PDA

View Full Version : Vertical Banding



Ian Johnson
09-30-2015, 2:18 PM
I always have vertical banding when rastering with my Gweike:

322509322510

What I've read is that it could be caused by power or mechanical problems. I'm thinking it probably isn't power, because the frequency of the banding is always the same no matter how fast it scans. That would seem to mean that the pulsing is related to the head's position rather than an external variation like inconsistent power. I've found plenty of posts saying that mechanical issues can cause vertical banding, but not much detail about what mechanical issues to look out for. What problems should I look for and how do they contribute to banding? I've seen belts mentioned, but aside from backlash at the ends of the move, how does tension contribute? I would expect something like dirty rails or bearings to create more irregular patterns.

Could it be something like the resolution of the drivers? I'm familiar with stepper drivers for 3D printers. If there is no gearing in the extruder and the drivers are set to 1/16 microstepping, there are few enough steps/mm that pauses between steps are just long enough to create tiny pulses in the flow of plastic which show as a pattern on the surface of the print. I can see a similar thing happening with the laser if the steps/mm are low enough.

Dan Hintz
10-01-2015, 7:48 AM
Vertical banding is almost always an issue with mechanical or dpi/resolution.

Dan Hintz
10-01-2015, 7:49 AM
BTW, I'd love to know what/how you're creating those pics... looks pretty cool.

Nicolas Silva
10-01-2015, 8:59 AM
Banding might be caused by a loose belt.

Ian Johnson
10-01-2015, 12:09 PM
BTW, I'd love to know what/how you're creating those pics... looks pretty cool.

I found a browser-based tool called Triangulator - http://goo.gl/d2xjI You download the source files, and open the page locally in a browser. As you click around on the image, it adds polygons and averages the values within it. You can click your way along features so it subdivides into smaller polygons where you need more detail. When you are done, it exports a SVG.

I thought it would be a cool variation on the 3D-engraved lithophane. The sharp transitions between polygons enhances the 3D effect, with the different layers visible at an angle through the transparent acrylic. It doesn't work quite the same as a traditional lithophane, since simple variations in material thickness don't affect the shade as much with transparent acrylic. It is more that the thickness affects the visibility of the engraved surface when viewed from the bottom. To really work, it engraved areas need to be frosted, but I would prefer not to have the banding.

I was wrong about the frequency. It does smooth out as speed increases. The example in the photo was run at 70mm/s so that it would engrave deep enough for the blacks. I can run faster in multiple passes and get a smoother result, but it is difficult to get a wide range of depth. I can adjust contrast between passes so previously engraved light areas go untouched, but it still tends to flatten out.

For DPI, is it a matter of matching up pixel size with dot size? I've tried a few DPI for a simple black square and didn't notice any change in frequency. Is it possible that PWM pulses can show if the movement is slow enough?

Kev Williams
10-01-2015, 2:10 PM
You guys really should do this test. I really have no clue as to the WHY my laser does this, but it DOES do it on purpose. And I'm betting many machines do the same.
Of note is my old ULS does NOT do this, but, it has NEVER had banding issues.

Here's the test. NOTE that I ran this test when my LS900 was a week old and even had a tech come check it out on warrantee.. This anomaly hasn't changed a bit in 11 years...

This is a piece of clear plex, 3-1/2" x 2-1/2". The test rectangles are 3-1/2" x .300" inch...

Resolution settings are 50 lines per inch Y, 400 dots per inch X. At 50 LPI the lines are .020" apart, easy to see what's going on.

First line, 100 speed, 100 power...

Second line, 50 speed, 60 power...And this is important, I ran it TWICE, and notice the laser precisely copied the first run.
--this means it CANNOT be a mechanical issue. It would never repeat this precisely if it was...

Third line, 25 speed, 30 power. At this speed the beam fluctuation is quite a bit less finally. But barely any change from 100 to 50 speed...

And the fourth line, to further prove it's not mechanical, is a .020" spaced horizonal hatch fill that's been vector engraved, 100 speed and 30 power.
--the lines are dead straight... (the bottom line is goofy because I pulled the plex out of the machine before it was done, then put it back-THAT is 'mechanical' error! ;) )

What I find a bit amusing is the first line of the raster cuts is dead straight, all others are wobbly..

SO, while I have no idea of the cause, the EFFECT is clear- In many places along the horizontal plane, the wavy cuts are going to overlap more so- and less so-
in several places. And these ovelaps, or lack thereof, is at least ONE source of vertical banding. In the top 2 lines, you can see 'banding' in the form of light and dark areas
due to the spacing of the cut paths... same thing happens at high res settings...

FWIW, the only sure cure I've found for banding is "less than black" settings, and running in photo mode. The dithering really helps...

Try it!

http://www.engraver1.com/erase2/band2.jpg
http://www.engraver1.com/erase2/band3.jpg

Dan Hintz
10-01-2015, 2:57 PM
What I find a bit amusing is the first line of the raster cuts is dead straight, all others are wobbly..

You need to find out how your steppers are controlled, i.e., is there any positional feedback. If there is, I can tell you what's likely happening...