Hobbing a worm gear Part 1 and Part 2

[kinetic (Steve)]

Tonight I performed a test to see how well the RA clutch works
on either side of the meridian.
Scope was balanced, clutch engaged and a logging set done
using K3's Drift Explorer.
I used my modded SPC900 at Newt focus on the 12"
First set was done in the western sky 'going downhill'
Second was done east 'going uphill'.

A slight tracking error rate was added also.
PE shape repeats with the same sine wave structure regardless
of which side of the meridian I am on.
So I take that to mean, the clutch is reasonably reliable when balanced.
I think I'm correct on this....

I then put my DSI II mono back in and did a set on NGC 7213, a Seyfert
galaxy in Grus.
Note the diffraction spike from Alpha Grus (out of field, top right)
PEC was on and all tracking error nulled out for the set.

Steve

Steve

[kinetic (Steve)]

While still continuing to refine PEC on the homemade RA gear
I have been almost completely around the circumference of the
gear following stars etc.
Arc sec per step size has not needed to be drastically altered so far.
(this is good news....this means the spacing around the gear of the
hobbed teeth is pretty accurate.
My only big PE is the non-concentric worm.
This, as in the last few posts, is manageable with Bartels PEC
turned on. Luckily the PE is cyclic and repeats almost identically.

However, to move forward from here requires many nights of
collecting accurate PE data with K3 drift Explorer and a spreadsheet.

This is the easy part. The hardest part has been getting an accurate
sync point when switching on PEC.
I have to toggle it on right at the correct position of an imaginary clock
face on the worm shaft.
The Bartels system even has a function to add this sync via software
and a pin on the parallel port. For now I just wanted some sort of
visual feedback that I was in the vicinity.

With the occasional resync needed from inside with remote desktop
this becomes impossible......enter the webcam.

I fitted a rough printed dial to the worm shaft and sat the dome
webcam looking at it.

All of this is now visible via remote desktop from inside and I can
call the PEC on /off via the ASCOM interface and custom LX200
commands.

It also lets me check visually there has been no drive trouble..handy.
(a grub screw came loose on the 60:1 gearbox output shaft and
the toothed belt pulley was slipping). I saw this via the webcam!

Pics attached. A screengrab of the dome PC with Cartes
and ASCOM connected driving the scope.
The live webcam view is top left.

Note the advertising on the webcam image. You just can't
get away from bloody advertisers can you

Steve

[kinetic (Steve)]

This is the latest refined curve of the worm gear.
This curve is synchronised in at the index point and
activated.

This curve has been refined from many guide cycles and has
been averaged using Mel's software.
His software can extract both drift and PE from a guide session.
It implements the drift component straight away unless you disable
it.

This PE is mainly caused by the non-concentricity of the worm.
Correction cancels most of this out but of course it would be good
to have this at a minimum to start with.
I have a way to do this. Luckily I stumbled on the way because
of how my Dad suggested I recover the worm from the hob.

In the first few worm/hob/blank tests I would 'recover' the worm by
boring it out very carefully and installing it on an 8mm shaft on
the worm carriage.

The simple act of boring something and parting it off releases tensions
in the steel that otherwise was rotating perfectly concentrically.

From that point on, the worm you think was perfect and concentric
is not! It has eccentricity now but only slightly.

Dad suggested eliminating the 'recover' process of boring the worm
and doing the whole turn down/cut acme thread/cut bearing shoulders/
hob a blank/part off process in one go.
No non-concentric error should happen. Or at least it would be far
less than the 'recover' process.

Finally, the method to minimise this error on an already made worm
sitting on the 8mm sleeve is to put a set of grub screws at 90 degree
intervals around the worm.
That way I can rotate the worm until the 'high spot' is at the 'low'
spot. This should cancel out most of the worm PE.

Edit: some recent results

Steve

[Spanrz (Brett)]

Steve, I see your getting some good results off the gear....

I have just purchased a worm gear arrangement, from Small Parts.
The price dropped to a good number ($130 odd) including express shipping for the set.
They are apparently "on sale" as they are in surplus, so I got it, in the right time
As before when I was looking, it was over $200 not including shipping.

I was going to wait till I got over the USA (planned trip in Jan-Feb) and purchase some gears over there, but my time is very limited on looking for the parts whilst there.
I have decided to buy these gears on sale to atleast get my brain out of "hold mode" as I was going nuts not doing anything about it.
Atleast I can now look for a gearhead / reduction box now, to reduce it to arc seconds.

Your above info, is invaluable as it's the exact info I am after.
I would have preferred to have a crack myself at a worm gear, but the infrastructure isn't there.
Only time will tell how accurate the gearing is.

The worm gear is 360 teeth. Giving 1 tooth per degree which = approx 239 seconds per tooth.
Now to spec the reduction. See how it goes.

[hikerbob (Bob)]

I've been back at it again.
I'm still not cutting threads of any merit so am sticking with recycling the worm thread from other components. The latest (and so far great) is the thread from a Trade Tools G clamp which sells for around $12.

The thread is not hard enough to do the initial cutting (it rounds over badly) but does seem fine for the final stages of cutting the worm gear. I've been roughing out the worm gear with a conventional tap of the same pitch and a slightly larger diameter then finishing it off with a slotted piece of the clamp thread.

I'm housing the worm in a piece of 25mm OD alloy tube with a 19mm ID and an 8mm shaft turned on at each end. Bearings and pullies from HE to suit and it's all pretty easy although I've not proven yet just how concentric the result is.

I've still got some cleaning up to do but this is looking promising.

Steve's mention of cork washers to allow for high load slippage solved the problem I'd had in working out how to do that. I had some old cork gasket material lining a toolbox which should be just right. I may still swap the tensioning screws over to grub screws.

Bob

[kinetic (Steve)]

Brilliant Bob,

How did you tackle end thrust with the HE?
That will be a big contributor to PE and backlash if you don't have
that taken care of.
The thread looks NICE!!!!
Oh and old kitchen table placemats are a good source of cork.
Tip em upside down and they are sometimes laminated with cork.
Wait till you feel how sublime the cork /pressure plate works...
That's thanks to Dave Gee btw...

Steve

[hikerbob (Bob)]

Steve I've not dealt with the end thrust issue yet. The bearing on the motor end of the worm is pushed up hard against an internal circlip and I've not doen the other end yet. I left some extra space in the tube at that end for that but anm still thinking about it.

Bob

[Rod]

Hi,

I hope this is the right place to ask this question. I have a bartels stepper driven scope which uses molded epoxy worm gears. These work OK but one day I would like to build something more accurate. I don't have a lathe so I wondered how good a gear I could produce with a jig like the one described here:

http://members.shaw.ca/gargwolanski/Gearcutter.html

Rather than use aluminium, I had thought about using plastic which would probably be quieter when paired with the stepper and metal worm.

What do you think?

Thanks,

Rod.

[kinetic (Steve)]

Originally Posted by Rod

Hi,

I hope this is the right place to ask this question. I have a bartels stepper driven scope which uses molded epoxy worm gears. These work OK but one day I would like to build something more accurate. I don't have a lathe so I wondered how good a gear I could produce with a jig like the one described here:
http://members.shaw.ca/gargwolanski/Gearcutter.html
Rather than use aluminium, I had thought about using plastic which would probably be quieter when paired with the stepper and metal worm.

What do you think?

Thanks,

Rod.

For what it's worth Rod, I think it might be a viable option if you
have a good set of bearings in the drill head with no end play.
As long as you can guarantee the hobber has no slop at all and
you can slow the speed down to a nice cut speed I can't see
why it wouldn't work.

You also need fine crossfeed and transverse feed controls.
That may be a harder thing to set up. You need to add fine depth
cuts.

Having said all of that, I'm speaking from a position of ignorance,
I haven't tried it.

As far as accuracy of the lathe-hob setup for cutting medium
accuracy R.A. worms :-

A few recent images with exceptional seeing validate it as
a good, cheap alternative to the commercial option.
That's always a good thing for us ATMers
Latest image of Eta Carina here:
http://www.iceinspace.com.au/forum/s...ad.php?t=51934

regards,
Steve

[kinetic (Steve)]

Last night while capturing this result the worm assembly made
a quite loud clunk and the live preview showed a star trail.
Lucky I was in the dome (nornally I image by remote desktop)
because I immediately stopped to investigate the noise.
I was sure the worm gear had just lost mesh and damaged the hob
surface but luckily it wasn't that bad.

First I suspected the nylock nut ( Pic 1) at the end of the thrust bearing
had , for some reason, wound itself tighter and made the worm
tighter......causing the toothed belt to jump a tooth.
When I pulled it all apart today I found the baseplate the worm
assembly sits on was to blame. Well I was to be more accurate.
I had tightened the bolt ( also shown in PIC 1) only 24 hrs earlier.
This and another 3 are what mounts the whole base plate
to the GEM and it sits on some very fine shims which mesh the
worm to the worm wheel.
I had forgotten that these were only nipped up and not excessively
tight. Of course, tightening one had meshed the worm in slightly more
and it had only driven itself to a slight high spot and got tight.
Luckily the first thing to give was the toothed belt jumping and
not a worm gear chewed out.

While apart I also noticed scrape marks from the clutch plate pressure
bolts which had chewed through the cork.(Pic 2/3)
I mounted the worm gear on the lathe index plate and re-faced this surface.(Pic 3)

All back together and meshed sweet, back in business.(Pic 4)
The advertising on the PEC index wheel is a bit greasy now and
should be replaced (PIC 5)

Edit: wow, I had trouble finding this old thread, I can't believe it's
got 7500 views!


Steve

[MrB (Simon)]

Glad there was no damage. Things like this always help out in the long run.... eg, discovering the prob with the clutch.

7500 views!? Popular thread!

[kinetic (Steve)]

Originally Posted by MrB

Glad there was no damage. Things like this always help out in the long run.... eg, discovering the prob with the clutch.

7500 views!? Popular thread!

The re-grease and re-seating the mesh certainly did no harm Simon.
I didn't lose the PEC index mark either so that's a bonus.
It wouldn't hurt to do some PEC curves anyway to see if anything
needs a tweak. I'm sure it should if mesh has altered.

Popular? yeah, maybe I should write an article

Steve

[kinetic (Steve)]

The drive train was tested last night in an all-nighter.
PEC curve is almost unchanged. The index point is exactly
at the same 'point on the clock-face' as before and the shape
of the correction curve as it was. Only a slight change in
it's magnitude.

Some results in this thread at post #13

Steve

[kinetic (Steve)]

Another clunk in the dark last night followed by another every
4.5mins....(the worm period)!

Last night the worm gear had returned to the spot where the
problem had been where I had the toothed belt jump a tooth.
I had marked the section with red pen just in case.
Seems like there is a high spot still that has been exposed
once again by a change in mesh. Thought it was fixed.
I will pull it all down again but this time I might try the
'paper between the mesh' method I have read in Peter's
Hypertune thread.
This will almost certainly introduce some backlash and need
a new PEC curve generated. I have had almost nil backlash with
this setup, the drive starts following stars immediately at 1524mm FL.
I'm reluctant to introduce backlash but if it removes a high spot and
protects the worm gear so be it.....

Steve

[kinetic (Steve)]

One of my good tech friends has been keeping an eye out
for gearbox reductions and clutches etc after I described my
successful worm drive to him.

He gave this to me and said, "is this any good to you?"

It has a 30v DC motor with a drive clutch straight on the
back, driving a seriously well made, heavy duty brass gear
train. And a position feedback just prior to the output shaft.
With a few simple mods like:

Stepper in place of DC motor,
Rotary feedback pot swapped for optical encoders,
Output shaft driving a worm/worm gear,

...it could easily be the drive for a rather large GEM.

One man's junk... eh

Steve

[kinetic (Steve)]

After a few nights of deep sky stuff and periodic random
clunks in the dark I think I stumbled on a big reason why
the drive has been doing this.......grease!
I've been driving the worm gear around well past the 'clunk'
high spot that happened weeks ago and the gear had driven
itself to an ungreased section! Doh!

I had not greased the entire 360 degrees of worm gear after the
clutch face trim on the lathe.
Obviously, once it had reached the un-lubricated section, only the
remaining grease on the worm itself was supplying the lube!
Worms and worm gears don't go well just metal on metal

Greased and solved last night in the wee small hours.
Another problem has been the index point is now wrong.
A few nights ago one of these clunk/ tight spots had caused the
toothed belt cog to turn about 2 hours on the index 'clockface'.
I have to re-do a PEC curve set and re-index the wheel.

Steve

[kinetic (Steve)]

Ok, after a few more hiccups and a week of gusty gully breezes
messing with PE refinement, I finally got a few sets to look at
in Excel and redid the PEC array in the Bartel drive.

The final night was above average seeing which helped a lot
to both quantify and test the PE.
My index point has now changed and I have to remember this when
I engage the error correction.

First pic is a few PE sets overlaid with the same index entry point.
You can see a bit of deliberate RA rate error (the downhill slope) on
all of them.

The next set is the result of half a dozen PE sets averaged and smoothed,
then fed into the Bartels PEC array. This set has zero RA error and
represents 2 complete cycles of the RA worm with PEC ON ,
then lastly , 1 complete RA worm cycle with it off.

I haven't worked out how to adjust all Excel values up and down so they
dip above and below zero (simple Excel maths formula stuff!) so I added
a line at value 18 because this is the index start.
So for all illustrative purposes this line represents zero.
Reading this it seems I have about +/- 4 pixels error in RA with Pec
turned on and +25 / - 18 pixels with it off.
At 1.14 arc sec per pixels this represents no larger than +/- 5 arc sec
of PE with it turned on. Now that got my attention because previously
I could only get it under +/- 10 on a good night and +/- 15 on average
nights when close to the celestial equator.

Pic 3 is a close up crop of centre stars in Omega Centauri with PEC
turned on.

Steve

[kinetic (Steve)]

Some more pics tonight with the refined PEC curve returning
consistent +/- 5 arc sec PE drift in RA.
Only problem was my index point has moved again! Grrr.
It was 3 o'clock on the index clockface and now is 4 o'clock.
Something that I can't fathom is shifting this and this time it's
not the worm grub screws.

Each set here was done right after a guide-save session in Bartels
scope drive and the curve smoothed.
Then dumped to the PEC.dat file...then finally used with NO GUIDING.

So it really is a custom PE for the section of the worm that has
already just been driven past...So technically it doesn't apply to
the section the capture is subsequently made on. If that makes sense.
In 3 sets , with 3 consistent PE figures under +/- 5 arc sec, this tells
me that the worm hob is fairly consistent around the circumference.

Steve

[kinetic (Steve)]

I've had some recent scares with the worm gear coming completely
out of mesh twice without warning.
The scope has tracked beautifully for hours then suddenly done an out
of mesh nosedive!
I've luckily been in the dome both times to catch it and the second
time last week I heard the worm scrape across the mountaintops of
the worm gear hob. A truly heartbreaking moment.
I thought I had definitely chewed out the hob surface both times.
I was at a loss to explain why this had happened because the worm
carriage is completely bolted down solidly and cannot drift out of mesh.
Secondly, the RA shaft cannot drift upwards and out of mesh in it's
bearings either. There had to be something in the system that was the
cause and I reckon I have found it.

Firstly I pulled it all down and was relieved to find no obvious damage to
the hob surface. What I did find though was the worm carriage had some
slight play in the brass bush on the drive end.
Not enough to cause a complete loss of mesh though so there had to
be something else.
What I found next was the worm when bolted down firmly had a
tendency to pull sideways to one edge of the pre-cut groove
shape of the worm gear.
I've attached a rough paint pic to illustrate this.
Even though the pre-groove should seat the worm snugly, if I let
it bolt down firmly to slightly one side as seen in the right side pic
of the pic, then only a substandard mesh has been achieved.
It felt like proper mesh but when combined with the play in the worm
bush it must have been enough to come completely out of mesh at
the point on the worm gear that was slightly lower. I know I have some
ever so slight non-concentricity of the worm gear.

Anyway...PIC 2 is a shot looking straight down the axis of the worm.
Even though it' slightly oblique an not a good focus, you can see
a slight misalignment of the worm in the pre-cut groove.

All bolted back together now and tracking fine. No obvious play in
the 360 dgerees of travel. (checked with a battery drill driving the
worm direct drive).

Hopefully back in business.

Steve

[Spanrz (Brett)]

My thoughts Steve, a ball roller (anti-friction bearing) to be used instead of a bush.
Or with new bushes, use Moly based grease.

Yes, I hate the sounds of metal grinding too....