Tails and Defects

[planters]

After tuning and tweaking the settings on a couple of low cost scanners, I have to say the performance is pretty impressive. They are rated at 30K and when set to aprox 10 degrees they produce a classic circle in square at 30K. When scaled up to 30 degrees the ILDA 12 pattern remains relatively unchanged except for a slight curving of the horizontal lines below the box, yet the circle goes completely haywire and forms a spiral. This larger pattern is produced running the scanner @ 27K and despite these changes the graphics produced look very similar except for size.

The only significant problem with the overall performance is that moving points/beams have significant tails and boxes/circles don't completely close. The tails are present at both size settings, but are longer at the larger size. They are polychromatic in that a white point leaves a white tail even though the green component is a little longer. The open objects are still open in the smaller image, but the gap becomes harder to see when the object is so small. Any suggestions in improving on these defects?

[james]

That has nothing to do with the accuracy to which your scanners are tuned. That is a function of the delay that exists between the color modulation signals and the scanner signals.

Scanners are not instant. They take a fraction of a second to translate the electrical signal into an actual position. Laser modulation is (almost) instant. DPSS laser might take a tiny fraction of a second to come on and there might also be some residual energy in the crystals after the pump diode is turned off.

So it is a necessary design element of any laser control system to be able to delay the color modulation signals behind the scanner position signals.

I don't know what system you are using to drive your laser projector. But there might be a way to adjust the time delay. There must be a time delay. Otherwise you would see much worse alignment than you are describing.

James.

[swamidog]

what software are you using? you should be able to control that with the color blanking settings in your software. you might also want to add a color correction board like this:

http://shop.stanwaxlaser.co.uk/dz-co...oard-252-p.asp

After tuning and tweaking the settings on a couple of low cost scanners, I have to say the performance is pretty impressive. They are rated at 30K and when set to aprox 10 degrees they produce a classic circle in square at 30K. When scaled up to 30 degrees the ILDA 12 pattern remains relatively unchanged except for a slight curving of the horizontal lines below the box, yet the circle goes completely haywire and forms a spiral. This larger pattern is produced running the scanner @ 27K and despite these changes the graphics produced look very similar except for size.

The only significant problem with the overall performance is that moving points/beams have significant tails and boxes/circles don't completely close. The tails are present at both size settings, but are longer at the larger size. They are polychromatic in that a white point leaves a white tail even though the green component is a little longer. The open objects are still open in the smaller image, but the gap becomes harder to see when the object is so small. Any suggestions in improving on these defects?

[planters]

I am using Beyond to control the projector, but I am unaware as to how to use the software functions to tune the delay to match the modulation. It was my understanding that the color correction board from Stanwax (I have one in my parts drawer) would allow me to equalize the delay between the lasers only. And so, I should be able to shorten the somewhat longer green (DPSS) tails to better match the red and blue, but that then the now perfectly white tails would remain otherwise unchanged. Is that wrong?

I will try the blanking setting in the set up menu, but it seems crude,ie 1,2,3 and 4. Any other less obvious adjustments to try as well?

[swamidog]

the stanwax board will allow you to control blanking rise and fall times for each individual color channel.

try the color blanking adjust in beyond (yes, that sounds very limited, but it's a start), then use the stanwax board to tweak things. it's not a surprise that you're seeing this after re-tuning your scanners.

I am using Beyond to control the projector, but I am unaware as to how to use the software functions to tune the delay to match the modulation. It was my understanding that the color correction board from Stanwax (I have one in my parts drawer) would allow me to equalize the delay between the lasers only. And so, I should be able to shorten the somewhat longer green (DPSS) tails to better match the red and blue, but that then the now perfectly white tails would remain otherwise unchanged. Is that wrong?

I will try the blanking setting in the set up menu, but it seems crude,ie 1,2,3 and 4. Any other less obvious adjustments to try as well?

[planters]

All right. That sounds like a plan. I'll post the results.

[norty303]

I am using Beyond to control the projector, but I am unaware as to how to use the software functions to tune the delay to match the modulation.

From memory, the blanking correction is in Projector settings. I think its a horizontal slider just below the scanspeed slider (although I may be thinking of LivePro here). It should jump out at you in one of the tabs now you know what you are looking for

[planters]

try the color blanking adjust in beyond (yes, that sounds very limited, but it's a start), then use the stanwax board to tweak things. it's not a surprise that you're seeing this after re-tuning your scanners.

From memory, the blanking correction is in Projector settings. I think its a horizontal slider just below the scanspeed slider

It's there and it works...kinda. Talk about crude. The choice runs from LONG tails on one side of a beam to short tails then to LONG tails on the opposite side of the beam and that's it; three choices. This seems odd when the other sliders on the same page allow one part in a thousand precision. The color differential remains and I am hoping the color correction board may deal with this as well as hopefully reduce the remaining tails that still exist. This will take a few days to test because I will have to open up the projector.

The other correction adjustments on the same Beyond page such as point numbers, spacing and anchors seem to have no detectable effect on the test patterns when they are changed. I can't seem to find a description of their operation in the help or tutorials. Any insights here as well?

[buffo]

It's there and it works...kinda. Talk about crude. The choice runs from LONG tails on one side of a beam to short tails then to LONG tails on the opposite side of the beam and that's it; three choices. This seems odd when the other sliders on the same page allow one part in a thousand precision.

If you have a QM-2000, I'm pretty sure you already have better than 1 part in a thousand resolution with the color shift control. (It should be per-point resolution, so at 30kpps, that would be 1 part in 30,000.)

The reason you can't get any better is because you are pushing your scanners far beyond what they are capable of, so the color signal hits the lasers well before the scanners can catch up. And even with single-point resolution, you can't be sure that you'll be dead-on for every frame, because at such wide scan angles the scanners are ballistic most of the time, and their position is subject to distortion, causing tails.

That's the short version anyway. Let me explain in more detail:

Earlier, you said that you displayed the 12/24K ILDA test pattern at 30 degrees optical (which is way, way more than even the best scanners are capable of!), and you were surprised that most of the pattern still looked OK, and only the center circle and a few short lines were distorted....

There's a good reason why that circle is distorted when you are scanning that wide... When the scanners are drawing that circle under normal scan angles (8 degrees or less), they are already ballistic. That is, they are under CONSTANT, MAXIMUM ACCELERATION. In fact, they are already distorting a good bit (3db, if I remember my math correctly).

Look closely at the ILDA test pattern in a frame editor. Do you see that center circle anywhere? No. What you actually see are 12 points in a dodecahedron (12-sided regular polygon) that lies completely outside the square. Yet when it's displayed, all you see is a smooth circle. (Or at least that's what you should see if your scanners are tuned correctly.)

This illustrates why the scanners are said to be ballistic. They never even get close to reaching the first point in that dodecahedron before they head off towards the next point. They are actually distorting that 12-sided figure into a smooth circle, but in a regular and predictable way. (Should be around 3db of distortion; they are "rounding off" those 12 corners, and staying completely inside the square.) They are right at the limit of their performance. Anything beyond that is only going to cause increased heat and wear. They won't move any faster.

The scanners race around that circle three times before moving off to draw the rest of the pattern. So when you push them too hard, the first part of the ILDA test pattern that starts to look bad is that central circle, because the individual paths start to separate, the circle gets squashed and twisted, and eventually it looks more like a spiral than anything else. This is a sure sign that you're over-driving your scanners.

Now, apart from the circle in the test pattern, there is another time when the scanners are typically moving at (or near) maximum speed. This is during the blanking jumps in a frame. This is when the laser is off, and you want to get the scanners moved to the next position as fast as possible before turning the laser back on to start drawing again. So the number of blanked points in a blanking jump is usually very low; just enough to get the scanners there. Most software will add a few "anchor" points at the end, right before the laser turns on, to allow the scanners to settle into position. But if you are pushing your galvos too hard (that is, scanning too wide), the scanners don't get into position fast enough, and the laser turns on while they are either still traveling to the end of the blanking jump, or while they are settling into position.

This is why adjusting the color shift setting doesn't work. It alters the laser timing one point at a time, but that's not good enough when the scanners are ballistic, because they could be in any position between two given points when the laser comes on. You need to give them enough time to settle before you turn the laser back on, but if you are scanning super wide, you can't predict when they will be ready.

OK, sure, if you had super-precise control (let's say 1 part in 500,000) of the color signal, you could eliminate the tails, but then that setting would be frame (and scan-angle) specific. As soon as you changed to another frame that had a slightly different spacing between the blanking jump points, then you'd need to change the color shift again by a tiny amount. And if you changed the scan angle, that would alter the timing again.

Bottom line: you adjust the color shift so that under normal scan angles (8 degrees), you don't have any tails in the test pattern, and the short vertical lines at the top evenly bisect the split horizontal lines. That will ensure that the color signal always comes on at the point after the scanners have settled (following a blanking jump), and it will cut the color signal off right before the scanners take off on a new blanking jump.

Once you've done that, you can increase the scan angle wider, because most artwork is not anywhere near as complex as the test pattern. But if you get to a point where even normal artwork starts to show tails, then you know you are pushing your scanners too hard.

Assuming you have everything tuned correctly and you absolutely have to scan wider, and also assuming that you only have tails on a few frames, you *may* be able to add extra anchor points to those troublesome frames (giving the scanners a few extra points worth of time to settle into position) and get by, but there is no guarantee. And in any case you'll still be running your scanners very hard, and they may or may not put up with it.

The color differential remains and I am hoping the color correction board may deal with this as well as hopefully reduce the remaining tails that still exist.

The color board will help if you have different modulation responses between the three lasers inside your projector. This manifests as a projector that has tails of one (or sometimes two) colors, and when you adjust the color shift, some of the tails go away but then the third color starts showing a tail. The color board allows you to adjust for these differences so they all respond the same.

But if you have white tails (all three colors showing the same tails) no matter what you set the color shift at, then you are scanning to wide. In this case the tails are not due to differing modulation responses but rather due to the scanners being too far behind because you are over-driving them.

Adam

[planters]

Thanks Adam,

All that is clear and I do not argue with any of it. Despite the very high resolution of the color shift, the control page and within it the color shift slider has only 6 positions. When the slider is moved from one position say 3 to 4 the tails, wherever they are seen, lengthen substantially and in exactly the same direction. When the slider is reversed from say 3 to 2 then the tails flip exactly 180 degrees and if continued to position 1 then they lengthen in this new direction. It seems to my mind that if I could adjust the slider to say 2.5 then the tails would largely disappear.

Regarding scan angle, 8 degrees is too small to be useful to me. What is the typical real world angle that most people scan at? Looking thru a couple dozen YouTube videos it seems that the typical projector is running at least at 30-40degrees. I am leaving out any video that may have substantial foreshortening and estimating the full angle from the projector as a point source.