How to cut metric threads on an imperial lathe and vice versa.

The Smart & Brown model A that I use is imperial, with a 6tpi leadscrew. Since getting a 125/127 mod1 gear on ebay and buying a number of mod1 gears to modify to fit the lathe, it can now produce metric threads from 0.75 to 3 pitch, which covers everything I would want to do at present.

Having made a new leadscrew nut, the lathe now has close to zero backlash between the leadscrew and saddle. That is fine, but there is still some lost motion between the spindle and the leadscrew. This means that I would not dare to keep the nut engaged and simply reverse the motor at the end of every cut.

I disengage the nut every time, the same as with imperial threads.

Impossible, it cannot be done, I hear you scream, hogwash, I say in reply.

A stop for the right hand side of the saddle is required if threading towards the chuck, or the left end if threading away. Set the stop a few thread pitches away from the start of the thread and back the saddle up to it at each pass. Choose a number on the threading dial and use it each time. The spindle and tool tip will synchronise every time.

I recommend using the same number on the threading dial every time as different makes of lathe have different numbers on their dials. The S & B that I use has 24 numbers, and I have a choice which other people would not have.

Am puzzled.

If the motor is restarted to take a second cut, a distance away from the work, surely, any backlash will be taken out.

Your method sounds like, effectively, doing the same thing

Move Saddle to known point, engage half nuts in the original place and take second cut, starting far enough away to remove any backlash.

Not knocking you, just sounds like we are having a violent agreement?. Just another way of skinning a cat.

Howard

I don't stop or reverse the lathe, I disengage the leadscrew nut at the end of the cut.

I forgot to mention that it is a good idea to make sure the thread dial marks line up with the datum line when the nut is engaged before doing any threading.

And to thoroughly silence your screaming disbelievers, suggest to them they should watch the oxtoolco video on YouTube entitled 'Metric Threading with an Inch Lead Screw'.

I never disengage the leadscrew, at the end of the thread I stop the lathe, turn the cross slide handwheel one full turn to withdraw the tool 2mm ( if that's sufficient ) then reverse the lathe just past the start of the thread, turn the cross slide handwheel one full turn in plus the next cut and proceed.

The lathe I use has no brake, it would require a long undercut to avoid tears. I did use that method some time back, when for some reason I wound the spindle by hand, zero runout on a coarse thread, probably.

Mart are you saying that you keep the lathe and screw turning, release the half-nuts, wind the saddle back to a stop then engage the half nuts again when the same number comes round on the dial?

I think that's exactly what he's saying. My first reaction (based on a blur of metric/imperial conversions) was that the idea was hogwash, or that it only worked in special cases.

As I'm not keen on proving Mart wrong mathematically, and it's far too hot and sticky today, I've cut a few  experimental threads instead. So far, Old Mart is right!

None of my books mention Mart's recommendation. There's a lot of detail about when you can and cannot use the various ticks on the dial with particular threads. I think it's because they concentrate on the special case whereby perhaps several points on a dial can be used to pick up a particular thread. Doing that saves time. What's not mentioned is that picking any one point and sticking to it always guarantees (I believe) that saddle and spindle are synchronised correctly, and it doesn't matter if the thread is metric or imperial or what the pitch is. The only disadvantage is the short extra wait while the dial comes back to the chosen position. But it's faster than keeping the leadscrew permanently engaged.

I'd prefer to prove mathematically it always works. Having failed miserably to do that, I can confirm the relatively few threads I've cut using Old Mart method, have all gone well. I think it's a winner!

Dave

Edited By SillyOldDuffer on 24/07/2019 10:51:27

Well I am no expert but I do not belive it will work for all thread pitches. Had a problem on my WM250 (3mm pitch leadscrew) trying to cut an M12 thread, I came up with this:

The screw-cutting thread indicator dial meshes with the leadscrew via a 30 tooth gear. To get the indicator dial to rotate once the carriage has to move 90mm (number of teeth x leadscrew pitch). Only those metric pitches that divide exactly into 90 will be able to use the dial when screw-cutting. Thus pitches of:-
0.5, 0.6, 1.0, 1.25, 1.5, 2.0, 2.5 and 3 will work, pitches of 0.7, 0.8 and 1.75 won′t work.
Replacing the 30 tooth gear with a 28 tooth gear would work for the missing pitches but not for all. With a 28 tooth gear the carriage has to move 84mm for a full turn of the indicator and 84 is divisible by 0.7, 0.8, and 1.75. So either change the gear, make an indicator that has both gears or just leave the leadscrew engaged and reverse the lathe.

I also made a flip-up tool holder which solves the problem.

John

If you have a 3mm pitch leadscrew, any of the pitches that are a factor of 3 do not require the dial at all. You can remove it and engage anywhere at all.

As long as you use a stop on the bed so the carriage position is known, and close the nuts when the correct mark lines up, the method will work all the time for any pitch of any system of units.

You should pay very close attention to old mart's advice viz: "it is a good idea to make sure the thread dial marks line up with the datum line when the nut is engaged before doing any threading".

That means you need to make sure the dial mark is lined up when you set the bed stop. You cannot just put the stop in a random place (well, you could, but then you'd have to make your own mark on the rotating part of the thread dial to align with the stationary fiduciary mark).

Well, I'm not afraid to say I'm still sceptical. Because the 127 gear is a prime number I can't see how the spindle and screw and dial can hold synch. I'm struggling with the maths but I'll get there sooner or later to prove or disprove it to my satisfaction.

One thing I will say is that it matters not at all how many graduations you have on your threading dial if you use the same one every time. You could have just one mark if one is all you'll use.

Ok here it is:

Let's say you have the lathe set for 6tpi, so the leadscrew turns once per turn of the spindle. I don't know how many teeth are on the dial gear but I'm guessing 24 (it doesn't matter for an example). To bring the dial one full turn the screw has to turn 24 times. Both of them have turned 24 full turns to bring the dial round one revolution so the spindle and screw are still exactly in sync.

Now you want to cut 4.5mm pitch so you fit a 127/135 combination (which by no accident is all you need to use to convert 6tpi to 4.5mm pitch)

6/25.4/127x135 = 4.5

So you have your saddle against the stop and it doesn't move because we're just watching to dial go round. Run the lathe and watch the dial perform one full turn. The screw has turned 24 times but spindle is turning at a rate of 135/127 to the screw which is 25.52 revolutions - that's just over half a turn out. The spindle and screw will be almost (but not quite) in sync every second turn (it will be 1/25th of a turn out).

Let's do the same calculation for 3 tpi by adding a 45/30 combination to the train. Remember we're not cutting just watching the dial turns vs spindle turns.

24 (turns of the screw) x (135/127) x (45/30) = 38.268 (turns of the spindle). So basically to bring the same number round on the dial the screw must turn 24 full revolutions but the spindle has turned 28.268 revolutions, which means it's more the 1/4 turn further round than when the mark last lined up. The screw and spindle are now miles out of sync, yet the carriage hasn't moved from the stop.

Unless there's some huge flaw in my thinking or there's some step not yet been given I don't see how anyone can use this system successfully.

I don't see any flaw in your thinking, it is exactly what I think. I can't see how it would work unless you are also counting spindle turns.

You don't need a threading dial, just marks on the end of the leadscrew will do. If those marks line up, the leadscrew is in exactly the same position each time.

If you also start from a stop, the saddle is in exactly the same position each time.

That is all true Mart, but the spindle must also be at the same point in the turn every time the mark lines up. That's where it will go wrong.

As I see it, the lead screw will be in the correct (same) orientation when the dial indicator is at the same number and the carriage will be in exactly the same spot every time if hard up against a stop. So it should work without problems - as long as the stop doesn’t get moved!

My lathe, with auto-trip on the long travel stops exactly the same place every time. I don’t need to even watch it as it cuts to a shoulder (except the tool would then be rubbing) so this (obviously going against a solid stop in the other direction) it will be no different.

I agree that backlash, if reversing the drive train, will not affect the cut - as long as withdrawing the cutter is carried out while reversing. It is not a case of ‘simply’ reversing the motor - the tool must be retracted while driving backwards - towards the starting point.

As Victor Meldrew would say, I don't believe it!

The OP claims he can cut metric threads on an imperial lathe with this technique, so let's have an example. Let's say the lathe with a 6 tpi lead-screw is set to cut M3 (1mm pitch)The ratio between spindle and lead-screw is (1/25.4)/(1/6), which is 1/4.2333. With the saddle against it's right hand stop, the lead-screw mark against its datum and the half nuts engaged, the tool is aligned with the thread. Now disengage the half nuts and turn the spindle until the lead-screw has turned one rev and the mark is against its datum again. The spindle has turned 4.2333 turns, so the tool cannot be aligned with the thread.

I've looked at the Oxtool video, and that is a totally different technique. He disengages the half nuts to stop the travel, backs out the tool, and stops the lathe sets the lathe in reverse, re-engages the half nut at the same thread indicator setting and reverse travels the saddle. This does work, because he is re-engaging the half nuts where they were.

Edited By duncan webster on 24/07/2019 23:50:59

A small correction in my post above, the formula should read 25.4/6/127x135 = 4.5

After all that there has been written about threading calculations, confusion persists. The dissenters are correct. Perhaps gear trains and their calculation confuses people. Although Pete and Duncan have already explained, and not wishing to steal their thunder, let's look at it another, possibly simpler, way.

Say you have a lathe with an 8 TPI leadscrew. Each leadscrew revolution moves the carriage 1/8", or 25.4 X 1/8 mm. Multiplying top and bottom by 5, to get the calculation into whole numbers, gives 127/40 mm. So, in order to cut, say, a 1mm pitch, the leadscrew must revolve 1 / (127/40) = 40/127 revs. This must, of course, happen each spindle revolution. This is where the changewheels come in, but we don't need to bother with that complexity. The problem is closing the half-nuts at the right spindle and leadscrew relationship. The point is that 40/127 is inconvenient because the smallest whole number that can be used to multiply this fraction by, in order to get a whole-number result, is 127. Only then will the leadscrew and spindle be correctly synchronised. So, the suggested method would work if one had a 127 tooth pinion on the thread dial indicator...

Edited By Kiwi Bloke on 25/07/2019 10:02:20

I tried out all of the metric threads that the lathe can produce with the gears that I bought in conjunction with the QCGB, and am afraid that 4 worked and 5 did not. The coarser pitched of the failures led me to believe that allowing the threading dial to run two or three turns before re engaging the nuts would have probably have worked. this would take up so much time that I have to admit failure. I shall keep a low profile for a time in case they come and take me to an old peoples home.

Looking back to the third post made by DC31k, the answer is demonstrated in the OXTOOL video on you tube.