Raster in x losing position

[bleeth]

I just got done cutting a lithophane file at 1.5 ips and decided to recut the same piece a tad deeper and so figured I would have no trouble cutting faster since there was so little being cut on this pass. I stepped it up to 2.5 and as the file was progressing with direction reversal every 5" x zero started shifting until at the end of the file it was over a 1/2" off. When it was over I checked for square and the y is still perfectly square. Obviously the force of the movement when reversing direction was "hammering" my positioning. I'm thinking that since there was so little opposing force (none really) from the object being cut it was caused by inertia at the end of the cut path and for whatever reason it was stronger when moving down (towards 0) then when away. Any ideas on how to prevent this?

Dave

[billp]

Dave,
I'm not sure what this "direction reversal" is. Please expalin...

[richards]

Dave,

Is the X-axis of your Shopbot on a steep slope where the X-axis is strongly affected by gravity? It doesn't seem possible that you would have creep towards one end of the machine unless there is some other factor involved.

I would write a simple sbp file that exercises the X-axis a hundred times or so similar to the following:

SA
JZ, 0.15 ' you don't have to actually cut

&COUNT = 50

START:
MX, 5.0
MX, 0.0
&COUNT = &COUNT - 1
IF &COUNT > 0 THEN GOTO START
'
JZ, 1.0
END

As the cutter moves back and forth, watch to see if it starts/stops at the same point each time it goes through the loop.

[bleeth]

Bill: The file was 5" along the x and 7" along the y(a typical 5 x 7 litho) cutting as a raster back and forth along the x and stepping up along the y and so the carriage was "reversing direction" at x,0 and x,5. By the time it worked it's way from y,0 to y,7 the damned thing turned my rectangle shaped file into a rough parallelogram tilting to the left over a 1/2 of ann inch!!

Mike: My bot is dead level. I'm anal about this. (Old boat builder you know.) When I square my y-axis I do it to 1/128 over 4 feet. (If I'm cutting panels I like them square.) I'll have a chance Sunday to run some more tests but between now and then all suggestions are welcome.
The change was very gradual and slight since the stepover was so small. It makes me wonder if the cg of the carriage with the motor on the left and the spindle cg slightly left of center it means that there is a very tiny moments of force factor that only comes into play with that kind of repetitive motion.
Dave

[rookie432]

Dave,
I have this exact same problem when cutting at too high a speed over a very short distance back and forth. I discovered this when doing some engravings on acrylic rowmark material. The fast cut back and forth on a raster in x over a short distance seems to be too much for the steppers to handle and postion gets lost. I assumed that this was the case and just slowed down my speed on short cut paths back and forth.
I would be very interested if the shopbot shouldnt be doing this and if it is something that is correctable but as of now I would believe that the new steppers on the alpha would be the proper solution.

[richards]

Dave,

I was joking about the slope; however, Bill J. might have a point about the Alpha. Since the speeds are higher, ramping plays a big part. In fact, I'm fairly certain that my alpha would never achieve top speed in so short a distance. He's probably right in suggesting that you slow down a bit. Alternatively, you might extend the ramp; but, that might negate the higher speed setting anyway.

[cnc_works]

Just a confirmation of sorts. Can't really contribute anything to a solution, but I was cutting a lithophane of similar size and my x shifted towards 0 by a few thou after the first inch, went another half inch, then shifted a few thou more and continued shifting just enough that I could see a staggered left border.

It might be something to do with ramping because it seems to me that I cut a larger lithophane, like 10" X 12", and didn't experience the same.

PRT96, PC router.

Donn

[Brady Watson]

Yes...I would most definately say that your loss of steps is a result of 'hard ramping'. Change your 'Minimum Distance to Check' down to .01 or .005 and adjust your 3D ramping as well.

-Brady

[bleeth]

My minimum Distance is currently set at .01 and 3d ramping is set at 100. The xy ramp speed is .70. What would you suggest for 3d ramping and what does that value refer to?
Thanks for all the input-I had a feeling I was just going to fast for the job-I have a picture in my mind of our old dog trying to come to a halt while running full tilt on a terrazzo floor!
The situation in general leads to questioning whether this is only prevelant in short runs or simply more exagerated in short runs? If I were doing a kitchen job with the typical 40-80 panels being cut out at a fairly aggressive speed with a compression bit would I be shifting x,0 there as well?
It's actually the first time I've tried to use that sort of speed for a 3d file. Your comment, Donn, about the larger size piece not having the same problem is encouraging-What speed did you run that one at?

Dave

[gerald_d]

For those mentioning Alpha's.....isn't the Alpha supposed to be immune to "creep"? This is one of its main selling points?

[billp]

Dave,
In another thread a few weeks ago I posted pix of two lithophanes I did of cats. they were both cut at 3IPS (after I had "cleared"out the first few lines to minimize the danger of bit breakage). I just went back and checked both pieces and I do not see any "creep" (if that's the correct term...).
I NEVER screw around with the ramping numbers ,(since like many of us I have NO idea HOW to...), so all of my settings would be the defaults for version 3.4.1
How did you hold your piece down? I used vacuum on both of mine...Any chance the piece itself moved instead of the toolpath?

[richards]

Gerald,
The problem, as I see it, has more to do with starting and stopping a stepper than it has to do with any particular type of machine. Because the alphas step faster, they depend on proper ramping to get to that speed. For a non-CNC example, no one would expect an automobile to accelerate from 0 to 60 mph instantly; rather, one would realize that the automobile starts at 0 mph and gradually increases speed until it reaches 60 mph. That 'ramp' requires both time and distance. Following that same thought process, no one would expect an automobile to instantly stop - that would be called a crash. The same basic situation occurs with stepper motors. They usually need to start slowly and gradually increase speed just like an automobile. Trying to get to speed too quickly is just like 'burning rubber' in an automobile, the tires are spinning but the car is not moving. In the stepper world, trying to start too fast usually results in missed steps. Just like an automobile, a stepper motor cannot stop instantly. If we try to stop an automobile too quickly, we usually 'skid', indicating that the wheels are not turning but the car is still moving. In the case of a stepper motor, the motor is being forced to turn by inertia even though the 'step pulses' from the controller have stopped.

The solution for automobiles and for stepper motors is to start slowly, ramp up to speed, travel the desired distance and then ramp down from speed.

[bleeth]

Bill: Thats what I thought (Your cutting speed) which is why I tried the higher speed. The piece was held down by four screwed down plywood clamps I made for the purpose and no way did it budge. But you did upgrade to the alpha and I'm using the PRT steppers and that may be where the difference lies. Also, did you cut that cute kitten with the longer dimension in the x direction? Did you cut lithos with your PRT at that speed?
My ramp numbers (above) are also what was already in version 3.41.

[billp]

Dave,
Yes the X axis is longer than the Y on the kitty litho. The fastest I remember doing them with the PRT was about 2", again after clearing out the first few lines. Possibly I also had better luck because my X length was a little longer? (8.5").

[gerald_d]

I'm with Bill - we never change the default ramp values either. (We are still in DOS)

[gerald_d]

Mike, the Alpha has position encoders, thus missed steps are supposed to be detected and compensated for.

[richards]

Gerald,
The encoders work within normal operating parameters. If I forget to program in an SO,1,1 command to start the router, and then start the program, you would think that the alpha would drive the z-axis to the table-top (0.0) and then stall (assuming no broken cutter). In any case, as soon as the z-axis stopped moving, the z-axis should read 0.0. Usually it is something much different than that (-0.25 to -0.65), indicating that, besides having egg on my face for not programming things properly, the parameters of the encoders were exceeded. The first time that happened (you would think that I'd learn after the first time), I removed the cutter and collet, thinking that the cutter had been pushed further into the collet, but the top side of the cutter was still flush with the top side of the collet, just as I had originally inserted it.

Just how accurate the encoder feedback is under 'almost' normal operating conditions is something that I've wondered about, but have no way of knowing with any certainty.

Also, so that no one thinks that I'm slamming the alpha or its encoders, every time that there has been a variance between what I thought should have happened and what actually happened has been the result of programming on my part - except a slight amount of jitter or roughness that sometimes occurs in a multi-axis move, usually during the ramp portion of the move. That jitter or roughness is much less than the 0.015 tolerance specified for the machine (showing that Shopbot has built a product better than they advertise it to be.)

[bleeth]

From what everyone says here and what I have since discovered in the last few minutes I would say that torquing of the carriage and lost steps from the force of the stops with insufficient ramping time at that speed is definately the problem.
The carriage is now out of square.
All parts cut previous to this part are perfect, including the recessed cutting area for the litho I pocketcleared in this part which started life as a piece of 1/2" corian. Since the adventure in question I just noticed a shifted x,0 after running c3 . Sure enough I'm now losing the rest of today to squaring the axis again.
Moral: Alpha may be able to handle 3ips in this type of cutting but PRT cannot.
Note to my friend in SA: I'll keep my spindle and wait with bated breath till I can justify Alpha upgrade. Tim Allen is right--"More Power"

ttfn

Dave

[bleeth]

OK_Square again but this time spent the time thinking about squaring jigs-Gerald's method of stop blocks in another thread looks good as long as the left y rail is welded to the carriage rails. Anybody else got any great ideas. It seems to me that as long as the entire system is bolted together once it's out of square you have to start almost from the beginning and, frankly, that is way to slow a method.

Dave

[richards]

I use a simple method with my 120/60 alpha. Just before power off, I send the carriages to a known rest position (-18,5), which puts both the x and y axis just a few inches from their respective physical end points. Then, before turning the alpha on, I check both sides of the x-axis with a pre-set combination square and a feeler guage. Most days I have to give a very slight nudge to one side or the other to get things perfectly lined up. As soon as I'm satisfied, I turn on the power, push the reset button and then run the C3 command (thank goodness for proximity switches, which I just started using a few weeks ago). Unless I loose power, the alpha stays perfectly aligned for the rest of the day.

(When I first got the machine, I fought alignment everyday, almost to the point of cussing - and that should tell you something coming from a Mormon. It seemed like I could never get the y-carriage to line up and then stay aligned. With my combination square method, panels are cut square - less than 1/32th deviation corner to corner on a 24 x 31 panel.)

[fleinbach]

The simplest method for squaring the Y carriage o the X-rail's is by using the stop blocks as you indicated Mike. But there is a simpler method then using a feeler gauge and combination square on one of the blocks every time you start the machine.
Once you get the Y-car square using your method take the time to make your stop blocks equal distance from the X-motors pinion gear. This will make your stop blocks square to the X-rails. Once this is done it is a simple matter to square the Y-car very quickly and easily every time you restart the machine. With the control box off pull the Y-car against the stop blocks. Now turn on the control box while holding the Y-car against the squared stop blocks. Even if the Y-car has a slight rack in it this will compensate and it will be square.
This method works flawlessly and quickly every time.

This was described in much greater detail here (http://www.talkshopbot.com/forum/cgi-bin/discus/show.cgi?tpc=7&post=17043#POST17043) back in September

[bleeth]

Unless I'm missing something,having the carriages perfectly square to the x-axis is a great start but does not guarentee squareness of the whole car yet. The y rails can still be out of square due to the sideways play in the bolt holes. A more practical jig seems to be one that screws or clamps down to the table square to the x rail with perpindicular arms extending upward so you can pull the rail against it and tighten everything down then.
As long as the rails are square to each other whether an axis from one motor or carriage end to the other is square or not is irrelevant.

[fleinbach]

Dave,

The Y-car rails need to be square to the X rails. The thread I linked to above from back in September describes several methods to obtain this.

This statement is absolutely correct.
"As long as the rails are square to each other whether an axis from one motor or carriage end to the other is square or not is irrelevant."
Even if one X-motors pinion gear axis was misaligned with the other it would not matter. All that would be important is that they hold the y-axis square to the x-axis. Once squareness is established you want to be able to obtain it quickly and easily each time you start up the ShopBot.

1. Using whatever method you prefer align the Y-axis square to the X-axis.
2. Next move the X-rail stop blocks until they touch the pinion gears. It does not matter whether the pinion gears are aligned or not. The X-motors pinion gears will be in the exact same spot every time it is against the stop blocks. If everything was square when they were set it will be square again.

Now, every time you pull the Y-car against the stop blocks the X and Y axis will be square. It will not matter even if the Y-car is slightly racked. The X-motors will hold everything square when they are turned own.

Using the stop blocks for this purpose is very quick and very easy once they have been set up.

[bleeth]

Frank: When I square up the y axis now I use the "clamp one side and nudge the other" system while trcking the spindle back and forth along the y axis and that's exactly where the problem lies. It is inherently slow to get it dialed in really close, particularly since tightening up the bolts can then nudge it back out of square. It's akin to setting up the blades on a planer the opld fashioned way with a dial caliper. What I'm trying to come up with is a jig to align the actual rail that, in effect, will accomplish what a magna set does for the planer or jointer.
Around my shop I keep a stack of "squaring triangles" with the centers cut out to assist in box assembly. I'm picturing a pair or more of these carefully attached to a rectangular plate the width of the table. You fix the plate to your square reference line and clamp the y rail to the squaring frames. Now turn your motors on, tighten everything down, unclamp and away you go. For normal daily checking either system should work but only checking the y rail directly will guarentee square 100% of the time.

On the other hand the spot welds Gerald used on his y car, after he was sure the rails were square, makes the carriage stop block method guarenteed.

By the way, I really like your spacer block add on. I think I like UHMW as the material.
Dave

[gerald_d]

Dave, in place of the spot welds, you can drill a dowel hole through the bolted joint and tap in a roll pin (http://www.prop-shop.com/g/a/asc7665.jpg). One at each bolted corner (4 total). Don't worry about perfect square before you put the pins in, just make sure that the y-rails are parallel to each other and at the correct distance from each other. After the pins are in, you theoretically have a hinged parallelogram, and then you square up and set the stop-blocks.

What is this about "magna's" and dial gauges for setting thicknesser blades? I just use a piece of MDF.