info
Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Gear Hobbing
131 views
Skip to first unread message
gareth
Feb 28, 2015, 7:54:41 AM2/28/15
to
Some of the early articles on gear-hobbing using a stepper
motor turning the workpiece that was sychronised to the
rotations of the hob used a series of belts and pulleys to up
the speed of the hob so that a sufficiently high rate of pulses
could be had to drive the stepper motor from an optical disc
Now, those of us with an interest in amateur radio will know of
the technique of using a PLL to resolve FM.
Now, the rate of variation of audio is several orders of
magnitude higher than the variations in speed of a cutter
when it is subject to the force of cutting, so it seems
to me that a PLL synchronised to an optical disc right
on the hob axis, without the multiplying effect of the
pulleys and belts, should do the same trick?
What is needed is to divide down in the first instance
so that the workpiece is turning at the same RPM as
the hob, and then divied down agin for the number of teeth.
This, of course, is for straight-cut spur gears for the division
ratios for helical-cut gears are a nightmare altogether, and
a second stepper motor is needed in any case for the feed.
(I refer to my widely-available spreadsheet, "Hobnail" that
calculates gear ratios for mechanical hobbing machines)
motor turning the workpiece that was sychronised to the
rotations of the hob used a series of belts and pulleys to up
the speed of the hob so that a sufficiently high rate of pulses
could be had to drive the stepper motor from an optical disc
Now, those of us with an interest in amateur radio will know of
the technique of using a PLL to resolve FM.
Now, the rate of variation of audio is several orders of
magnitude higher than the variations in speed of a cutter
when it is subject to the force of cutting, so it seems
to me that a PLL synchronised to an optical disc right
on the hob axis, without the multiplying effect of the
pulleys and belts, should do the same trick?
What is needed is to divide down in the first instance
so that the workpiece is turning at the same RPM as
the hob, and then divied down agin for the number of teeth.
This, of course, is for straight-cut spur gears for the division
ratios for helical-cut gears are a nightmare altogether, and
a second stepper motor is needed in any case for the feed.
(I refer to my widely-available spreadsheet, "Hobnail" that
calculates gear ratios for mechanical hobbing machines)
Andrew Mawson
Feb 28, 2015, 9:47:32 AM2/28/15
to
"gareth" wrote in message news:mcsdp5$6in$1...@dont-email.me...
Blimey! When the 'early articles' that I've read about hobbing were written,
the stepper motor hadn't even been conceived :)
Andrew
the stepper motor hadn't even been conceived :)
Andrew
gareth
Feb 28, 2015, 10:21:44 AM2/28/15
to
"Andrew Mawson" <andrew@please_remove_me.mawson.org.uk> wrote in message
news:mcskcp$uvd$1...@dont-email.me...
The stepper motor is actually an AC synchronous motor that has been adapted.
news:mcskcp$uvd$1...@dont-email.me...
Andrew Mawson
Feb 28, 2015, 11:07:05 AM2/28/15
to
"gareth" wrote in message news:mcsdp5$6in$1...@dont-email.me...
>
>
years ago :)
http://www.stevenson-engineers.co.uk/files/hob%20indexer2.jpg
Andrew
gareth
Feb 28, 2015, 11:31:32 AM2/28/15
to
"Andrew Mawson" <andrew@please_remove_me.mawson.org.uk> wrote in message
news:mcsp1u$io3$1...@dont-email.me...
>
> Gareth google John Stevenson and gear hobbing and you'll find he did it
> years ago :)
>
> http://www.stevenson-engineers.co.uk/files/hob%20indexer2.jpg
>
Actually, I've been in touch with the Three Amigos, John, Giles Parkes
> Gareth google John Stevenson and gear hobbing and you'll find he did it
> years ago :)
>
> http://www.stevenson-engineers.co.uk/files/hob%20indexer2.jpg
>
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.
Mark Rand
Feb 28, 2015, 5:25:13 PM2/28/15
to
On Sat, 28 Feb 2015 12:54:34 -0000, "gareth" <no....@thank.you.invalid>
wrote:
> Now, the rate of variation of audio is several orders of
> magnitude higher than the variations in speed of a cutter
> when it is subject to the force of cutting, so it seems
> to me that a PLL synchronised to an optical disc right
> on the hob axis, without the multiplying effect of the
> pulleys and belts, should do the same trick?
>
> What is needed is to divide down in the first instance
> so that the workpiece is turning at the same RPM as
> the hob, and then divied down agin for the number of teeth.
>
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
--
Mark Rand
RTFM
wrote:
> Now, the rate of variation of audio is several orders of
> magnitude higher than the variations in speed of a cutter
> when it is subject to the force of cutting, so it seems
> to me that a PLL synchronised to an optical disc right
> on the hob axis, without the multiplying effect of the
> pulleys and belts, should do the same trick?
>
> What is needed is to divide down in the first instance
> so that the workpiece is turning at the same RPM as
> the hob, and then divied down agin for the number of teeth.
>
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
--
Mark Rand
RTFM
Richard
Mar 1, 2015, 2:39:43 AM3/1/15
to
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
gareth
Mar 1, 2015, 5:48:06 AM3/1/15
to
"Mark Rand" <ra...@internettie.co.uk> wrote in message
news:p4e4fa542es58jr7a...@4ax.com...
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,
> 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.
Are you thinking of a traditional machanical hobber when you say that?
>
> 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.
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.
news:p4e4fa542es58jr7a...@4ax.com...
> On Sat, 28 Feb 2015 12:54:34 -0000, "gareth" <no....@thank.you.invalid>
> wrote:
>
>
>> Now, the rate of variation of audio is several orders of
>> magnitude higher than the variations in speed of a cutter
>> when it is subject to the force of cutting, so it seems
>> to me that a PLL synchronised to an optical disc right
>> on the hob axis, without the multiplying effect of the
>> pulleys and belts, should do the same trick?
>>
>> What is needed is to divide down in the first instance
>> so that the workpiece is turning at the same RPM as
>> the hob, and then divied down agin for the number of teeth.
>
> 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.
Sorry, misled you, for _TWO_ division chains are needed, one for the
> wrote:
>
>
>> Now, the rate of variation of audio is several orders of
>> magnitude higher than the variations in speed of a cutter
>> when it is subject to the force of cutting, so it seems
>> to me that a PLL synchronised to an optical disc right
>> on the hob axis, without the multiplying effect of the
>> pulleys and belts, should do the same trick?
>>
>> What is needed is to divide down in the first instance
>> so that the workpiece is turning at the same RPM as
>> the hob, and then divied down agin for the number of teeth.
>
> 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.
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,
> 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.
>
> 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.
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.
gareth
Mar 1, 2015, 5:54:14 AM3/1/15
to
"Richard" <rics...@gmail.com> wrote in message
news:clfu1t...@mid.individual.net...
> 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!
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 :-)
news:clfu1t...@mid.individual.net...
> 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!
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 :-)
gareth
Mar 1, 2015, 6:00:45 AM3/1/15
to
> 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.
Sorry, I don't follow that despite holding back my commentary on your
> 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.
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.
gareth
Mar 1, 2015, 6:02:07 AM3/1/15
to
"gareth" <no....@thank.you.invalid> wrote in message
news:mcuqnq$9o2$1...@dont-email.me...
(See later post !!!!)
news:mcuqnq$9o2$1...@dont-email.me...
>> 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.
>
> 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.
COMMENT RETRACTED!
>> 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.
>
> 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.
(See later post !!!!)
Richard
Mar 1, 2015, 10:05:15 AM3/1/15
to
relative to the gear face? I am assuming a helical hob of the type I
have bought.
Richard
gareth
Mar 1, 2015, 12:00:26 PM3/1/15
to
"Richard" <rics...@gmail.com> wrote in message
news:clgo58...@mid.individual.net...
case it was an attempt at a worm-wheel with the cut plunging
downwards and not transversely as for gears.
Richard
Mar 1, 2015, 3:20:57 PM3/1/15
to
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
gareth
Mar 1, 2015, 3:31:06 PM3/1/15
to
"Richard" <rics...@gmail.com> wrote in message
news:clhal7...@mid.individual.net...
> Ah now I understand "cut the teeth on the side". The two clamped together
> was the start of a method of backlash elimination?
Yes, that is right, for after cutting in tandem, then one would be spring
> was the start of a method of backlash elimination?
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.
Richard
Mar 2, 2015, 2:53:48 PM3/2/15
to
> 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.
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
paul mayhead
Apr 23, 2017, 3:18:03 PM4/23/17
to
replying to gareth, paul mayhead wrote:
Hi Gareth
How can I get a copy of Hobnail please?
--
for full context, visit http://www.polytechforum.com/modelengineering/gear-hobbing-54531-.htm
Hi Gareth
How can I get a copy of Hobnail please?
--
for full context, visit http://www.polytechforum.com/modelengineering/gear-hobbing-54531-.htm
Gareth's Downstairs Computer
Apr 23, 2017, 3:43:39 PM4/23/17
to
On 23/04/2017 20:18, paul mayhead wrote:
> 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!
> replying to gareth, paul mayhead wrote:
> Hi Gareth How can I get a copy of Hobnail please?
>
My Email address (when you demunge it) is valid.
Also, out of interest, what forum are you posting into?
Clrob3437
Dec 20, 2022, 3:32:17 PM12/20/22
to
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
--
For full context, visit https://www.polytechforum.com/modelengineering/gear-hobbing-54531-.htm
--
For full context, visit https://www.polytechforum.com/modelengineering/gear-hobbing-54531-.htm
gareth evans
Dec 20, 2022, 5:51:29 PM12/20/22
to
On 20/12/2022 20:32, Clrob3437 wrote:
> 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
> 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
>
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.
Clrob3437
Dec 21, 2022, 4:32:12 PM12/21/22
to
Hello Gareth. I am very pleased to hear from you. I too got to know Giles Parkes and what a gentleman he was. He arranged for me to take his place demonstrating the Jacobs Machine at the annual Midlands ME Exhibition. Up to that point I was not aware of Hobnail and struggled to find much information on determining change gears for helical hobbing and ended up writing an Excel spreadsheet myself based on continued fractions. I am sure that it's not as accomplished as yours. Can you tell me what mathematical method you used for Hobnail? I would e-mail you directly but as I am not a member of this forum I cannot see your address. Yes, I would very much appreciate a copy. My address is chris.r...@clara.co.uk. The rules may not allow this address to appear in the post so could the moderator please send it to Gareth if it must be removed ?
Thanks & regards, Chris.
Thanks & regards, Chris.
0 new messages