Erector set part number "FL" helical gear

[BGHansen]

Here’s another Erector set reproduction part string. This one is for part number ‘FL’ helical gear. I shot lots of photos so probably a couple of posts in this thread for the whole story. First the obligatory history lesson. . .

Gilbert introduced a number of parts to demonstrate different types of motion (like step pulley drive, cams, eccentrics, universal joints, etc.) in 1929. These parts were included in just two sets, the 1929 No. 9 and the 1929 No. 10 sets. The A.C. Gilbert Company had acquired the American Meccano Company in 1928 which I suspect lead to the addition of the drive train parts. The Meccano building system was far superior to Gilbert’s toy with many more gears, pulleys and other special parts. The Meccano sets came with a manual that specifically showed different types of motion and mechanisms (like miter gears, eccentric cranks, etc.). It was marketed as “Engineering for Boys”. My belief is Gilbert recognized that the Meccano sets were superior in this regard, so they added a similar set of what were called “Mechanical Wonders” demonstrations of motion and mechanisms in the larger sets.

OK, on to the ‘FL’ helical gear. I scanned the manual a couple of times and can find only ONE model that used these gears. I find it curious that the only model page showing the ‘FL’ helical gears (as labeled in the back of the manual in the separate parts pages with prices for additional parts) has them mislabeled as part number ‘FW’ (which was actually a slotted crank). In the grand scheme of things, it was just a kid’s toy. . .


Couple of original "FL" helical gears.



Only page I could find in the manual using the gears




Being a part sold in just the two largest sets from 1929 – 1932 makes this a fairly esoteric part. So why bother with something that has a world-wide demand of maybe 30 pieces? I’ve never cut a helical gear before and wanted to take a crack at making them.

Helical gears are typically made on a universal mill. Nope, don’t have one but I do have a 3-axis CNC mill with a 4th axis rotary table. The ‘FL’ gears meet at a 90-degree angle to each other, the gear’s teeth are cut on a 45-degree angle to their axis.

My plan was to mount the 4th axis on a 45-degree angle and hold the gear blanks in a 5C collet mounted to the 4th axis. The involute cutter would be held in an arbor in the spindle. Cutting motion along the blank would need to be on the X & Z axis with both moving at the same rate. On top of that, the 4th axis would need to rotate as the X/Z moved.

First step was to come up with a way to mount the 4th axis on a 45-degree angle. Yeah, wish I had Tormach’s Super Spacer with a tilting head at this point. I have a rotary table designed to mount either horizontally or vertically. My solution was to mount the rotary table on a 7” x 10” tilting table.

I ended up mounting a backstop to the tilting table to help locate the RT. Went to the Bridgeport and spotted, tap drilled and tapped holes for a backer plate. Then spotted and drilled holes in a piece of 4” x 10” x ¼” thick steel for the back stop. Bolted on the backstop and dry fitted the 4th axis on the bench. Held the 4th to the table with a couple of strap clamps.


Spotting, drilling tap holes and tapping the 7" x 10" tilting table



Spotting and drilling clearance holes in the backstop for the tilting table


Bolted the backstop to the table.



Mounted the 4th axis on the tilting table with a couple of strap clamps


Mounted a 5C collet chuck to the 4th axis for holding the gear blanks




Moved the assembly to the Tormach and trammed in the backstop. Then tilted the table to a 45. Checked the angle with a protractor and fine adjusted with shim stock between the fixture plate and tilting table.


Swept the face of the backstop and adjusted the table for parallel to the Y-axis.


Mounted the 4th axis and verified tram of the table surface in the Y and Z.



I have a few original ‘FL’ gears and a sketch of the dimensions. I consulted Machinery’s Handbook for the helical gear formulas which left my head swimming. As you can tell from my write up and methodology, I’m not a trained machinist. My dad was a shop teacher, but he didn’t spend much time in the shop with me. I had 4 or 5 shop classes in Junior High, experience and making mistakes has been my best teacher (and within the last 5 years some YouTube videos). Plus seeing how other guys do stuff on this forum! I figured why not forget the formulas and reverse engineer the set up.

The first hurdle was aligning the involute cutter to the center axis of the rotary table. I made an arbor out of ½” steel (diameter doesn’t matter) and turned a 90-degree point on the end. Plan was to mount the arbor on the 4th axis. I’d get the Y axis centered up by touching off on either side of the arbor and moving half-way in between. Get Z-height by eyeballing the center of the gear cutter to the point on the arbor. Get the X-axis by bringing the cutter up to the arbor’s point and jog until I pinned a piece of paper. Then X was adjusted by the radius of the cutter so the center of the cutter was on the center axis of the 4th axis. Zero’d out both X and Z at this point. All of the program moves would need to be coordinated between X and Z.



Turned a set-up arbor on the lathe to find the center of the 4th axis


Found Y by touching off on either side of the arbor


Found Z and X by eye and paper. Jogged the Z until the center of the cutter was centered on the tip of the arbor. Then jogged X until it pinned a piece of paper


Moved off in the Y and moved the X the radius of the cutter as the center of the cutter, not the tip, should be on the center of the 4th's axis. Jogged X & Z equal amounts to as close to the 5C collet chuck nut as I was comfortable with, then zero'd both X and Z. Cutting motion would be on a 45 as X & Z would move together. These axis were zero'd at the low point, involute cutter would start in space above the gear blank.




Next was figuring out how much the A-axis (4th axis rotary table) needed to rotate as the X and Z were moved along the gear blank. I mounted a pair of FL gears on an arbor and aligned them to each other. This made for a gear face about 5/8” long, figured the bigger the gear face, the better the measurement would be. I brought the X and Z down and adjusted the Y and A axes until the gear cutter was centered in the gear. Zero’d out the A-axis and noted the X/Z position.


Moved X & Z together near the top of the existing gears. Jogged the Y and A axis until the cutter appeared to be centered in the gear. Zero'd the A-axis at this point and noted the X/Z coordinates


Marked a gear helix with a Sharpie so I was tracking just one tooth




Then backed off the Y-axis and jogged X and Z down to the bottom of the gear. I’d marked one tooth of the gear with a Sharpie; rotated the A-axis and jogged Y until the gear cutter was centered in the gear. Noted the A-axis move relative to the X/Z move (X/Z moved from 0.770” to 0.448” while the A-axis rotated 100 degrees).



Moved down the gear with like moves in the X & Z to near the bottom. Rotated the A to an even 100 degrees as it was close to the bottom. Then jogged X, Y and Z until the cutter appeared to be centered on the same tooth.


Noted that the A-axis rotated 100 degrees with X & Z moves from 0.770" to 0.448". Plan was to start the X & Z at 1.5"; math worked out to an A-axis rotation of 465.8 degrees over that move in X/Z.



I figured on cutting about 1.5” of helical at a time which would give me two gears. Did the math which told me a move of 1.500” in the X & Z would require a 465.8-degree angle move on the A-axis. I measured the depth of cut in the existing ‘FL’ gear at around 0.070”.


Nearing the photo limit, so on to part 2. . .

Bruce

 

[BGHansen]

Part 2 of my saga. . .

Blanks were turned from 0.625” diameter brass to an OD of 0.588”. The gears go on a standard Erector set 5/32” axle, the blanks were faced, center drilled and drilled with a #21 drill.


Turning gear blanks from 5/8" brass. Face, center drill, axle hole drill and turn OD to size





Wrote a G-code routine to do the moves. Took about 7 minutes to cut a gear on the mill.


Set up on the Tormach


I'm still fascinated by the CNC doing its thing. Just so cool to see the blank rotate as the X & Z move.



Mill work finished




Machined gear blanks went back to the lathe where a hub and set screw were added. I make heavy use of hardened drill bushings when making these parts. Erector set parts usually used 6-32 screws to fasten to an axle. I have a number of hardened drill bushings at different spacing (depending on the part) for knocking in the tap drill holes. Slip a drill bushing over the part, drill the tap drill hole with a hand drill, then power tap with a cordless drill. Part and clean up the parted end on a bench grinder with a Scotchbrite wheel.


Mounted the finished gear blank in the lathe and turned a hub to 5/16". Slipped on a hardened drill bushing and drilled a 6-32 tap hole


Power tapping with a cordless drill and parting





It worked out pretty well. I was working with a couple of lengths of brass. One was on the mill getting the gear cut while I was at the lathe finishing up the parts. Great when a plan comes together (successfully)!


Couple hours work.


My reproduction on the left, original part on the right. I'll hit the repro's with a propane torch to add some patina. They're all super shiny at this point.



In retrospect, I’ll consult Machinery’s Handbook again and see how close my reverse engineering came out. Biggest head-scratcher for me was the helix angle calculation; how much to rotate the A-axis relative to the X/Z movement. I might not have the luxury of an existing gear if I ever make another helical gear.

On one hand, it seems like the face of the gear could be modeled as a flat piece of paper with the width being the circumference of the gear blank. The cutter should make one revolution around the circumference of the gear blank as the X & Z move a distance of “Sin (45) *(gear circumference)”. I did this calculation using the bottom and the top of the gear tooth and got angles of 542 and 520 degrees. So, go at the center of the gear depth of 530 degrees? See why my head was swimming!

Plus side of the CNC set up is I can change a couple of numbers in the routine and do some experimenting. The teeth on my gears are a little sharper than the original part, so know I’m off a little but not enough to scrap them out (it was a kid’s toy. . . ).

I’ll let you all how I do on eBay with these. My Erector reproduction parts competition back in the day used to get $45 a pair for these. I made the 20 pictured in about 3 hours, so worth the time if I get that price. Plus, I can now say I’ve cut a helical gear!



Thanks for looking, Bruce

 

[Lo-Fi]

FFR: Something to note is that the helix angle of a helical gear - and the calculations - are all based around the pitch diameter, not the external diameter. This makes them notoriously difficult to reverse engineer, particularly when the tooth profiles have been shifted for a particular application.

Love the work, great use of CNC!

 

[benmychree]

When I cut helical gears, I take an approximation of the spiral angle at the pitch diameter, do the math to convert it to a spiral lead, gear up the universal mill and dividing head and mount the sample gear in the dividing head and trace one tooth with a test indicator to see if the lead is correct; if it is leading or lagging, I change the gear train accordingly, I can usually come within a few changes before it is perfect.

 

[BGHansen]

The G-code program is below. I turned some aluminum down to 0.588" as it's cheaper to experiment with than brass. I can change the spiral angle by changing one variable (#101). I tried 500 degrees and 400 degrees as an experiment. Cut fine but didn't "want" to mesh at a 90 like the 466 degree angle. Looks like the 466 degrees I measured is pretty dang close. I'm sure there's an on-line calculator from a gear company that could give me the actual numbers without pulling out my hair trying to understand Machinery's Handbook.

Another "creative liberty" I took was with the size of the involute cutter. I couldn't find a #34 diametral pitch cutter. I went with the closest I could find which was a 0.7 mm modular pitch 12-13 tooth cutter. A 0.7 mm module pitch converts to a diametral pitch of 36.3. A 0.75 mm module pitch would have been closer (converts to 33.9 diametral pitch), but I couldn't find one of those either. On the plus side, it is a part for a kid's toy and if anyone notices the VERY subtle differences, they're up to no good! Mine mesh perfectly with the original parts and the center to center works great in the model.

Bruce




;Using G59 offset
;use the 4th axis set at a 45 degree angle
;use a 1/2" steel rod with a 90 on the end for coordinate set up
;touch either side of the rod to find Y
;mount the involute cutter and adjust Z until the cutter is centered on the tip, ZERO Z
;jog X until a piece of paper is pinned to the set up tool. ZERO X
;move off center on Y and adjust X more negative by the radius of the involute cutter.
;X and Z will move together for a 45 degree move along the gear blank.

;NOTE: Cutting a 12-tooth gear ONLY. Need to modify the REPEAT command for a different
;number of teeth. ALSO, the 4th is rotated to the next gear position in the A100 subroutine.
;This is currently set to 390 or 30 degrees advanced from the previous cut. The advance should
;be 360/number of teeth.
;NOTE that 390 was used because of the helix angle of 466 degrees. The 4th could be returned
;to ZERO or 30 (for the next tooth of a 12-tooth gear), but that's an extra rotation of the table.
;If the helix angle is less than 180, do a move directly to the next tooth.

(tool 4 gear cutter)

G59 G90 G94 G91.1 G40 G49 G17 G80 G92.1
G20 G90

G17 G90
G20 G59

#100 = -1.00 (Y-axis coordinate to engage cutter into the gear blank)
#101 = 466 (axis rotation on the A-axis during the programmed X/Z move)

T4 M6 G43
S2000 M3 (spindle on)
M9 (coolant off)

G94 F8 (feed rate 8 ipm)

(subroutine for gear cutting)
O100 sub

G1 Y [#100 -0.2]
G0 X1.5 Z1.5
G1 Y #100
G1 X0 Z0 A #101
G0 Y [#100 -0.2]
G0 A390 (index 4th axis to next tooth)
G92 A0 (Zero 4th axis)
G0 X1.5 Z1.5

O100 endsub


(Start of MAIN PROGRAM)

(Loop 12 times to cut 12 teeth)

G92 A0 (make sure A-axis / 4th is at ZERO)
G0 Z 4 (make sure Y/X are clear first)
G0 X 3 (make sure Y is still clear)
G0 Y [#100 - 0.2] (move Y to 0.2" from the gear blank)
G0 X 1.5 Z 1.5 (move to starting location of X and Z)

O110 repeat [12]

O100 call

O110 endrepeat

M5 (spindle off)
G0 X 4
G0 Y 1

M30

 

[BGHansen]

FFR: Something to note is that the helix angle of a helical gear - and the calculations - are all based around the pitch diameter, not the external diameter. This makes them notoriously difficult to reverse engineer, particularly when the tooth profiles have been shifted for a particular application.

Love the work, great use of CNC!

It is a game changer for stuff like this. It is so much easier to modify and develop stuff. I have a couple of jobs that involve cutting 50-tooth and a 75-tooth gears. It goes OK with a manual dividing head, but throwing a stack of blanks on the table and letting the CNC do it will be much more productive.

Bruce

 

[benmychree]

The method that I use for figuring the lead of an existing spiral, such as a gear tooth, is to try to measure the angle of tooth from the centerline of the part and the OD measurement
The lead equals the circumfrence of the part divided by the tangent of the spiral angle. In example, a part with a diameter of 3.863 and a spiral angle of 57 degrees, the circumfrence would be 12.136" and the tangent of 57 degrees is 1.5398, which when diveded, gives a lead of 7.881" loking at B&S table of leads, the closest is 7.883, close enough! Change gears are 100T on table screw, driving a 48T on the compound bracket with a 44T ganged with it driving a 86 T gear on the dividing head's worm shaft.