Electronic Leadscrew for the New Lathe

[vtsteam]

I just spent a discouraging evening looking up electronic leadscrews on the internet. They don't seem to work very well, and, they cost multi $, or both. Or they need a big computer, too. Any suggestions to get past both of those problems would be greatly appreciated!

I read that standalone units don't make sense, and that a supposedly DIY open source unpopulated standalone board costs $250.

I read that a full fledged CNC program can work (and actually, more cheaply, since I already have steppers, stepper drivers, and spare older computers with parallel ports).

BUT a computer and associated gear and wiring is practically the size footprint of the lathe.

BUT also the single pulse per revolution of the spindle in most stepper CNCs is NOT accurate.

BUT some say it is and the reason it isn't accurate is because Mach 3 has bugs in the threading portion of the code.

But TurboCNC has been used for threading for over a decade.

But is it accurate enough?

But LinuxCNC uses a quadrature signal besides the single pulse, so maybe it is acceptably accurate.

But GRBL a simple CNC program that can run with very simple compact computers like Raspberry Pi to control an Arduino has no lathe threading capability.

Most threads devoted to this kind of thing seem to deteriorate after the first two or three pages into arguments about closed loop and open loop theories and the number of pulses per revolution or inch, etc. and end up nowhere.

Here's what I'd like:

A small box that had a small display and a few buttons, that would hook to a sensor on the headstock, and to a driver for the stepper motor, and would allow you to enter the feed rate desired, direction, and if needed threading input requirements. That's all. And that there was a board available for doing that for about $100 or less.

Is there such a thing? My guess at least from reading so far, is no.

Yes?

[Brass_Machine]

I did see something like that awhile back. It was essentially a cnc "lite".

I will see if I can find it again.

However, why not use an arduino and make your own?

[Brass_Machine]

I can't understand what he is saying... but something like this? Arduino control...

[bertie_bassett]

Ahh the good old ELS problem.  I remember getting just as frustrated trying to find a simple solution.

What do you actually want it to do though??

Do you want it to do the whole threading job?? Or do you want to do the threading yourself and just have the speed controlled electronically ?

If you want to do it yourself, all you need to replicate electronically is the change gears,  so a simple spped control circuit would be the main part, tie it to the headstock speed with a comparator and bobs your uncle!  Then for threading either keep the half nuts engaged or use a thread dial.

If you want the electronics to do the whole job, then you'll need to have something keeping track or where everything is plus drive 2steppers, so then you getting into big enclosures with lots of parts. In fact you'll be jusst pn the edge of a full cnc machine.

Personally I'm going to go stick with change gears for now ( if i ever make any). And in the future have a simple speed controlled lead screw 

[lordedmond]

This may be a completely dumb question but

If you use the system as Bertie speced out you have to reverse the lathe , so with a single encoder pulse from the headstock how does the system know how to reverse the lead screw ?

In my mind a single pulse will give position but not direction , but I may be wrong again


Stuart

[SwarfnStuff]

I subscribe to the KISS principle wherever possible.  If I were to go ELS I think Bertie's idea would be my choice so I would like more info from Bertie, AKA explain a bit more please B. This is one of those good to have mods in my one day do list. 

[PekkaNF]

Damn you VT

Few years back I did some armchair engineering and went trough the whole google over and found arguments/flames/missing the point and you name. It came pretty obvious that: Whole lot is pretty simple at first sight, but you need to make and informed decision on following approaches:
* KISS = limited, but least you have some chance in success. When it is comes mostly programmable project there is a great temptation to add way too many conflicting features. Deside what you want and stick to it. I went other way and didn't complete it.
* Do you want only to eliminate change gears, or also with spindle start/stop? How about automatic threading with multipass? Do you have temptation to make taper and taper threads....see where I'm getting?
* Mechanical or system inaccuracies - acme screw, backlash, lightweight? Or ballscrew and great stiffness.
* Brute force or a lot of fiddling, esitmation and cheesy math. Like this spindle speed caculation: Do you have enough spindle gear inertia and stiff (AC drive or constant rpm servo) to keep the spindle near constant velocity and can get by with one pulse per revolution. Or do you have lightweigh mechanics and fluctuating DC-drive and really need all the pulses you can get and processor to cope with vartying spindle speed and beeffy servodrive to keep up the feed with this fluctuation. Like shooting moving target that chances it's speed all the time?

The big question that dawned to me is: If you lathe and gear is big enough to make electronics and calculus simple it is likely to have threading feeds. If your lathe does not have it you have to make it CNC way and then this ecological niece is not that big.

Anyway, I don't see any fundamental reason it would not work simple if carefully designed. Therefore I'm tooling up slowly to make one of my lathe "ready" for some sort of ELS.

Pretty sure my ramblings will not add much, but here is one eager watcher.

Pekka

[JHovel]

What's wrong eith the ELS here http://autoartisans.com/ELS/
What made you say "they don't seem to work very well"?
I've built it and converted my lathe. Not only is it dead easy, but it also has a port to connect to a CNC controller if you want to go that way for some jobs. And you can continue to ue the lathe manually.
When threading, it does not reverse the spindle. It retracts the tool and reverses it quickly back to the starting point, feeds it back in with a setable increment and runs down the thread again. It moves the tool along the flank of the thread so it cuts only on one side - the 'proper' way.
I lave the fact that it has its own look-up tables for just about all standard threads, can do taper threads, turn standard tapers  and can broach without the spindle running.
What more do you need?
Joe

[John Rudd]

Full kit available here....

http://medw.co.uk/wiki/index.php?page=ELS+Price+List

[John Stevenson]

Been thru this and touched on most of what has been mentioned here.
Got the tee shirt, been sick all over it and it's now a shop rag.

Unfortunately got to go out today so wont be able to post until later tonight but there are answers.

[RussellT]

I made an electronic taper turning device which used a geartooth position sensor to detect the spindle speed and then stepped the top slide motor every x pulses for each tooth.  The software was very simple and runs on an old laptop but would easily fit on a Pic or similar.

The difficult bit in extending that idea is picking up the thread on subsequent cuts.  You could use a gear with a wide tooth to detect spindle position as used on some engines for crankshaft position sensing but the software would be more complicated.

My initial thought would be to use a stop to set a consistent start point and then start the cut on the wide tooth.

From a software point of view it would be easier to make it work with two sensors.

Russell

[vtsteam]

Wow, I thought I'd wake up to maybe 2 replies.

Answers to various posts:

I don't mind a simple threading function. I don't plan to add an X stepper, and I don't lathe thread often.

A straightforward lathe (the one I'm building) with conventional acme screws. Anti-backlash nuts or spring or weight per ancient TurboCNC practice.

Reasonable practical thread accuracy for model engineering, not commercial/NASA/theoretical micron air bearing
ultimate spare no expense, etc. hooha.

Under $100 for a functional board so a 175 UK pound board, plus accessories doesn't fit spec.

I don't want to design and code something. I want to finish my lathe and build engines. Building the lathe is fun and rewarding, but it is in the way of building engines. I don't want now start also learning enough to design a circuit and program a microprocessor and have that get in the way of finishing the lathe so I can build engines. Hope that makes sense....

I don't mind soldering/populating a simple board kit if absolutely necessary, and if $100 or less. Or connecting up existing components (ie. an arduino, RPi, breakout board, stepper driver, steppers, power supply -- and I already have all of those components, or an encoder, which I don't have yet)

What I mean is the thread controller board -- whatever that is.

I have no prejudice against what is called a "Full CNC" solution vs an "Electronic Leadscrew" solution other than the size and comlpexity of an old honkin computer and monitor and cables all over the place in a tiny overcrowded shop on a small lathe. Actually, some of the electronic leadscrews actually look as big and messy as a full CNC rig so this may not even make sense.

The closest parallel to what I want is the ancient TurboCNC lathe at DAK, and if it worked well enough for reasonable threads and could run on a small Arduino sized board and interface with a couple line LCD would completely do it for me. I've used TurboCNC otherwise, and I speak DOS, so no problems at all and no prejudices about lack of bells and whistles.

John, I trust you know all of this, have been through it and probably do have a solution or the reasons to stop wishful thinking and get on with it.

Also, one more thing. I don't mind a mechanical rather than electronic leadscrew.

I just don't want to have to read a chart in tiny numbers in dim light on the left hand side of the lathe inside a cover with my head turned sideways, memorize those numbers on a particular line, find a wrench to fit gear retaining bolts, grab ahold of blackened oily gears remove them, loosen a greasy cast iron banjo and have it flop onto my thumb, put keyed spacers into the right hubs for doubled up gears, figure out which way they face -- behind or in front of each other, figure out which of three slots they go in, get that all wrong, remove the gears, wipe safety glasses with grease, apply to forhead, try gears again in a different slot, put a piece of paper in between gears to set correct spacing try to hold it in place while adjusting banjo, sliding gear in place, and holding a wrench to tighten simultaneously with only two hands, etc.

Then when finished threading, repeating the process to return to fine feed.

Is there an easier way -- and I don't want to make a conventional QC gearbox -- which project would seem to delay building the lathe to build engines even further.

Maybe something with timing belts to at least get rid of the grease?

Or what about a master thread deal, like Unimats had?

[vtsteam]

Somebody needs to come out with a 5 volt single board computer the size of a credit card that runs DOS natively and has a CNC compatible parallel port and a B/W LCD display capability. Doesn't even have to be more than 100 Mhz  and look like a 486.

Or better yet, we need a DOS tablet, with a parallel port! 

[John Rudd]

Rasberry pie......small board computer.....Doable?
I'd have thought so....

[vtsteam]

John, I've got one but the available CNC package, a GRBL interpreter which sends instructions to an Arduino board running GRBL itself, does not do threading.

The (or a) problem probably lies in the fact that communication is via USB, and so realtime monitoring and adjustment would be slower than some kind of parallel interface with lots of data in and out lines.

One other reason (said a GRBL developer) that threading is also a low priority, honest, I read this online, is because "lathes are more dangerous than milling machines," and the code would need to be more robust. These guys are 3D printer oriented.

Apparently this hasn't stopped other CNC programmers, and I'd hate to be locked inside the cabinet with a running Fanuc machining center.

Heading out now to cut some firewood wth the chainsaw......truly....


 

[awemawson]

So it IS a lathe that you are building, not a ladder! The dog'll be really pissed off 


ps no problem threading using TurboCNC - converted my first Denford ORAC CNC lathe that way an awfully long time ago !

[bertie_bassett]

my ideas are purely in my head and have never ben tested, so bear that in mind!

with regard to reversing the lead screw with the spindle :- a spare contact on the motor contactor or switch can easily be used to tell the stepper to reverse.

or if you have a thread dial indicator thingy, just leave the spindle running, release half nuts at end of thread, manually wind back the carriage, adjust cutting depth and reengage half nuts when dial tells you to.

seems simple in my head.

other musings:

iv often wondered if a stepper were to be driven by the spindle, could the resultant voltage produced be enough to use as a speed reference? plus the stepper could also be used to drive the head for indexing or very slow speeds??

[vtsteam]

Eric, Something like that, yes, looks promising.

Russell, sounds like a good principle and idea, but I need something that is working now. 

JHovel, there's a thread in this forum devoted to the problem of inconsistent threads w/single spindle point pickup-- don't have the address of the thread offhand, search if interested... 

John S. figured you'd know..... 

Pekka, I think I might have already touched on most of the questions you asked so far, but simple threading, acme feedscrew, and the spindle motor will be a 2.25 hp (mfr rating) DC treadmill motor and  DanFoss Cycletrol 150 controller. It does still have the big cast iron flywheel on the motor and, if it would be an advantage, I could retain that (though I hadn't planned on it). This is for my new abuilding 9"x12" heavy mini-lathe, I mean dog ladder. 

Bertie, some interesting ideas there.    But not a working circuit/software/firmware reality.

Andrew, lathe? What lathe? This is just a what if kinda thing.........

[bertie_bassett]

this any good to you?

 http://www.homemetalshopclub.org/projects/electronic_lead_screw/els.html

[PekkaNF]

The more I think this at my point of view the more I'm inclined to build "simplified" ELS type feed to chooce the feeds and rely gears on threading, very tempted to calculate simple norton gearbox for 90% of the three/four mostly used lead.

Normal feed is not that demanding on lead accuracy and does not need all calculation for threading start/approach stuff and no temptation to try out any automatic cycles.

There is one manufacturer that uses timing belt on threading least one Wabeco and I have seen timing belt on one Austrian lathe (it also has a "tow hich" to feed the tail stock).
http://www.lathes.co.uk/wabeco/img18.gif
But I don't see really any benefit on timing belt spagethi.

Pekka

[vtsteam]

Already saw it last night Bertie. Nice theory and prototype info, but is there a board or kit available?

I like the one Eric posted a video for, since it seems to use off the shelf Arduino and Arduino Shield, but much more info needed. Software available? How well does it cut threads? Does it require that Shumatch DRO in the video to work?

Pekka, if I have to mess with gears for threading, might as well go whole hog and skip the electronics just for feed.

I still do wonder about the master thread follower type (unimat style) as possibly the simplest mechanical type.

Then I start thinking take it off the lathe altogether and just make some kind of hand cranked threading machine. Seems like a lot of people just crank the spindle by hand anyway, so why even thread on a powered lathe then?

I guess a master screw and follower wouldn't be practical for making long pieces, leadscrews and such, but my lathe is going to be 12" between centers, so would I really need big threading capacity anyway?

[bertie_bassett]

im sure iv seen something similar that was complete and working, don't think it was a kit, but showed exactly what was srequired . unfortunalty iv no idea where I saw it, might have it saved on the old laptop, will take a look tomorrow if I get a chance

[PekkaNF]

Pekka, if I have to mess with gears for threading, might as well go whole hog and skip the electronics just for feed.

jea but, no but,
feeds are used pretty much all the time
threading only max. once per part
AND you want to use them in alternating order....no need to change gears (yuk) until thread pitch needs to be changed.

I could setle electric motor wth variable feed (mm/r) but rpm of it must be ofcourse controlled by spindle rpm. There is allways manual control like in manual crankking, but pretty often it would be nice to "dial in" feed, even if it is off by 5-10%.

I saw a little writeup on MEW I think few years back. Someone tried two popular ELS systems and he was able to produce thread, but they did not were really that impressive and he needed to get the parameters close to right. Least that was my impression. I desided it wasn't for me. Looked like it made easily thread looking form, but did not produce goog quality threads consistently.

I still do wonder about the master thread follower type (unimat style) as possibly the simplest mechanical type.

http://img18.photobucket.com/albums/v55/EPAIII/SetUpOnTable.jpg
http://bbs.homeshopmachinist.net/threads/5095-Thoughts-on-a-fusee/page4

If you anly need short threads I can see the attraction to that threading "formers and followers" method.

Then I start thinking take it off the lathe altogether and just make some kind of hand cranked threading machine. Seems like a lot of people just crank the spindle by hand anyway, so why even thread on a powered lathe then?

I remember seeing a picture of crude "screw lathe" that was used on mass proction around WWII. One machine more and you loose "register". But it would me interesting to see.

Pekka

[Country Bubba]

Somebody needs to come out with a 5 volt single board computer the size of a credit card that runs DOS natively and has a CNC compatible parallel port and a B/W LCD display capability. Doesn't even have to be more than 100 Mhz  and look like a 486.

Or better yet, we need a DOS tablet, with a parallel port! 

Hey Steve, how about this box for dos??

http://www.roboard.com/ncbox-189.html

And it can be found here:

http://www.robotshop.com/en/roboard-ncbox-189-cnc-machine-controller.html

And their located in your part of the world!

Robotshop inc.
555 VT Route 78 suite 367
Swanton, Vermont, USA, 05488

[John Stevenson]

Somebody needs to come out with a 5 volt single board computer the size of a credit card that runs DOS natively and has a CNC compatible parallel port and a B/W LCD display capability. Doesn't even have to be more than 100 Mhz  and look like a 486.

Or better yet, we need a DOS tablet, with a parallel port! 

Hey Steve, how about this box for dos??

http://www.roboard.com/ncbox-189.html

And it can be found here:

http://www.robotshop.com/en/roboard-ncbox-189-cnc-machine-controller.html

And their located in your part of the world!

Robotshop inc.
555 VT Route 78 suite 367
Swanton, Vermont, USA, 05488

No mouse terminal ?