Question on using stock leadscrews with CNC

[Titaniumboy]

Hello,

My first post here. I just found this forum a week ago, and I'm amazed at the level of skill here. Many, many impressive projects ranging from CNC conversions to custom-built bench mills and lathes. I can't believe how many hours I've already spent looking over some of the threads here.

I have never owned a mill or lathe (other than the small Sears lathe that I bought off of a buddy for $100 about 10 years ago, but have never used). I did take some machining classes through the local community college in another state about 12-15 years ago, but that was a long time ago, and it was very hard to build any skills with only touching the machines once every week.

I am currently trying to decide between an X2, X3, or RF45 clone. The prices for the first two would be $391 and $800, respectively (after the 20% Harbor Freight coupon). The RF45 clone looks like it can be delivered for $1550 from Lathemasters, but they appear to be out of stock at this moment. So the X3 is about double the price of the X2, and the RF45 is about double the price of the X3 (and quadruple the price of the X2). Currently leaning toward the X3, but that could change. I realize that these machines have serious limitations, and that each would need a complete rebuild as shown here: http://www.arceurotrade.co.uk/projec...on%20Guide.pdf

My current plan is to take some more machining classes at one of the local community college (by the way, it was very difficult finding any community colleges that even had a machining technology department anymore in this area), and get a solid background in manual machining. I think it would be useful to know how to run a mill manually before attempting a CNC conversion. I would think that a CNC conversion would not begin before next fall.

My use of the mill would be entirely for home hobby use. Home repairs, interesting projects found in Home Shop Machinist, boat and motorcycle upgrades, etc. Eventually it would be nice to be able to mill an AR lower from an 80% forging or 0% aluminum forging, or even mill an AR lower directly from aluminum billet. I envision milling materials such as aluminum, steel, SS, titanium, and plastic.

Anyway, back to question in the title of this post. I see an incredible amount of effort on mill CNC conversions revolve around reducing the backlash of the existing stock leadscrews, with most people apparently going for ballscrew upgrades, with some people settling for ACME screw upgrades. I tried searching for an answer to this, but didn't have much success:

1. Has anyone had very much success doing a mill CNC conversion using the stock leadscrews on the 3 chinese mills mentioned above?

2. Also, I'm confused about all of the attention focused on backlash reduction. Has anyone had any success using table positioning (from DRO encoders or something else?) to enable tighter tolerances from the stock leadscrews?

3. How well does the backlash adjustment feature in Mach3 work with stock leadscrews? ACME screws? Ballscrews?

Thanks for any comments.

[LongRat]

3 very different machines you list there, with different approaches required to CNC each. The small travels of the X2 and the fact that going would be slow in steel and Ti would limit your options based on what you want to make with it. If you have space for either of the other 2, I would be leaning towards them.
As for CNC, the smaller X2 with no vertical leadscrew is arguably the hardest one to convert, although kits are available. And it is bar far the most converted machine so there is a lot of information. To answer your questions in my opinion:

1. There are certainly a lot of X2 users running stock ACME screws, mainly as it is the cheapest machine and the trickiest to squeeze the common 16mm ballscrews into. It can yield very good results, if you bear in mind you may have to adjust out the backlash occaisionally as the nuts wear. Probably less people using stock screws on the bigger machines. The price of ballscrews is a smaller fraction of the machine price in their case, and the performance to be gained potentially is even larger, so it makes sense.

2. You can get excellent accuracy with the stock screws and DROs, but that doesn't help when it is running under CNC. Mach can not take information from linear scales for positioning. Then you have to use software compensation, which assumes equal backlash throughout the travel range and can do undesirable things when you want to do 3D contouring work and are using CV mode in Mach. Plus of course the table is not rigidly constrained on the axis if there is backlash, even with compensation turned on, so this will compromise surface finish.

3. I guess point 2 answers this, personally I haven't had too much luck with backlash comp and things got much better with it turned off and with double nuts on the ballscrews to remove it mechanically at source.

[Crevice Reamer]

Hi Titaniumboy. Welcome to the Zone!

Firstly, I recommend that you watch the excellent MIT videos:

http://techtv.mit.edu/genres/24-how-...machine-shop-1

Secondly, here is some basic CNC info to peruse:

help with old parts

CR.

[Titaniumboy]

LongRat,

Thanks for the reply.

When you say that a lot of X2 users are using the stock ACME threads, does that only mean manual X2's?

What, specifically, are the results when using these stock ACME threads after a CNC conversion?

Yeah, there really is a lot of differences between the X2, X2, and RF45 clone. I'm about 80% settled on the X3, with the X2 and RF45 clone each having about a 10% chance. I really think the X2 is too small (although hossmachine sure seems to make one sing - I suspect this is much more a testament to his machining skills than to the capabilities of the X2), and the RF45 clone just seems really large and heavy to put in a crowded garage.

It's a shame that Mach3 can't accept inputs from linear scales. This table position feedback, along with the proper software, should be able to work wonders compensating for things such as backlash and leadscrew wear.


Crevice Reamer,

Thanks for the links. I'm in the middle of viewing the excellent MIT videos. Also thanks for the link to your "CNC for Dummies" post. I needed that! Some really excellent information.

[Crevice Reamer]

Here's a test video of X2 with stock screw and G540 / 387 motor doing 180 IPM:

File failed to upload? Invalid file?

This calls for further investigation.

CR.

[HimyKabibble]

Originally Posted by Titaniumboy LongRat, Thanks for the reply. When you say that a lot of X2 users are using the stock ACME threads, does that only mean manual X2's? What, specifically, are the results when using these stock ACME threads after a CNC conversion? Yeah, there really is a lot of differences between the X2, X2, and RF45 clone. I'm about 80% settled on the X3, with the X2 and RF45 clone each having about a 10% chance. I really think the X2 is too small (although hossmachine sure seems to make one sing - I suspect this is much more a testament to his machining skills than to the capabilities of the X2), and the RF45 clone just seems really large and heavy to put in a crowded garage. It's a shame that Mach3 can't accept inputs from linear scales. This table position feedback, along with the proper software, should be able to work wonders compensating for things such as backlash and leadscrew wear. Crevice Reamer, Thanks for the links. I'm in the middle of viewing the excellent MIT videos. Also thanks for the link to your "CNC for Dummies" post. I needed that! Some really excellent information.

"It's a shame that Mach3 can't accept inputs from linear scales. This table position feedback, along with the proper software, should be able to work wonders compensating for things such as backlash and leadscrew wear." - That is absolutely wrong! Software CANNOT "fix" backlash. The machining forces themselves will move the tool around within the limits of the backlash, and it is *impossible* for the software to correct for it completely, with or without positional feedback. All the software can do is mitigate the effects under certain circumstances. If you want real accuracy and consistency, you MUST have a mechanically "tight" machine. There is no way around this.

Regards,
Ray L.

[LongRat]

Yeah, that's the problem. Ultimately the only way to get really great results under all conditions is to get rid of backlash mechanically.
There are several people running CNC X2s with stock scrrews. I can't comment on the performance as I went straight to ballscrews, but there are posts on the zone.

[Titaniumboy]

.

[Andre' B]

Originally Posted by HimyKabibble "It's a shame that Mach3 can't accept inputs from linear scales. This table position feedback, along with the proper software, should be able to work wonders compensating for things such as backlash and leadscrew wear." - That is absolutely wrong! Software CANNOT "fix" backlash. The machining forces themselves will move the tool around within the limits of the backlash, and it is *impossible* for the software to correct for it completely, with or without positional feedback. All the software can do is mitigate the effects under certain circumstances. If you want real accuracy and consistency, you MUST have a mechanically "tight" machine. There is no way around this. Regards, Ray L.

Yep.

Even without machining forces.
Say the scales are reporting a position a 0.001 or 2 away from the target position.
And the backlash is all wound up in to opposite direction.
The drive puts some power to the motor to move it in the desired direction and the motor and screw start moving but because of the backlash the scale has not moved. So the control thinks there is enough load on the motor that it cannot move and so increases the power to the motor. This repeats until the system hits the other end of the backlash and the scale moves, problem is by this time it is going very fast and blows right by the target position. The control then starts putting power to the motor in the opposite direction but the same thing happens and you end can end up with some very violent oscillations.

Many sinker EDMs have position feedback from a linear scale but they also have separate velocity feedback direct from the motor or ball screw. That velocity info allows the controller to better deal with the backlash between the motor and the position scale. Even so when the ball screw wears to the point of having more then about 0.0003” slop you have to rebuild or replace the screw. And in a sinker EDM there is very little load on the servos, just the mass of the table and oil.


Edit:
The only way to make it work is to open up the permited position error and reduce the speed that the motor controls respond to errors this makes for a very sluggish system. You also need to program the CNC so that it basicly simulates the way a person turning the handles on a manual machine would do things, and that eliminates CAM generated programs and 3D surfacing stuff.