Galvo head models?

[Dannym (Danny Miller)]

I’m seeing that a lot of what you can do with a MOPA depends on the galvo head capabilities.

I have a SG7110 and most often raster at 7000mm/sec on an FL160 110mmx110mm, which is about the max speed I can go.

I can command it faster, but then you start to see overburning on the line starts because the mirrors aren’t quite up to the commanded speed.

Actually some of the quality problems I still have may well be due to this still being more significant than I realized.

Going slower requires proportionately reducing the frequency to avoid melting the surface, but that really digs into the runtime.

I’ve wanted to get an SG7210 head which is supposed to be an upgrade but I’m a little confused because I see there are A,B, and C suffixes, and no suffix. I just got in one which was supposed to have silicon mirrors but the badge said K9 mirrors, which is not advised for my 300W MOPA as they have poor thermal conductivity and can heat up and expand. Maybe won’t happen while holding back the wattage so I don’t melt the coin, but it would make the machine fail whenever I do engage the full power so probably not an upgrade. I had to return it without trying it.

I can see that, for the same galvo head, if I change from a 110x110 field to a 200x200 field lens, the mm/s capability should nearly double, and reduce for a smaller field lens, because the mirrors actually only know the angular speed, they don’t care what the lens does with it after to make a linear surface speed out of it. But a larger field lens has a proportionately larger spot size- but that may actually help if the spot size is smaller than you need for the detail level in your work. Switching to a 200x200 would not only go 2x the speed, but you’d double the line interval and take 1/2 the number lines per pass, so 4x faster speed.

Anyhow, point being, I see there’s a limited number of 10mm aperture plug-and-play galvo heads here, and I likely have “the low end”, SG7210 is “+100 more than” an SG7110, but the specs for AliExpress SG7210 head ads go ALL over the place and I don’t know if it’s because of the A/B/C/(null) suffix, or if silicon mirrors might be heavier and fundamentally slower, or if their “max marking speed” makes no sense at all because it doesn’t specify it in angular units OR specify the lens size used to get that surface mm/sec figure.

This is a very frustrating market- I’m not sure I’ll find anyone here who has upgraded to even one other head on the same machine, thus have actually seen the difference. Pretty much given up hope of finding anything really comprehensive like comparing ALL the heads on the same MOPA with the same lens.

[fins]

IIRC, the A, B and C suffixes are for the different red light configuations. When you get in to the kW teritory, you have to start looking at higher end heads like scanlab et al.

[jkwilborn (Jack Wilborn)]

I’ve tried to research galvo heads. And I’m not sure if I’m following you, but you can decide that.

@Dannym

mirrors actually only know the angular speed

Yes, but the mirrors have no smarts. What you need to examine is how the control board tells the mirrors to move. In my view, it’s a lot like spending time on the tires when it’s the whole operation of everything.

Do different lenses have different jump/delay defaults?

I’d have to contest this as a longer lens has a longer line of travel, so an abnormality in the galvo should be seen much easier with a longer lens. [scan-speed-adjustment] This is about surface scan speed, but the emissions of a longer beam would show more detail. Just thought I’d throw this in, what do you think?

What we care about is surface speed. Frequency gives you a pulses/s rate, scan speed is figured out by how the controller is configured. I have a spread sheet that covers speed and pules/s for my MOPA that was put together by Russ Sadler. When I got my MOPA, he sent it to me.

The control board eventually controls surface speed by knowing what lens you use. This is loaded when you load a specific device.

@Dannym

“max marking speed”

Yes, they need to be specified in angular abilities or the data isn’t worth much. I think I had the same issue you’re having.

That’s my 2¢ worth.

[Dannym (Danny Miller)]

Well, the more I use the MOPA, I may have a 300W, but I’m so far from being able to max out the power that I might not be in a different boat with a 100W or even a 60W.

The limit is scan speed. More specifically, I’ve got everything tuned out but I can see it jumped to a new line, waited long enough to become stable, but when asked for higher speeds and scaling up freq proportionately, the accel distance becomes significant and the first few pulses overlap and cut way deeper than they should, the “burn out”. It doesn’t seem to occur on stops because the system just keeps up the constant speed and pulses until the end of the line, then just stops pulsing and lets the mirror overshoot. This can’t be systematically corrected with increasing Ton delay. Ton is calibrated correctly and this might make it seem better in some cases but worsen others.

On my SG7110, 6000mm/sec on an F290 lens looks good. More and that burn-out starts showing up more. 3300mm/sec on an F160 lens is basically the same accel problem.

What I’m trying to figure out is how much better the SG7210 is. Or the other various galvos.

But, to be honest, the bottleneck doesn’t seem to be with the galvo as much as it seems.

On the initial passes, with 100% of the coin getting hit, it takes like a second per pass.

But, the further down it goes, with high level of detail resulting in many starts and stops on each raster line, it slows to a crawl. Which is surprising, actually. The level of detail should not slow down the scan lines.

I had to wonder, is it really slowing down the scan speed for each line, or pausing before each line to buffer it? Almost certainly has to be the second- if you slowed the line as it cut, unless you also reduced the frequency, the line would be seriously melted by pulse overlaps.

I took a rough estimate- a lot of on-off content slows the pass from about 1 sec to 8 sec. That pretty much dominates everything else.

Am I wrong in seeing it may be pointless to go with a faster scan head if there’s a major bottleneck in the detail that is capping the engraving speed <1000 mm/sec once we get into high detail?

[fins]

I think you are running into limitations of the controller/driver combination. My understanding of the layer between the usb in and fpga out would suggest a fundemental timing limit for setup and teardown of different operations. my approach is binning by path length and batching settings based on that.

[Dannym (Danny Miller)]

This is where I think the bigger fish is a switch from the BJJCZ LMCV4 (“EZCAD2”) to the BJJCZ DLC2 (“EZCAD3”) controller, which is an expensive upgrade but has vastly better specs, but from what I was told, Lightburn isn’t able to make use of it yet and it’s basically not going to perform any better- yet.