseph Noci 1 on 07/08/2018 20:33:13

[Michael GilliganParticipant @michaelgilligan61133]

Posted by XD 351 on 07/08/2018 19:05:33:

Doesn't give any plans but does show his general set up and software he is using .

.

Another thank you

MichaelG.

[John PParticipant @johnp77052]

Hi Michael
Back in issue MEW 193 i wrote the article Gear hobbing in the mill,the system designed by Richard Bartlett (Compucut) is not at all similar to the published hobbing unit in Mew 108 as it remains in sync when the spindle is stopped .The article shows the constuction of the spindle ,the offset design keeps the table in close to the column up to the maximum size of gear that can be cut.I don't know if the unit is still available from Richard, info@compucutters.co.uk
The photo shows the the unit as supplied you just dial in the the tooth count.
Although you are only interested in making spur gears the unit as shown is only a little more work to be able to produce helical gears.The unit can be mounted at either end of the milling table making it easy to produce small left and right hand helical gears with high angles as seen in the photo's .
Arc Euro used to sell the hobs but they are no longer available which is a shame as they were very good cutters at a reasonable price.Would be interested to know where you are getting these from.
John

[Author]

Posts