resolution error?

[KarmaJon]

I am having significant resolution errors regarding automeasuring.

The machine automatically measures my piece as being 33.95" long by 5.44" wide. The actual piece is 34.30" wide by 5.634 inches wide. Errors by hand micrometer notwithstanding, this descrepency seems rather significant.

Auto Measured Difference
33.95" 34.300" - 0.350"
5.44" 5.634" -0.194"

I will call CW when they are open. but I was just wondering if somebody had some idea where I may look.

PMS... That having been said, I have worked on more than my fair share of Mori Sekie's and Fanucs (automated machinery), and I am suprised that somebody had the "Spheroids-of-sufficient-volume" to attempt to make an automated machine for the home hobbyist. It's construction is a simple elegance, and THAT requires a LOT of work. Hat's off and good luck to LHR! I'm a guitar maker, and I'm going to push this baby to it's limit!<grin?<can't wait for the probe!>

[KarmaJon]

Ok, I had a belt break. It ran into the side of the machine. Perhaps the rubbing was the cause of the resolution error measuring the length.

[Jeff_Birt]

Don't forget it is opticaly locating the board edges (in the X adn Y axis) and its perception of what constitutes an edge can be hinderd by board defects, very dark wood, or a dirty board sensor. The small roller in the front of the machine is teh board position sendor for the X-axis, any board defect on this part of the board could cause problems too.

A better test for resolution in the X-axis might be to manully jog the board by a certain amount referring to the LCD readout and then to meausure the actual displacment of the board.

[KarmaJon]

I'll do that when I get the new belts. I can't believe it's only 10 bucks for a pair of them! Yeah, it wasn't the best piece of wood that I used to test the measuring skills of the machine. The Y axis had high repeatability, I think the x-axis issue could have been related to the belts. Oh, well, I have a week or so to figure that out before my probe comes! YAY! :o

[KarmaJon]

Ok, got my new belts today. X-axis repeatability is HIGH, however, not accurate.

I repeatedly get a measurement of 28.68 for a board that truly measures 29.5". I'll try different length boards to see if it's an offset problem versus a scaling problem.

I was taught that repeatability is more important that accuracy, now the question is how to correct for the inaccuracy! :wink:

[KarmaJon]

Ok, I've tried several different pieces of wood, and some pretty good solid stuff, too!

There appears to be no ryhme or reason, no offset OR scaling error, and this baffles me!

I'll call CW, but I'll try to keep the thread alive so that whatever my answer may be, others will find something useful from it.

[Malfunction]

keep us posted :D

[KarmaJon]

A better test for resolution in the X-axis might be to manully jog the board by a certain amount referring to the LCD readout and then to meausure the actual displacment of the board.

Ok, I'll volunteer as the most inept person in the forum. How do you make this thing jog? I can't seem to do it.

[BobHill]

Jon,

Jogging is done width wise with up and down arrows on the LCD keyboard, and length wise with the left and right arrows. You can only jog when the message area tells you that you can.

Bob

[KarmaJon]

How do I get it to let me jog?

The only place I can find to jog is to jog a piece into place when I carve a pattern. I'm trying to jog it while I get a readout of distance traveled so I can check for accuracy.

[The Bard]

when you save a design, sometimes you have "autojog" as a default setting.

if you change that, you can manually jog the piece.

At least, I think that's how it works. I'm more of a hardware person.

[KarmaJon]

Accuracy on MY machine seems to be in the 97.8% range with regards to the X axis.

Y axis is tight. I don't know how to tes Z axis.

I'll call CW tomorrow and see what they have to say. I'll keep you guys posted.

[Jeff_Birt]

KarmaJon, try putting a couple pieces of black tape at each end of the board so that the board sensor will see the change in contrast between the board and the tape as the edge. Measure the distance between the pieces of tape and compare this to the machines reading. I think this will remove any parallax error in the measurment of the boards edge (in other words at some point past the edge of the board the contrast will have changed sufficntly enough to register as the 'edge', but it will not be the true edge). I also remember something in the manual about calibrating the machine, but it required the 3/8" bit to do so.

[KarmaJon]

Thanks, Jeff, I gave that a go, but no luck. However, I LIKE the way you think... it certainly DOES cut one possibility out of the equation!

Apparantly, I'm the only one having this issue, because I can't find anything else posted.

I made a drawing of various lengths of lines cut at 0.031 deep with a 1/16 inch carving bit in various places on the board in an attempt to determine any possible mechanical binding or slipping.

The lines cut by the CW were consistently 102.2% longer than the lines I drew in the Designer program along the X axis. (This rules out slipping or binding.)

The boards that I MEASURE are are consistently 103% larger than the CW measures it. (which COULD have pointed to slippage or binding.)

I did a long rectangle, and it DID go back to it's origin, so the CW IS repeatable (VERY important!), it "thinks" it's doing the correct thing.

Logic dictates that since it measures things (x-axis) smaller than they are, and it cuts things (x-axis) longer than what I drew, then the machine is reading things in increments 2-3% smaller than they really are along the x-axis. This holds true for a 1" cut, or an 18" cut.

I'm including a file I used to do a cutting accuracy check for anybody who would be curious to give it a try on their own. Some of the cuts at the end are at a small angle, but not so much as to cause such an error (Do your trig!)

A calibration may be in order, but the error is NOT an offset error. An offset error would correct for a consistent magnitude of error. For instance, if my measurements were always off by 1/2" whether I cut 2" or 30" lines. Mine always seem to be off by the same percentage, the magnitude of which increases proportional to length.

(They say that consistency is the hobgoblin of little minds, but I'm never sure how to interpret that!)


The board is 29 x 3 x 3/4.

[KarmaJon]

My machine reminds me of my wife: Nothing but trouble since I got her, but, God help me, I LOVE her! :)

[Malfunction]

My machine reminds me of my wife: Nothing but trouble since I got her, but, God help me, I LOVE her!

lol, too funny

[Greybeard]

KarmaJon - In your first post you said the sizes of both x and y were wrong, ( your figures seemed to suggest a similar error in each), but in a later post you said the y axis was "tight". do you mean this literally, or is the y axis ok ?
The fact that you get a negative input error and a positive output error, of the same size it seems, suggests to me a possible computational fault in the cw.
You have a sensor plus software plus stepper giving it the input information, but the software plus stepper producing the output. This rules out the sensor being the culprit, and the stepper is incapable of a +ve and -ve percentage error you have.
This only leaves a software problem.
I'm put in mind of bi directional counters and adc chips - those forms of logic circuit that may be used in both input and output circuits. A fault in such a circuit might produce the symptoms you have.
Over to the cw team I think.
Regards
John

[Ron Baird]

Sounds like you need to do a calibration proceedure from your keypad.
Also check your head crank down pressure (should be 80 lbs).
Take a board 10 or 12 inches long measure the length then turn it 90 degrees (sideways) and measure the width and see if the mesurements are the same. Make sure your board is not warped.

[KarmaJon]

KarmaJon - In your first post you said the sizes of both x and y were wrong, ( your figures seemed to suggest a similar error in each), but in a later post you said the y axis was "tight". do you mean this literally, or is the y axis ok ?.......

My first measurments were done on some scrap pine with rounded edges, I think that this led to the earlier inaccuracy of the Y measurement. Every time I measure wood with sharp edges, the Y axis is tight, and by tight I mean accurate AND repeatable.

As far as computational errors, that seems reasonable. I didn't know if it was a servo with an encoder, or a stepper motor set up.



Sounds like you need to do a calibration proceedure from your keypad.
Also check your head crank down pressure (should be 80 lbs).
Take a board 10 or 12 inches long measure the length then turn it 90 degrees (sideways) and measure the width and see if the mesurements are the same. Make sure your board is not warped.

A calibration may be in order. Lately, I've been having to push down on the head while I crank it, lest I get a "Clear Board Sensor" error. The boards I'm using are really good, and four of them were recently planed. I'll look into the calibration.

Thanks for the input and info, Grey and Ron!

[Ron Baird]

See my post on head pressure and check it. Clear board senror usually means the sensor is dirty or not getting high enough readings (0-160). Dark wood can effect low readings also. Head preasure effects a lot of things proper length is one. Also make sure the little brass tracking wheel is clean. The calibration proceedure was developed to be able to recalibrate length and width problems like you are having. They developed it because my m/c was cutting short and then measuring even shorter both ways. This is to keep your m/c calibrated without sending it it C/W.

[pwngsd]

That's a good one! <G>

[KarmaJon]

See my post on head pressure and check it.

Well, all my pressures seem reasonably ok, now. I lubed the moving head parts and the clutch clicks at the appropriate pressure. That took care of the "Clear Board Sensor" error. I still have the same measuring/cutting issues. I've done a couple little pictures, and they've turned out good. Being off in one direction by only 2 or 3% isn't that big of a deal for that stuff. But I'm going to have to tighten it up a lot more and be repeatable before I take carbide to my Bubinga!

[Ron Baird]

Have you done a calibration test on it?
That may correct the shortness problem.
If it doesn't you may have to send it to CW to be recalibrated.
It actually measures length by several different methods so it can be a complicated problem. what I have told you to check are the most common causes.
Ron

[KarmaJon]

Yeah, I appreciate the help, Ron. The only calibration test I can find involves some joining bit... I'll let ya guys know what CW tells me...

[Ron Baird]

Yes you need a 3/8 " strait or spiral bit to do a calibration. I use a solid carbide down spiral bit to do jointing, it does a better job than a strait bit but you will probably need a 3/8 to1/2 collet to put it in the bit adapter.
Ron

[KarmaJon]

Ok, I'm on to something. I removed the O-ring on my brass Board Sensor, and my length measurments are REALLY close now! I mean, we're talking 0.001% over the length of a 22" board!


In the drawings, they don't have an O-ring. Do you guys have an O-ring on the LCD side of your knurled brass knob? My brass knob has a groove on either side of the knurling, and an O-ring on the LCD side groove.

Mine had an O-ring, and it seemed a little larger in diameter than the knurled knob diameter. If you guys DO have an O-ring, does it seem noticeable larger than the knurled knob?

I'm not saying this is definitive, yet, but it makes sense that a measuring surface with a larger than specified diameter, and therefore, a larger than specified circumference, would result in measurements of boards shorter than they really are, and cuts into boards that are longer than those specified in the program.

[Ron Baird]

Yes the O-ring should be there. It is to prevent slippag and to overcome a problem where sometimes the brass wheel sits in on a toothand rocks back and forth causing a kind of vibration in the sensor circuits. If your that close a calibration funktion will probably fix it. Get a jointing bit it is good for bringing the board downto the exact size and squaring it.
Ron

[KarmaJon]

Yes the O-ring should be there. It is to prevent slippag and to overcome a problem where sometimes the brass wheel sits in on a toothand rocks back and forth causing a kind of vibration in the sensor circuits. If your that close a calibration funktion will probably fix it. Get a jointing bit it is good for bringing the board downto the exact size and squaring it.
Ron

Yeah, I bought a double sided 3/8 in 1/4 shank bit today for just such an occasion.! Right now I'm scanning, but I'll cancel that operation and put the 3/8 joiner in to calibrate it. Are you saying that the brass toothand doesn't neccesarily contact the board, and that it's the O-ring that does the work? Thanks for you input in advance!

[Ron Baird]

As far as I can tell they both touch the board, the O-ring compresses enough to let the brass wheel touch the board. I have seen brass wheel tracks on some soft woods but I think on harder wood the wheel might not dig into the wood enough to keep from sliding a bit. I was also told that they had a condition where the wheel sometimes sits in a position where it is on the top of a tooth and it rocks back and forth enough to cause a minute viabration in the sensor circuit that caused large problems.
The O-ring somehow dampens that effect so if you remove it you may cause other conditions. If doing a calibration does not correct it you may have to have CW calibtate it. As I said befor there are many things that can effect length problems. A good test is to take a 12" x 10" approx. board measure its length with the m/c then turn it side ways (90 degrees) and measure width and they should be the same (within .05"). Hope this clears it up for you.
Ron

[KarmaJon]

As far as I can tell they both touch the board, the O-ring compresses enough to let the brass wheel touch the board. I have seen brass wheel tracks on some soft woods but I think on harder wood the wheel might not dig into the wood enough to keep from sliding a bit. I was also told that they had a condition where the wheel sometimes sits in a position where it is on the top of a tooth and it rocks back and forth enough to cause a minute viabration in the sensor circuit that caused large problems.
The O-ring somehow dampens that effect so if you remove it you may cause other conditions. If doing a calibration does not correct it you may have to have CW calibtate it. As I said befor there are many things that can effect length problems. A good test is to take a 12" x 10" approx. board measure its length with the m/c then turn it side ways (90 degrees) and measure width and they should be the same (within .05"). Hope this clears it up for you.
Ron

For what it's worth, my O-ring seems "fatter" than the brass knob. Noticably so, which is not something I would anticipate from an issue that's causing a 2-3% error. I could see where running on the brass knob alone could cause problems. As far as the 12X10 test goes, the only piece of wood I have that wide that I would sacrifice 10" of is some crappy pine, and I wouldn't trust the results anyway!

Anywho, I purchased a Frued 3/8" straight double fluted bit with a 1/4" shank to do a calibration test. Oh, and I put the O-ring back on. However, I don't have an extra collet, and removing the 1/8" bit from the collet that I received with the machine is proving to be a daunting task!

Chris from CW contacted me today, and he is having me conduct a calibration test with the 3/8 bit. I think we're on the path to solving this issue which seems, thus far, to have stricken me solely! :shock: Thanks for you input, Ron. It was one of your posts in another thread that really helped me zero in on my problem.

[GAK]

I have resolution errors with my y and z axis.

When I attemp to cut a 10.062" square (.062" is half of the 1/8" bit width), the x-axis is cut at 10", but the y-axis is cut at 10 3/32" -- In other words, I get a rectangle instead of a square. The y-axis error is proportional i.e., the y-axis on a 5" square would be 5 3/64".

When using any bit, the depth of cut is .01" deeper than the software bit program setting.

Yes, I have calibrated the CW using a 3/8" bit. There was no difference in the resolution errors after the calibration.

I have tried tape, different boards, etc. without any difference in the resolution errors.

I am trying to make templates for my wife's quilting projects, and a perfect square is mandatory, as is the bit depth for engraving fine lines.

Is there a way to change the machince default setting on the x, y or z axis?

I would appreciate it if some CW users would cut the attached file using the 1/8" straight bit.

The CW machine should produced a 10" square cut .004" deep. Carving time is just a couple of minutes.

[Ron Baird]

I am glad to help you out hope you get it fixed I have had that problem and my m/c is good now. The board does not hav to be that wide but it hasto be under booth top rollers when you turn it sideways. Another thing that will effect length is the board has to be the same width for the full length. allways check that the sliding guide is not pressed against the board so tight that it stops the board from moving on the sand belts.
The screws on the bit adapters should be locktighted, soto loosen the screws heat the adapter a lillte with something like a heat gun to soften the locktight.
Ron

[KarmaJon]

Well, I'm trying some other tests. I ordered some new belts and some new collets, but it's the Chinese New-Year and I'm told it will be a while before they arrive.