okay...what's your backlash?

[grizzlymog]

I'm trying to get an idea of what sort of accuracy shoptask/shopmaster users are getting in practice for leadscew and ballscrew setups. So what are you folks measuring for backlash in X,Y, and Z?

I just tightened the crossfeed/longfeed bearings down on my old smithy to as best as it gets and I'm getting .023" on the crossfeed and .036 on the longfeed for backlash. Ugh.

[neilw20]

About 0.0002" x and y, and head nod (an X3 sticky quirk) of about 0.0015.
Repeating in Z to 0.0002" with the column bolted to the brick wall,
after going below the position by the head nod amount and coming back up to make it settle. Programmed backlash of sorts?
Spindle expansion cold to hot is about 0.002"
It is easier in metric, but these are figures you might comprehend.

[grizzlymog]

Quite good Neil, thank you. I assume that you have ballscrews installed. Did you have to do much tweaking to achieve these figures? Metric is comprehensible to me. In the Canadian oilfield we intermix hectares and meters for maps with feet for footage calls in wells with acre-feet for volumes. Go figure...

Originally Posted by neilw20 About 0.0002" x and y, and head nod (an X3 sticky quirk) of about 0.0015. Repeating in Z to 0.0002" with the column bolted to the brick wall, after going below the position by the head nod amount and coming back up to make it settle. Programmed backlash of sorts? Spindle expansion cold to hot is about 0.002" It is easier in metric, but these are figures you might comprehend.

[James H.]

Backlash on these machines is tricky. I have an 09 tripower, it came with standard screws. (edited to add, Its actually an 08, but I did not receive it until 09)
Out of the crate backlash was nearly .040" on x and y. After lots of tuneing and tweaking it came down to around .025". The bronze antibacklash nut was pretty much useless.
If you adjusted the backlash out cold, the nut would heat up later and become too tight, result lost steps.
If you adjusted so it became just right after warm up it would be sloppy during the beginning of the program.
There was no happy medium since the backlash was essentially proportional and dynamic depending on what your program was doing in the moment.
So I ordered Ball screws and nuts from Mcmastercarr. Did not know where else beside shoptask to get them besides Mcmaster at the time. There was so much negative feed back about service I decided not to rely on shoptask for support or service.
Now I have ball screws installed and double nutted with a preload. Backlash drops to .015" on the x and .001" on the Y. Z was .007. That is all out of memory which does not always work right... .
Now for some more tweaks.
The Z is a single nut that has a natural backlash of .003", but total backlash was .007" the spindle plate arm that connects the screw to the spindle flexed .004". an even longer arm comes off the back to connect the DRO. When the spindle is on that arm shakes like a diving board and the DRO is only good within .01" because of the hi speed flexing of the vibrating arm.
Solutions. Mill new Spindle bottom plate and support arms. 1.75 thick aluminum plate. Arm flex down to nearly nothing. Total measurable backlash .003" which comes from the single nut setup.
Y Axis. The screw is aligned centrally and the support blocks are fairly good and solid. With the double ball nut total backlash .001" Which comes mostly from the thrust bearings and support plate.

Now the really ugly X. Pretty much all of the .015" backlash here comes from machine flex. Even with no load on the table and the gibs adjusted not too tight there are flex points in the system as follows.

The support plate for the X axis thrust bearings flexes .004" which you can literally see with your naked. eye. Its not loose it simply is not thick enough and has not support on one side. Solution fabricate a heavy duty three point brace that bolts to the table cross member, the Way casting, and finally the flexing plate. First brace I made went from the plate to the casting, but it was not stiff enough.
So there is still .011" backlash left to deal with. Poking around with my dial gauge I found the table actually had negligible rocking on the ways that was surprisingly true and good. After much head scratching I found out the flex was coming from a twisting action. between the screw nuts and the various pieces that connect them to the table. The screw is so far removed from the table that if you straightened out the maze of metal that connects them the screw would be over a foot away from the nearest x way.
All that metal flexes quite a bit and this reads and backlash. Solution for me, bypass three separate brackets with one extremely heavy duty one that bolts the bottom of the table that happens to have a nice flat bottom.
This bracket ended up being 1.5" thick steel block that rests inside the screw instead going long on the outside and reaching back up to the screw.
The stiffer arrangement dropped the backlash another .008".
I have not been able to get it any better despite a few other dead end attempts as stiffening up the machine.
final results after many weeks of messing with it. Z.003" Y.001" X.003" The Z could be further tweeked with a preloaded ball nut. I could probably get it down to .001" also.
I would never buy another shop task machine, but I cut my teeth on it and learned a great deal by addressing so many issues. I had thought though the project was what I would be cutting. Turns out the project I work on the most is the machine itself.
The mechanical issues go far beyond Backlash. Bushings, bearing, gears, motors, Capacitors. Everything has either had to be remade, replaced, re tuned, or repaired.

That is my experience so far.

James

[neilw20]

Now the wall shakes a bit. When the cement turns to dust, I will worry further.
Added bonus is the light fitting.

[instructor37]

Originally Posted by grizzlymog I'm trying to get an idea of what sort of accuracy shoptask/shopmaster users are getting in practice for leadscew and ballscrew setups. So what are you folks measuring for backlash in X,Y, and Z? I just tightened the crossfeed/longfeed bearings down on my old smithy to as best as it gets and I'm getting .023" on the crossfeed and .036 on the longfeed for backlash. Ugh.

I think the more important question is what are you looking for in a machine. The numbers NeilW posts are very good. If you read his other posts, you will see he is very knowledgable about machines and electronics, so he is going to be on the top end of the spectrum. Others will be on the bottom end. To find a machine with Neil's spec out of the box would be in the stratosphere price wise. I think with either acme screws or the standard rolled thread ball screws Shopmaster sells, you will be in the 0.004-0.006" backlash range. With a software that has backlash compensation, these figures will give you some good results- probably in the 0.001" repeatablility range.

[grizzlymog]

Thanks guys for your thoughtful and useful comments. It sounds as though users have different experiences with the initial quality of the machines. Whether or not backlash is consistent is a big deal! I'm trying to compare Shoptask to other 3-1 machines to try and calibrate a purchase/no-purchase decision. Space is an option, otherwise I would likely get a big hunk of American iron in here.

I've owned a smithy 1720 for about 10 years and use it only occasionally so it's not been too bad except that the mill is near useless. This is a model that Smithy discontinued years ago, but is still sold on Harbor Freight. Even Smithy has their standards! Say what you want about them, but the support is excellent. I ordered some parts last week and they were shipped out the same day. Anyway, the machine is what it is and I've experienced the quality of the Chinese machines. A new smithy doesn't offer me much more than this old one, except for a slightly more functional mill head I suppose and easier gear changes.

I recently began thinking about a CNC conversion of the smithy, and that led me to CNCZONE. Light machines mean light cuts so it takes much longer to get stuff done. Normally this is not a big deal for me, but I figured CNC would be ideal for the repetitive, mind-numbing tasks such as threading.

After doing a lot of reading and research, I figure I could spend ~ $1500 or less on a quality 2-axis conversion of the smithy and have a functional CNC lathe. Then I came across the info for the new patriot VFD and liked the features is has. I could throw hundreds more at the smithy conversion to implement a VFD drive for the spindle...but still I'm left with a useless mill head.

At this point I'm trying to decide on spending a little $$ on the devil I know for a basic CNC upgrade, or spend much more on a new wildcard with potentially much more functionality so I appreciate all your comments. Thanks!

[smallblock]

Originally Posted by grizzlymog A new smithy doesn't offer me much more than this old one, except for a slightly more functional mill head I suppose and easier gear changes. After doing a lot of reading and research, I figure I could spend ~ $1500 or less on a quality 2-axis conversion of the smithy and have a functional CNC lathe. Then I came across the info for the new patriot VFD and liked the features is has. I could throw hundreds more at the smithy conversion to implement a VFD drive for the spindle...but still I'm left with a useless mill head. Thanks!

If you CNC'd the Smithy and just tossed the mill head away, you would have a lathe about the same size as the Shoptask. Then you could buy a small mill, but then you would need another CNC system and your space requirements would go way up. Maybe you could sell the Smithy on Craigslist and start fresh.

[grizzlymog]

I've decided to CNC the smithy. I'll let you know how(if?) it turns out. I'm going to start slow and use the existing lead screws and see how that goes. Now that I have the machine completely apart, I've found several areas of slop that I've been able to adjust considerably that will help with the backlash. If it's still too sloppy, then I'll just go with some anti-backlash ballscrews.

I may try some mods to make the the mill head more stable like a bridgemill, but won't throw a lot of $$ at it. I will still have $$ left over to buy a small X* mill if I need/want to. Unfortunately, I don't have a masonry wall to bolt it too like Neil!

Cheers - Paul

[neilw20]

You could bolt it to a dead car on blocks with a 6 foot, 2" pole.

[grizzlymog]

...or perhaps to my mother-in-law!

[Mkroe]

So on the Patriot/ Eldorado Mill. I have done extensive changes to make it CNC worthy. As someone else said in an earlier post. I have spent most of the time making parts, brackets and things to get this going. My backlash is now approx 0.007, per axis. This is due to the pulleys and belts. I only have 0.001 in the ball screws that I used. Unfortunately, using backlash compensation in MAch3 is somewhat useless since this puts M# in Exact Stop mode when compensation is on. similar to Exact stop settings for corners when in CV mode. The problem is stopping and starting the mass in direction changes while cutting a heavy(weight wise) load, the steppers will stall or lose steps. Now don't get me wrong in saying that with care and very slow feed rates and shallow passes these problems can for most part be avoided. Unfortunately for me What I need to make would take Days to do. Still takes 8 to 10 hours to cut the plates I need from steel. at 6 IPM and 0.050 passes with 0.100 step overs just for the roughing. The finish passes I can run 10 to 12 IPM, but again I'm, only taking about 0.015 passes with 0.010 step overs. Must change endmills at least once during the roughing as they are usually worn out by midpoint of the process. If nothing else must re zero Z to account for wear.

I am making clamshell type molds for pouring Aluminum Cast parts. Getting approx 1000 pours to a set vs 4 to 5 pours to a sand cast mold. These are for a friends business.

Well if you have any questions let me know.

MK