PDA

View Full Version : Servo electronic threading



localele
27th August 2012, 06:16 PM
Thought I would share this link about a servo driven electronic threading conversion. Would seem to be a nice addition.
These guys also do some very nice restoration jobs on Hardinge lathes. I would like to get their painter to do my ute .
Babin Machine Tool (http://www.babinmachine.com/index.php?HLVELECLEAD)

Bryan
28th August 2012, 12:22 PM
Do some Googling before doing business with that guy. Caveat emptor.

localele
28th August 2012, 04:10 PM
Hi Bryan, I did some googling and came up with a few links to a Different person. Not related at all to the link I posted as far as I could tell.

Machtool
28th August 2012, 06:51 PM
Yep, That one is Paul Babin, no relation to the Al, Allen, Mike, Hillary, LAtool Babin.

Paul Babin seems to have a good reputation, with the guys that have done business with him, at the P.M Forum.

In summary.
Babin from Connecticut = bad.
Babin from Massachusetts = good.

Phil.

Greg Q
28th August 2012, 10:05 PM
I like the idea of an electronic leadscrew control. A guy in Canada sells an inexpensive ELS kit for stepper or servo power.

Taken to its logical conclusion it would be great to have an infinite speed ballscrew with a jog/move control to allow for "manual" feeding...you could then eliminate half nuts and the rack.

Eisen in Taiwan make an HLV-H clone with such a system

CTL-618DT Super High Precision Toolroom Lathe with Digital Threading - Eisen Machinery Inc (http://www.eisenm.com/ctl-618dt-super-high-precision-toolroom-lathe-with-digital-threading.html)

Bryan
28th August 2012, 11:21 PM
My mistake. Thanks guys for clearing that up.

Greg Q
28th August 2012, 11:33 PM
Yeah, if I was Paul in Mass. I think I'd rename my business to "The Good Babin"

GQ

RayG
29th August 2012, 12:00 AM
Taken to its logical conclusion ....[/URL]

Hmmm that would be a full CNC lathe I guess...:)

I also like the electronic lead screw idea, I was at [URL="http://www.schachermayer.at/"]Schachermayer's (http://www.eisenm.com/ctl-618dt-super-high-precision-toolroom-lathe-with-digital-threading.html) in Linz yesterday eyeing off a CNC lathe, a Weiler similar to this one WEILER Praktikant VC Plus (http://www.schachermayer.at/maschinen/metall/?s=19785)

Electronic Lead screw is a nice idea and very likely do-able as as retro fit for quite a few machines...

Regards
Ray

Greg Q
29th August 2012, 12:10 AM
I was considering doing an ELS on my lathe but then stumbled upon the genuine gears that I was missing on PomBay for next to nowt. To complete my set I only need a 40 tooth, but it can wait til the Deckel is going as I have the 1.5 modul cutter set on the shelf here. The Canadian ELS also has provision for a powered cross slide which enables taper cuts and power tool retract for fast threading.

Electronic Lead Screw Main Page (http://www.autoartisans.com/ELS/)

Greg

RayG
29th August 2012, 07:38 PM
Hi GQ,

Thanks for that link, I hadn't seen it before, looks like a good project. :2tsup:

Regards
Ray

cba_melbourne
29th August 2012, 11:27 PM
Greg, the Canadian ELS is perfect for threading things like fasteners. And at the same time provides an infinite number of automatic feeds via the half nuts. But it is useless to make acurate threads like leadscrews.

This is so for two reasons. Once because of recursive rounding errors by the software. It is only very little every revolution, but if you make a leadscrew it may easily be half a turn out of pitch after 100 turns. Imagine the leadscrew of your lathe being 1/2 turn out of pitch from end to end..... The other reason is that the pitch linearity from turn to turn is dependant on the cutting forces and the spindle motor speed regulation. This is because the main spindle has only one syncronisation pulse per revolution. The electronics can only detect a syncronisation error between main spindle and leadscrew once every full main spindle revolution. Imagine the spindle slowing down during a cut, but the leadscrew continuing at a steady speed. Then the error is detected and the leadscrew slowed down.... the result is a slightly drunken thread. Both problems will not matter in most practical cases, but you will never be able to make an accurate spare leadscrew with this ELS.

It is surprising how difficult it is to design an electronic equivalent to a humble gear drive that is even remotely as accurate! Chris

jhovel
29th August 2012, 11:48 PM
Ray & Greg: that's the ELS (Electronic Lead Screw) kit I built for my feedscrew-less lathe. Just not quite yet fitted....
Chris: the recursive error thing has been solved some time ago after a very very very long discussion on the ELS Yahoo group with lots of experts contributing to the algorithm correction. The current processor code allows you to do accurate long threads.
There was a comparison article in one of the model engineering magazines (I think it was Model Engineer's Workshop) that reported that the British equivalent device (not kit) was able to long threads and the ELS kit was not - but they based the comparison on an older version of of the software. That was a pity.

The ELS kit, by the way, is sold by the co-ordinator of the open source ELS project that ran for quite a long time on the CNC Zone forum a few years back. His name is John Dammeyer and he is very easy to get along with and most helpful in every aspect of building or implementing the ELS.

One of the biggest advantages of the kit for me is that it can drive both Z and Y axes on a lathe and allow you to make taper and parallel threads in any pitch or TPI irrespoective of the leadscrew pitch, do accurate tapers and automate repetitive tasks without all the steps in software that CNC would require. It also has a CNC interface that allows Mach 3, Turbo CNC or whatever to interface and drive the axes motor controllers without having to have two sets of these if one wants to do both.
I like the idea of a stand-alone solution a lot for a lathe.
Cheers,
Joe

cba_melbourne
31st August 2012, 12:02 AM
.....Chris: the recursive error thing has been solved some time ago after a very very very long discussion on the ELS Yahoo group with lots of experts contributing to the algorithm correction. The current processor code allows you to do accurate long threads.
There was a comparison article in one of the model engineering magazines (I think it was Model Engineer's Workshop) that reported that the British equivalent device (not kit) was able to long threads and the ELS kit was not - but they based the comparison on an older version of of the software. That was a pity......


Joe, I understand the recursive error is due to floating point math erors with the 8bit PIC. It can only be solved with a more powerful processor. The other problem was due to the 1 pulse per revolution spindle encoder. Again, the 8bit PIC is simply too slow to handle multiple pulses from a real encoder. See also John Dammeyer's messages 5042 and 5053 in the ELS group. It would be really great news if John found a solution - but I do not think he had a breakthrough yet. I have been very interested in John's ELS from the beginnings, even took out a "circuit cellar" subscription only to get my hands on his detailed article on the ELS project. I never built one, exactly because of the above issues. But would be glad to reconsider if one day it can cut threads with an accuracy that approaches what a simple gear driven leadscrew is capable of.


The English ELS you mention was the "Frog", sold as ready made ELS for the little Sherline lathe. It suffered from the same single pulse encoder problems though. And before this, a certain Fritz Link had published the plans and diagrams to make an ELS for an Emco Compact 8 lathe, using a fast PLL circuit with technology from the late 70's with homemade multipulse encoders, this was accurate but rather complex....

Anyway, you now built one of John's ELS and I hope you will be posting pics and let us know your experience with it. Chris

MuellerNick
31st August 2012, 08:08 AM
Once because of recursive rounding errors by the software. It is only very little every revolution, but if you make a leadscrew it may easily be half a turn out of pitch after 100 turns.

That problem has been solved looooong before by Mr. Bresenham (http://en.wikipedia.org/wiki/Bresenham%27s_line_algorithm). You don't even need FP, integer does the job perfectly.
Though I don't know how ELS does it.


Nick

cba_melbourne
31st August 2012, 10:39 AM
That problem has been solved looooong before by Mr. Bresenham (http://en.wikipedia.org/wiki/Bresenham%27s_line_algorithm). You don't even need FP, integer does the job perfectly.
Though I don't know how ELS does it.


Nick

Nick, this is not a fundamental problem, it all has been solved long ago for industrial grade CNC by throwing money at it.

The ELS in question is a very low cost open source project dating back to 2005. see here:
Electronic Lead Screw Main Page (http://www.autoartisans.com/ELS/)
It can be built for under AU$500 (the kit is $250, then you need some extra bits like stepper motor and enclosure etc). It is aimed at smaller lathes like the South Bend 9", which either never had a quick change gearbox, or where change gears are incomplete or went missing. To keep this project affordable for the backyard workshops, John had to compromise. The 8 bit processor just cannot do the job fast enogh to accomodate the requirements for making long threads wit accuracy. Other than that, for the modest outlay it is a very powerful ELS than can create threads of any pitch (metric or imperial or module) on any lathe equipped with a leadscrew. Unlike a real CNC conversion, the machine remains in every respect a manual lathe. The ELS simply takes the place of a (very complete) Norton gearbox. Chris

MuellerNick
31st August 2012, 05:26 PM
Nick, this is not a fundamental problem, it all has been solved long ago for industrial grade CNC by throwing money at it.


I knew the ELS-project even before it worked,
You don't have to throw money at that problem, brain cells are way more appropriate.
Did you real the link? The algorithm is fast, as it doesn't need any floating point, so there are no rounding problems.


Nick

cba_melbourne
31st August 2012, 07:18 PM
I think John Dammeyer is well aware of the Bresenham algorithm and has been using it for his ELS code. There are limits to what brain cells alone can do. But is it not fascinating, how difficult it is to replace something as plain and simple as old fashioned thread cutting gears in a lathe?

RayG
31st August 2012, 07:34 PM
Hi Nick,

I think it would be a nice upgrade to that ELS kit to re-design it to use a dspic (which of course has on chip quadrature encoder interface)..

Do you know if anyone has done it already?

Regards
Ray

localele
31st August 2012, 07:40 PM
You guys are losing me with all the terminology but here is the logical conclusion.Made in Taiwan.

cyclematic toolroom lathe 工具車床、小型車床、精密車床、工具車床 韻光機械 (http://www.cyclematic.com/ctl-618e.htm)

MuellerNick
31st August 2012, 11:21 PM
I think it would be a nice upgrade to that ELS kit to re-design it to use a dspic

I have no clue how the software is implemented. I just wanted to say that there is a clever way to solve that specific problem. Clever vs. brute force.

I have been a professional software developer for about 20 years. And micro-controllers are damned fascinating!


Nick