PDA

View Full Version : Mercury scaling factor



Jerry Allen
03-31-2007, 7:25 PM
I have had my Merc 25 going on 2 1/2 years now and had deliberately ignored the scaling factor label inside the front door which was applied by GCC at the factory. The reason I ignored it was that it said the suggested scaling was 3/1000 although the driver has separate settings for X and Y. I had meant to ask but it kept slipping my mind.
Today I made a 10 by 10 grid of 1 inch squares, engraved it on a sheet of birch ply, and then measured it. Sure enough, it was out 3/1000 in X. Y was right on. So I set the scaling in the driver as a default and tested some old designs which I was having trouble with. Amazing difference, especially on larger pieces.
In most cases you wouldn't notice the error. But when you are doing a wooden clock, for instance, or fitting a interlocking part the difference can be noticeable.

Jim A. Walters
04-01-2007, 6:29 PM
I had to cut an acrylic circle that was 15" in diameter. The size was rather critical, so I cut quite a few sheets of cardboard before I gave the acrylic a try. I believe mine is 0.003" per inch also.

Jerry Allen
04-02-2007, 7:41 AM
Jim,
Did you set the scaling factor? It's on the Advance tab in the driver.
I wish I had checked this earlier.
I'm attaching the grid I used.

Richard Rumancik
04-04-2007, 10:09 AM
Jerry

Glad to hear that the use of the scaling factor improved your results. For my Mercury, I found the scaling factor to be an unsatisfactory solution. Here is why:

I need to use the laser to do very accurate cuts. If a part is to be 24.00 plus or minus .020 long then I need to be sure that there is no built-in repetitive error in the machine. There are enough other sources of error in fabrication that will eat up my tolerance budget. (For example, kerf width variations, material variations, affect of air assist, direction of cutting, etc). The scaling factor should theoretically allow the repetitive error in the machine to be eliminated.

By looking at the driver, you would think that you could increase the scale of a part in 0.1% increments (1/1000). But I have found that a 1/1000 and a 2/1000 scale factor result in the same amount of expansion. Similarly, a 3/1000 and 4/1000 have the same effect on the length. So only odd numbered scale factors are actually "addressable".

As a real world example, I mark two parallel lines 24.00" apart in CorelDraw. When I mark a piece of cardstock I measure the lines to be 23.95" apart. This calls for a scale factor of 24.00/23.95 = 1.00208. Round this off and select 2/1000 scaling. But this scaling factor is not addressable. If I use it, I will actually get a 1/1000 scaling factor which will make my lines 23.974 apart. A slight improvement, but not good enough. If I use 3/1000 scaling factor, they are then 24.022 apart. (It jumps .048".) So the scaling factor causes my part to dance around the required value but I can't actually hit the mark of 24.00.

I'd be interested if you could try this experiment for yourself and see what you get on your machine:
Draw two vertical lines .20 long 24.00" apart in Corel.
Send to laser.
Move lines down .22" and set scale factor to +1/1000
Send to laser.
Move lines down .22" and set scale factor to +2/1000
Send to laser.
Repeat as needed and observe the results. (In my case 1/1000 and 2/1000 give the same results.)

Perhaps you were lucky that your machine error just happened to coincide with a scaling factor that works. Mine is between addressable correction factors.

Even if all the .1% increments actually worked, they would still not be quite fine enough for my purposes. On 24 inches, 0.1% is .024" which is quite a large change. I think that the scaling factor should be in 0.02% increments to be useful. This would allow me to tweak to within .005" on a 24" length. Right now it is closer to .050" increments on 24" which is not useful.

I also think it would have been better if it was set up in the firmware rather than in the driver. Then you could probably do better than .005" on 24. GCC told me that the reason for the coarse increment was a "Windows limitation". At the firmware level you could probably set a very accurate correction factor.

I am using the 1.74 driver and V2.20-02 firmware. Can you tell me what you have?

I know not everyone cares if their laser is cutting accurately as for many decorative items and signs it doesn't matter. But if there is anyone in the group using an Epilog or ULS, I'd like to know if they use a scaling factor system or something similar to tweak the machine accuracy. If so, does it work?

Rodne Gold
04-04-2007, 11:07 AM
You could do corrections in software extremely accurately by stretching or compressing the whole drawing prior to printing

Mike Null
04-04-2007, 11:49 AM
That's my method.

Jerry Allen
04-04-2007, 4:54 PM
Richard,
That's pretty old firmware and driver. Mine is 2.24/3.57.
I'll check your request tomorrow. Doing taxes today.

Rodney and Mike,
I guess that would work. I just would rather not have something else to remember to do and undo if a change is needed. My scaling factor is in X only. Y is right on. I forget minor details enough as is, and frequently move things around when cutting. I figure if the feature is there, it's worth using. If it wasn't, I'd just scale the drawing in X.

Richard Rumancik
04-04-2007, 8:23 PM
I did not say it in my post, but I have had to resort to Rodne's method to get accurate parts, namely, scaling the part in CorelDraw. I do not like to do this because it means a lot of backtracking if I need to do any changes.

With my method, I save a copy of the file to a new name, and then cut from the scaled (distorted) drawing. I usually have to sort some vectors to optimize cutting order etc. as I get into production and make other tweaks. However, I can't (shouldn't) edit the geometry in this file as it is distorted geometry. If edits are needed, I have to go back and edit the main file geometry. Then I end up having to repeat the scaling operation, re-doing the plot order, and remembering any other tweaks. It works, but it is a hassle and requires more files to track.

Like Jerry, my Y-axis is bang-on. But my X-axis is .21% undersize. If I could have just entered this value once and for all in the firmware setup or driver, it would save me a lot of trouble.

Jerry Allen
04-05-2007, 11:57 AM
Richard,
Performed the test as requested.
I carried the test through 6/1000.
0,1= no change
2,3 = right on
4 = over +.07
5,6 = over +.14
Seems to be another software anomaly at 5/1000.
I also scaled X using Transform in Draw at 100.3% which was right on. No surprise there.

Bottom line is, regardless of the method or machine, if you want to know why your critical dimensions are not 100% correct on critical items, you need to run a test. I thought 10" apart was bad, but it's way worse at 24" with no correction. Am I wrong to assume that this must happen on all machines?

Rodne Gold
04-05-2007, 3:21 PM
Just about all CnC machinery or XY tables will have the ability to correct axis scaling , all the other stuff we use does.
Biggest problems with unscaled axes is when nesting the same item in a sheet and some are rotated. Nothing is quite right.

Richard Rumancik
04-05-2007, 4:42 PM
Jerry

Your results seem a bit different as to grouping than mine, but you can still see the large increments in size. I use a .21% scaling factor in CorelDraw which I determined based on a 24" distance. The further apart your test lines, the more accurate your correction factor will be. Of course it's necessary to have a way of measuring accurately. I use a 24" machinist's rule and a 10X loupe to measure my lines. A measuring tape doesn't have adequate accuracy for this.

Some people will never have a problem if their machine is .050" out on 24". It depends on what you are doing with your laser.

I am hoping the ULS and Epilog owners will have some input on this and tell us how scaling errors are corrected on those machines.

Jerry Allen
04-06-2007, 10:02 AM
Richard,

I think the fact that there is no consistency in the setting for the driver and that it must be checked is the issue here. Mine happens to work okay but until you replied, it never would have occured to me that 2/1000 would work too.
I did use a 24" machinists scale and a 24" sheet of ply. That meant my 24" mark was off the board which is why I only gave 2 place dimensions.

You should probably should start this thread over without Mercury in the title to get more varied input.

I also noticed in another thread that you might have a 4" lens you don't need. I have a new 2.5" I don't use. PM me about the lens if interested in a possible trade or sale.