galvo mirror mounting theory question

[buffo]

Thanks. But, I take issue with your back of the envelope.

Hi Eric;

Actually, it was in reply to Norty's question about air resistance. But no matter...

Even a 1cm wide mirror scanning a across a 30 degree optical angle (15 degree mechanical) will have a maximum tip speed = to 1cm x 3.14/24 = 0.13cm x the frequency.

My calculation specified small-step bandwidth. The bandwidth for a 30-degree step is easily an order of magnitude lower.

Put simply - the widest scan angle that you can still correctly scan the center circle of the ILDA test pattern at 30K is the small-signal limit. It's probably somewhere less than 3 degrees optical, which would be 1.5 degrees mechanical.

Plugging that into your formula above yields 1 cm x 3.14/240 x 2500, or around 33 cm/second.

Admittedly though, that seems very slow; it's less than 1 MPH. Not exactly sure where the error is though - that may be correct.

My original calculation was based on an assumption of linear travel rather than an actual measurement of angle traversed, so that may be it. But this seems to show that the scanners are actually moving a lot slower, which doesn't make much sense since you can't actually see them moving with the naked eye. (Unless the distance is too small and the frequency too high for our eyes to notice? That might be it...)

If the full angle scanned can actually be scanned at 2,500Hz full cycle then you get 6.25M/sec. peak and 3.13M/sec. average. Right?

Ahhh, sort of.?. Ignoring for a moment the fact that you can't actually get 2500 Hz on full scan angle, if you assume 30 degree scanning (15 degrees mechanical), that is exactly 1 order of magnitude larger than my example. So you'd end up at 327.1 cm/sec, or 3.3 meters per second average. Not sure where the 3.13 came from, but it's close...

At the same time the acceleration forces I estimate will be on the order of 6,500Gs; 6.25M/sec. x 1/(1/10,000 sec.).

Ok, again ignoring the fact that this is far beyond the capabilities of any scanner, I'm wondering why you chose 1/10,000 second as the time interval for something with a period of .0004 seconds?

I agree that if this were possible, it would lead to some incredible acceleration numbers. But it's not possible - at least not yet.

Are these scanners actually capable of 2,500 Hz full cycle/full range?

No scanners are capable of this. And I don't know if we'll ever see ones that can do that in our lifetimes.

Let's start with the current state-of-the-art, which is the Cambridge 6215HC - currently rated at 60Kpps with small mirrors at 8 degrees optical. But that is not the same thing as 60,000 Hz at 4 degree mechanical angle. What it does mean is that the circle of the ILDA test pattern will be just touching the square at 8 degrees optical. That's it.

We know that circle is made up of 12 points, so take 60K and divide by 12, and you get 5000. That's the small-signal bandwidth limit. (Because if you try to go much wider, that circle tears itself apart.) So as long as you are under 3 degrees scan angle (which represents the size of the circle relative to the rest of the pattern at 8 degrees), you can more-or-less reliably scan something. Granted, there is still a lot of distortion, because that circle should actually be a dodecahedron that lies outside the square, but the points are rounded off and shrunk inwards a good bit by the scanners running flat-out.

Note that the small-step bandwidth limit doubles as the scan speed doubles, but it's always at least an order of magnitude less. (To be precise, it's exactly 1/12th the scan speed.)

Now, Bill's statements about his scanners suggest a maximum improvement of 30-50% in scanning speed over the best Cambridge has to offer. So with Bill's new scanners, it's reasonable that we could see 90Kpps scan speeds at 8 degrees optical. But that will only increase the small-step bandwidth up to 7500 Hz. And the full-travel bandwidth will be at least an order of magnitude below that. Thus we still have a long way to go...

The problem is that we're stuck with this ugly definition of "speed" measured in points per second. Because it's really only relevent to a specific set of conditions. (Namely, when scanning the ILDA test pattern correctly at 8 degrees optical.) And converting that to more usable units is difficult and cumbersome. I guess the real truth is that scanners are a lot slower than people realize.

Adam

[planters]

you can't actually see them moving with the naked eye. (Unless the distance is too small and the frequency too high for our eyes to notice? That might be it...)

Yes I'm almost certain that is the reason. You can see them, but your eye does not isolate them at any one spot long enough to register and focus and so that is why they appear blurred at the edge.

why you chose 1/10,000 second as the time interval

I chose this interval because in a 2500 Hz full cycle (5000Hz 1/2 cycle) the mirror must accelerate from "rest" to the peak speed of 6.25M/sec 1/2 way through this 1/2 cycle and then immediately begin decelerating to "rest" stop. There are 4 acceleration/deceleration periods within one full cycle.

6500 G's may seem like a lot, but at the small scale of the mirror edge it really isn't. The edge of a 1" Dremel disc experiences 13,000 G's

No scanners are capable of this. And I don't know if we'll ever see ones that can do that in our lifetimes.

Actually MEMS (DLPs) can probably do this now. The limit is their phasing. Right now they act incoherently like a vast array of independent little mirrors. I believe a mirror smaller than about 1/2 mm will produce diffraction effects on the order of 1mrad, but to go really fast you must go small. I think the next phase will be (like with the CPU's) parallel beams and multiple scanners/MEMS rather than ever faster single channel projectors.

The problem is that we're stuck with this ugly definition of "speed" measured in points per second. Because it's really only relevent to a specific set of conditions. (Namely, when scanning the ILDA test pattern correctly at 8 degrees optical.) And converting that to more usable units is difficult and cumbersome. I guess the real truth is that scanners are a lot slower than people realize.

I agree with you. These KPPS are only useful for apple to apple comparisons. I also agree with Bill and would rather see resonant frequency, bit depth and Watts.

[mixedgas]

I'd rather have resonant frequency, peak watts, small and large square wave response on a digitizing scope, plus a FFT of a step,the ILDA test pattern, the Laser Media test pattern and the 10x10 grid. Plus I'd run the standard test show I use. And I'd still not have enough data.

Eric, do you have a bench signal generator with adjustable square waves? You should try this.

I do what Cambridge taught me to do. Put in a square wave, adjust the tuning, and make measurements.
Once someone shows you what and where to look, many, many things show up as measurable. Is it something you can learn to do from a book? Well partially. This is one case where instructor lead learning is needed. With the hardware, with a tweeker tool in hand.

Buffo, while I'm sure Cambridge feels a major disconnect from the laser show community, ask their marketing folks if they will come to SELEM?
NELEM if it happens is certainly close enough for them.

In the past, they sent a sales engineer, a marketing person, a single axis test bench, and a scope. They handed out nice ~30 page books on scanner setup, some of it clearly taken from their internal classes. They then taught a class, and stuck around for semi-private individual training later. I still have my booklet. I've posted parts of it. But that was 15 years ago.

Most people read the Camb tuning book, think they have grasped it, and never try the scope method. Either that or they don't zoom in on the right part of the waveform. Again, instructor lead learning helps here.

As for the scope method.
Does it get you images? Roughly. Does it save time, YES!! Does it do the final tuning to match ILDA Stardard? No. Will you be tuned better for a given Angle, yes, given patience and pratice. It tells you one heck of a lot about small signal jump time, and how your controls interact. When you can actually see how one turn of a pot really affects your output, doing the tuning the next time is much easier.

Once you've done the small signal response, your much better off. Then you can switch back and forth from ILDA to Square wave.

Then you run the finished product's feedback signals and current signals into the scope in X-Y mode. You then watch the show in laser while watching it on the scope. It gets very instructive, very fast, as you can see where the amp is expending energy. You then see where the image wastes scanner time.

Before some one says, You can do this without the external wave gen.. Dont. Don't chime in and say, I can make the square wave as a frame. Well, you can, and it will help you, somewhat. However in most modern software, you can't turn off various auto-magical corrections and timing jitter added by the show software.

I've always wanted to build the scanner measurement software tool. I have a VERY good idea how to do it. But I have no funding, little time, and I have 532 to make. The hardware costs almost nothing to do, I'd just need a good programer and time to work with them.

Steve

[planters]

Cambridge contacted re. NELEM and depending on the "agenda" they are considering it.

[Bradfo69]

Sigh... stumbling across threads like these, while very interesting to read, are always glaring evidence of how little about this stuff I'll ever have a chance of understanding or, be able to participate in an intelligent conversation about.

Carry on...

[planters]

Sigh... stumbling across threads like these, while very interesting to read, are always glaring evidence of how little about this stuff I'll ever have a chance of understanding or, be able to participate in an intelligent conversation about.

Carry on...

Any thoughts about cumming to Boston on the 26th?

[Bradfo69]

Still giving it some consideration but, found out my banquet captain comes home from Florida on the 26th so, it still may prove difficult. How close to where ever this is, is there a train station? (not to derail the thread - you can pm or put answer in the meet and greet thread) Train might be a good solution for sleep if I need to leave late Friday night.

[planters]

I believe Krazer lives right there and might have this for you. I would PM him. Nice hotels right across the street though.

[buffo]

I chose this interval because in a 2500 Hz full cycle (5000Hz 1/2 cycle) the mirror must accelerate from "rest" to the peak speed of 6.25M/sec 1/2 way through this 1/2 cycle and then immediately begin decelerating to "rest" stop. There are 4 acceleration/deceleration periods within one full cycle.

Ahhh... Ok, I'm following you now. Never really thought about it that way before.

6500 G's may seem like a lot, but at the small scale of the mirror edge it really isn't. The edge of a 1" Dremel disc experiences 13,000 G's

Wow... I had no idea. But yeah, the math checks out. (Guess I shouldn't be quite so angry when those little cut-off discs fly apart now, should I?)

Actually MEMS (DLPs) can probably do this now. The limit is their phasing.

What is the maximum deflection angle of a MEMS though? I though it was much less than 1 degree. If you could get 15 degrees out of one, then that might be worth looking into. (True, you'd still have the phasing problem though.)

Steve: If you've got a contact for someone at Cambridge, shoot it to me in a PM and I'll talk to them about SELEM.

Adam

[mixedgas]

Steve: If you've got a contact for someone at Cambridge, shoot it to me in a PM and I'll talk to them about SELEM.

Adam[/QUOTE]

My inside sales contact left for a job in the Government.

Sorry,

Steve