Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!

  • 119 Replies
  • 47878 Views
  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #80 on: 20 / January / 2008, 02:53:50 »
    Advertisements
    Deleted
    « Last Edit: 22 / April / 2008, 07:12:57 by Barney Fife »
    [acseven/admin commented out: please refrain from more direct offensive language to any user. FW complaints to me] I felt it imperative to withdraw my TOTAL participation. Nobody has my permission, nor the right, to reinstate MY posts. Make-do with my quoted text in others' replies only. Bye

    *

    Offline fudgey

    • *****
    • 1705
    • a570is
  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #81 on: 20 / January / 2008, 06:52:20 »
    Quote
    Interesting test! Could you post some downsized images to show how you determine this? I think others would find it interesting (as well as myself).

    Sure... see the attached images.

    First, 10pros.F2.6_zoom1x_Tseq.77.474us.jpg is an image taken using a fast sequence (to show you where all the LEDs really are). The bright one in the middle is a reference diode, it's powered on constantly and doesn't blink. To the left of the reference lamp there is another reference diode, which is invisible as it should be, since it's never powered. Note that not all LEDs are equally bright, this is mostly because of the fact that the LEDs are not ideally positioned and their viewing angle is about 30 deg.

    10pros.F2.6_zoom1x_Tseq.1007.16us.jpg shows a neat and clean measurement with four bright LEDs and  two dim ones: shutter was closed before the one on the left was done with it's blink time and the one on the right probably started to shine just before or while the shutter was opening.

    Quote
    Have you taken into account the LED's own cycle time? As with all devices, they all suffer from a "lag" of one sort or another. You might want to research online the response time of the particular LEDs that you are using. This could help you to determine a little more accuracy by taking their "fudge factor" into account.

    Yes, as I said above, I believe the LEDs to be fast enough. But this is the one thing I haven't actually measured. The manufacturer's data sheet promises a 20 ns typical speed of response (exponential time constant), from which I figured they would quite likely be faster than my dead time between consecutive LED blinks (1.3 us).

    Quote
    This is why testing for such high speeds is proving difficult. I still haven't pulled my old silly-scope out of storage (not only because it's -20 F. outside lately) because I am certain that the phosphor lag on the display would again prove useless for a test of this nature.

    I did some initial experiments using an old analog scope (not quite as old as yours I think, from the 80's), but the phosphor was way too laggy for any accuracy (a more cunning source signal setup than DC could've helped but I figured it wasn't worth the trouble).

    Quote
    One more thing. Have you tried using long-zoom settings and using the aperture override too, to see if you can obtain even faster speeds at f/11 (or higher?).

    I think I already said (at least I ment to :D) I didn't try any aperture overrides yet. By long-zoom do you mean that it would be more likely to get the camera go to F8.0 at full 4x zoom than to get past F8.0 at any zoom position?

    Quote
    Also, I'm not sure what you mean when you say, "Especially the shots taken with 100% image size suffer from significant shutter distortion." Since none of these cameras use a focal-plane shutter there are no shutter distortion effects. What are you referring to here?

    I sort of assumed the shutter would be a mechanical obstacle even if it isn't positioned in front of the sensor like it conveniently is positioned in cameras where you can change the lense. What led me to think so was the fact that the camera makes noises when taking pictures and that looking up the lense, there clearly is a shutter. But if that's there just for dark-frames and previews and the real shutter just an electrical signal to the sensor, I'm obviously mistaken.

    You'll have to forgive me for actually not knowing much of anything about photography or camera optics ;)

    In any case, from those initial results it definitely looks like I can capture the LED "motion" more accurately if I take the camera further ie only use the very center of the lense, instead of taking it up close and filling the entire photo. This made me believe that the thing I see in the lense indeed is the shutter, and that it is a hole that rapidly widens, stays open for a  defined time and then closes.

    Take a look at 100pros.F2.6_zoom1x_Tseq.1007.16us.jpg. It has only one or two bright LEDs, followed by four or three dim ones. But then there's a lonely LED on the upper left corner, which is clearly a little bit active. I'd call that shutter distortion.

    Quote
    (p.s. Have you thought about focusing with the room lights on first? Big Grin The more "point source" your lights the more easy it will be to have their light show up on the CCD's pixels and therefor be more accurate to tell when they are fully lit or only beginning or ending their ramping or decaying light cycle.)

    Actually I think I did that in most photos and locked the focus manually to prevent the camera from messing it up once it was dark. Focus wasn't perfect but not the biggest imperfection here.

    I really am assuming the LED response time is insignificant. One more thing making me thing it's not a LED lag issue is something that can be seen in the actual images: on images which have more than two dim LEDs, the unexpected dim ones are those that are supposed to shine next, not ones which have just blinked (except the one dim LED before the first bright one, of course). The further in the future their time to shine is, the dimmer they appear.

    This is shown in 100pros.F8.0_zoom4x_Tseq.296.22us.jpg: the brightest LED is in top-left, the one that was powered before it is on it's left and is very dim. Then there are about 10 LEDs which were powered after the brightest one, all decreasing in brightness in the photo.

    You've probably noticed I didn't actually include pictures of the fastest shutter speed measurement (F8.0 zoom 1x). That's because the LEDs aren't quite bright enough for crystal clear images at those speeds and I figured they wouldn't be good for  introductory purposes.

    *

    Offline fudgey

    • *****
    • 1705
    • a570is
  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #82 on: 20 / January / 2008, 09:15:32 »
    I did some more tests, now with aperture override with maximum shutter speed (1/100k Tv override), similarly to the previous measurements. I only shot with 10% image area since those results were easier to interpret earlier.

    I tried both ends of zoom (1x and 4x) with 3 different Av settings:
    F8.0 (no Av override, camera limit)
    F10.07 (Av override)
    F16.00 (Av override).

    For both zoom positions, the aperture override seems to work beyond F10, since F10.07 the LEDs are significantly dimmer than at F8.0 and at F16.00 they are significantly dimmer than at F10.07.

    For 1x zoom, the images are way too noisy to tell whether shutter speed changes with aperture (the number of visible LEDs does go down, but that's mostly because they disappear into noise; I'm already digging the LEDs from the image by increasing gamma). I need to look into using higher forward current for the LEDs to make them brighter. Using ISO1600 might help a bit as well(?).

    For 4x zoom, the shutter is slower, a slower blink sequence is used and consequently noise is less of a problem. It seems pretty clear that shutter speed increases with smaller aperture (larger Av).

    My shutter time limit estimates (for 10% central image area) for 4x zoom using aperture override on the A570IS are:

    F8.0:  220 us (1/4500 s)
    F10.07: 170 us (1/6000 s)
    F16.00: 75 us (1/13500 s)

    Note that the first (F/8) one should be the same as one of my previous measurements (that time I got 210 us, which is virtually identical). This time I took 10 photos per measurement and the results were pretty consistent. There is some room for interpretation of dim LEDs, though.

    Please note that I'm not saying F16 would be any different from F11, I only tested these three settings, and it seems F10 is not the limit at either zoom position.

    *

    Offline e2b

    • *
    • 37
  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #83 on: 20 / January / 2008, 09:46:34 »
    fudgey,
    You might be interested in these previous posts I made in another thread about shutter speed tests. They give results of some tests on relationships between shutter speed, aperture and focal length for A610 and S3.
    http://chdk.setepontos.com/index.php/topic,232.msg1707.html#msg1707
    http://chdk.setepontos.com/index.php/topic,232.msg1787.html#msg1787
    http://chdk.setepontos.com/index.php/topic,232.msg1790.html#msg1790

    I made some different experiments today, attempting to measure shutters speeds by taking pictures of a CRT display (1024 x 768, 60Hz refresh). I set the display to show a series of black and white lines, approximately 3 mm thick (on a screen 21 cm high), and sloping at about 1 degree. With both the A610 and S3 (at F8) exposure seems pretty exact at 16/100k sec (which produces just one single white streak with a length approximately equal to the width of the screen). At 8/100k sec the exposure seems a bit longer than the setting. At 4/100k sec the exposure appears much less than half the value at 8.100k, and there is no change for settings faster than 4/100k.

    I am still in some doubt about these results as they could be affected by persistence of the phosphors, but these results are very similar to what I had estimated with the completely different method based on observing shifts in the histogram peaks.



  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #84 on: 20 / January / 2008, 11:47:59 »
    Deleted
    « Last Edit: 22 / April / 2008, 07:13:35 by Barney Fife »
    [acseven/admin commented out: please refrain from more direct offensive language to any user. FW complaints to me] I felt it imperative to withdraw my TOTAL participation. Nobody has my permission, nor the right, to reinstate MY posts. Make-do with my quoted text in others' replies only. Bye

    *

    Offline e2b

    • *
    • 37
  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #85 on: 20 / January / 2008, 20:06:13 »
    e2b? I'm curious on why you would create a display of white and black lines at an angle like that? Does that make it easier to count the lines?

    Barney, as you supposed, the sloping lines are just to make it easier to count lines on the old low-quality CRT, which was never very sharp. I had previously made a few test pictures on a better-quality CRT display on which I could just distinguish lines, but even then I found a sloping line made it a bit easier (this CRT was not available yesterday). I think that lines would be easiest to distinguish on a CRT with a Trinitron tube (which almost makes me wish I hadn't given mine away!)

    Howver, there is also a problem with certain ways that CCDs are read-out. Some of them are read in linear banks or bands. Which, surprisingly, can cause focal-plane shutter distortions too in an electronic shutter.

    I gather that the current 'ideal' sensor (regarding shutter speed) is a 'Frame Transfer CCD', which dumps the whole set of sensor information into a duplicate storage area, from where it is read for processing. This kind of sensor doesn't need a mechanical shutter at all, and doesn't involve sequential reading from the main sensor, but is bigger and costs more.

  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #86 on: 20 / January / 2008, 23:22:47 »
    Deleted
    « Last Edit: 22 / April / 2008, 07:14:03 by Barney Fife »
    [acseven/admin commented out: please refrain from more direct offensive language to any user. FW complaints to me] I felt it imperative to withdraw my TOTAL participation. Nobody has my permission, nor the right, to reinstate MY posts. Make-do with my quoted text in others' replies only. Bye

    *

    Offline fudgey

    • *****
    • 1705
    • a570is
  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #87 on: 21 / January / 2008, 12:45:23 »
    Quote
    Fudgey, thanks for posting all that. MOST interesting. It's odd how at such fast speeds everyone is having a difficult time trying to pin-down exactly what is happening when. I sometimes even suspect it might have to deal with ambient camera temperatures affecting shutter speeds.

    I believe these extreme shutter speeds may very well be affected by many things. Actually there's one thing here I didn't mention earlier because I wanted to double check that my reference LED drive current is ripple free (it is, to a sufficient extent): out of 10 consecutive shots taken with the serial shutter on a stand, using extreme overrides, the reference LED brightness has noticeable variations shot to shot. I.e. the camera doesn't perform identically from shot to shot.

    Quote
    From what I understand, the shutters in these cameras are both electronic and aperture working together. There should be no lowering of light levels at the edges of the frame as the aperture opens and closes to expose the CCD (which in turn is also turned on and off rapidly in sync with the opening and closing of the leaf-shutter). This is why these types of shutters are so highly prized in the range-finder style of cameras that don't depend on the slow focal-plane shutter with all their inherent exposure problems of a narrow slit moving slow over the recording media. In a leaf-shutter system, the concentrically arranged blades of the shutter open and close concentrically, just like an iris-diaphragm (and indeed in digital cameras they are usually one in the same). This diaphragm/shutter assembly is situated in the optics so that as it opens and closes all light from the lens passes through it evenly. Meaning there should be no vignetting in any area of the exposed frame. And ... just as you can set your aperture for f/2.x or f/8.0 you will see that the full frame is evenly lit, no vignetting. So the same will hold true when it is also acting as a shutter.

    I'm still not quite convinced that just having a (cheap!) leaf shutter would remove distortion entirely if there aren't enough leaves. But if the shutter doesn't actually move while taking a photo, I'll have to agree of course.

    Quote
    Howver, there is also a problem with certain ways that CCDs are read-out. Some of them are read in linear banks or bands. Which, surprisingly, can cause focal-plane shutter distortions too in an electronic shutter. I've seen some images from digicams similar to the helicopter photo that I posted, where instead of the whole blade being distorted, an airplane's prop is cut up into small bands of curved distortion, as each band of the CCD is read out too slowly. It all depends on how the data is pushed off of the CCD as it is being read. I'll have to try to find that image online again to show others an example of this electronic shutter distotion (I may have even saved it to my hard-drive, I can repost it if I find it). I don't think this affects these cameras though. I've found no evidence of this in any high-speed images I've taken to date.

    This would indeed explain a lot...

    Quote
    Now, onto the interpretation of the partially lit LEDs ... this is indeed a head-scratcher. The first thought that came to mind is that at such high speeds, the capacitance of the wires themselves are causing the delay in the LED response time. As voltage is applied to a wire, it slowly ramps up in voltage until the wire's own capacitance has been reached and then the LED lights proportionately as the peak voltage happens. Then the reverse takes place when it is turned off. If those LEDs do indeed have such a short response time, then it can only be a capacitance problem that is causing the slowly lighting and dimming of the lights.

    My setup is low-impedance, the LEDs have 150 ohms in parallel to weed out effects of parasitics. I've measured the LED voltage and current and they seemed nice and sharp. Also, I measured the timing of consecutive lamps and they were as expected as well. Keep in mind that the requirement is actually just 1 us i.e. 1 MHz, dead slow for digital electronics :D.

    But to be extra sure, I took a photodiode I had lying around (SFH203P if you must know, specified at 5 ns rise/fall in certain conditions, so I knew it could be configured to not be the weak link). I measured a 200 ns rise/fall time from all 14 LEDs  and didn't care to dig in deeper to see how much faster than that they really are since that's quite a bit more than required already.


    Oh and LED alignment really is a pain, but not really an issue here since I can always take a shot where all LEDs get the same amount of blinks to show which ones are not pointed perfectly. Also, the LEDs obviously have significant manufacturing tolerances in brightness (+200%/-50% wouldn't be surprising, these aren't binned parts) so even perfect alignment is not going to perfect the setup. I just always need to use some consideration before jumping into conclusions.

    Quote
    I find it interesting that you aren't getting the higher shutter speeds that I suspect are happening in my S3. At such high speeds, I'm thinking that not only is it dependent on very minor differences in the camera, and what settings are used (as you have seen they are also f/stop dependent), but that each make and model of camera might be able to also affect what shutter speeds are available. I'm still convinced, by looking at histograms and other tests I've done that I can get at least 1/40,000th of a second, and perhaps even 1/64,000th of a second on the S3.

    To remind, I haven't actually found the very fastest speeds yet since my LEDs are still a bit too dim. Tripling the output is probably dead easy by increasing the parallel impedance of the LEDs (since now I know they can be slowed down without affecting the measurement and have the setup to recheck that easily), but I haven't gotten around to doing that yet.

    Quote
    Looking forward to seeing any other tests you might find (if you tackle it more). And thanks for posting what you have so far. It's nice to know that I'm not alone in trying to find an easy and definitive way to test exactly what is going on. If nothing else, I'm so glad to now have shutter speeds that will stop a hummingbird's wings without using flash. I'm an available light addict. Nothing I hate more than ruining an image by having to result to using flash. These new shutter speeds will open up nature-photography doors that I didn't even get to peek through in the past.

    Actually I already modified the LED blinker software a bit. It now blinks only 12 of the LEDs, in 3 groups of 4 LEDs. Each group forms a square, and one LED from each group is lit simultaneously (with what looks like sub-100ns sync as they are always in the same port of the uC). This way the blinker "draws" three rectangles in perfect sync (upper left corner LEDs first, then upper right LEDs, then bottom right, bottom left and back to top left etc).

    I haven't tested it with the camera yet (just 4 LEDs in the sequence may be too few to really get any results), but I'm hoping to get images of the three squares out of phase to shed some more light to this issue and to convince you this is real :P


  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #88 on: 23 / January / 2008, 18:47:59 »
    This are the results of my attempts to photograph a lighting of a cigarette lighter at high speeds. It's a shame that the sensitivity falls so low with speed - hence the noise. Funny thing is that i had to repeat the lighting on and on and so i got a blister on my thumb. I took about 300 pictures on continuous shooting mode. Only few of them actually hit the moment when anything interesting was happening. Hope you'll like it - try for yourselves !

  • Publish
    Re: Whoa! ****MAJOR**** High-Speed Shutter-Speed Discovery!
    « Reply #89 on: 24 / January / 2008, 01:23:09 »
    Deleted
    « Last Edit: 22 / April / 2008, 07:14:36 by Barney Fife »
    [acseven/admin commented out: please refrain from more direct offensive language to any user. FW complaints to me] I felt it imperative to withdraw my TOTAL participation. Nobody has my permission, nor the right, to reinstate MY posts. Make-do with my quoted text in others' replies only. Bye

     

    Related Topics