Bed image doesn't refresh


#1

New Pro unit getting my calibration treatment (finding the usable edges, checking how far off the image is vs. the laser, etc). So I’m using two one-inch lines to find the corners.

When I’m in the app, and I’ve run the score of the two lines in the corner and then go to move them to another corner, the bed image doesn’t refresh. Hitting the gear icon and clicking on it nothing happens (no noises or head movement or image change) it looks like this after “refresh bed image”:

Turning the GF off and then back on (with head positioned under the lid camera) results in an unchanging “Calibrating”. The only way to clear the calibrating message is to kill the app window in Chrome, turn the machine off and then turn it back on and let the head do its dance before starting the app in Chrome again.

Then I reload the design and this is the image I see:

The camera is capable of getting a nice clear image, just not in any circumstance except a refresh & reboot of the machine.

This was from 5:50pm to 6:15pm EST today.


#2

What is your procedure for finding the usable edges?


#3

Just remember: usable edges move base on cut/engrave settings. The faster the speed, the wider the margins.

Also I find that calibration is much longer if you have materials in the bed. In some cases of you have scrap in there, the rectangular cutouts will confuse the camera and calibration might be impossible. I don’t think that’s what you’re describing, but if it were me I’d probably do this stuff with an empty bed.


#4

Visually represented:

This is a comparison of 3 manual engravings of a 1/2" wide, 100% black square (just the first few passes are shown). The black areas are motion without laser power. The grayscale areas represent % of laser power from 0-100% white. Blue indicates no travel.

They are at 100, 500 and 1000 GF Zooms™, from top to bottom respectively. (The top is at 125 LPI, and the two below it are at 340 LPI. That’s why the spacing between rows is different.)

At 100, there is very little accel/decel distance. At 500 it is close to a 1/4". At 1000, it is around 3/4".

(Note the brick layering effect. This is an artifact of the way GF modulates laser power. I’m drafting a summary of my findings on that, which is where these images come from.)

EDIT: Forgot to mention: The image is scaled at 1 pixel per step on each axis.


How to save from the glowforge app
#5

Nice work, there!


#6

I have a small cutline about an inch & a half long. One vertical and one horizontal. Set it to score at 10 pews so it marks the masking but doesn’t cut thru. I move the ends of both off the piece until the gray no-go zone pops. Then I move the horizontal one to one corner and then use the arrow keys to step it down one click at a time. Once I get the “no artwork” I leave it and do the vertical one - moving it to the side until it triggers the Ready (no artwork disappears). Then I grab the horizontal one again and tick it one up with the arrows to make up for the one down I did before that took it off the board. Then I run the score.

I repeat for the other 3 corners.

Then I mark the tray’s plastic edges with tape & sharpie so I know where the cuts are going to go.

I do the same with a high speed engrave (1x1 square) and a slow speed engrave (really have to ratchet down the power on this one so I don’t waste a sheet of material).

You lose an amazing amount of room with engraves. I was surprised at how much in fact.

Which is a given when I’m doing multiple jobs on the same piece of material. In this case, there were no cuts in the material - it was virgin PG. Having to remove the material everytime it decides to do a calibration is impractical. Calibration on power up works fine if I move the head.

It’s the failure to re-scan the bed and show me a clear picture that’s causing issues.


#7

I would suspect poor WiFi bandwidth—particularly upstream speed—as a likely culprit.

If you try a speed test with a laptop there, be sure to disable 5GHz WiFi on the laptop so you force a 2.4GHz connection. Then pay most attention to the upstream speed. This is commonly a small fraction of the download speed.


#8

Same setup that worked fine for 9 months of PRU use. Unless the Pro has weaker antennas or some other change that makes for more marginal signal access.

BTW, the speed is 28/6 - down vs up.


#9

Yeah, probably something else.


#10

I think this is related to the post you made here: Bed image doesn’t refresh. I’ll leave it open so we can revisit once we get the other one straightened out!


#11

Did you hear the one about the programmer who got stuck in the shower?

The instructions on the shampoo bottle said “lather, rinse, repeat”.


#12

Snortle! :smile:


#13

Looks like it’s resolved. I finally got the opportunity to try it again and it’s fine.

Interesting things I found though - now my center placement is dead on (last week the test showed I needed to move things 1 click right and 2 down). I started today with last week’s indicated correction and it was off low & to the right :slight_smile:

The other thing I found is there’s no difference in where the no-go zone appears with 500 or 100 zooms on a score. It’s also dead on repeatable at the edges when moving by a known number of clicks.

The last interesting thing I found is that my usable width is 19 5/16" wide and 10 15/16" tall. Not quite the 19.5x11 that’s supposed to be there so it’s good to measure so as not to design something that ends up a hair to big :smile:

You can close this.


#14