Page 2 of 2 FirstFirst 12
Results 11 to 15 of 15

Thread: Advanced Knife Edging

  1. #11
    Join Date
    Feb 2008
    Location
    Northern Indiana
    Posts
    921

    Default

    This thread has got me really excited. So, I did a few simple tests with stuff I had around. First test was putting the beam of a m140 through a 30deg prism and shinning it on a target ~50 feet away. I measured the location of the spot a .5A then 1A then 1.5 amps. The the spot moved 3/16 of an inch from .5 to 1A then another 3/16 from 1 to 1.5A . This shows the wavelength is changing with current. However the diffraction amount is very small. So, I am not sure how, using a prism, we could get enough diffraction to get the beams to overlap in a small distance?

    Next I tried a diffraction grating with too different m140's. The wavelength of these two were very different when measured at the same current. i didn't calibrate the grating with a HeNe so I don't know exactly what the WL were on the m140's. But it proves there is a significant deviation on the m140's.

    Planters, is it possible the variation you see after the prisms is more due to the different angles the beams are entering the prism than due to different wavelengths? If your combiner is set to overlap the spots at a distance all the individual beams would be converging at some small angle before the prism.

  2. #12
    Join Date
    Feb 2011
    Location
    New Hampshire
    Posts
    3,513

    Default

    Firstly, the shift of refraction angle with wavelength is the fundamental property of the prism. I mean that's what they do. I believe krazer's graph and your results passing the single beam thru the prism verifies this effect even though your distance isn't calibrated. I looked into this about a year age for the red diodes and decided to stay away from diffraction gratings because of the losses that are unavoidable from the energy that is spilled into the higher orders (minimal 25%)

    What you might also confirm is that if you reduce the refraction angle to say 1/2 the value of your first test, by rotating the prism, then does the movement with current also decrease by 1/2? I think it will. If I am correct then the problem of current vs redirection of the beam can be reduced to an arbitrarily low level. What needs to be determined is how many of the diodes can be overlapped on the prisms before the required divergence exceeds the acceptable limit. I am overlapping three and will check the shift in the far field with current. By selecting diodes from my "pool" I find a maximum of 11nm ( but range from 1nm up to that maximum ) at the same current levels and this is due to a fixed property associated with each diode. This is similar in magnitude to the current dependent wavelength CHANGE.

    Planters, is it possible the variation you see after the prisms is more due to the different angles the beams are entering the prism than due to different wavelengths? If your combiner is set to overlap the spots at a distance all the individual beams would be converging at some small angle before the prism.
    I am trying to understand this process, but I think that if we eliminate the current effect for the purposes of this argument then starting with identical diodes with identical wavelengths that are adjusted to converge on some distant point on a far field projection screen they will have a steadily decreasing spread along this path. If we intercept this array along this path and expand the width of the array by say a factor of two with a prism pair in order to reduce the far field divergence then the convergence angle from the prisms onward will have to double to converge at the same distance. Now if the prism pair is adjusted so that there is some residual dispersion then these identical diodes will still converge at the same distance, but be laterally displaced (obviously). The width at the prisms remains unchanged. If however, the wavelengths of the diodes are different then the dispersion will cause the spots at the far field distance to no longer co align and to correct this the beams out of the diodes will have to be redirected slightly. If selected correctly then the diodes on each side of the centroid at the prisms will need to be redirected toward the centroid and so reduce the width of the array. If not trying to pack in a ton of diodes then the dispersion caused by the prism can be set low and the effect of wavelength shift will also be low and maybe low enough so that the current effects can be ignored.

  3. #13
    Join Date
    Feb 2008
    Location
    Northern Indiana
    Posts
    921

    Default

    i think we are thinking of several different thing here. I'm a visual person and not the most articulate writer. So, I made a win paint to show what Im thinking
    Click image for larger version. 

Name:	prism.png 
Views:	50 
Size:	16.1 KB 
ID:	35936

    The top illustration is what I thought your setup may have looked like. The three beams are set to overlap at some far distance. I exaggerated this in the picture to show the possible reason for the large difference in diffracted (i think refracted is the proper term?) angles.

    Next is the the concern that the change in current and respective WL change will "throw off" the far field spot. This should not be a problem with two opposite prisms because the small change (greatly exaggerated in the middle illustration) is " self corrected" to some extent. When one prism refracts less the second does also. So, the output of the prism has a small lateral change (also exaggerated) but the spot should stay "close" at the far field. And this is evident by those using prisms on the blue in an RGB setup not losing alignment at different Diode currents.

    The bottom illustration shows my experiment with a single prism at different currents. This shows what krazer measured, increasing WL with increasing I. If we go by his graph the difference should be around 5nm. The total distance translated in my experiment was ~ 3/8inch from .5 - 1.5A at a 50 foot distance from the prisms. So even if we increase the angle of the prism it seems as though the diodes would have to be very far from the prism to get overlap.

    Not sure... am i missing something?

  4. #14
    Join Date
    Feb 2011
    Location
    New Hampshire
    Posts
    3,513

    Default

    Thanks for the illustration, but your description was good and I did understand it the way your illustration presents it. When the prism pair is used as it is normally"intended" the compensating angles leave the in/out beams parallel and the effects of dispersion are compensated and the only effect of a wavelength shift will be a miniscule change in the magnification ratio and of no concern. My set up allows the user to set a variable net dispersion. I agree with your second paragraph and would add that the spots would stay very, very close. The question is how much residual dispersion can be permitted before the shift becomes problematic. ANY allowable dispersion will permit the beams out of the diodes to be packed tighter. My set up is pretty big and the baseline to the prisms is a little longer than average and so the effect of any enhanced convergence onto the prism will be greater than average and this helps also.

    What is the refraction angle between the in/out beams in your experiment (in my case I measured it to be about 8 degrees)? Can you try my suggestion and see what happens when you cut this angle to say 1/2 of its present value? In any case, 3/8 of an inch at 600 inches is 0.6mrad and this is on the order of the divergence from a pretty good projector; especially if you are talking about the TOTAL, three color, overlapping RGB divergence. Also, what happens to the individual spot dimensions of the corrected beam at 50 feet? As the diode recruits additional multimode regions with increasing current, the spots tend to swell and how does this compare in magnitude? If the shift can be limited to a level that leaves it within the overlapping smear that the swelling and contracting three colors present in the far field then you may be able to add more diodes to the array by taking advantage of their inherent wavelength variation and combining them in the order that the dispersion compensates.

    Is this helping or am I just making this muddier?
    Last edited by planters; 12-01-2012 at 09:02.

  5. #15
    Join Date
    Feb 2008
    Location
    Northern Indiana
    Posts
    921

    Default

    Ok i did this quick before leaving for work. (have to be to work in 15min)

    i started with the angle at 30deg. At 50' the spot was just under .5 inches wide at .5A. At 1.5A the spot probably got a little bigger but was hard to measure due to the brightness. It did however move over ~1/8 inch.

    I wanted to set the angle to 15deg but the min i could do with this prism angle and ref index is 20deg. At .5A the spot was just under .75 inches. At 1.5A the spot was just over .75inches. It moved approximately 3/16 measured on one side and ~1/8 on the other. So i guess the deviation would be the average of 3/16 and 1/8.

    So on first look its not behaving like we would think. The magnification is ~2/3 less, but the ratio of how much it moved(refracted) is close to the same. Actually a bit more. I'm scratching my head.

    tomorrow when i have more time i will try to measure at 100 feet and 20 and 40deg deviation on the prisms. Also, I will use calipers to try and get more accurate data.


    I am still missing the big picture here I think. With such a small deviation in refraction how much can we expect to gain in beam overlap with a prism?

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •