Threading problem, threads to long

[JoeCB]

Very interesting and educational discussion here, thanks to all that contributed. It appears that we are heading to the source of the problem... the lead screw pitch.
Now, just for info, I will contribute my experience with cutting metric on an imperial lathe. If you are fortunate enough to have an Atlas/ Craftsman lathe with QC box AND the accompanying "Quick Change Attachment Book" along with 57 pages of supplemental gear set-ups for feds from 0.8000 to .0010 / inch there is a page of metric thread pitches ranging from .25 to 7.0.
The Atlas has an 8 tip lead screw. The metric pitch chart gear trains, produce only approximate metric pitches, yet the degree of error is very slight, typically ranging from 0.0002 to 0.0005" per inch of lead. I find this small error acceptable for the normal short thread lengths encountered in most work. All this using the full set of change gears in conjunction with the QC box options. Very handy from the "ole" Atlas 10 " .
Attached is a scan of the "Metric Pitch Chart" , for info

Joe B

[sch]

I think the lathe has a 10TPI leadscrew, looking at the pix given by the OP, doubt there is more than 0.01" error in the OP measurement, not enough
to make it metric. Looking at the gears he used 50 driving an 85T , this ratio is 0,58824 where the 18TPI is 0.055555". So each turn of the spindle
would be divided by the 10 of the lead screw giving a thread pitch of 0.058824, a significant discrepancy of 0.0032 per thread which is not too far
from what the OP pix show when the threaded aluminum is compared with the barrel. If the OP has a 47T gear to use instead of 50T this would be
pretty close enough. Alternatively a 90T gear with the 50T would give the exact 0.55555 ratio needed.

Note that the last two gears in the train given are both 60T so do not change the ratio fixed by the 50T/85T. If the gears listed in his table are all he has
then MAYBE some combo of these might be devised to get the 0.5555 final ratio needed, or at least a closer ratio than the 0.588 of the 50/85 but at 10:30 pm this will be left for others to calculate. It is not a leadscrew or compound problem it is a lousy set of change gears.

[The Artful Bodger]

I discovered (some time ago) a simple way to make plastic gears.


DSCN0196 by aardvark_akubra, on Flickr

A plastic blank is mounted on the lathe spindle and a steel gear of the required gear 'family' is mounted on a spindle on the tool post. The size of the steel gear is not important.

The method is really simple, the lathe is started and the plastic blank brought into contact with the steel gear which begins to turn. Applying a little pressure causes a rumbling noise but after a minute or two that subsides. Apparently the friction of the steel gear running over the blank has heated the plastic which becomes softer. More pressure is applied and again held until the rumbling subsides. This sequence is repeated until teeth are formed of the required depth.

That's it, except for trimming the edges of the plastic gear.

A few notes, the number of teeth is determined by the diameter of the plastic blank which equates quite closely to the PCD. Any size steel gear will create a plastic gear with the correct tooth form regardless of the number of teeth on the steel gear or the diameter of the blank. It is important to withdraw the steel gear from contact before stopping the spindle, otherwise the plastic gear will be deformed where it cools against the gear. Not all plastics are suitable but it is not hard to experiment.

[epicfail48]

Unless i missed something somewhere, this thread is 5 pages in with plenty of somewhat out there suggestions, yet the pitch of the leadscrew has yet to be accurately measured

Seems like it would be a lot faster to do whats already been suggested and set a dial indicator on the carriage and see how far it moves in a single turn of the lead screw. If it moves .1", then its a 10tpi leadscrew and the problem lies elsewhere. If it moves a different amount, the leadscrew is at fault. Seems a lot faster than counting gear teeth or cutting a %#&load of test threads

[nickel-city-fab]

I do not have a solution on this, I have tried many many times.
......
Another thing, The half nut is bending the lead screw when engaged, the top of the nut touch first and bends the led screw down.

I would fix this first, then try threading again in the normal way.

[RB211]

I have disengaged the half nut while threading metric threads on my inch lathe. If I started on a tick mark on the threading dial, disengages then kill the spindle motor, let it coast to the stop, I simply reverse the lathe and allow the threading dial to come back to my original tick mark and re-engage without issue with timing. Just do not let the threading dial make 1 complete revolution without reversing the same amount.

[Johannes Frank]

Unless i missed something somewhere, this thread is 5 pages in with plenty of somewhat out there suggestions, yet the pitch of the leadscrew has yet to be accurately measured

Seems like it would be a lot faster to do whats already been suggested and set a dial indicator on the carriage and see how far it moves in a single turn of the lead screw. If it moves .1", then its a 10tpi leadscrew and the problem lies elsewhere. If it moves a different amount, the leadscrew is at fault. Seems a lot faster than counting gear teeth or cutting a %#&load of test threads

I tested this, I marked on the chuck and then turned it 18 times with the 60-60-85-50 gears installed. The late has a digital readout on the Y and X . After 18 turn the readout read 25,760 mm (the readout is set at mm) Now I do not know if this is normal but one inch is 25,4 so there is a difference of 0.360 mm witch is maybe not so much and can be from the free play in the carriage?

[epicfail48]

I tested this, I marked on the chuck and then turned it 18 times with the 60-60-85-50 gears installed. The late has a digital readout on the Y and X . After 18 turn the readout read 25,760 mm (the readout is set at mm) Now I do not know if this is normal but one inch is 25,4 so there is a difference of 0.360 mm witch is maybe not so much and can be from the free play in the carriage?

Too many other variables being brought into play. Ignore the headstock and DRO. Put a mark at 12'o'clock on the leadscrew, engage the halfnut, then rotate the chuck by hand just enough to get the mark on the leadscrew to rotate 360 degrees, and measure how far the carriage has moved. Pay no attention to how much the headstock needs to rotate to get it there, thats not relevant to what youre trying to figure out. You only want to be paying attention to how much the leadscrew rotates and how much the carriage moves


Ive got a dial indicator set up to measure the carriages movement, the halfnuts are locked to the leadscrew, backlash is taken up, and theres a black sharpie mark on top of the leadscrew


Ive rotated the headstock by hand to put the leadscrew through 1 full rotation. Dont know how much the headstock rotated, dont care, that doesnt matter, all im paying attention to is the leadscrew rotating. As you can see, my carriage moved .056" in 1 revolution of the leadscrew, which corresponds to an 18tpi thread (well, technically 18tpi is .055" pitch, but i doubt i got the mark perfectly back in the same spot). On your lathe, if you do this test and you get .100" movement in 1 rotation of the leadscrew, youve got a 10tpi thread and your threading problem is somewhere else in the machine. If you get .098", then the guess that your leadscrew is actually metric with a 2.5mm pitch is a lot more likely. I would actually recommend running the leadscrew through 5 rotations, to make the measured numbers larger and show any error better. If you rotate the leadscrew 5 times, 10tpi will mean .500" movement, vs .492" if its a 2.5mm pitch

[luthor]

There is room for error in this method as you have noted, it would be quicker and more accurate to just measure it properly over either 6 inches or 150mm, the OP should have done that 5 pages back.

[Johannes Frank]

The thread charts on the front of this lathe DO show both English and metric threads. And the specifications shown by the sellers DO state that the lead screw is an inch (English) one.

But neither the photo posted here nor any of the images of this lathe I can find on the internet show any way to select between those two types of threads (English and metric). If the lead screw is indeed a 10 TPI pitch, then there must be some way to convert the rotation of that ENGLISH screw to a metric rate.

So, I must ask the OP, is there any selector knob or lever on the side of back or top or inside of the machine where you do that selection?

If there is no such selector, then there must be some kind of compromise gearing inside of the head stock that produces fair to good approximations of both English and metric threads; NEITHER of which is actually correct. And do not tell me that no manufacturer would do that. The Chinese factories have been making quill down-feeds on the inexpensive milling machines that are marked with inch divisions (0.001") while actually traveling just 25.0 mm for 1000 of them instead of the correct 25.4 mm. They have been doing this for years and show no signs of stopping.

I would perform some very careful measurements of all of the thread pitches on this machine, both English and metric. I am betting BOTH are inaccurate.

Caveat emptor!

As I can see the only way to convert between mm and inch is just by changing gears. Now I ordered this lathe specially with a inch lead screw and I know I would only be cutting in inch for gunsmithing. I marked on the chuck and rotate it 18 turns, on the digital readout it sad 25,760 mm so there is a difference but that could just be the slack on the carriage?

I m starting to think I have a bad set of gears or this lathe is just what it is? The factory say everything is normal and they have no explanation on this situation.? I have talked to two gunsmith here and we cannot find what the heck is going on;-)

To be clear, I´m cutting with 60 degree cutting steel from Sandvik I have also grind my own to test if that would change anything, same result.

I made a copy of the lead screw and it seams it is 10tpi screw? but this is not so accurate if there are a small increase in the screw it can show up when cutting treads
So bad gears are a question.

[Johannes Frank]

Too many other variables being brought into play. Ignore the headstock and DRO. Put a mark at 12'o'clock on the leadscrew, engage the halfnut, then rotate the chuck by hand just enough to get the mark on the leadscrew to rotate 360 degrees, and measure how far the carriage has moved. Pay no attention to how much the headstock needs to rotate to get it there, thats not relevant to what youre trying to figure out. You only want to be paying attention to how much the leadscrew rotates and how much the carriage moves


Ive got a dial indicator set up to measure the carriages movement, the halfnuts are locked to the leadscrew, backlash is taken up, and theres a black sharpie mark on top of the leadscrew


Ive rotated the headstock by hand to put the leadscrew through 1 full rotation. Dont know how much the headstock rotated, dont care, that doesnt matter, all im paying attention to is the leadscrew rotating. As you can see, my carriage moved .056" in 1 revolution of the leadscrew, which corresponds to an 18tpi thread (well, technically 18tpi is .055" pitch, but i doubt i got the mark perfectly back in the same spot). On your lathe, if you do this test and you get .100" movement in 1 rotation of the leadscrew, youve got a 10tpi thread and your threading problem is somewhere else in the machine. If you get .098", then the guess that your leadscrew is actually metric with a 2.5mm pitch is a lot more likely. I would actually recommend running the leadscrew through 5 rotations, to make the measured numbers larger and show any error better. If you rotate the leadscrew 5 times, 10tpi will mean .500" movement, vs .492" if its a 2.5mm pitch

Thank you I will do this and post the result in a few hours.

[Johannes Frank]

Thank you I will do this and post the result in a few hours.

Well I did this, I clamped a digital caliber on the lathe ( do not have such a nice digital micrometer;-) I did this twice on two different location, result are the same the lead screw turned 360 degree and moved the carriage 0,1000 inch 25,4 mm or 3/32 in all the tests,


image widget

[Peter.]

So the leadscrew is 10tpi and you have a gearing issue.

Do you have 40t or 48t on the spindle?

[Richard P Wilson]

Well, if you are now absolutely certain that the leadscrew is 10TPI, and there is no inherent problem such as end play in he leadscrew bearings, the only other thing that can give the odd results you are getting is the gearing. Check absolutely everything, count that the teeth on each gear match what is stamped on them, then work out the ratio between the spindle and leadscrew. For your 18tpi thread, it needs to be a reduction of 10 -18 overall. Set up what you think is the correct gear train. 18 turns of the spindle should give 10 turns of the leadscrew. Its easy to check, just time consuming.

[nickel-city-fab]

Well, if you are now absolutely certain that the leadscrew is 10TPI, and there is no inherent problem such as end play in he leadscrew bearings, the only other thing that can give the odd results you are getting is the gearing. Check absolutely everything, count that the teeth on each gear match what is stamped on them, then work out the ratio between the spindle and leadscrew. For your 18tpi thread, it needs to be a reduction of 10 -18 overall. Set up what you think is the correct gear train. 18 turns of the spindle should give 10 turns of the leadscrew. Its easy to check, just time consuming.

Wel, it is good to know the problem isn't the screw itself. However, I still maintain that The half-nuts bending the screw (mentioned on page 4) Is a problem