Lathe Change Gears - What Are The Best Combinations?

[epanzella]

I have the same lathe as you. I needed a thread that wasn't in the chart. That's when I learned that those 6 gears plus the gear box gives more than 1,000 different thread pitches. I keep all the possibilities in a spreadsheet for reference.
Dave

Dave,
I also have a chart that shows an incredible amount of combinations for the G4003G. I found it when I needed 27tpi threads for a project and my thread chart skipped over that one. I keep a copy of it with my lathe manual.

 

[stupoty]

I use feed rates between 20th and the finest I have which is 2.5th (i think), the most common is about 4 to 6 th per rev.

If you have enough to keep them in pre arranged groups that helps reduce the time to swap them if you want to change the feed rate a bit. Maybe even with some paint marker on them for the order.

Stuart

 

[BarnyardEngineering]

For known threads, the best combinations would be the ones recommended by the lathe manufacturer.

For unknown threads, to me the best answer would be the simplest solution. In other words, the simplest combinations of gears.

Now if you have an 8TPI lead screw and need to cut 4TPI threads, how do you do it without spinning the lead screw 2X as fast as the spindle? Only thing I can think if is to turn the spindle 1/2 as fast as the lead screw...

 

[jimbojones132]

There are a number of limitations on the actual gear combinations that can be used. Just because a gearing combination is mathematically feasible doesnt mean its physically feasible, safe or easy to use. The following observations are based on actual usage and not simply opinion:

- you don't want to use a 3-change gear combo to make a thread. Mathematically and physically its possible...but not wise. Reason is that you have NO failsafe in case the gears get jammed up...all 3 gears are in a direct-drive scenario and if there is a jam, you'll end up busting many things. There is a reason why the gear changes on the lathe only show 4 or 5-change gear combos as the manufacturer has a double use for those keyed collars that the gears ride on 1) to provide a method of mounting the gear(s) 2) 2-gear set adjacent on any one of these collars ALSO provides a 'sacrificial part' situation in that if the gears jam up, the collar will shear in half between the adjacent gears and possibly save (some) of the lathe.

- there are physical limitations on the TOTAL length of the gear combinations and if they will successfully reach from the Spindle to the Leadscrew as too few teeth wont reach and too many will exceed the maximum reach of the change gear bar.

- 5-gear gearing combinations require clearance between the rows of gears in order to prevent the wrong set of gears from meshing

If you've got a gear change program, you also need to add the above considerations in to it for it to isolate the 'usable' gear change combos from the 'mathematical' gear change combos. You'll see that that are many less combinations available but indeed, there are still plenty more than what the manufacturer claims (especially on the Metric side)

 

[Wreck™Wreck]

The lathe can handle 4 or 5 gears at a time which is something like 264,000 combinations.



Good to know. Do you have a gut feel for when close is close enough? The example you give of 0.0005 for 24 TPI is a 1.2% error which it seems is not good enough. Should I be looking at 0.5% error? 0.25%?

Thanks!

A small lead thread is difficult to measure directly, one may easily measure the PD over wires which is very straight forward and inexpensive.
Measuring the actual lead is a different matter, ring gauges work well for this purpose and make the parts suitable for use.

If one obsesses over numbers do a simple lead screw test if the lathe is small enough. Engage the half nuts at the desired lead and rotate the spindle by hand until the lash is removed, mount a dial indicator against the carriage and rotate the spindle by hand one revolution using another indicator to produce 1 revolution. The difference is the error.

 

[matridium]

I have the same lathe as you. I needed a thread that wasn't in the chart. That's when I learned that those 6 gears plus the gear box gives more than 1,000 different thread pitches. I keep all the possibilities in a spreadsheet for reference.
Dave

I'm new to the forum, However I see I have the same lathe as you do. I too need to cut threads that are not on the chart. Do you happen to have the spreadsheet with the thread pitch and gear combo available to share?

 

[joshua43214]

/Playing the math nerd
You have 14 gears, it looks like there are 5 gears in the gear train, the order in which the gears are chosen matters, and the same gear can not be reused.
So, the total number of combinations is n!/[r!(n-r)!] = 14!/[5!(14-5)!] = 240240. Multiply that by the number of lever positions in the gear box (the A, B, C lever looks like a partial gear box), and you get 720720 thread combinations.

I wrote a Python script last for someone here a week or so ago who was trying to cut a thread not listed on his lathe, he had a much more complicated system and it took over 2 seconds to calculate the output. Thought something was wrong until I checked the combinations lol.

I suggest you leave the gear train in whatever position lets you switch between roughing and finishing cuts using just the lever.
For cutting threads not listed, I agree with the suggestion that running the gear box slower is better than running it faster.

 

[shooter123456]

/Playing the math nerd
You have 14 gears, it looks like there are 5 gears in the gear train, the order in which the gears are chosen matters, and the same gear can not be reused.
So, the total number of combinations is n!/[r!(n-r)!] = 14!/[5!(14-5)!] = 240240. Multiply that by the number of lever positions in the gear box (the A, B, C lever looks like a partial gear box), and you get 720720 thread combinations.

I wrote a Python script last for someone here a week or so ago who was trying to cut a thread not listed on his lathe, he had a much more complicated system and it took over 2 seconds to calculate the output. Thought something was wrong until I checked the combinations lol.

I suggest you leave the gear train in whatever position lets you switch between roughing and finishing cuts using just the lever.
For cutting threads not listed, I agree with the suggestion that running the gear box slower is better than running it faster.

Aren't a large number of those permutations impossible because of the size differences in the gears?

 

[joshua43214]

Aren't a large number of those permutations impossible because of the size differences in the gears?

For sure some would. The actual number would depend on the gears available, worst case would be ~75% would not work. That still leaves you with ~50,000 combinations times the gear lever positions. Though that does bring up the possibility that some all large gear combos might be able to reach the drive gear at the spindle by only using 2 gear sets in the drive train (the lead screw would run in reverse) adding another bunch of combinations.

Just a fun exercise either way. I took so much math for fun when I went back to school that I would up getting a second major in it. Took so much stats I could have taken a minor, but I didn't want to get trapped into being an analyst. I far prefer working in the lab to working at the computer. These days I only spend about 10% of my time writing scripts and doing statistical analysis - and all that is non-publishable and used for optimization of lab protocols.

 

[epanzella]

It's fairly simple to come up with other combos if you have the gears. Need 32TPI? Just cut the 16 tpi feed speed by half. I needed 27 tpi but my QC box wouldn't do it. I set up for 18 tpi and used a 40t to 60t gear combo to get 27tpi.