Programs don't keep their position

[eklug]

I work in a fairly large moldings shop. We supply many of the large RV companys. For RV fascia we need to carve the ends and sides out of wrapped mdf blocks to accept the vertical or horizontal carriers. We have 2 shopbots and the operators are fairly new at it. They are not very computer literate and have a difficult time setting up the programs. I create the actual programs but they need to position them every time they run. I used to run one of the shopbots but about a month ago they moved me into the office to be a full time programmer/ CAD drafter. When I used to run the blocks I would run the setup pieces,measure the distances I needed to adjust to get everything centered, etc. And in 10 minuntes I would be ready to run. I didn't know why from one day to the next the programs would need to be repositioned but it didnt take me long to fix so I didnt worry about it. Well now that Im the programmer I want it so that once the program is set up it stays that way.
I have some ideas about why the programs don't stay set up. Either operator error (bits incorrectly sized for example) but this can't be the reason every day and I was sure I was doing everything right when I ran the machine yet the setups needed re-done then also.
The limit switches moving slightly and throwing off the position when the sb is homed every morning. Occasionally I find the screws will loosen and a switch will be rotated out of place. But again I dont think this would be the cause every day.
The sb is just not heavy duty enough to run many reps for hours a day. The machines currently run about 57 hours a week. (double that when and if we get operators on 2nd shift). Some of our block orders are for 600+ blocks. We are very behind mostly due to the inexperience of our operators and the long set up times. The sbs are at capacity, actually above it, and if the operators didnt need to do the same setups day after day it would greatly help.

Does someone have any help on this matter? Why don't our programs stay postioned accuratly from one day to the next?

thanks
emily

p.s I can post a photo of how the sb is set up to run these corner blocks if it would help.

[richards]

Emily,
You need a true reference point on your table. It can be as simple as a dot from a pen, or a small hole drilled with a 1/8-inch bit, etc. It doesn't matter where the mark/hole is, for instance, I have a mark/hole on my table that is 1,1 from the limit switches. After I zero the machine, I give the M2, 1, 1 command to position an indicator (1/2-inch shaft turned to a V-point) over the mark/hole. Then I slowly lower the indicator until it just touches the mark/hole. There should be equal clearance all around the indicator. If everything matches up (and it usually does), I give a JZ, 11.55, 5.5 command, which moves the spindle to my "real" 0,0 position. Then I give a Z2 command to zero the X and Y axis. On the other hand, if the indicator is not exactly on at 1,1 I move it 0.025" at a time until it is where I want it, and then give the command to set that exact position to 1,1 (I think it is the VA command, but I'm at a computer customer's business right now where we only run Linux, so I can't load SB3 to check things out.)

With a reference point that is independent of the proximity switches, I can easily get to a know location, even if the switches or the targets have been bumped out of position.

[eklug]

Thanks Mike I will try that out. Even though it adds a few more steps to the morning routine it will hopefully save us time from not having to continually re-position the programs.

[ckurak]

Emily,

Try a little bit of clear silicone on the side of the proximity switches to hold them in position. This should keep them from moving.

However, Mike's suggestion is more accurate than the limit switches. My method (learned from another ShopBotter) will keep the switches in place so you can then perform Mike's method.

Also, I have heard about writing a program that will set the X & Y positions in a similar fashion to the Z-Zero plate program. Does anyone have any code and/or pictures of such a setup?

I hope you can get this solved quickly and effectively.

Charles

[jay_p]

Emily,

I have had employees who were trying to clean the machine bang the sensors with a vacuum nozzle. I ended up making acrylic shields to protect the sensors from well intentioned staff.

Jay

[Brady Watson]

A couple things...The prox switches that SB supplies also come with 8/32 thread screws & nuts. These are not extra parts meant to stay in the bag! - Prox target angle plates are meant to be thru bolted with this screws. That means drilling thru your rails to do it. This should eliminate knocking them out of whack.

Secondly, the prox switches can be very accurate IF you grind the raised '|' marks off of the heads of the target bolts. Get them nice and flat, then adjust them for gap. Test them over and over again to make sure that they are hitting the targets with good repeatability.

Adding to what Mike wrote, very few know that they can write their OWN custom cuts (IE just like C2 and C3)...You could make your own let's say C7 that moves the tool from the home position (0,0) that the C3 routine defined, over to your new desired 0,0 position that lines up with your purpose built jig(s) perfectly. I made 2 custom cut files for myself. One C7, moves the tool to 24,12, which is just right for my long arms when changing tools and a C8 that runs the tool to X 120, when I want to clear off the table of cut parts and lay down another sheet of material. Make the tool work for YOU! Who's the robot here anyway?


-B

[jay_p]

Brady,

The targets were bolted. It was the sensor itself with a bolt and with silicone that was getting whacked. When you have 20 year olds working for you anything is possible even when well intentioned.

Jay

[paco]

The machine with all it's components can be test for reliability if in doubt.

It's a little more trouble (I need to chuck in a blank bit) but I prefer a solid aluminum block to zero (X Y Y) my tool just the way I do it for my Z with the Z plate. So far my test reveal a 0.002" accuracy and repeatability.

I don't thrust my Bot when it come to accurate (and expensive) or long process when they aren't done from start to end. Because it can have problem with communication and because it's flexible (very light framed). I'm not saying that SB3 is absolutely not reliable but I'm just better than it is at making sure everything is right on!

These days I'm carving large piece of both Birch and MDF and at each block I calibrate the tool, note down the X & Y location so whatever happen, I can resume... within 0.002". Those carving are long to process (about half a day each) and the Birch version has about 16' of solid glued up wood (about 7" X 5" X 64"; at about 5.00$can/feet) in it. For those I prefer to home from the center and to zero (Z) from the spoilboard (always the same spot) to avoid any bad reference with some 1/4" to 1/2" of material all around so that I just can be off.

Emily, we (I) would need to see and know more about your setup to help you achieve better morning start up.

[beacon14]

Emily,

I wrote a calibration routine that routs an "L" shaped slot exactly 1" away from the edges of the material (I use scrap material to calibrate). The operator measures the actual distances with a digital caliper and enters the measurements. The file then replaces the prox switch values with corrected values, re-zeroes the x and y axes, and routs another slot at 2" from the edge. 9 times out of 10 this second slot is within a couple thousandths, but if not the program will repeat.

BTW just rezeroing using the prox switches gets it close but the first slot never measures 1.000 even though it theoretically should if nothing has changed since the last calibration.

[richards]

I've found that temperature plays a significant role in accuracy. Right now it's about 3:15 a.m. here in Salt Lake and the temperature is in the low teens. About twelve hours from now the temperature in my unheated shop will be closer to 40-50 degrees (depending on how direct sun light warms up the large dark garage door on the south side of the shop - and how hard the wind in blowing through the cracks). It's not unusual to have a 50-degree F. change in temperature on any given day of the year.

My Alpha has a twelve foot by seven foot table with a little more than ten feet by five feet cutting area. That steel expands and contracts throughout the day. Normally, I try to cut parts that have to fit tightly together as close together (time wise) as possible - so that the expansion/contraction of the machine won't produce out-of-tolerance parts. (By the way, it's great to own a CNC machine that cost under $20,000 including spindle and vacuum, that has high enough accuracy that I have to consider temperature as a factor in part size.)

[jhicks]

We find that the Y gantry will become "out of square" from one day to the next simply due to turning the motors off at night. We have clear calibration marks on the bottom and top X rails we use to calibrate both ends of the Y rails each morning and often during the day. Its interesting that unless we bang too hard on material or experience an unusual interruption for any reason, the motors stay tight and square all day. Once we turn the software off, and restart, we can see as much as 1/8" variation OFF square on the new start up.
We have not welded the gantry but have triple checked all gears for slop and all is right & tight but the release of the pinions and reengagement are just enough variation to cause this variation over the 10' length of the table.
I guess the message is, things do happen, get loose, or move ot of position so one needs a quick and accurate calibration check as frequently as needed but definately at start up.
All we do to square up is turn off motors, reposition to the calibration marks, turn on motors and they hold position till shutdown.
Each machine seems to have its own personality so the new employees/operators need to develop an intimate relationship.

[fleinbach]

Jerry,

Rather then aligning with marks as you've described check out my explination on using the stop blocks here. This method is very quick and always acurate.
http://www.talkshopbot.com/forum/show.cgi?tpc=7&post=43809#POST43809

[jhicks]

Thanks Frank, I used to have stop bolt heads and indeed thats a good option. The potential problem is with the slightest or vibration of those bolt heads turning can set you up for problems.
With an absolute square score line on both top and bottom x rails, we feel more comfortable and its a quick verification and calibration method as well.
As we all know, its whatever works for you and there are many ways to 'skin the cat" but thanks for the feedback.
Hope all is well out east.