Thread: cutting a 73 tooth gear

At the time (1971) I was specializing in EDM work. I had two Model 242 Excello machines, each with a rotating spindle.

To lock the spindle in a specific position it was necessary to tram it. Once you unlocked it, there was no way to regain position without tramming it again.

I wanted to be able to re-position after spinning, and I wanted to be able to move to a specific angle for other things. In other words, I wanted to be able to index the spindle.

I spent one week considering all known methods. Shot pin, optical ring, worm gear, etc. All had problems that made them impractical.

I then decided a face gear solved many of the problems, but was too limited on the numbers of positions. After a day or so I devised the idea of several stacked face gears with different numbers of teeth. (I didn't know that it was patented in the Netherlands about the time I was born in 1934, I had never seen one.)

For a week I spent all hours of the day an night with an old Frieden rotary calculator trying to come up with differential combinations that would be easy to use. I got a lot of flak from my dad for wasting so much time (he was working for me at the time) instead of being out in the shop producing.

After a week of non-stop calculations and reams of notes on my desk, I finally said there are no combinations that are easy to use. I was using using whole numbers of teeth (integers) because that's the only way face gears were made, otherwise how could you change positions? All of combinations required calculations to convert fractions of the circle to an angular equivalent.

I'm sitting at my desk about 10 AM with a nice warm sunshine on my back, and getting pissed for spending so much time and getting nowhere. Face gears are the only way to lock very precise locations in a the small space available, so that's what I wanted to use.

I said to myself, I'm not gonna waste all that time, I'm gonna FIND a way to make this work. I took a different approach. What If I made the first pitch 60 minutes and the second pitch 61 minutes. The difference of 1 minute is easy to read. Well, 61 minutes leaves you with an odd sized tooth when you get back to your start. OK, that will fit into a corresponding small space on the matching set, but what happens when you move one pitch? Interference on the tooth next to the starting place. OK, get rid of that tooth, then what? After a couple of minutes, I realized that once you did this 60 times, you have a cleared area on the opposing face gear of about 60 degrees and you have indexed 1 degree, then what? Hmmmm... go back to your starting point and do it again but since you already cleared the area, you lose no more teeth, and there's a lot left to give good position.

Basically, that gives you a ratcheting action in one minute increments, just by keeping the odd tooth in the cleared open area of the opposing gear. WOW ... NEAT

Problem solved. Make the next pitch 61 minutes one second and do the same thing.

A one second indexer with only 3 pitches and it reads direct.

That thought process took me 10 minutes.

Bottom line: One week to create (for my first time) a differential serrated tooth indexer, one week to decide it can't be done to get the results I wanted using whole numbers of teeth, ten minutes to get around that.

So the mechanism has 3 elements. Interlocking face surfaces (about 2,000 years old) The differential principle (don't know how old that is) and the odd number of teeth with a cleared space on the opposing gear allowing multiple engaging locations, a first. Take any one of those three out of the mechanism, and it can no longer move exactly one arc second and be direct reading. Thus, it is basic.

I went out to the shop and told my dad what could be done. He said "There must be something wrong with that or someone else would have already done it."

The rest is history.

Tough nut? Definitely, otherwise it would have been done a long time ago by the many talented people who tried.

Proud of it am I? You bet. This also explains my logo... any angle out of an opened circle.

Cut a section out of an old tape measure that's exactly 18 1/4 inches long. Tape it into a loop, or better, silver solder the ends together. Use another tape measure to check that the markings on either side of the junction are lined up exactly. Make a disc 5 13/16 inches diameter to press this loop over, and index every 1/4 inch. Makes 73 divisions.

You could mount a point on the mill table to pivot this disc on, and mount something to use as a pointer that just barely clears the disc's rim. You can then drill a 73 hole pattern in the disc by carefully aligning every 1/4 inch mark by eye. If you want or need to use a larger indexing disc, go with a loop of tape 36 1/2 inches long, make the disc twice as large in diameter, and index every half inch.

I made a sprocket using a loop of chain pressed over an mdf disc as an indexing plate. That worked well, and I then made a second, smaller sprocket using another mdf disc with loop of chain. Both turned out to be more accurate than the sprockets they replaced. For an odd number of divisions, you wouldn't be able to use the standard link to join the ends of the chain. You would have to push out a pin and link two same links together. Not really a problem, just a bit of fiddling. Using bicycle chain, you'd get a disc about a foot diameter for 73 divisions.

Either method gives you a way of adding a 73 hole pattern to your dividing head.

I use one of these.

http://www.divisionmaster.com

Maths is a thing of the past now.

John S.

Here's some pics of using a section of bandsaw blade to get the divisions you need.
http://photobucket.com/albums/1003/winchman/DWheel/
http://photobucket.com/albums/1003/w...egree%20Wheel/

You don't even need to solder the blade. Just machine the recess so it snaps in place.

The pawl you use for indexing needs to be hardened. I used music wire lashed to a strip of hacksaw blade. The blade served as the spring.

Roger

yes, i am still looking at this
everyday. thanks to everyone
that replied. i have given up on
this project.
wlbrown

wlbrown... Out of curiosity, why have you given up?

Winchman.... Interesting setup. Clever.

Since we are showing final results of systems, I'll show a set of Newbould Indexer plates. There are 4 plates in a set. They are molded plastic, and will lock any angle you set within .0004" on a 6" circle.

I made the mold myself.

My last order from Timken for these was for 400 sets at $160 each set. They made their own mounting setup to suit their purpose. There is no more economical way to index. You can set any angle at random, there is NO backlash, and they will NEVER wear out.

What is the URL for your forum?

What is the URL of whose forum?

I use a rotary table for most of my projects like this. I have a couple of indexing heads but rotary tables just seem easier to use. I have an Excel spreadsheet that I made to help to the index calcs, show what plate to use, number of turns, etc. The math is easy enough to do but the spreadsheet was fun to make. Hope it helps. engineerjack@hotmail.com