My first post...great site and forum!

I am thinking of a 2-axis retrofit on my old 1J. I bought Chris Bruno's plans and think I am going to do something close to that. 1200 oz/in or so steppers and Geckos. Probably Mach2/3 for control. Get that cooking and then do the Z-axis...

The thing I am waffling on right now is whether or not to fit ballscrews. What kind of problems can I expect without them? What suggestions do y'all have for Ballscrew kits if I do decide to get them? Anybody make a kit for the 9"x42" BP Series 1? I have no problem fabricating the motor mounts and such, but for some reason, I would like the ballscrews to be as "bolt-on" as possible.

TIA,
Wally

I wouldn't even think of NOT fitting ball screws if you are going to do CNC.

Turn the handle on a zero backlash ball screw system and then do the same with an ACME threaded system with similar zero backlash. The friction difference is quite noticeable as will be the accuracy.

Generally speaking, lead screws do not wear evenly. Thus, keeping backlash comp under control is all but impossible with worn screws, regardless of the efforts/compromises made to "adjust" for the wear.

There are several bolt-in ball screw setups out there. HIWIN makes them in rolled and ground versions. Rolled is plenty good enugh for hobby use. Ground screw is the definite choice for precise milling, especially if you plan to do 3D profile milling of intricate details.

I seem to recall that the rolled versions are available for around $600-$700/set. A billet yoke might be extra and will add to rigidity and accuracy while doing heavy cutting.

The HIWIN and other ball screw retrofit sources can be linked off of the ajaxcnc.com website from their components page (I think).

Turn the handle on a zero backlash ball screw system and then do the same with an ACME threaded system with similar zero backlash. The friction difference is quite noticeable as will be the accuracy.


You can get acme that's more accurate than ballscrews, so that's not necessarily true. Either come in many different grades of accuracy.

Gerry: I'd expect that to match the 0.0005" or 0.0003" published lead error of HIWIN ground and rolled (respectively), one would have to use a ground acme thread. How costly these are is something I can not discuss intelligently.

Yes, accuracy could be there if the Acme were ground instead of simpy cut or machined thread. But I"d still be inclined to think that friction of an acme thread/nut would be higher than a ball screw, especially if nut preloading were done so as to eliminate backlash.

We just did a tune-up of our rolled thread ball screw in our BPT EZTRAK. With the screw loaded with o'size balls for true zero lash (slight ball preload) and the incorporation of a 480lb preload, true ABEC 7 ball screw support bearings, our turing friction is now as low if not LOWER than the OEM setup and we're seeing no perceptible backlash.

Granted, we didn't have an acme to compare against initially but rolling friction is typically lower than sliding friction. Ultimately, friction (and resultant wear) is a factor with CNC's, especialy as you try to run at high inches/minute.

Wally:

Two cents from a relative Noob with an operational conversion.

I have not actually tried acme screws since I replaced mine. But, I hear that several people use them with CNC and do just fine. You will probably have to fabricate split zero backlash nuts. Does that make sense?

The big question is how accurate do you need your work to really be? I'm of the opinion that a whole lot of stuff can be +/- .005 - .008. I think you can get that with an acme screw and a good mapping exercise with your software. Remember the software can compensate for a lot of this too. Including the backlash. Some creative programming might be able to compensate for the the rest.

On the other hand, this could get old real fast........

Here's some of what you are trying to avoid:

-- Biting into the work and jamming the tool when directions are reversed under CNC control. Tools will get broken, work gets pulled out of clamps, etc. This is probably most pronounced when transferring to climb milling. This will also result in surface irregularities and just plain ugly work.

-- Accuracy issues as you can imagine.

-- Chatter as the work can bounce off the tool as allowed by any slop in the screw. This can be resolved with a different feed rate of course.

-- Irregular drive motor performance / stalling as you move onto less worn parts of the screw.

But these are all kinda worst case scenarios. There are people who have made big router tables using threaded rod and home brew anti back lash nuts.

Yes, I think you can get a kit for your machine. I got mine from Rockford Ball Screw. Rolled, not ground. I have seen better yokes, but this one works fine.

I have heard good things about those HIWIN screws as well.

Hope that helps.

W

Yes, I think you can get a kit for your machine. I got mine from Rockford Ball Screw. Rolled, not ground. I have seen better yokes, but this one works fine.

I have heard good things about those HIWIN screws as well.

Hope that helps.

Where did you buy your Rockford Screws from and how much were they? Do you have a picture comparing the OEM Bridgeport yoke to the Rockford yoke? What are better yokes (elrod)?

Gerry: I'd expect that to match the 0.0005" or 0.0003" published lead error of HIWIN ground and rolled (respectively), one would have to use a ground acme thread. How costly these are is something I can not discuss intelligently.


But you went onto recommend rolled ballscrews, didn't you?

I bought my Rockfords directly from Rockford. I think you can probably special order them from MSC or McMaster too.

$350 for the X assembly
$225 for the Y assembly
$75 for the yoke

From the looks of it the Elrod yoke is beefier. Honestly, I don't think mine is giving me any problems. I think it could if I was hogging out steel with a big roughing end mill.

Note that the Rockford yoke is for the Rockford screws. You are not going to be able to use the Elrod yoke with the Rockford screws from what I know. Different bolt paterns on the nuts.

I have pictures of the new yoke but they are on film. When I get a chance to scan them I'll upload. But the Rockford and stock yokes are about the same. I can take a picture of the stock yoke too. It's in the garage on a shelf.

I have a journal of the conversion so far at:

http://www.cnczone.com/forums/showthread.php?t=8749

GER21 posted: "But you went onto recommend rolled ballscrews, didn't you?"

Not really and I appologize if I gave that imression.

My machine HAS rolled screws which is what it came with. Since we didn't have the $$$ to upgrade, we merely tuned it up with oversize balls in the screw and some horribly expensive ball screw bearings.

We cheat with lead error. How so??? We "cheated up" the lead and backlash comp for deadly accuracy in a 3", +/-0.5" Radius where we want to do real precise profile milling - response is hugely distorted beyond that range.

When we want to do work over larger span, we reload a more proper lead and backlash comp table. Real PITA but thats all we can do with the budget we have at this time. You can do it rather easily as it is a DOS based system and we know our way around that system fairly well.

When time and $$$ permit, I plan to upgrade from EZTAK (pretty much obsolete but it works OK) to a more modern system (considering several) that has modern servo feedback, autotuning and higher resolution.

By doing a laser scan of the table and upgrading to a system that will allow laser comp by section, we can "map" the lead of our rolled screw and let the computer take care of any lead error - this will let us keep the rolled thread screw for a bit of a cost savings. Isn't modern technology great???

The alternative is to go to a ground screw which I did consider and would prefer. Problem is, the EZTRAK uses a special end that isn't compatible with any off the shelf ground ball screw kits I found. I just don't want to throw out my drives, screws et al.

HIWIN's are available off the shelf in ground or rolled thread but they bolt in to manual machines and mine differs substantially due to it being an EZTRAK.

HIWIN screws are in mine now. They work great. Only reason to change is that we're trying to turn the BPT into a Haas for a fraction of the cost.

Reason: to machine some master profiles that we use to grind cams with in-house. We're within 0.0001 of a CNC ground part now which you can't see in the finish ground part.

These are the DIY things that a semi-retired engineer does when one has the know-how and access to bearing/ball screw rework technologies. Besides, it is really cool to be able to have someone say "you can't do that" and then see them utter in amazement, "how the %#@@ did you do that with a Bridgeport???" after you do....

Life don't get much better when something like that happens.

Well....

I'm two parts into this now and I can say confidently that the rolled screws are fine. I am not trying to make a Haas out of my 1982 2J either. But it is well within the +/- .003 range or better. That is good enough for a bearing journal so that's good enough for what I see that I might be doing.

So...I am going to come right out and say that rolled screws are a good thing for the average used piece of bridgeport iron and the average tolerances that are really needed. In my humble noob opinion of course.

-jd

Here is a scan of a ROCKford sales flyer for S1STD Ballscrew conversion.
Yes, ground are a lot more accurate than rolled but at a price!

George

Nc Cams,
How does one map the lead error of the ballscrew? Is there an alternative to a laser inferometer? How can large distances be mapped without accumulating errors in the measurment?

You can map the table by using a set of JOB blocks as the standard - say 4" long.

Find the lead error at one end and then move the standard down the travel 4" at a time, rezeroing of the prior endpoint.

Thus, measure from 0.000 to 4.000 find error. THen, move the standard so the end that was at 0.000 is now at 4" and then see the lead error at 8" and so on down the table.

Time consuming but 4" is 4" and if you do it carefully, you should get lead mapped to within a few tenths or so all the way down IF YOU ARE CAREFUL AND DILIGENT. When it is all said and done, have somebody come in with a laser and have it done. Cheaper and faster in the long run and it is traceable as a standard if you get someone who is properly qualified.

You can also make a stepped gage and have it CMM'd. Faster to do but still involves $$$'s.

NOW, once you have the date, what do you do with it??? Unless you have the ability to input a lead profile into your software and for it to fit a curve to it, you're probably wasting time.

nc cams,
thanks for the info. I figured it was something as simple as that. What about using a higher class ballscrew to map it, ie have a stage with a servo. move (via computer) the stage a certain increment, then move (via computer) the other ballscrew the same increment, and use a dial indicator to measure the difference. At least the mapping measurements couldnt be worse than the better ballscrew. backlash should be taken out if moved in the same direction. I imagine you could automate this with a digital indicator. Or I guess you could skip that and just use glass scales to do this as well.

Jimbo: you're mixing things up a bit which can get you confused

Backlash is lost motion, period.

Lead error is deviation from desired true position assuming you have NO lost motion.

Yes, if you go in 1 direction backlash SHOULDN"T be an issue - but it is.

Why? the tool is going to shove back against the cutter and size control is diminished as is finish, accuracy, etc. The tool moves as if the lead at any specific point were off.

Thus, you do need to eliminate backlash if you want high accuracy - this is where ball screw and ball screw bearing preload comes into play (no pun intended).

Once you eliminate backlash, you only have to deal with (localized) lead error.

Thus. if you have no backlash, and tell the screw to move the table 4.000" and you have a net lead error of +0.0005", you'll move 4.0005 when you think you moved 4.000.

Now, if you stack backlash onto that (say 0.0005), you now have to subtract the backlash from net movement because you expected to move 4.000 but you lost 0.0005 so you actually move 3.9995.

Yes you can comp the daylights out of B/L or lead error but, once you have a system that has both minimized, you get spoiled and you'll never be satisfied with one that is run of the mill... It is a dream to hit the numbers rather than to creep up on stuff and do the "oops, went too far".

nc cams,
yes, I agree backlash still needs to be accounted for even after the ballscrew has been mapped. The backlash I was referring to was using the two ballscrews side be side during a mapping process, maybe with some tension against the ballnuts and always moving the same direction should eliminate (most) backlash from the mapping numbers.

My application was more for improving the lead error of rolled ballscrews over a large distance for a cnc router. With these cheaper ballscrews it is difficult to remove all the backlash or it will become difficult to turn. I do agree that backlash needs to be taken out of the equation or the mapping numbers will not be consistent.

thanks for the very useful info.

May have posted this before - please excuse the repost if it occurs.

First some basics:

Rolling friction is always less than sliding friction

Break away torque is always higher than torque needed to maintain motion in rolling environment.

Higher acceleration generates more force/resistance than and vice versa.

Gib clearance will create lost motion - tite gibs will generate huge back lash problems as screw needs to input high force and stretch until it finally moves stuff....

Hence, if you have a properly preloaded ball screw and a properly preloaded ball screw bearing and your gibs are properly adjusted, your friction will not be appreciably higher.

When you start to cobble up ball bearings and/or tapered rollers (ergo use them improperly) you will see a friction issue.

We did a moderate preload on our rolled screws via o/s balls and added true ball screw bearings (read that expensive). Oem bearings had 70-100 lb preload, the true screw bearings we replace them with had 480lb.

Result: our break away torque is nearly the same or lower than stock bearings but the continuous torque is LOWER (much smoother bearing due to internal load path of A/C versus deep groove ball bearings it replaced). Lead error is 0.001 worst case at 4 positions we mapped it at when we had machine tuned. WE can CLOSELY approach the accuracy of a ground part when we slowly and carefully cut a master profile with our cnc mill.

Backlash is miniscule (something like 0.0002 comp max on one axis and 0.0001 on the other). We see 0.0001 or so "flat" if we look hard via ultra precise dial indicator at direction change.

Lead error is 0.001 or less over 4 inch segment checked at 4 positions along the X axis. Almost but not quite as good as the neighbor's VMC but a fraction of the cost.....

Our mods were pretty much bolt-ins asside from the ball screw preloading. It wasn't cheap but they served our needs quite well.