question for machintek, others welcome

[Sparky_NY]

I want to confrim what normal behavior is on my V2E3 bridgeport. I have retrofitted drives and mach control. Encoders are 2048 and X2 in the drive, not at X4.

Using a tenth indicator, if I move up to a point from a distance of say 50 thou or more away, it stops within a tenth or two of the commanded position.

If I do tiny steps, using the mdi or my pendant, I get what appears to be "sticktion" With steps set for a 1/2 thou, one step may move 2 tenths, the next step 8 tenths, the next 1 tenth, the next 7 or 8 and so on.

This appears to be sticktion, which a tiny move of a 1/2 thou does not overcome, just "tensions" things. The next step "breaks" things free. I tried the same with steps set to one thou and it behaves similar, always within 3-4 tenths of target.

I am finding positioning and backlash within a couple tenths max error, usually a tenth provided the move is a "few" thousanths (5 or more) up to any distance and at any speed.

I have verified the encoder error is at zero during each 1/2 thou step and watched the ballscrew increment each time while watching the tenth indicator. I found some of the famous 20tac47 ballscrew support bearings for a super deal and have them installed. I have also experimented with gib adjustment but it does not have a effect unless they are overtightened.

The behavior is the same on X and Y.

Moves this small would not be made during machining so would only cause a potential problem when trying to locate a edge with a edge finder.

Is this normal behavior? I am pretty sure its just normal sticktion, but want to comfirm.

[machintek]

Please remember that this is a S1STD with a CNC retrofit done by Bridgeport. It does have a belt drive and dovetail ways. What shape are the ways? How tight are the gibs? What way oil are you using?
My way of setting backlash: Lock the knee.
Increase the backlash by .003 inches. Cut a inside circle with a light cut. Now run a dial indicator around the inside. You can measure each step and know exactly how much you have to take out (reduce) the backlash of each axis. If you do not over comp the backlash it is difficult to measure a flat. Also dynamic backlash is different from static backlash.
By the way, the older box way Bridgeort VMC had stiction to the tune of .001. I would increa. se the oiling for the first year to minimize/overcome this. Once well broken in the oiling could be dailed back.

George

[Sparky_NY]

I realize the V2E3 is basically a standard series one machine, retrofitted by the factory. They did use the good stuff though, NSK C3 grade ballscrews. I am using standard Vacra#2 way oil. The gibs are set to the .0005 factory spec, maybe a little looser (.001 max) to help sticktion. The ways are chromed and like new. I have upgraded to those $800 a pair 20TAC47 preloaded ballscrew support bearings. (they were on ebay recently for $105 a pair!). I have been keeping the machine over oiled for now because it sat unused for a few years. All oil lines seem to be functioning well.

I am not really concerned about backlash, it appears to be around 2 tenths which is fantastic for a used machine of this type. That is measuring backlash with moves of 50 thou or in that neighborhood to eliminate the effects of stiction, using jo-blocks and a tenth indicator. I am not compensating for backlash at all at this point.

I am presently getting about a half thou to 8 tenths stiction. I am reasonably positive this is stiction. It appears only during tiny moves (under 10 thou). Any other move is within a couple tenths of target location. It is these tiny step moves in the same direction that I was wondering about. Lets say stepping in a half thou at a time accurately.

So, by your answer, I am assuming a thou, give or take is not unusual for sticktion which prohibits making accurate tiny step moves. (tenths to a thou)

I am happy with the performance but was very curious as to what normal was. Your years of experience gives that sense of "normal" that I lack and that is not documented.

Is there a way oil that would make a noticeable difference in stiction?

Thanks Much!

[machintek]

Maybe some drive tuning can help. Can make it a bit tighter.
Do you have a scope?

George

[Sparky_NY]

Yes, I have a scope and the drives have software that can issue step commands and display the response on the computer screen the same as a scope display.

BUT, I don't believe drive tuning would do anything. I have removed the drive belt cover and carefully watched the actual rotation degrees at the ballscrew while issuing 1/'2 thou steps. The screw rotates the same amount each time but the actual movement on the indicator varies. Also, each movement goes to zero error on the encoder very fast.

Another "test" I did was to issue a 1/2 thou step, notice the indicator only moved 2 tenths, give the table a sharp "thump" with my fist and the indicator moves the remaining distance for the correct 1/2 thou step. You can observe the table tensioning on the first half thou step, it is a "soft" type movement, the next half thou step is crisp because the tension then exceeds the sticktion and the table "breaks" free. This is when I loosened the gibs some and over oiled to see the results, no change. A much lower friction way oil would probably help but I doubt such a thing exists. (vacra #2 now)

I am convinced it is sticktion. Previously reading your method of setting backlash and the difference between static and dynamic methods confirmed to me the existence of sticktion in the real world.

Just the other night, I cut a circular pocket for a ball bearing bore for a friend.
(just over a inch). I "snuck" up on the final dimension in several passes, measuring each time. I found that I could take a couple tenths cut accurately with no problem. BUT, that is dynamic, not static again. The sticktion is already broken once the table is in motion.

The real question is, what is to be expected? What is normal sticktion? The original boss 8 specs don't help much here, .0001 display increment, .001 accuracy over the saddle. Nothing in the original spec says that just because the display and commands can increment in .0001 steps that the table will also step that small accurately. (in fact, the .001 spec implys that it won't ! ) Its the old game of resolution vs. accuracy.

So, dynamically I can position within a couple tenths every time, anywhere. Tiny static steps will put me within a half thou. I am thinking this is all that can be expected of the hardware. I am not complaining, this is darned good. I just wondered what "normal" is for these machines.

[mactec54]

Hi Sparky_NY

When your machine is cutting a part you are not having a problem, so how warm is it were your machine is I found you need to keep them around 66/68deg so if you are not in this temperature range you are wasting your time, trying to test the movement of your machine, also I found if you run a warm up program for the X & Y axes this can help as well, this needs to run for about 20min, do this before testing & you should get no more than .0002 of sticktion, You have to think you are trying to move a 350lb table .0001 & unless it is floating on the oil & the temperature is just right it is not going to happen

[Sparky_NY]

Interesting thought, and makes a lot of sense.

My garage temp stays between 60 and 68 at all times so that should be good.
A warm up run to get things loosened up, the oil circulated etc makes a lot of sense, I will have to try that.

I have also given some thought to wiping the ways clean, then wiping them with a coat of mobil one, working it in a bit then testing. This would just be a experiment but a interesting one. Mobil one is quite efficient at lowering friction. The dial indicator would tell the story for sticktion.

Originally Posted by mactec54 Hi Sparky_NY When your machine is cutting a part you are not having a problem, so how warm is it were your machine is I found you need to keep them around 66/68deg so if you are not in this temperature range you are wasting your time, trying to test the movement of your machine, also I found if you run a warm up program for the X & Y axes this can help as well, this needs to run for about 20min, do this before testing & you should get no more than .0002 of sticktion, You have to think you are trying to move a 350lb table .0001 & unless it is floating on the oil & the temperature is just right it is not going to happen

[Sparky_NY]

I ran across a interesting spec the other night, for a V2Xt dx32 which appears to be pretty much identical hardware to the V2E3, with a more modern controller.

Input resolution .0001, Servo Resolution .0001, repeatability .0008, and positioning accuracy .001

From these specs it appears the factory recognizes the effects of stiction also!

I am within a couple tenths on everything except tiny steps of less than a thousanth. Guess I shouldn't expect any better.

Originally Posted by machintek Please remember that this is a S1STD with a CNC retrofit done by Bridgeport. It does have a belt drive and dovetail ways. What shape are the ways? How tight are the gibs? What way oil are you using? My way of setting backlash: Lock the knee. Increase the backlash by .003 inches. Cut a inside circle with a light cut. Now run a dial indicator around the inside. You can measure each step and know exactly how much you have to take out (reduce) the backlash of each axis. If you do not over comp the backlash it is difficult to measure a flat. Also dynamic backlash is different from static backlash. By the way, the older box way Bridgeort VMC had stiction to the tune of .001. I would increa. se the oiling for the first year to minimize/overcome this. Once well broken in the oiling could be dailed back. George