WideMove tuning?

[The_Doctor]

I was looking at some old pictures and found these:

They make it clear that the output signal matches the scanned image, so I assume that this is not a mechanical flaw, as the driver 'knows' about that distortion, and is not trying to correct for it as it would if it were mechanical (and in which case I imagine the correction attempts would show up as important clues, differences in those two output shapes).

So my question is: is this some kind of hysteresis deliberately added electronically to reduce jitter or ringing that might otherwise occur at small signal levels? And if so, is there anyone here who knows how to tune WideMoves to reduce it more convincingly? It seems to me that if there is hysteresis it may be reducible, and ideally I'd like a stronger galvo drive to overcome this small-signal error. I don't mind if it slightly overdrives the scanners, as it would not be a general condition, I'd still operate well within safe limits, I just want to see if they can ruthlessly persue exact position even when the signal is reduced to smaller levels than those which show this error. The obvious similarity between output signal and scanned image implies that WideMoves can perform a lot better than these are doing, because the scanned image is a very good copy of the output signal. Can anyone tell me what to do to bring the best out of them?

[The_Doctor]

Never mind, ignore this thread unless you're really interested in WideMoves. It's amazing how many memories fall away in 6 years. I went Googling for 'WideMove gain damping', and found huge detail in discussions I'd had with Zoof and DZ and others in Stanwax's 'Oh Bollox' thread, as well as on alt.lasers.

To cut it short, it comes down to whether I'm willing to compromise their virginal state and start tweaking the few controls offered. I predict that some improvement will be had if I increase input voltages, reduce 'scan size' control, and boost both 'servo gain' and 'damping' controls. As there are things I don't know about the innards it will be worth testing a reduction of both of those too. Finally, another small problem may be fixed as Zoof mentioned, by making sure than both scanners run at same speed.

If anyone's REALLY interested in WideMoves, let me know ASAP because I can either sell them on in their virginal state, or start playing merry hell with them. As I want to buy some 6215's (another thread mentions that), I want to sell if I can, complete with the platform I built specially for these WideMoves, to make full use of their wide angles. They're ok at those wide angles. Any flaws I describe exist at small angles only, at very small signals. So long as I don't have to follow up trying to cure buyer remorse, I will sell. Given the amount of detail I forgot mentioning, I doubt I could ever be accused of not describing these things clearly enough, warts and all. I don't use eBay to sell, haven't for years, but anyone who buys these will have no unexpected troubles if they search for posts about WideMoves around here, most of them are likely mine... So enough from me. Just let me know if you want them, make me some modest offer I'm unlikely to refuse, after asking me for all images, data files, so you know about the platform as well as the scanners. Don't leave it too long if you want me to lay off those tweak pots!

[-bart-]

What sizes are the mirrors to accomodate these angles ?

[The_Doctor]

What sizes are the mirrors to accomodate these angles ?

5mm * 12mm according to Medialas Widemove info, for 90° with a 2mm wide beam. (Coating specified to withstand 10W power at that width, 15W at 3mm).
Playing with SketchUp or any other geometry-based tool will verify their claims for 90° and 2mm. My scanner platform will let them do wider, given a thin DPSS beam. I aimed to allow the full capability of the scanners, given that wide moves were the whole point.

[mixedgas]

Before you assume your scanners are not optimally tuned, what is your criteria for evaluation? At what Angle? There are better ways to evaluate, ie ILDA pattern, Laser Media Pattern, small square wave step, or large square wave step. Traditionally if you are not using an ILDA pattern you tune for damping and servo gain using a square wave input (impulse) and a scope on the feedback. However you must then switch to something like Laser Media Pattern, or overlaying both X and Y channels on the scope, to ensure velocity matching.

Distortion in that circle can come from a lack of power supply or amplifier compliance, as well as other issues, such as differences in tuning "speed" from axis to axis.


The pictures attached show the effect of the damping potentiometer, they are reversed in order by the browser. They show ringing, slightly under-damped, and properly damped. The last pic is from a Cambridge manual, showing Position Sensor output and Galvo current for a small step square-wave and large step square-wave. (credit to CHAN and CambTech for the pics)

Steve

[The_Doctor]

Before you assume your scanners are not optimally tuned...

Before I get into anything else, I ask you to answer me on one thing. Do you recognise that my assumption is that they ARE correctly tuned, and that this is the sole reason that I spent so long asking questions and testing with various inputs instead of tweaking things I didn't understand?

[mixedgas]

The assumption is simple. If you have a small signal tracking error in the loop, you check it with the 30 to 50 Hz square wave and watch the leading edge of the feedback on the scope, while watching the two dots formed on the projection screen. On the scope, the leading edge should curve over sharply and consistently with less then one or two faint cycles of ringing. If there is position jitter or hysteresis in the system the scope trace will show it.

If the scale factor is set correctly you then overlay the traces of both scanners responding to the same square wave input and look for differences.
If matched correctly for velocity and error feedback, the traces will overlay.

Unless your circle in the picture is starting to approach the upper bandwidth limit of the scanners, you will not find the problem by eye.

I'm just suggesting the same procedure the factory would do, so its your ball.

I learned long ago that the scope/damping method gets me to a given hardware problem quicker then the various test patterns. It allows me to see small changes in the response as I adjust the potentiometers. It also allows me to isolate the different feedback loops on the amp as I probe them. Dual trace scopes make this easy.

I'd also open the lids on the Wide Moves, Wide Moves are pretty much Eye Magic clones, and look for dust/flakes from the magnets on the position sensor or between the stator and rotor. Before I did that step, I would check out the response on the scope, first.

You need to leave the sensor scale controls alone, as well as any DC offsets in the system.

Perhaps you should voice your complaints to a factory engineer.

Your amp topology should differ, You may have PID or PD or PII depending on how the loops are set up, however this is the best instruction you will get on the subject, read the sections in the back, "Tuning Strategy" which have the simulated scope traces:

http://www.google.com/url?sa=t&rct=j...pDmOW3ddUR36mw



Steve

[The_Doctor]

You didn't answer my question. The more I post the more you think I do not understand. As I'm ok with that assumption... (I'll repeat that bit to head off any smart remarks about it.) As I'm ok with that assumption, and want to sell the scanners, it is best that I do not open them (which requires removal of mirrors), or tweak anything. The best thing I can do is nothing. That way I don't risk devaluing them. Whatever quirks they have are small, consistent since new, and have not changed, and most people have wondered why I ever thought there was a problem, so maybe there isn't, much.

As to looking at things not apparent by eye, that's not the issue, I'm interested in what I thought was an abberation that WAS visible. As it is, it affects only small signals, so as Florian Rotter (a laser systems engineer who sold them to me) said, it is not a great concern.

If you're interested, have a look at the AVI attached. It's an XviD made from JPEG images, one frame per degree step in a rotated line. It shows the effect that got me started in questions and tests. I've decided to drop all opinions I ever had about it, though I'm interested in those others may have. For myself, all I know is, if I am going to spend lots of time on technical matters, it will be those I have stated (other thread, which you also saw) are my main aim in living: programing a phase modulation synthesiser. Anything that gets in the way of that will be pushed aside (as gently as possible), as I don't know how much longer I have to live, and it may not be enough to finish the job.

Incidentally, I agree with you about scope/damping rather than test patterns. Strangely, when I said similar things, everyone started demanding that I use standard test patterns!

[mixedgas]

All I see in the rotated line is a velocity mismatch and hysteresis in one axis. What I need to know is the scan angle and frequency of the line waveform when that movie was taken. A critical piece of information you are not including with these images.

Test patterns let you find Gross errors, the small, subtle, stuff needs a scope on the feedback.

Can you scan the Laser Media Pattern for me? It is for finding the subtle errors.

Steve

[The_Doctor]

All I see in the rotated line is a velocity mismatch and hysteresis in one axis.

Same here, the hysteresis I worked out, it was Zoof who explained to me the velocity mismatch. I might have figured it out, but I didn't.

What I need to know is the scan angle and frequency of the line waveform when that movie was taken. A critical piece of information you are not including with these images.

Offhand, I don't remember. I tried to look up old files to see if I could find it. Scan angle is easiest, based on me being in same room with no major shift in fittings, I can gauge that within reason: 27° optical.
Speed is less certain (I can find no wave or script to generate it, from that experiment), but the slower the speed, the more pronounced the effect. As I wanted to photograph a revealing test case, I'd have chosen the slowest speed that neither eye nor camera would see as flickering or incomplete. Estimate a few tens of Hz for the draw, the apparent rotation of the drawn line was a lot slower.

Can you scan the Laser Media Pattern for me? It is for finding the subtle errors.

Sorry, I haven't got anything like enough set up for that. Lasers in boxes, scanner to one side of room with thin layer of dust, cables required in use doing other stuff (and the only way I ever used an ILDA pattern was when James Lehman converted it to WAV file for me, I never had that other one). I've been working on audio and coding for months here.

Anyway, as you can see now, the angle is fairly small, hence my realisation some time ago that to do what I want, I need scanners intended for high accuracy and smaller angles. Truth is, I sorely overestimated how much angle I really needed in the first place! WideMoves with truly wide angles don't show this geometry error, but to use them at those angles I'd need a much bigger wall, and I'd then be scanning much more slowly than I want. There was a lot more wrong with my choice in scanners for my needs, than there is with the scanners. This is why I want to sell them, (unmolested and mounted in what is a very nice box of thick Al and acetal plastic), having figured out what I should have done the first time.

EDIT: I found the original WAV file. It was a very musical 440Hz, faster than I remembered. Basically a triangle wave with 10 samples of dwell time, all at 44100 Hz sample rate, AM modulated by slow sinewaves in quadrature for each channel. Which wasn't exactly how it was constructed, but that's the effect of it.