Daydreaming about an Electronic Leadscrew

[jpanhalt]

The leadscrews on my lathes are imperial thread.  I would like to cut metric threads.  I know there are conversion gears for some lathes, like older Hardinge, for which one of the pair is 127 teeth and the other is 125 teeth (or similar).  The little lathe I want to convert is a Prazi SD300 (attached).  It's a bit unusual in not having half-nuts or threading dial, but can still be used for threading with various change gears. It's still a PITA to do.  The leadscrew is 20 tpi (0.05" per turn).  My first thought was to make an electronic half-nut and threading dial, which should be relatively easy.  Here's the daydream part: If going to that trouble, why not design for real metric too? 

My rough plan is to use cog belts and a stepper motor to drive the leadscrew with a rotary encoder coupled to the spindle.  I will probably have an encoder on the leadscrew as well, either for feedback or simply for doing what a threading dial does.

I haven't found any cogged/timing pulley that has 127 teeth.  Can anyone suggest another simple way that will give a very close approximate to metric? I haven't worked out the allowable tolerance for various metric threads, but read that 1% error per thread (i.e., 0.01mm per 1mm pitch, or 0.05mm off at each end for 10 threads) is OK for a home workshop.

[tpowell1830]

There are several YTers that developed an electronic lead screw but I liked this guys approach.

[Andy Watson]

I've been watching the Electronic Lead Screw project , perhaps one day  I will get around to doing something about it. In the meantime, I have both 21 and 63 tooth gears in my set of change wheels. If you only require occasional use, could you 3D print a 127 gear?

[jpanhalt]

Thanks for the speedy replies.  I am not too worried about the coding.  I did some of that several years ago to coordinate a stepper motor to a rotary encoder.  I used an optical one (lost in memory) and a magnetic one (AMS5048). 

My main concern was mechanical for getting as close to 1.27 to 1 (or whatever).  Of course, doing metric threads on imperial lathes is not new, and I found a table some generous person posted years ago (attachment1).  W= gear on spindle (Werk) that drives Z1;  Z1 and Z2 on a keyed shaft (white gears) and turn together; L = leadscrew (attachment2).  The gear combination shown for 1mm pitch came up to 1.000606, which is less error than I suspect for the leadscrew when cutting a 20 tpi thread.   Hope my math is right. 

I need to start playing with a spreadsheet.  What are good hobbyist sources for cogged pulleys?  I am not totally adverse to gears, but belt drives are quiet.

[james_s]

I need to start playing with a spreadsheet.  What are good hobbyist sources for cogged pulleys?  I am not totally adverse to gears, but belt drives are quiet.

I would try McMaster-Carr, they're my go-to for mechanical bits and pieces.

[jpanhalt]

Thank you , James.  I always check McMaster-Carr.  When I lived on the East side of Cleveland, I could easily drive to their will call desk and get same day without shipping.  It's a great resource, but expensive.

[jpanhalt]

As the title says,"Daydreaming."
This project has been in the back of my mind since Fall as I finished up my other project (a SmartRoast thermometer).  I had not searched very hard for information, but the 4000 pulses per revolution in the YT seemed a bit high.  On further search, I found this site: https://www.rocketronics.de/en/els/faq/#:~:text=The%20pulse%20rate%20of%20the%20encoder%20should%20be%20400.

300 euro for a system?  Not bad.  Anyone with Rocketronics.de experience?  BTW, it suggests an encoder with only 400 pulses per revolution.  That sounds closer to what I would have guessed.

[Benta]

Far too much effort, there's a much simpler trick.
I have a somewhat larger lathe with metric leadscrew, but that's just the "inverse" of your issue with the same solution.
The problem is, that a 127-tooth gear wheel won't fit on the end of the spindle (or on the banjo), it's simply too large.
The solution is to use a 66/52 transposer pair, which will fit.
That gives a ratio of 1.2692, which is close enough to 1.27 for all uses I've ever had.

Now, whether Prazi offers those, I don't know, but 33/52, 66/26 or 33/26 transposing will work the same.

EDIT: OK, it seems Prazi don't.

Apparently you have a complete set, which is 30, 35, 40, 45 .50, 55, 60, 65, 75.
The missing 52- and 66-tooth gear wheels shouldn't be too hard to find.
The specs for the SD300 are pretty standard:
Material: Nylon/Delrin, most probably reinforced.
Module: 1
Width: 8 mm
Width at hub: 10 mm
Pressure angle: 20 degrees.
No idea about hole size, but that should be really simple to measure on your machine.

Cheers.

[jpanhalt]

Thank you, Benta.  We seem to share a deep love for machines.

Neither the 66 nor 52 tooth gears were available from Prazi or its successor(s), but they may not be hard to find.  The other problem is the absence of a threading dial/half-nuts and stopping before hitting the spindle  (without a groove).  When a groove is possible, I use it.  Otherwise, I modified the crossfeed to mount my tool post on the extreme rear, machine off the back, and reverse the headstock.  Let me add, I have a real, quite capable lathe in storage.  The Prazi was a stop-gap measure.  Unfortunately, "stop-gap" has turned into 20 years.  ):  A high priority is to bring the "real" lathe out of storage this Summer.  It's a Smart and Brown British lathe from 1966 and is a joy to use, once you get used to it -- typical British.    The little Prazi is in my basement, which is warm.  The Smart & Brown is in the barn.

My main reason for this dream is to flesh out a project that will keep me occupied next Winter.  The days just fly by when I am doing a project, making parts, and writing code.

Regards, John

[langwadt]

Thank you, Benta.  We seem to share a deep love for machines.

Neither the 66 nor 52 tooth gears were available from Prazi or its successor(s), but they may not be hard to find.  The other problem is the absence of a threading dial/half-nuts and stopping before hitting the spindle  (without a groove).  When a groove is possible, I use it.  Otherwise, I modified the crossfeed to mount my tool post on the extreme rear, machine off the back, and reverse the headstock.  Let me add, I have a real, quite capable lathe in storage.  The Prazi was a stop-gap measure.  Unfortunately, "stop-gap" has turned into 20 years.  ):  A high priority is to bring the "real" lathe out of storage this Summer.  It's a Smart and Brown British lathe from 1966 and is a joy to use, once you get used to it -- typical British.    The little Prazi is in my basement, which is warm.  The Smart & Brown is in the barn.

My main reason for this dream is to flesh out a project that will keep me occupied next Winter.  The days just fly by when I am doing a project, making parts, and writing code.

Regards, John

with an "electronic" lead screw it is easy to add an auto stop feature so the carriage stops at a preset position, once you reverse the spindle it can then pickup at the right spot to stay in sync

[jpanhalt]

Yep, in fact, I built something like that soon after I got it based on a Digimatic (Mitutoyo) digital dial caliper that had a modified SPI output.  In those days, I had no idea how to read SPI, so I simply zeroed the dial and looked for when the signal went negative.  Crude, but it worked.  On/Off is a typical SPDT relay circuit that holds on until the coil current is interrupted.  Hence, it was easy to stop. 

One of my goals for the ELS is of course to do the same.  Another thing that can be done is to taper a portion of the last thread with a servo/stepper on the cross slide.  I am not going to go that far.  I just do a rough approximation manually when I don't use a groove.  I mentioned earlier cutting threads from the back side when not using a groove at the end.  In that case, tapering the last thread comes naturally as you cannot feed into a piece that is not turning.

Edit:  This bolt was done that way (attachment)

[jpanhalt]

Moving Forward Slowly

I am looking for the leadscrew drive now and am considering a servo drive instead of stepper.  EXmotion(an Emerson company) has "All-in-One" BLDC 24V motors and drive for about $200 or less.

Links:
Selection: https://www.ezmotion.co/
User's manual: https://media.ezmotion.co/product-documents/user-guides/MMS-MMP7xx-x1_User_Guide_r1.0_5_6.pdf

Does anyone have experience with them? 

I am considering something in the 50W to 77W range.  The spindle motor is about 184W DC @ 135V.  It's a very small lathe.  The leadscrew knob radius is about 22mm and is easy to turn.

[Doctorandus_P]

For mechanical transmission, you will need a gear with 127 teeth, because it is the smallest common denominator between Metric and Banana units. Any other ratio will cause rounding errors. In your first post you write that 1% pitch deviation is acceptable for you, but that only works as long as you keep the half nut engaged. This method can not be used with a lathe treading dial. At the moment you disengage the half nut and the threading dial starts turning, you start accumulating errors, and for cutting thread (in maybe 10 passes, and delay's in between) the error will be too big and you will not be able to cut your thread because the passes do not line up. If your transmission ratio is not exact, you always have to keep the half nut engaged and turn the whole thing in reverse to "unwind" the accumulated error for the next pass.

For software, I saw the video from Clough42, and he uses some beefy TI processor and (presumably) complex calculations, but I have not looked at his firmware. A simple and very effective approach is to use a variant of the Bresenham line drawing algorithm. With this algorithm you can make any transmission that can be expressed as a division of integers (the result does not have to be an integer, but can be an rational number) without any error. The bresenham algorithm is exact! And it is also very low on computing power.

For your motor. I suggest to start with a stepper motor for your first projects. You can get more performance out of servo motors, but they are also more dificult to control. It's quite likely you have to set PID parameters to get a stable response form servo motors, and a stepper motor is plenty good enough for this application. You can get the most power out of any given motor at high speed. If a motor is blocked, it can still deliver torque, but it can't deliver power. Stepper motors generally run quite well upto about 600rpm (For Nema23 sized motors). You can use the existing gears to make an (approximately 1:10 to 1:30) transmission between your motor and the lead screw. You can also use a stepper motor with built in gear or make something with a timing belt if that is easier. The exact ratio is not important, you can simply put the exact gear ratio in the Bresenham algorithm.

Also note that gears always have an integer number of tooth. There are no gears with fractional teeth.  This means that any transmission made with gears (or timing belts etc) can be modeled exactly with the Bresenham algorithm and without accumulating error over time.

Edit: "irrational" -> "rational" (Math classes were a long long time ago. even in another century

[jpanhalt]

Thank you for that advice.  I was aware of the use 127/125/100 in larger lathes, e.g., Hardinge HLVH.  I do have a good lathe* (not Hardinge) stored in my barn.  A project this Summer is to fix up an area for it and the mill that is rodent proof.  This Prazi/MD300 is tiny and those larger gears in module 1 won't fit without modification.  So, I bought a 47/37 pair that will fit.  Ratio: 1.27027..., which is pretty close.

I thought about a stepper and watched the Clutch42 comparisons.  In the latest YT I saw, he attached a NEMA 23 integrated servo and driver from Stepperonline and decided to keep it:
[plain]https://www.youtube.com/watch?v=clgpZJFoqIQ [/plain] 
Last night, I bought the same one for about $100 including DHL shipping.  I did firmware for a stepper a few years ago and wanted a new challenge.  Coding is actually a Winter 2023 project.  I wanted to  get a start on it now, particularly the mechanical part as it is awfully cold in the barn in the Winter.

John

*That lathe, a Smart & Brown Model A, which I got in the early 80's is a joy to thread with.  It was built in approximately 1966 and is in great shape.  Unfortunately, it is imperial and Whitworth to boot.   

[langwadt]

.. Stepper motors generally run quite well upto about 600rpm (For Nema23 sized motors). You can use the existing gears to make an (approximately 1:10 to 1:30) transmission between your motor and the lead screw. You can also use a stepper motor with built in gear or make something with a timing belt if that is easier. The exact ratio is not important, you can simply put the exact gear ratio in the Bresenham algorithm.

most small lathe leadscrews are 1-2mm pitch so for most thread sizes the ratio is closer to 1:2

[jpanhalt]

That's correct.  My Prazi is 20 tpi (1.27mm), and everything runs off the lead screw all of the time.  There is no rack pinion for rapid movement, nor are there half-nuts..

[sokoloff]

With this algorithm you can make any transmission that can be expressed as a division of integers (the result does not have to be an integer, but can be an irrational number) without any error.

The division of two integers always (definitionally) results in a rational number.

[Benta]

So, I bought a 47/37 pair that will fit.  Ratio: 1.27027..., which is pretty close.

Nice one! It's better than the 66/52 approximation on my lathe.

[jpanhalt]

It's also available on eBay in a 3D printed plastic version.  I got a steel version from China (considerably cheaper than the 3D print).  They are cast, so center may be a little off.  Years ago, I asked how to center a gear in a 4-jaw for on-center boring.  The answer I got was to use copper strips.  I use 0.025".

[langwadt]

It's also available on eBay in a 3D printed plastic version.  I got a steel version from China (considerably cheaper than the 3D print).  They are cast, so center may be a little off.  Years ago, I asked how to center a gear in a 4-jaw for on-center boring.  The answer I got was to use copper strips.  I use 0.025".

sorta related, https://youtu.be/1paTlW-W9Zo

[Benta]

sorta related, https://youtu.be/1paTlW-W9Zo

Not really, unless you have a very big wallet.

[Benta]

It's also available on eBay in a 3D printed plastic version.  I got a steel version from China (considerably cheaper than the 3D print).  They are cast, so center may be a little off.  Years ago, I asked how to center a gear in a 4-jaw for on-center boring.  The answer I got was to use copper strips.  I use 0.025".

Forget 3D, it has no strength. Like I wrote, the original Prazi gears are reinforced Nylon/Delrin.
Cast iron is a good choice, it meshes well with the Delrin.
Concerning copper: yeah, but aluminium will work just as well and is easier to find.

Can you post a link to the gear supplier?

[langwadt]

sorta related, https://youtu.be/1paTlW-W9Zo

Not really, unless you have a very big wallet.

sure if you do it in steel, he started with 3D printed

[jpanhalt]

@ Benta, Here's the original listing: https://www.ebay.com/itm/174747512509
There are at least 2  that offer the same pricing.

Only the larger gears are hardened. The sizes I ordered are soft castings.  Should be here in about a week.

[Benta]

Wow, dirt cheap to say the least.
You'll probably need to bore them out and mount them on a bushing. Personally, I'd just braze them together after that and use the set as a permanent metric/inch transposer. Easier than making keyways etc.
I look forward to seeing your results.