Oh Bollox

[The_Doctor]

I got the idea of hysteresis when seeing that the error in my rotating line doesn't appear smoothly rising out of each axis it passes though, there's a kind of 'sticky' quality, like when pulling a magnet away from steel. I think the WideMoves have at least two different problems, perhaps.

The speeds of the scanners might be different and I like this idea, I'll look into this closely soon. Speed difference and 'stickiness' will combine to form the separation of draw and return draw. Note the asymmetry in my rotating line AVI, especially at the horizontal and vertical orientations where the stickiness is very magnet-like, causing early collapse into single-axis drawing, and a tendency to retain that state as rotation continues to use both axes again. Stepping through frame by frame helps, each shows one degree of rotation.

There are two things that should minimise the speed difference effect in my WAV file though, one is the dwell points, the other is interpolation needed to make the wave. I have no 'points' at all except those vertices at the ends of the drawn line. The dwell points ensure that both scanners are stationary for a moment, at the same time. Then, the interpolation means they draw steadily, in effect a lot of very tiny steps. I'm not saying that different servo gain or mechanical inertias aren't having an effect, but I am saying this implies they were not tuned well together at the factory. Individually, perhaps, but not together.

What line geometry does the Laser Media test pattern trace? I'm not sure I can interpret this from an ILDA file as I have nothing that uses those. I can easily emulate it exactly with a WAV file prepared by Lua script, if I know what it must draw, exactly, and how it traces between draws. A simple diagram will help a lot.

I recognise the problem of gain fighting damping. So far I doubt that's an issue in this case. I've also thought that it might be worth risking a bit of that, and running at slower speeds to prevent overheat. This could reduce the effect of that stickiness problem. I'll look at the difference in speeds first though. Am I right in thinking that gain adjustment can be used to compensate for slight differences in inertia? I think that's what it has to do to make sure the speeds are the same for the same input, because it's not actually the speed, so much as the acceleration, that matters.

Edit: Surreal habit of mine, topping three pages in a row now...

[Zoof]

See the pics below. The first one shows the hysteresis file that I found useful (scanned @ 20k, 40 deg). The lines never cross the center point, they always get pulled back to the sides again. The bottom image (green) is the lasermedia image (borrowed it from JoJo, thanks) it also shows our separation problem.

The other two images are basically the same idea. Those helped me identify the problems of friction in a speaker scanner setup. Note that the idea is similar, lines that go to the center and get pulled away again. In this case the never got to the center because of friction making the mirror stop early because it is a feed-forward system. Your description of 'stickiness' reminds me of what I saw with my speaker system. Note that the diagonals are also separated (phase shift). This was due to different inertia of the mechanical mirror attachments. Very hard to control, esp. the mirror lever arm length which had the most significant effect on the phase shift and picture size.

I mentioned the gain fighting because it seems like one can get an ilda test pattern showing (almost) properly for different combinations for servo and damping gains. Only then, the speeds are set different for the galvos and we get the issue at hand. The lasemedia and hysteresis images are apparently much more sensitive and allow you to find the same setting for damping and gain for both galvos.

[The_Doctor]

Thankyou, those will be useful, especially the first two. What limits the speed of each draw, if anything, and are there any pauses at vertices for dwell points?

There's a big difference between letting the scanners make their way between vertices as fast as they can jump, and drawing the lines with interpolation to regulate the speed. What's important is how badly the WideMoves showed problems despite both dwell points AND interpolator regulation, i.e. no hard fast jumps anywhere in my WAV file signal.

The stickiness problem is often worse with lower scan speeds too, which indicates that inertia is not the cause of this error. Inertia based errors would increase with scan rate, wouldn't they? I was using the momentum of the scanners to partially overcome the stickiness at moderately fast speeds. That's why the stickiness shows mostly where that momentum is forced to change, and most of all, when the scanners are stopping or starting their motion.

Friction is an interesting point, I think there does appear to be friction in WideMoves. That too could maybe be overcome to some extent by extra gain and damping (while accepting lower scan speed), as you say, several combinations of both can result in a decent ILDA pattern.

[Zoof]

to give you an idea; the lines from center to side are interpolated and have 15 points. All points are spaced the same. The corners have dwell points, not sure how many.
Unfortunately the files I have are not in ilda but some other binary format supplied with the easylase testprog. Maybe someone can dig up the corresponding ilda files.

[The_Doctor]

Nice. I can calculate it from that info now. Dwell point duration is less important, it's enough to know that there is a moment to allow the scanners to settle.

I'm currently drafting some SMT pad layouts ready for convenient design use but I'll give full time and attention to trying these patterns and some tuning soon.

[Zoof]

doc, FYI
30-40 deg angle, around 12k

http://www.youtube.com/watch?v=FrxfdZR2lKw

[The_Doctor]

Thankyou. Something odd about that though, it appears to be passing around a centre instead of rotating on one. I've not tried that on WideMoves but I think it might mask the sticky-problem to some extent. Might be a useful test because it could separate two possible causes of the problem (gain difference and zero-motion stickiness), indicating which is dominant.

I remember using a set of speakers as galvos once. Those had several sources of line separation because of the loose quality of materials used, but this stick effect is peculiar to WideMoves, it's not a phase or gain thing. While that might be there too, the straight line in my AVI file shows it takes three frames (three degrees) of rotation to even start to turn, and the straight line suddenly veering to the new angle in the next few frames clearly shows a continued tendency to stick to a nonmoving state as the signal increases. There's a very obvious stepped threshold there that has nothing to do with phase errors.

The diagonal stage definitely indicates some phase error but I'm more concerned with that low-motion threshold problem. I think only a dramatic increase in gain and damping, and accepting much lower speeds, can improve this. That's a risky strategy in galvos with no heatsink block, which is what drives me to sound out this issue extensively before trying that. It's just as likely to kill them as make them more usable.


Edit:
http://www.photonlexicon.com/forums/...2&d=1174575932
New link to that AVI file for convenience.[/QUOTE]

More:
http://www.lostgallifreyan.pwp.bluey...le,%20Wide.avi
http://www.lostgallifreyan.pwp.bluey...,%20Narrow.avi

Those two AVI's show a rotating pentacle. I mentioned how the line passing round a centre will mask the sticky effect in a way that won't happen if it rotates directly on that centre. These files show this to be true. The second is scanned at the same speed, on a smaller angle, and shows the problem dramatically. That stepped jump in the lines as they pass through vertical and horizontal isn't a phase problem, it's something else. A phase problem wouldn't have that obvious discontinuity.

[Zoof]

Thankyou. Something odd about that though, it appears to be passing around a centre instead of rotating on one.

I noticed that too, the source comes from the fact that the line isn't drawn straight through the center of the frame. But the frame is rotated around its center. So the line doesn't rotate around its own middle.

[The_Doctor]

Edited previous post.. Added stuff. I haven't done new tests yet, I've still got enough material from earlier ones to use to show the problem. Your file reminded me of that, so it is a useful way to separate the two possible causes. This isn't phase/gain related. That exists too, but appears only to contribute to a little rotational asymettry, and even most of that is due to one galvo being 'sticker' than the other.

I ought to add that during long sessions I noticed the stickiness varying slightly, it was sometimes very minimal. I never figured out what was changing to cause that though. I'd considered a change in the viscocity of oil in the bearings, but I'm not convinced it's that. I never added any oil to try to change this either, I want to understand more before trying to change anything.

[dsli_jon]

I ought to add that during long sessions I noticed the stickiness varying slightly, it was sometimes very minimal

Hey Doc -

Greetings from 'over the other side of the pond' (and a tad bit SOUTH, too!)

...Read your entire set of posts - in BOTH threads - and have a silly question: have you ever taken the cover off of these things? (just the galvo 'heads') - if I am correct, they should just 'slide off' in the direction TOWARDS the mirror...

IF you CAN (and are willing) could you do this, and take clear, internal-detail pix and post them (or, if you prefer, PM-me, and we can e-mail...) - reason I ask, is two-fold...

#1, Not sure if you are familiar (likely ARE) with the 'Greek' scanners from "Eye Magic" - aka: EMS, (3080s, 4000s etc) Infinitys, etc...- though I have consistently been told by MediaLas (and others) that the 'WideMove XT' is NOT BASED ON, OR MADE BY Eye Magic out of Greece... (????????) Hmm, well, though the connector is a bit different, the BODY has GOT TO HAVE been made nearly-identical (down to the bearings) thru SOME sort of 'co-operation', at least... no way this is 'coincidence'...



(L: EMS 4Ks from OSLS.com; R: "M9024" (WM's, duh) from jmlaser.com)

And from everything I understand, (heartily open to correction, here...) EMS was the originator of this body-design, NOT MediaLas - MY GUESS, is that Tom (EMS) sold either 'rights-to-develop' off his new 'body-design' and MediaLas modified / and or integrated their design for the unique characteristics that make them 'wide-move's vs straight EMS - (ie driver, position detector and no-mechanical 'stop' differences, etc..) OR - they are, in fact, just 'modified' EMS 4Ks - dunno, but IF you can post pix of the INSIDE of the 'WideMove's, I can likely discern more about this possible connection...

WHY is this relevant to your post?

Because, #2, IF, IN FACT, these share INTERNALS - specifically the MAGNET - EMS went thru about a year of HUGE issues worldwide with magnet DELAMINATION (nickel-coating would 'blister' after time / heavy use and eventually 'flake' and cause mild-to-severe 'aberations' - JUST LIKE what you were describing ((sorry, I tried to view the vid clip, but even with the most recent version of Quicktime Pro (I am on a Mac) I can't watch it - just shows up 'all-white')) - the issues would come and go, etc.. - BUT - HOORAY! EMS now uses a COMPLETELY different magnet, that CANNOT 'delaminate', etc) - this COULD be the problem with the WideMoves IF they use those same / similar magnets and/or if they are more than a year old / MediaLas did not choose to 'update' to the new magnets...

Now, I concede that all of what I just typed could be pure rubbish, IF WideMoves share ABSOLUTELY ZERO technology with EMS - including that problematic magnet-design - and/or EVERY PAIR, brand new or barely-used, of WideMoves behave this way, well, then it would not likely be due to overheating / overscanning / delaminated/blistered magnets - however, it could be a position-detector issue... there were also a few, rare little other 'bugs' with EMS 'gen 1 and 2' designs, that have all been corrected in the current-selling 'gen 3s'...

Anyhoo, not trying to be a pain in the arse but if you have postable / emailable pix of the INSIDE of your XTs, those would be very helpful to see, cause we do not (currently) have any WideMoves TO check / report...

Aside from that now-fixed issue, just 'FYI', EMS 4000s are PHENOMINAL GALVOS - 45-50K+ w/ 3mm mirs. - especially at higher detail / scan-angles / even giant images (> 150%!) - I was REAL keen to follow any EMS improvements / comeback...and they did it! Add to that Karls' (Banthai) comments about how 'TOUGH WideMoves' are - which also re-inforces my belief that they ARE 'offspring' of EMS, cause EMS always were VERY, VERY 'tough' - ie: you can actually grab ahold of the shaft, while scanning, and STOP the galvo - and NOTHING BLOWS UP!!! (except you now have a mirror embedded in your finger! )

'WHY would you WANT to do that??', well, imagine: at a gig, in the middle of scanning a high-detail image; suddenly - projector power gets knocked out, and your EMS go from 50K pps to ZERO in a snap - and THEN - power comes back on, and 'cause you didn't have time' to kill your projections during the 'hiccup', scan-signals are still going to the amps - and the EMS JUST PICK-UP WHERE THEY LEFT OFF AND KEEP RIGHT ON GOING!!!!! ...Try THAT with 6215s!!

We get some very-high detail out of them.. this shot isn't even the 'worst' one we've done (and yes, the tuning was not perfect, here - it was FREEZING out that night and even in the Show Truck, it affected the tuning a bit) - sometimes, client's will only let you go SO FAR in 'simplifying' their logo / graphics for a show - you try and explain, but they just say 'we want it' - so, you do the best you can...These guys were very happy with theirs...thanks to EMS!

Lemme know about the pix... Cheers..

- J