Wednesday, October 23, 2019

New Shop Tool - Old, Obscure Dividing Head

I've needed to create a gear to finish the hand lever collet closer attachment.  This meant I needed a good dividing head (well, I had the small rotary table, and could have done just fine with that one, because a 40 tooth gear is perfectly divisible on a rotary).  I ordered an old dividing head off of eBay. It's time to get it identified.

It appears to be labelled as a "Republic Tools", and the "1161" prefix stamped into the spindle was also stamped into the bottom of the casting (matching numbers... that's a good thing).  It rotates very smoothly, so it's been well cared for.

Typically, a dividing head is a 40:1 ratio, meaning you turn the handle 40 times and the spindle turns once.  This one looks different.  It felt like a 48:1 with the brief check I ran.  I will have to verify this, but it felt less like a 40:1.

The spindle through hole is about 1" (0.923" in the back, probably just an awkward angle and not accurate in any form of the word, and about 1.050" in the front).  There is no real taper on the bore of the spindle, so it's not 3C compatible.  It is definitely not 5C, or a morse taper.   That is going to force me to use the threads for any indexing purposes.

The spindle thread is 1 3/4-8 TPI, so it's an older thread that I might have to build an adapter or a backplate.  I'd like an adapter do I could potentially move from the lathe to this without removing the workpiece from the chuck itself, then things are more likely to be concentric if I need to change machinery between turns.

My brain went immediately to trying the infamous 127 tooth gear so I could potentially cut metric threads on the lathe, and that lead me to check the diameter of the hole plates :

At 4.875" on the outside diameter, if I ran a hole pattern around a 4.5" diameter circle (14.137155" of travel along that diameter) and divided that 127 times, we'd have 0.11131618" between holes along that path.  Seeing as that hole pattern would not have enough material between holes (let's call that "meat"), it could not be a 127 straight-line sequence.  I would have to offset the holes to get a good pattern into at least two rows, but considering 127 is a prime number, it should be three rows (or you end up with two holes on the same path right at the end).  Is that doable?  Absolutely.  I might just have a shop put the holes in the plate for me, though the accuracy isn't as critical for them because I'd be using that to bootstrap another 127 hole plate.  If another shop did it for me via CNC, I'd not need to bootstrap it - it would be accurate enough.  If I do it myself, I would use the index plate to make a second index plate (causing it to be much better in accuracy on the second one at 48 times, if the ratio is 48), and then I'd use the second index plate to create a third index plate.  This would improve the accuracy 48*48, by 2304 times.

Anyway, I need to adapt a chuck to it.  The chuck is a K72-80 (80mm) independent 4 jaw chuck.  I cut a backplate out of some cast iron and bored it to 1.595".

I used a threading tool to cut some basic threads (not big enough), and then used a boring bar at as much of an angle as I could manage to hog out as much thread material as I could.  I grabbed my 1 3/4-8 TPI tap, and man-handled that thing through the bore to get my thread.  This caused the bore and the face to be perfectly aligned (centers even).

While there, I faced it down to match the chuck.  I had to check that the thread was correct, so I spun the dividing head into the thread so I didn't have to re-clock the thread when I removed it..

Successful in the test fit, next up was to identify the bolt hole information so I can drill the holes and mill out relief for the socket head bolts to hold both together.

At 2.25" from bolt to bolt, on a three hole pattern, I could then start some calculations to convert that into usable information.  Frankly, I need the bolt radius.  I started out doing some serious math to determine the diameter/radius.  I spent a half hour doing math using trigonometry before I thought, "Why?"

I ran and grabbed the machinists handbook, looked up the circular segmentation chart, and on a 1" diameter piece of stock, the 3-hole pattern had segments of 0.866025".  Here's where we get tricky.  Because the table listed a diameter of 1, it should be a simple algebraic format :

Segment_Distance=Diameter * Segment_Value

Quick flip of terms to reverse what we need (we need the diameter, not the segment length), followed by isolation of terms :


Add our values :

The 2.598, is the diameter of the bolt circle.  The radius (distance from the center) is what I need - to know how far to move it once I have it centered.  Dividing by 2 gives a radius of 1.299".  I grabbed the calipers and set them to 1.299" and lined it up with the center.... viola! That's what I need.

I jigged up the scariest setup of my life on a mill.  It was a matter of using an angle plate (on 1-2-3 blocks to get it tall enough), and a lot of sketchy clamps.  I even had to clamp the back plate to the indexing head (only once it was in position).

I could mill the bolt head pockets, but I could not drill the bolt holes (the mini mill doesn't have quite the work space to use a drill chick and an indexing head on it's side.  Once I had the pockets, I could use a collet to hold a center drill to get those started, and then finished the through holes on the drill press.

Now, it's still too tight to install.  The back plate went into the freezer for a few hours, and the last 20 minutes of the freezer, the chuck went into the oven.  They all came out at the same time.  The different temperatures on the material gave enough expansion/contraction distance that it could be easily bolted together.

With that mated up, I could reverse chuck it into the lathe and face off the back, then add the clearance for the thread.  I first parted it off :

Then I faced it and clearanced the thread :

And finally installed it :

I'd call that a success.  I can now cut the gears (if the table is large enough to hold the dividing head, gear, and tailstock.  I did cut my first gear, but the setup time was extraordinary.  So, I had to change the 4 jaw independent chuck for a three jaw self-centering.  The only one I could find was a 4", and I didn't have a chunk of 4" cast iron for a back plate.  You don't want to try to tap the threads in good steel for a 1 3/4-8 TPI thread.  Trust me, the part just twisted in the jaws.  The only 1 3/4-8 TPI pre-formed back plate is actually much larger - I could only find it for a 6" chuck, so I ordered the 6" back plate.  Here's where things get tricky.

When you attach a chuck to a lathe, you have to machine the backplate on the lathe to maintain concentricity.  It is an absolute must.  In this case, it's going on a dividing head, so if you machine it on a lathe, you will end up with the chuck center line offset from the dividing head center line.  Simply put, you have to machine the backplate on the device it is to be used for.  Have you ever machined a backplate for a dividing head?  No?  Here's what to do.

Taking a 6.25" back plate to be used on a 4" chuck, you need to hog off a lot of material.  Also, I needed about 3/4" of depth, and this back plate is 1.25" deep.  I used the lathe to break down as much of the outside surface as I could.  You must leave it oversized, because you will still machine the plate on the device it will be used on.  So, after the lathe, I'd removed a bit of material.

After removing that, I could put the backplate on the dividing head backwards.  This would allow me to face off the rear surface that mates up against the dividing head.

Once that is surface is machined, I pulled the backplate off and turned it around.  I could then machine the outside edge down to match the chuck.  In this case, the chuck was 3.948", and at this point the back plate was 4.250" in diameter.  I machined off 0.030" using the mill.  You must be careful here.  As the end mill rotates, the direction where it meets the backplate must tighten the backplate instead of loosening.  The speed of the dividing head is minimal, so we don't need to worry much about having it round.  It's nice to do, though, for a finish.

With the outside edge, you can now machine the mating surface for the chuck.  This should be done very slowly - it is the same principle as doing it on a lathe, but the part isn't rotating, and you are taking it off with an end mill.  Remember, when you take off 0.125", it removes twice that because it takes the 0.125" off the other side, so it's really taking 0.250".  Take your time.

After the boss for the chuck has been cut, you can surface the boss so it is deep enough.  At this time, you can rotate the dividing head and drill your holes for mounting the chuck.  I flipped the back plate again, so that the indexing-side surface was on the outside.  This allowed me to bevel the rear flange for clearance on the screw caps.

With that complete, it is now time to pull the plate off and install the chuck.  That completes the installation of a new chuck to a dividing head.