Thread pitches and change gears

[paradox_pete]

(not a tapered thread,)

I believe it should be possible to cut a tapered thread by offsetting the tailstock as you would to turn a taper. That's the only way to single point a tapered thread that I've come up with. I'd be curious to know if there are other ways.

Your ELS might also benefit from a 48 tpi setting, I believe you'll find 48 tpi in both the UNC and UNF standards.

-Pete

 

[WobblyHand]

I had made a request on github for a modification to the Clough42 code to include the ability for a user to enter a custom thread pitch. One of the contributors did just that, allowing entry of custom metric thread pitches. It will only allow for two decimal places, so .01mm resolution and won't hit inch pitches closely enough for my needs. My intention is to dig into his code and determine how he accomplished this. The stumbling block is the Clough42 display resolution which only allows two decimal places. There are ways to overcome this but it requires the coding knowledge to accomplish them.

If you permitted a user entry such as this in your code, it would allow creation of custom pitches without the need for anticipating some weird pitch requirement in the future and cluttering up your pitch table.

The equations are straight forward to determine the values for a particular thread. Where I found it tricky was making sure I had integer values for the ratio. At least for my system, I need integer values - since I have discrete counts for both the stepper and for the rotary spindle encoders. I have yet to find a fully automated way to guarantee good solutions. I tried using a gcd function, but it didn't seem to be correct 100% of the time, even though I could find a manual solution. Maybe I am thinking about it the wrong way - I need to revisit it. Would make life easier to be able to enter oddball pitches, rather than having 150 element pitch table to scroll through (and still not have the pitch you are looking for).

 

[markba633csi]

If one had to accept some error, I would say one would want an error of less than 0.5 percent for coarse threads and maybe 0.3 percent for fine threads
and this would apply to short threads not long ones- short being something like 0.5 inch or less
Just a WAG

 

[WobblyHand]

I believe it should be possible to cut a tapered thread by offsetting the tailstock as you would to turn a taper. That's the only way to single point a tapered thread that I've come up with. I'd be curious to know if there are other ways.

Your ELS might also benefit from a 48 tpi setting, I believe you'll find 48 tpi in both the UNC and UNF standards.

-Pete

Can't argue with you about this in general. Issue is more of a choice for me to 1) have a ridiculously long thread table that every single time one must scroll through, or 2) make it easy to enter in a relatively uncommon thread. Both involve a lot of coding for me that doesn't quite fit into my existing architecture. Both are solvable, simply haven't put the time into evaluating it. Currently if I have need for a custom thread, I solve for N & D and make a custom build for that thread, then I revert the build. Since I have the source code (and I authored the code and understand it) I can do this easily. But it is admittedly a poor solution.

 

[WobblyHand]

I would say one would want an error of less than 0.5 percent for coarse threads and maybe 0.3 percent for fine threads
and this would apply to short threads not long ones- short being something like 0.5 inch or less

In general, I have been able to find rather exact solutions to the problem, at least on my lathe, that are good to better than six decimal places, usually eight.

 

[DLF]

I had made a request on github for a modification to the Clough42 code to include the ability for a user to enter a custom thread pitch. One of the contributors did just that, allowing entry of custom metric thread pitches. It will only allow for two decimal places, so .01mm resolution and won't hit inch pitches closely enough for my needs. My intention is to dig into his code and determine how he accomplished this. The stumbling block is the Clough42 display resolution which only allows two decimal places. There are ways to overcome this but it requires the coding knowledge to accomplish them.

If you permitted a user entry such as this in your code, it would allow creation of custom pitches without the need for anticipating some weird pitch requirement in the future and cluttering up your pitch table.

I do not know what Clough42 has done, but my original question was also in that direction.

I made an excell sheet which for any given thread pitch (or TPI) will calculate all change gear combinations and will specify the correct QCGB settings which will produce that pitch

It gives 3 kind of values: perfect pitch with 0% error, approx pitches which are within +/-0.5% error and approx pitches with +/-1% pitch error.

And I was wondering if 0.5% error resp. 1% error would be too much for a usable thread pitch.

Sent from my SM-G973U1 using Tapatalk

 

[MrWhoopee]

My faux pas --- But 27 is a number that is used in our pipe thread series, and very few lathes can cut it, Axelson is the only one that I've seen, 19 TPI is used for pipe threads by the Brits.

My SB Heavy 10 cuts 27 TPI, as well as 11-1/2, which are both NPT threads that have been eliminated from many modern lathes.

 

[DLF]

I have updated my code to work with +/-0.3% error (yellow lines) and +/-0.5% error (white lines). Green lines mean perfect pitch value.

My Paulimot PM5000 (metric version of the PM-1440E-LB) will cut 19 TPI perfectly, but for 29 TPI there is a very close approximation (+0.04% deviation).

And that is exactly the question: will such approximations work?

 

[RJSakowski]

I have a thread pitch table for the G0602/G0755 that @WobblyHand mentioned which gives the possible thread pitches which are available using the stock gears. The example of 19tpi given can be achieved using stock gears to within .08%. The gears required would be the 100 , 127,48, and 46 t. with a gear box setting of IIIA.

I offer this only as an example of what can be done with creative use of existing tools. It more than likely would never be needed.

As to what degree the actual pitch has to match, that is a good question and the answer was given by @Asm109 above. Metric and inch threads can be interchanged for some threads. The 10-32 inch thread fasteners will thread into an M5-.8mm nut. OTOH, an M5-.8mm fastener may or may not fit a 10-32 nut, depending on thread class. I suspect the difference is more about the difference in diameters rather than the pitch difference of .79% though. For this example, the threads will be a mismatch of 1/5th of a thread after 25 threads or .78". For a 1/4" nut, the mismatch would 4.9% of a thread.

My expectation is that if I could get the threads to match within 1%, It would probably make a workable fastener.

 

[WobblyHand]

I have updated my code to work with +/-0.3% error (yellow lines) and +/-0.5% error (white lines). Green lines mean perfect pitch value.

My Paulimot PM5000 (metric version of the PM-1440E-LB) will cut 19 TPI perfectly, but for 29 TPI there is a very close approximation (+0.04% deviation).

And that is exactly the question: will such approximations work?

I don't quite yet know how to calculate it, but, the answer is in the class of the thread fit. This indirectly gives the maximum amount of slop in a thread. Basically it depends on the class of thread fit, and the error. Obviously for short thread lengths, the error will not be able to accumulate enough to bind, and for longer lengths, say 20 threads, it will. True answer depends on the exact dimensions of the screw and or nut and the error.