Connect with us

Converting a 200W discharge lamp video projector to LED

Discussion in 'Electronic Repair' started by N_Cook, Sep 27, 2013.

Scroll to continue with content
  1. N_Cook

    N_Cook Guest

    Assuming over-riding the opto couplers to falsely confirm to the system
    micro that the arc is struck and the lamp is lit (maybe requiring a
    delay) firstly, with the lamp ps disconnected.
    Then I intend in the first instance to buy 5x 1.2W white LEDs, 3500K, 20
    degree, for proof of concept. Assuming that sort of works then get
    perhaps 10 more, going down to 2700K or 3000K or perhaps (unlikely)
    4000K and lenses to colimate to 2 degrees. Set inside a reversed conical
    silvered glass of an ex-lamp to direct spillover light into the
    colourwheel/light tunnel aperture.
    The intended LEDs are 11x10mm footprint so can be mounted quite close to
    the colourwheel(for 5 anyway) on a spherical back mount. I may as well
    retain the original fans, perhapps knocked back a bit for less noise
    intrusion later on.

    When coming to scaling up I originally was thinking of using a sectored
    curvi-linear silvered reflector from PIR units (with faned air cooling)
    but have since come across 2 degree lens converters for these LEDs so
    may as well go with them and shine directly from a larger spherical
    backing mount, directly to the colourwheel aperture.

    I'd be interested in any suggestions or comments other than of the I
    would not bother type of replies. Anyone happen to know what the light
    wastage proportion is of a non-ideal paraboloid reflector and non-point
    source discharge lamp is? I'm aware proper LED projectors have active
    drives to RGB LEDs and not colour wheels but there are a lot of
    ink-jet-printer-syndrome surplus HD video projectors around with too
    expensive short-arc lamp costs to replace

    Some bods been here before with converting a couple of types of
    discharge lamp converters

    If anyone is interested, my exploration of inside a standard domestic
    GU10 LED lamp (to see if they were all in series or mixed series/parallel)
    Breaking in:- hold the bulb in a glove and heat the dome cover with
    "low" temp hot air and prize off with a needle. With old soldering iron
    destroy the epoxy join between the , not obvious as silvered, pcb to the
    lamp housing. The slight greeen colouration is due to the reflection of
    the green dye of the pcb which is not glass fibre reinforced it seems,
    maybe epoxy substrate only. Prize the pcb away.
    No glass breakage at any stage.
    Overlay of this one marked JH-GU10-20
    HV ac side 1M//0.33uF 400V dropper and 1/4W resistor size
    fuse/inductor/fuseable resistor? pink colour with red black brown, or
    reverse order, colour bands, about 0.4R to small SMD MB6S bridge
    LV quasi-DC side SMD 510R dropper to 20 LEDs in series.
    White ceramic cap is cemented to the glass of the lamp.
    Bench ps 50V across LED string and 510R all LEDs just lit
    54V and 0.5V over 510R and about 2.6V over each LED some sort of low
    level brightness.
    With 75% mains (240V that is) 6.5V DVM dc over 510R
    or 5.5V DVM ac over 510R
    100% mains 9.1V "DC" or 7.2V as "AC" reading over the 510R
  2. Have you calculated whether you can cram enough LEDs into that space to get
    the same brightness level as the discharge lamp? Or are you expecting to learn
    from the test?

    I'd like to point out that is no such thing as -- nor can there be -- a white
    LED. LEDs are necessarily limited to a narrow band of wavelengths. * All (???)
    white LEDs are (I assume) a blue LED with a yellow-fluorescing phosphor. **

    This /looks/ white to the eye, but the red and green wavelengths needed for
    color reproduction aren't present.

    Unless your white LEDs contain red, green, and blue LEDs, I don't think this
    is going to work.

    * This is actually a good thing if one is trying to match a specific color
    space, and you can manufacture LEDs whose wavelengths correspond to the three

    ** I'm thinking of indicator lights and such. Lamps to replace incandescent
    lighting would necessarily have to put out red and green, or colors wouldn't
    look right.
  3. N_Cook

    N_Cook Guest

    Its not possible to do any calculation as to brightness because it is
    impossible to find the true loss of light from a conventional
    distributed ie not point source discharge lamp source and non ideal
    reflector, I would guess that 2/3 of the rated light does not get into
    the apaature and then how to calculat the proportion that is at such an
    angle to the light tunnel to the active chip and multiple reflections
    that little of that gets to where it is wanted, axially along the 2
    inches or so of narrow diameter light pipe .

    As for colour rendition it is unlikely to be any worse than the current
    situation of having to place a rose-pink filter over the projector lens
    to get some red into the image as the lamp must be too far into the blue
    end of the spectrum tio be compensated for in the setable timing of the

    So in both cases very much a suck it and see, seat-of-ones pants
    situation, but worth a go, if you've seen the price of these supposed
    replacement discharge lamps
  4. N_Cook

    N_Cook Guest

    I've had another look at the product data and the 2 degree lenses are 4
    degree , in normal terminology

    I don't know what chromaticity means but for the 3500 deg K version a Cx
    of about .4 and Cy of about .39
    A bit more graphic the spectrum is continuous and smooth "bell curve"
    peak shifted 40nm from 550nm of the standard eye response curve to 590nm
    and the 50% points broader apart at 150nm compared to 100nm of the eye
    and a 50% down peak at 460nm which I suppose is the potential bugbear
    for such a lamp conversion
  5. "N_Cook" wrote in message
    I don't know what chromaticity means but for the 3500 deg K version a Cx
    of about .4 and Cy of about .39
    A bit more graphic the spectrum is continuous and smooth "bell curve"
    peak shifted 40nm from 550nm of the standard eye response curve to 590nm
    and the 50% points broader apart at 150nm compared to 100nm of the eye
    and a 50% down peak at 460nm which I suppose is the potential bugbear
    for such a lamp conversion

    Could you send me the data sheet, or its URL? I'd like to take a look.
  6. N_Cook

    N_Cook Guest

    LEDs, I did not find a uk/us URL
  7. Could you send me the data sheet, or its URL? I'm not an expert on this subject (though I do know a little). My gut reaction
    is this...

    The color reproduction index is only 80. That's poor. That doesn't mean that
    the LED won't work -- but it will probably need appropriate filtering that
    won't be easy to achieve.

    The spectral emission (p11) isn't particularly flat -- and definitely not
    smooth -- no doubt one of the reasons for the poor CRI.

    I don't think you're going to get pleasing results.
  8. N_Cook

    N_Cook Guest

    All the discharge lamps in video projectors , I/ve seen have an
    electrode and squashed , not optically conductive glass, axial exactly
    in line to where you want the light to go.
    With directional 20 degree central LEDs, and lensed 4 degree ones
    off-axis. will direct most of the energy directly to the half inch
    aperature , without any reflectors . Reflectors just to mop up spill over .
    I can see some high power red LEDs being added to counter the blue sub=peak
  9. Phil Allison

    Phil Allison Guest

    "William Sommerwanker"

    ** As fucking usual, the Somerwanker fool just makes stuff up.

    Anyone can Google "white led" and get the facts.

    There is plenty of green orange and red in the light from regular white

    ..... Phil
  10. "Arfa Daily" wrote in message
    You've never seen a '70s RCA set? Brown was about the only color it /could/
    produce (along with some blues and yellows, if I recall correctly).

    Brown is actually a very dark red.
  11. N_Cook

    N_Cook Guest

    Have you tried the video projector use rather than Gobo type things
    where you are not trying to squeeze light along a small aperature light
    tunnel? I've not found the data out there but I suspect the vast
    majority of the light available to a gobo setup just does not get into a
    light tunnel setup , so if you can direct all your LED lamps into that
    tunnel then the overall requirement is much lower than normal ratings
    would suggest.
    The use of this video projector is for text and graphics so colour
    rendition of the likes of flesh tones is not too critical, very rarely
    showing any video as such.
  12. N_Cook

    N_Cook Guest

    Intended use in the main part is for projecting text and graphics so as
    long as there is a colour difference, any colour difference almost,
    rather than correct colour rendering that is all that is required , a
    rare pic with a green flesh tone or something does not really matter too
  13. N_Cook

    N_Cook Guest

    I've moved away from thinking about using lenses as they restrict the
    footprint size. If I double stack the LEDs then a spacing of centres
    about 8.5mm is possible and so 14 LEDs in a 32mm diameter. So can be
    quite close to the light tunnel and 20 degree beam spread is fine and
    outer LEDs approach angle will still only be 30 degrees or so, and a
    reasonable proportion of that will reflect only a few times and
    substantially get to the DLP chip. Will still try 5 LEDs initially.
    Will power up individually at only 100mA or so and set the angles of
    each individual LED for maximum brightness at a simulated window with a
    photodiode , then wire all in series and make a more substantial back
    mount before transfering to the projector
  14. **What a waste of time and effort. The best LEDs are approximately
    similar efficiency to that of halide lamps. As others have stated, the
    big problem will be that you are substituting a compact light source
    with a rather diffuse one. The optics are not designed for such use.
  15. N_Cook

    N_Cook Guest

    But the lamps used in video projectors are ,in effect, not compact. The
    direct light path from the reasonably compact source is blocked by an
    electrode and non optical structural glass lump, relying on the mirror
    surface of the light tunnel / light pipe to average out/balance-up the
    light coming in at all sorts of angles, off the parabaloid reflector
  16. "Arfa Daily" wrote in message
    I'm not sure about that.

    When I said "very dark", I meant having a low value. The chromaticity diagram
    does not include value -- only hue and chroma. "Brown" is how the eye
    interprets reds of low value.
  17. **NO. By the time you try to shove 200 Watts of LEDs into the enclosure
    (including apprpriate heat sinking) You're not going to be able to focus
    the whole thing properly. It's a daft idea, unless you are prepared to
    use MUCH less LED power (say 15 Watts) and a consequent huge drop in

    BTW: The light from a parabolic reflector does not come off at all sorts
    of angles.
  18. N_Cook

    N_Cook Guest

    That Osram with built in lens is discontinued, only came out 3 years ago.
    I'm getting 7 of the LCW W5SM, white 2700K 120 degree and matching 7 off
    6 degree hex shaped lenses.
    The off the shelf 7 cell hex cluster lens is for parallel , not focused.
    Using a 5 to 6 inch ball as a jig I'll combine the 7 as a focused cell.
    With the lenses at 85% transmission, brings the 75 lumen per LED down to
    64 lumen, so x 7 =450 lumen. Some heatsinky type protrusions added to
    the LEDs to catch the fanned air.
    When new the projector was rated 2000 lumen, although still taking 200W
    , the light output is now much less. Combined with the poor optics of
    these lamp setups I suspect fully directed into the light pipe 450 lm is
    not much different to the present discharge lamp situation. Still plenty
    of room to add another 6 plus lenses around the periphery to bring up to
    900 lm.
    Playing around with a scrapped colour wheel dicroic disc and assorted
    white LEDs the red transmission , to eye anyway , is a lot lower than G
    and B.
    So I will get a number of red 100mA 5mm , 15 degree standard size LEDs
    to add red, perhaps 6 at the interstices of the lenses, mounted to the
    rear and between the SMD LEDs plus maybe another 6 around the periphery
    to infill.
    Good progress with the silvered "cone" for mop-up, front of PAR lamp
    removed and bulb removed, about 3/4 way through grinding through the
    thick glass with cintrided disc, to remove the barrel part. Previous
    attempt with thinner glass photoflood failed. If I was brave or had a
    load of these sort of lamps, I would try the old bottle cutter routine,
    freezer spray and a ring of "fuse" wire around , fired up for the cut
  19. N_Cook

    N_Cook Guest

    Well that was very successful, a job I've never done before. Not a cone
    but the nearest I could find with
    a good silvered internal surface and right sort of dimensions. The
    thinnest part of the glass 4.5mm thickening to 7mm as 12 flutes around
    the stem, but a neat ground-glass cut.
    So I have a spillover reflector with 70mm internal diameter available
    for the LED assembly, down to 21mm diameter opening for the colour wheel
    aperature (from memory about 12mm) and 43mm axially.
    It will be a couple of weeks before I can get some time to convert the
    projector . But in the meantime anyone any ideas how to measure the
    intensity of the LED assembly at the axis and focus of the array, with
    any sort of accuracy (no known "standard candles" etc in my
    possession)?, for anyone else coming down the same path. My uncalibrated
    luxmeter , I doubt goes that high plus probably would melt, as would any
    of the plastic neutral density filter I have and placed in the path of
    that beam
  20. Guest

    How about some photos of what you are doing, most of what you are describing is WAY over my knowledge of optics and lenses.
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day