Pin router fixturing for escape wheels and larger gears

[john e]

For this work, I made three variants of the fixtures. The first is in this post. (I will put the pics in first, then edit like crazy to keep under 30 minutes.)

Pic 1. This is the first attempt. The center hole pattern is setup so that you put a hole into the work, then rotate the pin for that hole depending on how many teeth you have. For example, 60 holes in the jig is useful for 60 teeth, 30 teeth, 20 teeth, etc. The pins to either side are to capture the plate with the tooth profile. The sizing of the plate limits the lower size of the gear.

Pic 2 has a piece of lucite in place, and the escape wheel pattern template. Note the rotation arrows indicate the direction the fixture should be run, as going the other way can kill teeth. (experience). Also note that reversing that plate gives me the 3rd wheel pattern as well. Notice that I had to mill out some thickness on that template, I ran out of headroom between the pin router bridge and the fixture.

I put two indexing holes into the gear blank, as the template gets in the way, so really, the holes on the top half of the fixture are pretty useless. Luckily, I just luuuv drilling lots of holes.. I also make sure that the indexing holes in the gear will be in a location where the gear cutouts will go.

Pic 3 is the lucite escape gear in the fixture as viewed from below.

Pic 4 is what happens when you get too thin on the gear. I made a template to pin rout the holes, and the bit broke a section of the gear away. If you look closely, the 3rd wheel outer ring of yellowheart is comprised of 10 glued sections so that the grain is in the strong direction for tooth forces. Unfortunately, it is in the wrong orientation for the router forces. This is where I said to myself...self, what about gluing up multiple layers like a plywood setup to strenghten both directions.

Pic 5 shows what happens when the base of the teeth and the glue joint coincide..heartache number 2.

Pic 6 was accidentally put in, I dont' know how to delete the pic. This is the walnut pallet assembly, the brass is a 1/4 dia insert tapped 8/32 for an allen setscrew used to push some leather against the tube at the arbor. Boyer has that hole as tapped directly into the birch plywood, and walnut is incapable of supporting tapped threads that way, hence the brass insert.

Pic 7 is the final escape wheel. Note that the mahogany teeth have a much larger thickness of complete mahogany, the teeth don't break at the root.

John

 

[john e]

The second fixture was designed to be a bit more flexible in terms of size, it allows smaller gears to be cut. Pics first, then text..

Pic 1 is the stage I use to drill the holes around the indexing plate.
Pic 2 is a finished plate.

I use an excel spreadsheet to calculate the number of turns I have to rotate the rotary handle to get the number of holes per 360 degrees. I can put any number into the excel, so can make any indexing wheel I wish.

Pic 3 has the indexing wheel pinned to a wheel blank, there is a thin layer of scrap below the blank, and on the top is the tooth profile plate. If you look carefully, there is a half inch aluminum rod glued into place, more on that from the following pics.

Pic 4 shows the result of bit chattering against the work. I used a 1/8th bit and a 1/8th pin, so it was one pass. Every so often, the work would chatter, so would kill the tooth being cut. There was no process method that would stop that occasional problem, and I'd hate it to happen on tooth 64 of a 64 tooth gear.

Pic 5 shows what was happening as a result of excessive force used to prevent chatter. I was bending the index pin in the plywood holes of the blank and the fixture. That clearly doesn't work.
Pic 6 shows what happens when I do a first pass using a pin that was 3/16th first with an 1/8th bit, then a second pass using the 1/8th pin. There was much less force as a result, and I made good looking teeth.

Pic 7 shows what happens long term to the indexing pin and the plywood hole in the gear blank. It ovalled significantly, so the teeth started to warp. If you look back at pic 3, that is what the aluminum insert is for. I drill a hole in that, and the indexing pin no longer can tip to the side.

Pic 8 shows what is needed for different wheels. In this case, I made two profiles, one for a 42 tooth and another for a 35 tooth wheel. Note also that the valley profile is not like the Boyer teeth, these were actually for some wheel repair gears. The roots had to be cleaned square as the router cannot do that.

For those gears, I used 3 small brads to lock the blank to the indexing wheel. The final gears have no inner cutouts, so I could not use a locking pin as per the previous gears.

John

 

[john e]

The third fixture was a bit more robust. I used it for the final gears of my Tempo clock.
Pic 1 shows that I pinned the index plate to the top of the work. What is missing from this pic is the plate I put over that to catch the holes of the index plate. If you look carefully, you can see the pins I put in at 3 and 9 o'clock.

Pic 2 is the third wheel. For that one, I used the bottom plate for the indexing, it also has holes.

Pic 3 is the second wheel halfway through. Note that the cuts do not go through the full thickness, as that wheel is 3/4 inch thick. The 1/8th bit I use can only go 1/2 inch.
What I did was cut all the teeth partway through, and then remove the blank from the fixture, flip it over, and use the tooth surfaces against the pin of the pin router to complete the teeth. If the alignment is very good and the pin is exactly the right thickness as the router bit cutting diameter, in theory it should just trim the rest of the wood flush to the already cut teeth. It actually worked.

Pic 4 is the finished 2nd wheel. Walnut center, and maple on both sides.

John

 

[David S]

John I am not a woodworker and am trying to figure out how I could employ what you have shared on the metal working side of things.

Very nice work and thank you for sharing.

David

 

[john e]

John I am not a woodworker and am trying to figure out how I could employ what you have shared on the metal working side of things.

Very nice work and thank you for sharing.

David

Thanks.

The fixturing of the wood is not good enough for metal I suspect. The only way I could see these methods as being applicable to metal is with a wire EDM. A follower system can be used with gap spark feedback controlling the velocity of cut, but even that is only a gnat's hair away from a full CNC method. Trying to use fixtures of this variety with metal would really scare me, even aluminum would play gotcha....and win.

The only thing I've been able to transfer over to metalwork is the excel spreadsheet and rotary stage method of creating indexing wheels. That came in handy for my work in brass on the unimat as well as a pair of watchmakers lathes. While a 60 count indexing wheel is divisible by quite a few numbers, try as I might, I couldn't get an even number for a 35 or 42 tooth gear. Go figure.

I've attached the pic from my presentation, I hope it's clear enough.
The rotary stage I used has a wheel for rotation, it shows degrees and smaller hash marks, 20 marks per degree. This sheet gives me the indications for a specific number of teeth. The tooth number is in the second column at the top, and the sheet gives every location for 360 degrees. The second column gives the angle, the third gives the integer part of the angle, the fourth column converts the remainder of the angle into "tics" or hash marks.

This example was for the 35 tooth wheel I repaired for a friend.

John

 

[Tinker Dwight]

Making things like 35 or 42 is not too hard.
35 is easiest. You make 2 wheels, one with 5 holes and the other with
7 holes, both at the same radius.
On the 7 hole you pick a reference.
On the 5 hole, you have a index hole that is aligned
along the radial line of one of the five holes.
You can now create the 35 locations by adding or
subtracting combination of the two by placing an alignment pin
through one hole on each the 5 and 7 at the same time.
Depending on where you aligned the two disk, you can
get all the possible teeth locations.
No need to drill 35 holes.
Basically you are doing (( a * ( 1/7 ) + b * ( 1/5 ) ) * 360 modulo 360
It is based on breaking it up into the prime factors. You can do a
21 tooth this way if needed, with a 3 and 7.
You can't do a 19 or 23 as they are already prime.
I'd need to think a little more about 42 holes but I
think you can make one disk with holes spaced at
2 holes and one with 21 holes or maybe a 18 and 7.
It makes sense to number the holes in each and
make a chart for the alignment pin and reference
pin, for each tooth( 1 to 35 for the 35 tooth ).
Tinker Dwight

 

[john e]

That is totally possible. I recall reading about this method, I forgot what book it was in.

Two issues I've considered:

1. When cutting wooden teeth with the pin router, I found that the order or sequence of tooth cutting made a difference. My escape wheel for example, was not happy if I cut teeth such that I was pulling out of a complete tip, but happy driving into the next valley completing a tip. A lot had to do with the cut rotation and feed direction, as well as grain orientation. My escape wheel currently uses 10 segments for the teeth such that the grain angle never crosses the tooth edge, but my 3rd wheel was a three layer laminate and the outer layer grains do become sensitive to bit cut/feed orientation as well as grain; a definite learning curve.. I suspect that it would be possible to carefully cut the teeth in a pseudo-random pattern, like every fifth tooth, then a shift of 360/5 for the next group of 7. I could even setup such that a simple lever and pin arrangement catches one set of holes, with the other set catching a hole in the work.

2. Double holes and pins means double the tolerance error built into the gear. If my holes have a tolerance of 3 mils on center, two holes would add to a possible 6 mil error. At one inch radius, that is 6 milliradians, or about a third of a degree. You're intent on making me work for accuracy, aren't you? I mean, granted, it's wood we're talking about..but hey, if ya can measure it....


If I make a single indexing plate with 5, 6, 7, 8, 9, 11, 13 holes, and can drill one hole in the work, this becomes very very simple. I pin the work in the 5 hole radius for example, then pin the 7 hole using the locking arm, I get 35. If I use the locking arm at 9, I'll get 45..6,and 7, 42.

If I drop the pins to 1/16th dia, I can squeeze the indexing wheel diameter, but at the expense of holding strength. But If I make the base fixture by press fitting a Tee shaped arbor from the bottom side to prevent pullout, I can devise a clamping mechanism on top, perhaps a 10/32 or 8/32 tapped thread for a knurled locking nut..
Also, then the outer pinning arm can be made with a tapered pin for engagement accuracy and ease of setup. It also eliminates all holes from the baseplate other than the center arbor tee, the work cutting clearance slot, and a pair of holes for securing the locking arm pair.

I like discussing with you, thanks. I think I'm gonna build a new metal plate very soon.. Will take plenty of pics.

Also, I think I'll make the indexing wheels so that I can add them to the unimat over at the clock school, we occasionally work on wooden works. This indexing scheme would be great for cases where the cutting bit axis is parallel to the face, as long as the work is clamped between a pair of sacrificial blanks to prevent tearout.

John

 

[Jim DuBois]

There are a couple of ways to prevent or minimize "tearout" on cross grained teeth. One of which is to sandwich the wheel blank in between 2 pieces of disposable materials, such as 1/8"-1/4" plastic, or ¼" wood, clamp it tightly, and cut the teeth. This is successful only if the wood is not overly dry, or the cut is not made too fast, or the cutters are not dull. Keep in mind any time the cutting force on the area of wood being machined exceeds its tensile strength the wood will crumble or chunk out.

A second way is to stabilize and strengthen the wood before commencing cutting. For years bowl turners have used Polyethylene glycol for this purpose. (PEG) is a polyether compound with many applications from industrial manufacturing to medicine. PEG has also been used to preserve objects that have been salvaged from underwater. It replaces water in wooden objects, making the wood dimensionally stable and preventing warping or shrinking of the wood when it dries. In addition, PEG is used when working with green wood as a stabilizer, and to prevent shrinkage. It also adds a certain degree of strength and seems to give the wood a less rough feel.

Problems seen in this thread are suggestive of why wood wheels were gang sawn back in the day. There has been a fair amount written of this process by George Bruno and Ken Roberts, see Ken's book Eli Terry and the Connecticut Clock for additional details. http://articles.courant.com/2003-02-01/news/0302011632_1_eli-terry-clock-case-bruno

I am a proponent of CNC work when possible. In years past CNC could be both expensive and complex. Today, not the case. A 3 axis CNC router can be built using eBay available parts and hardware from readily available sources for very little money, like under $100 with careful shopping. My first CNC machine I built I had under $50 in it when I started cutting gears on it. Rex Swensen http://members.optusnet.com.au/~swensenr/index.html offers a lot of free SW as well as advice and suggestions in regards to CNC clockwork. I have built machines as small as 2"x2"x2" travel, and machines as large as 48"x48"x16" using surplus parts and often junk yard parts. Accuracy is a bit dependent upon ones mechanical skill as well materials used, but accuracy of .005" or less is pretty easy to achieve, with .001" being reasonable, and .0005" possible using off the shelf surplus stuff. SW can be free. Ball screws are required in my opinion if you want real accuracy.

Given a desired wheel dimensions, it can be drawn in a matter of minutes, converted into machine code and cut (routed) in a very few more minutes.

Regarding creating index wheels of unusual tooth counts a good method to create such wheels involves the use of "plumbers tape" . http://www.suremarineservice.com/images/products/detail/3051t23stainlessplumberstape.1.jpg I am speaking of the steel tape that has holes punched in it. The holes are quite accurate. To make an index wheel of say 23 or 37 or what ever it is just a matter of counting out the holes, making a ring out of the tape with the appropriate number of holes, fit it to a properly sized plywood disk, put the disk on your indexing arbor, fit a stop pin, and index and cut away…. Or you could use a simple free DOS gear cutting program and a $5 stepper motor, an $8 driver, a very old and most likely free computer, and cut any number of teeth from 1 to maybe 280,000…..with no additional effort….of course you may need to create profile cutters.

 

[john e]

Thanks Jim

What you call disposable materials, I called sacrificial. Same idea.

What would a PEG soaked piece of walnut, maple, or mahogany look like? Does it change the color or appearance?

I'll check out the book and link when I get a chance, thank you.

CNC? That's cheating. It's akin to those fixtures they showed on how it's made, where the operator loaded the bottom plate, then locked the parts with two fixtures, dropped the top plate onto it, then added the 4 nuts...where's the challenge in that?? I mean, so what if it takes less than 30 seconds to assemble a mantle clock.

I run a pair of 11 axis machines, so can really appreciate the technology. I'm building an Wire EDM unit for my own use, so I may be giving you a ping for some materials advice..

Ball screws are interesting beasts. The stuff I'm doing now is wrestling with accuracy issues. Over a span of about 100 mm, we're getting repeatable errors on the order of 7 microns. I'm not sure if it's the screw machining or one oversized ball. Luckily, the encoders we use have 10 to 100 nanometer accuracies, so we can correct.

Do you have a favorite source for quadrature encoder to usb interfaces? While I do use steppers with microstepping chips, I'm still a DC servo guy at heart. Plus, I've got about 6 quad encoders in my bin, I'd hate to just toss em.

John

 

[Jim DuBois]

John, you are certainly well ahead of me when you are concerned about micron accuracy. I have done all my stuff on the cheap, none of it servo drives, so i can't help on the encoders and the like. Always have done stepper motors, they meet all my needs as they exist today and stay within .001" accuracy. And my most sophisticated machine is 4 axis. I would love to have 5, 6, or 7 axis machines, but that is not what I have and I am very much afraid I would have a great deal to learn if I were to buy or make a many axis device. Most of my work is 2 1/2 axis work to be entirely truthful.

Regards PEG, it tends to make what ever wood it is on look "wet". On light woods it will add a very slight green tinge. Assuming you plan to shellac/lacquer/varnish the finished gears the PEG should not show IME. I have not used any in 30 years so I am not current on its use. When I did use it, it worked well, but the wood needed to be soaked in it for some time. Thicker blocks of wood take longer....pretty much expected. Some users say it works best on green woods, not seasoned...but they are turners and seasoned wood in lumps has usually already cracked, which if immersed in PEG while green, is minimized.

Regards the wire EDM, there are 2 machines I have always wanted, a wire edm and a water jet. Neither is really practical for garage shop use, but I did look over the plans offered in Home Shop Machinist? to make a small unit.....I got sidetracked and never went forward on that....is that set of plans your starting point for the wire EDM or do you have a better set of plans or ideas?

 

[john e]

Micron and sub micron accuracy is at work..at home, or play as I refer to it, I'd be happy with half mil or mil accuracy for normal machining, but for wire edm, I would like to make finished surfaces which do not require additional polishing. Microwire EDM requires speeds in the micron per second range, so that kinda limits how coarse I can go and not stall the EDM process with shorts.

I'll be perusing the home shop machinist plans you mention, thanks.

My setup is being made from scratch. The first thing I found was that the electronics associated with the state of the art microwire edm research papers was amazingly rough and rudimentary. They talk about nanosecond style pulses, yet build setups that couldn't get out of the way of the goodyear blimp electrically.

I've been getting stuff like rails, bearings, acme's from mcmaster carr. Slowly of course, but eventually...

My goal is to finish it before I die of old age..the race is on.

John

 

[Tinker Dwight]

On high count wheels, one can make two rows to keep
them from overlapping. There will always be two that
end on the same row though ( feature of prime numbers )
I'd built a XY table for putting over flow wires on PC boards.
I'd not used ball screws, instead I'd used these bronze
pieces that were squeezed onto the screw ( can't recall
the name. )
I used DC servos and encoders. My encoder resolution
was 0.0002 inch. I did several calibration runs with
a laser and was never off by more than 0.00023.
This was for a travel of 24 inches.
At the time I looked into various options and didn't
like ball screws. I felt it better to just use a bigger motor
to overcome the increased friction.
That was about 20 years ago. My budget could afford
some nice stuff.
I made my own electronics for ramp and position control
that I feed to the servo amps.
I recall my safety check on current control to the servos.
I'd place a pencil in the path and only made a small dent
in the wood. I felt that was about right to keep from
cutting off a finger. The machine still had two switches
that needed to be held before moving or firing the laser.
Tinker Dwight

 

[john e]

My linear will be about 4 inches max travel. I purchased 1/2 alum ceramic coated rods for the linear, but the slide bearings I bought were very sloppy, useless for my needs. I may just drill brass and slit/clamp to suit my tolerance.

The rotary needs to be accurate enough for roughly .1 mil at 4 inches. I'll just do a simple tapered cone bearing set, then rotate it to machine finish the top surface flat. For drive, I'll use a slide with stainless bands along the edge of the rotary table, and figure out how many counts for 360.

EDM really involves no forces, so that makes the design much easier. Only problem is, it doesn't cut wood so well..

Jim,
the Bruno article was great. Black cherry for gears, laurel for the pinions, nice. I'll have to look for some.

He reminds me of the Balzer Family Clock Works. Three generations working on tower clocks, and clock repair. Very nice, visited them last year. Such wonderful people, you could tell they were really into it. He made about 100 wooden gear clocks he had planned on selling retail, IIRC he made them well enough that they were 8 day units.



John

 

[Jim DuBois]

John, I really do recommend ballscrews. Mc Master - Carr has them, the sweet spot is the 5/8" diameter size....price wise. I have used them in several projects, when I can't find an appropriate size or pitch from surplus sources. As you are aware Acme will always have more backlash issues, as well as require a lot more driving force to move a load. Or to move less than a load, rapidly. http://www.mcmaster.com/#ball-screws/=x3f7xj The couple of times I built something with acme, using anti backlash nuts I ended up converting to ball screws in the end. With encoders the backlash issues can be minimized but it adds processing time and overall complexity to the "solution". For wire EDM there is almost no cutting force on any axis so some fair amount of my concerns regarding acme screws is not relevant...

Just saw your follow up post John. Yes, the Balzers do nice work. I have been into tower clocks for a long time, nothing like them of course. I have 1 small E Howard T&S at the moment and 3 time only street clock/small tower clock movements, as well as a stack of NOS E Howard tower clock parts. Something to play with in my spare time. I was not aware Balzer's were doing WW these days....that is interesting.....

 

[john e]

I did notice that sweetspot on the 5/8th size, really weird.. I guess there's quantity in that specific size, who knows.
I was thinking of lightly preloading the linear acme because of the zero force nature of EDM. Of course, acceleration forces are not exactly large at a micron per second top end speeds.

I believe Balzer did his WW several years ago, and I think the project was discontinued. I don't know what his plans in that regard are at this time.

He was totally into clock work, very passionate about it.. He lives and breathes gravity escapements.. All the cutting machinery is in his basement..major cool. The really great thing, his wife really knows her way around clocks as well. They were just so friendly.

John

 

[Jim DuBois]

Since some of the discussion previously was around asymmetrical tooth profiles here is an accurate rendition of both a pinion tooth profile and a wheel profile as created about 200 years ago by one of the Terrys.

 

[john e]

Since some of the discussion previously was around asymmetrical tooth profiles here is an accurate rendition of both a pinion tooth profile and a wheel profile as created about 200 years ago by one of the Terrys.

Very nice. It's clear that they are running the pinion tooth profile as parallel sides for strength. A hypocycloidal necks down, and they are taking material strength into account. It looks like they've designed the wheel teeth to mesh as best as possible, it seems like first contact will be long before they reach the arbor common line. I'm sure the wheel velocities are not constant like an involute, but man, it really looks good on paper.

We learn so much from the biggies of the past, no? I've a collection of lathe books, the best is from 1942. I'm probably one of the few to think of Goodrich as a "page turner".. Even with some of his errors.. And De Carle's , lathe book from 52.

edit: It's interesting to see that they control the wheel tooth end profile by "radius typical" and having the .070 width and wheel radius. That makes the engagement pretty dependent on the wear of the teeth. The drawing shows a tooth just leaving engagement but pretty well rotated. That really makes the final assembly after repair important, as one has to really feel out the meshing on the repaired teeth. From the drawing, it looks like the best option during repair is to use a sanding wheel to re-create that outer tooth radius exactly, then sand the teeth lead and trail to match the tip width, then radius the teeth. (I put a small sanding drum in the drill press, and put the wheel on an arbor flat to the table so it rotates freely, then slowly walk it to the drum until it hits all the teeth equally.) I don't believe I took any pics of this operation.

I've attached another on my presentation pics, defining the tooth profile for the hypocycloidal pinion. Those engaged in this thread understand it quite well, I show it for any lurkers who are interested. The red dashed lines are the drive face geometry, the yellow dashed is the "looks good" trailing face.
John