scanner options?

[cipher0]

So what you're essentially saying is DT or any other chinese scanners are a lot slower than they claim to be in the specified angle and everyone who has told me otherwise till now is either lying or doesn't know what he's talking about himself?

Makes we wonder if i want to dive any deeper in this field. There have been quite a number of people saying DT30 is 30 kpps at 8 degrees and DT40 is 40 kpps at 8 degrees. Everyone who ive talked to till now actually from both emails and chinese sellers (the latter actually say 30 or 40 is the 'normal' speed and there is also 'max' speed).

[The_Doctor]

Well, this is why I wanted people to show me that the DT40's and other scanners could do the rotating line test as I described it. Turns out the Lasermedia pattern has this exact test built into it as Bill describes it, having diagonals with a draw both ways over the same line. Clearly amps have something to play in this too, although I need to understand more before I can separate that from the effect of the larger mirrors in Bill's second video. (Note that static display of Lasermedia won't show this, you have to set it slowly rotating as well, or at least set it to some arbitrary rotated but fixed angle. It seems to me that this may be a good way to test scanners for their change from small step to large step response, by doing that then increasing scan angle until the bowing/fishing thing happens.)

[Galvonaut]

I'll be more than happy to see scanners sold as per spec. Crap galvanometers sold over spec. are the reason I've bought used CT6800s - I hear that they are trusted and an industry standard company and hope to get some decent results from them.

I might try the diagonal test on the various galvos I have here once I have the CTs in a working box. It sounds like an interesting test.

Keith

[Pangolin]

So what you're essentially saying is DT or any other chinese scanners are a lot slower than they claim to be in the specified angle and everyone who has told me otherwise till now is either lying or doesn't know what he's talking about himself?

No. What I am saying is that if the scanner looks like a Cambridge model 6800, it will work like a Cambridge model 6800. This scanner is built in a specific way. The rotor diameter, coil construction, number of turns on the coil, etc. all play a role.

Moreover, what I am saying is that the 6800 is a 30K scanner. They work great at 30K. But when you force them to go faster (and when I say "them" I mean any 30K scanner) then you'll be sacrificing *something*.

The first thing you'll be sacrificing is large-signal performance. If you watch the video I made for North (and at the end, Doc with the rotating diagonal lines), you'll see that when the pattern is small, the lines stay absolutely straight, but the lines don't become bent until the pattern is made much bigger. If you tune 30K scanners to faster-than-30K, the bending lines will start bending at smaller angles. You'll "leave the small signal domain" at a smaller angle.

Another probable thing you'll sacrifice is image quality. Any spring/mass system (like the rotor-shaft-mirror system) has at least one resonant frequency. Once a rotor starts resonating, it will distort the projected image. At 30K, the -3db bandwidth is 2.5kHz. If you've got resonances at 3.5kHz you might not notice them. But if you tune to 40K (which is 3.3kHz) then you'll start to excite those resonances. I can show this very clearly in the projected images produced with DT40s. I can also show -- and have many times -- that when you REMOVE THE NOTCH FILTER from the DT-40 amps, and actually tune the amps back to 30K, the projected images look much better.

So, the main point is that tuning a scanner that was designed to go 30K to faster-than-30K will involve sacrifices. Just like clocking a CPU at a rate faster than it was meant to go will sacrifice something (like reliability, or at least convenience, since you'll have to go through a lot more trouble to keep the CPU cool).

[Pangolin]

the reason I've bought used CT6800s - I hear that they are trusted and an industry standard company and hope to get some decent results from them.

The 6800 was a great scanner. It was the industry standard up until around 1999. Used ones will work well, and likely the only problems you'd have are the bearings. If you purchased ones made around 1996 or later (the first two digits of the serial number is the year it was made), then they have ceramic bearings which should have a longer service life.

With the 6800 the real wildcard is the amp. The original CB6580 design from 1992, and the VERY FIRST '671 design worked well. But in around 1999 and later, they made a change to the components and got real aggressive with the servo in an attempt to give more performance, but the result was less performance. The "diagonal lines" test we've been talking about will be more troublesome if you're using '671 drivers (those are the surface mount drivers) made in 1998 or newer. Although I pleaded with them to change it, Cambridge did not change components back to more reasonable values until around 2002 or later, and as far as I know only corrected it in the '673 dual axis amps.

Anyway Doc's suggested test could indeed be very useful. I've never really used the LM test pattern in that specific way until Doc raised the question online, and even when I used it, I only did so out of convenience of an easy way to demonstrate what Doc was talking about. Prior to that I only used lines or a scope.

So this could be a new test -- make the LM test pattern rotate and increase the size until the lines are no longer single-lines. That will give an INDICATION as to the angle at which you enter the large-signal domain.

Incidentally Keith, if you "force vector interpolation" of any test pattern, then the rules are different. As far as I know, LDS, LSX, Phoenix, Dynamics, LSP and all of the other LDS variance are really like "vector interpolation by default" and you have to really force it to "no interpolation" type deal. QS is kind of the opposite. It will assume point-oriented input unless you force it to be vector. But in any case, if the computer thinks it's supposed to be a vector-interpolated image, then it may add additional points, effectively slowing down lines, which will corrupt the test. It's possible to force vector interpolation on a pattern to the point where you can make it full size and still will not bow the lines. This is what Doc did in a screen shot he showed. So make sure when you're messing with test patterns, you tell the computer "no interpolation" (in whatever way you have to tell it that, for the particular software you're using).

Bill

[Pangolin]

I remember that 1Khz or so was the limit on WideMoves, which clearly relates to the 18-24 kpps spec given depending on angle.

Actually Doc, 1kHz -3db bandwidth is 12K. See the beginning of my video where I describe what this whole "K" concept means and where it came from (hint: 12 points in the ILDA test pattern).

[The_Doctor]

Actually Doc, 1kHz -3db bandwidth is 12K. See the beginning of my video where I describe what this whole "K" concept means and where it came from (hint: 12 points in the ILDA test pattern).

Point taken. Linear relation! I think my error came from not being sure what my scan angle was when I was pushing for the fastest speed 7 years ago. I know it got to more than 1KHz but I can't remember how much so I cited the only certain memory I had, adjusted downwards to avoid what I thought might otherwise be exaggeration.

The video with the K bit? Big video, half an hour? I saw two shorter ones, but not the big one yet. About points, I'm guessing this is a bit like a MIDI sequencer in that timing of events, and magnitude of those events, combine to form the pattern by directing the mass in motion, and that what Steve Roberts called 'pulling points' to correct for overshoot or poor line weight is like 'clock moving' a MIDI sequence event. If this is correct I may be closer to understanding than I thought. If not, I'd better watch that video.

EDIT: One thing I know: in music, small timing details adjusted by moving an event a few clocks can be important for the 'feel' of a note, the sense of mass it has. In other words even in the absense of mass our brains expect mass-like behaviour to be behind a musical event, and are not fully satisfied that it feels real unless the timing matches the mass it thinks must be moving! In a piano sound on my synthesiser, the attack speed was critical for exactly this reason. FM never sounds exactly like a piano, but my sound feels more like one that most samplers do, because of this detail. Not so relevant maybe, but I thought it might be interesting to describe this..

[andyf97]

This made me wonder something; Is the best quality car a Rolls Royce; its certainly not the fastest around corners but would keep up with the most on the highway.

There are scanners with different speeds, from 15 to 60 kpps from what i know.

But which are the 'best' quality scanners? Im not talking about the speed, but overall quality.

For 40 kpps i know there are few options: DT40, DT40W and PT40. Which one is better?
For 60 kpps i've heard about 'CT60', but cant find information on the internet. Maybe it can display more points, but that aside maybe quality is worse compared to DT40 and maybe DT30 is better quality than PT40.

What do you guys think? I want as much points as possible, but i also want smooth scanning.

[The_Doctor]

Anyway Doc's suggested test could indeed be very useful. I've never really used the LM test pattern in that specific way until Doc raised the question online, and even when I used it, I only did so out of convenience of an easy way to demonstrate what Doc was talking about. Prior to that I only used lines or a scope.

So this could be a new test -- make the LM test pattern rotate and increase the size until the lines are no longer single-lines. That will give an INDICATION as to the angle at which you enter the large-signal domain.

That reminds me of something. Steve Roberts (mixedgas here) said that what I saw was ballistic response, which I'm guessing is related to large step response. I disagreed with him (and we now know that it was a sticky resistance to starting from rest in the Widemove case), but in many situations he'd be right when it appeared as your video shows it in the 506 when NOT using an interpolator. In other words his judgement may have been caused by not realising that I was using one. As I knew I was, this was one reason I knew the Widemoves were doing something a scanner shouldn't, because keeping it tracking an interpolated signal meant it was never getting a signal that could throw the scanner hard during the test.

In other words, this test may be extra-useful. If an interpolator is NOT used, then bowing/fishing reveals onset of large step responses when increasing scan speed or angle. If it IS used, bowing/fishing may indicate a reluctance to start moving due to friction or high viscosity of lubricant or contamination in a bearing. This is beginning to sound like a nice dagnostic tool.

There was also the thing that Zoof told me, that if the bowing/fishing thing happened even on perfect 45° diagonals, for example even when the Lasermedia pattern is in normal fixed upright position, the damping and gain of two scanners is not matched for both, but this line abberation is presumably the oldest known diagnostic it allows. We now have two more.

[cipher0]

Sorry Bill, your original post sounded like that to me.

Okay, DT is based on another scanner (which is meant to run at 30K). So what? Just because something is based on something else, doesnt make it a "copy", but a "derivative". If it is exactly the same, then you could call it a copy.
Now if there is evidence that the chinese haven't changed and improved the original scanner at all and DT30 is essentially the same, then thats another story and then please link me to some evidence so Ill make better decisions.

So, the main point is that tuning a scanner that was designed to go 30K to faster-than-30K will involve sacrifices. Just like clocking a CPU at a rate faster than it was meant to go will sacrifice something (like reliability, or at least convenience, since you'll have to go through a lot more trouble to keep the CPU cool).

[/quote]
Overclocking a CPU is risky, but if done right it works. If this point is about scanners like DT40, then I only see this as a thing to take into account for people building projectors. Not really a problem.

I have a set of those DT40 that I plan to replace with the new EMS-8000 (60k) from Eyemagic. So maybe if you think they are sufficient for you, it would be a great deal

Where can I learn more?