Is there a known issue with smaller frames?

Bar wrote:

There isn¢t a way to look it up afterwards :confused:

We really should save it in the maslow.yaml along with the anchor locations.

David Lang

2 Likes

I just re-ran calibration.

My worktable is not perfectly rectangular, but is roughly 8ft 1in x 4ft 1in. I have carriage bolts at the four corners to anchor the Maslow 4; these are inset between 1 and 2 inches from the various edges (depending on how much overhang the tabletop had in that particular area). Since my set-up isn’t perfectly rectangular, I measured the two lengths and two widths between each pair of carriage bolts and averaged them to create the values I fed into the Maslow’s config. settings for calibration.


Next, since the parts I’m trying to make right now aren’t that big (about 17” x 5”), I decided to calibrate a smaller portion of my table than what maybe I could have tried to calibrate. I ran a 5x5 calibration on a 38” (aka. 965.2mm) x 21” (aka 533.4mm) area.

Following the calibration, I got this message:

Here’s the dialogue box from the Maslow connection:

Calibration complete
Calibration values:
Fitness: 0.78986314910065
Maslow_tlX: 17.6
Maslow_tlY: 1145.9
Maslow_trX: 2412.8
Maslow_trY: 1188.1
Maslow_blX: 0.0
Maslow_blY: 0.0
Maslow_brX: 2414.8
Maslow_brY: 0.0
A command to save these values has been successfully sent for you. Please check for any error messages.

I think this value in the high 0.7 range is what I’ve been getting previously as well, but the shape of my parts were (by eye and with rough measurements) coming out fairly accurate so I didn’t think twice about it. I haven’t done any hard-core accuracy testing yet and I’ve also only been cutting (well, “attempting to cut” more often than not, admittedly) one piece at a time and keeping that small piece centered in my workspace to keep things in the best possible location for the machine’s accuracy (or what I think to be the best location given the various other forum posts I’ve read).

Note: Following this calibration, I successfully jogged around in various directions moving by 1", 0.1", and 2" at a time with no “slack cables” issue. Also, this successful calibration message and successful jogging is similar to my “usual” experience pre-cutting; I’m not sure what triggered the slack cables the other day after the two failed attempts to finish cutting my 1 piece. First, I want to see if this cal. is good or bad though, and then I can try more cutting and give you move feedback on all that.

  1. I suppose my first question I should ask is: what is a “good” fitness value? Is the fitness value similar to an R-Squared value where we’d be looking for something in the high 90’s (like 0.98 or 0.99)? If so, then maybe my fitness values are actually pretty bad?? Again, not sure what measuring stick to compare my 0.79 value to!

  2. If 0.79 is decent, then I can do some cutting later this week to test some accuracy instead of trying to cut my parts out. If anyone has an “accuracy test file” with lines spaced a certain distance away or something like that I’d be happy to try it or I’ll just sketch something up quickly. Would be cool if we all had access to a standard one though for the sake of aligning troubleshooting steps and results from the distributed sample pool that is the Maslow 4 install base to reduce variables (apologies if that exists and I’ve just missed it!). For me this would be great because I might think to cut and measure lines or boxes but maybe circles are more interesting or maybe the fit between joining parts is more interesting/useful to get feedback on and I certainly wouldn’t know how to best test for accuracy in a way that is most meaningful to you and the other “super users” here on the forums! Having a file we can all run for accuracy tests and then a drawing of exactly what measurements to take and report on could allow the community to create a pretty awesome database for gathering insights from where things like set-up size or anchor types (thinking of rigidity) or thickness of cutting material would be the variables while all the tests and measurement data would, in this case, be the constants. The way microscopes have “masters” to calibrate against is an example of the type of thing I’m thinking of (example link).

Note: my firmware is 0.85.1.

Full text from Maslow serial:
Maslow-serial 30DEC2024 MD.log (8.0 KB)

Also - thanks so much for the time, consideration, and assistance!!

1 Like

0.79 is absolutely decent

1 Like

This sounds to me like it could be an encoder connection issue

MattD wrote:

  1. I suppose my first question I should ask is: what is a “good” fitness value? Is the fitness value similar to an R-Squared value where we’d be looking for something in the high 90’s (like 0.98 or 0.99)? If so, then maybe my fitness values are actually pretty bad?? Again, not sure what measuring stick to compare my 0.79 value to!

fitness is 1/average error so the higher the better and it can go well above one

0.79 is very good for the current state of the software.

David Lang

2 Likes

Great thanks for the more detailed info! This, and Bar’s earlier reply, makes me feel like my issues are not necessarily due to the size of my setup so I’ll try some more cuts and record what happens with a bit more diligence and report back/continue to leverage old forum posts.

Happy New Year everyone!!

1 Like