• Optics question (Fresnel equations)

    From Rhydian@21:1/5 to All on Thu Feb 10 15:02:56 2022
    Hi,

    I'm building an optical instrument that points a 850nm LED at a boundary between two materials at an oblique angle, and measures the (specular) reflection with a photodiode at the same (opposite) angle.

    The first few prototypes are working well but I want to compare the
    performance I'm getting with the theoretical limits. My starting point
    is the Fresnel equations, but the part I'm having trouble with is that
    they give separate results for the s and p polarizations. How do I
    combine the two into a total reflected power?

    As the incident angle approaches the critical angle for total reflection,
    both the s and p numbers approach unity, so clearly I can't just sum
    them, or take the vector sum, or I would get an answer greater than 1.
    Average? Use the highest of the two?

    I'm assuming here that the photodiode detector (Osram SFH2700) has a
    response that's insensitive to polarization, but happy to be corrected on
    this point.

    I have a copy of "Building Electro-Optical Systems" but there's clearly something I'm missing. Google is not much help either, it finds pretty-
    much exactly the same question (but for microwaves rather than IR) from
    two years ago, and no replies.

    TIA

    Rhydian
    (who should probably have paid more attention in electromagnetics classes
    30 years ago)

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Phil Hobbs@21:1/5 to Rhydian on Thu Feb 10 11:35:31 2022
    XPost: sci.optics

    Rhydian wrote:
    Hi,

    I'm building an optical instrument that points a 850nm LED at a boundary between two materials at an oblique angle, and measures the (specular) reflection with a photodiode at the same (opposite) angle.

    The first few prototypes are working well but I want to compare the performance I'm getting with the theoretical limits. My starting point
    is the Fresnel equations, but the part I'm having trouble with is that
    they give separate results for the s and p polarizations. How do I
    combine the two into a total reflected power?

    As the incident angle approaches the critical angle for total reflection, both the s and p numbers approach unity, so clearly I can't just sum
    them, or take the vector sum, or I would get an answer greater than 1. Average? Use the highest of the two?

    I'm assuming here that the photodiode detector (Osram SFH2700) has a
    response that's insensitive to polarization, but happy to be corrected on this point.

    I have a copy of "Building Electro-Optical Systems" but there's clearly something I'm missing. Google is not much help either, it finds pretty-
    much exactly the same question (but for microwaves rather than IR) from
    two years ago, and no replies.

    TIA

    Rhydian
    (who should probably have paid more attention in electromagnetics classes
    30 years ago)


    You just treat the two polarizations independently and add up the
    photocurrents when you're done.

    LEDs are pretty well unpolarized when you look at them from a distance.

    There are polarization effects with angle, due to the Fresnel
    reflections from the top surface. If the LED has a flat top facet,
    p-polarized light escapes better, so there's a tendency for the light to
    be somewhat radially-polarized. Textured surfaces and lensed packages
    smear that out pretty well, though, so to leading order your LED should
    be unpolarized.

    Thus, it's a good guess to assume the LED light has equal amounts of s-
    and p-polarized light. These don't interfere, so the total photocurrent
    is just the sum of the s and p photocurrents.

    Cheers

    Phil Hobbs

    --
    Dr Philip C D Hobbs
    Principal Consultant
    ElectroOptical Innovations LLC / Hobbs ElectroOptics
    Optics, Electro-optics, Photonics, Analog Electronics
    Briarcliff Manor NY 10510

    http://electrooptical.net
    http://hobbs-eo.com

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Tauno Voipio@21:1/5 to Rhydian on Thu Feb 10 18:52:38 2022
    On 10.2.22 17.02, Rhydian wrote:
    Hi,

    I'm building an optical instrument that points a 850nm LED at a boundary between two materials at an oblique angle, and measures the (specular) reflection with a photodiode at the same (opposite) angle.

    The first few prototypes are working well but I want to compare the performance I'm getting with the theoretical limits. My starting point
    is the Fresnel equations, but the part I'm having trouble with is that
    they give separate results for the s and p polarizations. How do I
    combine the two into a total reflected power?

    As the incident angle approaches the critical angle for total reflection, both the s and p numbers approach unity, so clearly I can't just sum
    them, or take the vector sum, or I would get an answer greater than 1. Average? Use the highest of the two?

    I'm assuming here that the photodiode detector (Osram SFH2700) has a
    response that's insensitive to polarization, but happy to be corrected on this point.

    I have a copy of "Building Electro-Optical Systems" but there's clearly something I'm missing. Google is not much help either, it finds pretty-
    much exactly the same question (but for microwaves rather than IR) from
    two years ago, and no replies.

    TIA

    Rhydian
    (who should probably have paid more attention in electromagnetics classes
    30 years ago)

    Building a refractometer?

    --

    -TV

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Rhydian@21:1/5 to Phil Hobbs on Thu Feb 10 23:13:29 2022
    XPost: sci.optics

    On Thu, 10 Feb 2022 11:35:31 -0500, Phil Hobbs wrote:

    Rhydian wrote:
    Hi,

    I'm building an optical instrument that points a 850nm LED at a
    boundary between two materials at an oblique angle, and measures the
    (specular) reflection with a photodiode at the same (opposite) angle.

    The first few prototypes are working well but I want to compare the
    performance I'm getting with the theoretical limits. My starting point
    is the Fresnel equations, but the part I'm having trouble with is that
    they give separate results for the s and p polarizations. How do I
    combine the two into a total reflected power?

    As the incident angle approaches the critical angle for total
    reflection,
    both the s and p numbers approach unity, so clearly I can't just sum
    them, or take the vector sum, or I would get an answer greater than 1.
    Average? Use the highest of the two?

    I'm assuming here that the photodiode detector (Osram SFH2700) has a
    response that's insensitive to polarization, but happy to be corrected
    on this point.

    I have a copy of "Building Electro-Optical Systems" but there's clearly
    something I'm missing. Google is not much help either, it finds
    pretty- much exactly the same question (but for microwaves rather than
    IR) from two years ago, and no replies.

    TIA

    Rhydian (who should probably have paid more attention in
    electromagnetics classes 30 years ago)


    You just treat the two polarizations independently and add up the photocurrents when you're done.

    LEDs are pretty well unpolarized when you look at them from a distance.

    There are polarization effects with angle, due to the Fresnel
    reflections from the top surface. If the LED has a flat top facet, p-polarized light escapes better, so there's a tendency for the light to
    be somewhat radially-polarized. Textured surfaces and lensed packages
    smear that out pretty well, though, so to leading order your LED should
    be unpolarized.

    Thus, it's a good guess to assume the LED light has equal amounts of s-
    and p-polarized light. These don't interfere, so the total photocurrent
    is just the sum of the s and p photocurrents.

    Cheers

    Phil Hobbs

    OK, thanks, makes sense now.

    The LED is an Osram SFH4050, the top surface is slightly frosted so as
    you say, hopefully I can just treat it as 50:50 split between s and p polarization.

    One piece of odd behaviour I did see with this LED - I assumed the output
    power would be roughly linear with current, and lose efficiency and tail
    off as the die heated up. But going up in 50 uA steps to about 5 mA (max
    is 100) there's a noticeable upward curve. At first I thought I'd
    somehow screwed up the photodiode amp, but I tested it on an Ophir Nova
    II and got the same results. I don't remember seeing this before with
    other LEDs.

    So long as the output power is long-term stable to within a few dB it
    won't matter (there isn't space for a monitor photodiode in the design).
    I will put a few of them on continuously for a few months, just to check.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Phil Hobbs@21:1/5 to Rhydian on Thu Feb 10 18:57:22 2022
    XPost: sci.optics

    Rhydian wrote:
    On Thu, 10 Feb 2022 11:35:31 -0500, Phil Hobbs wrote:

    Rhydian wrote:
    Hi,

    I'm building an optical instrument that points a 850nm LED at a
    boundary between two materials at an oblique angle, and measures the
    (specular) reflection with a photodiode at the same (opposite) angle.

    The first few prototypes are working well but I want to compare the
    performance I'm getting with the theoretical limits. My starting point
    is the Fresnel equations, but the part I'm having trouble with is that
    they give separate results for the s and p polarizations. How do I
    combine the two into a total reflected power?

    As the incident angle approaches the critical angle for total
    reflection,
    both the s and p numbers approach unity, so clearly I can't just sum
    them, or take the vector sum, or I would get an answer greater than 1.
    Average? Use the highest of the two?

    I'm assuming here that the photodiode detector (Osram SFH2700) has a
    response that's insensitive to polarization, but happy to be corrected
    on this point.

    I have a copy of "Building Electro-Optical Systems" but there's clearly
    something I'm missing. Google is not much help either, it finds
    pretty- much exactly the same question (but for microwaves rather than
    IR) from two years ago, and no replies.

    TIA

    Rhydian (who should probably have paid more attention in
    electromagnetics classes 30 years ago)


    You just treat the two polarizations independently and add up the
    photocurrents when you're done.

    LEDs are pretty well unpolarized when you look at them from a distance.

    There are polarization effects with angle, due to the Fresnel
    reflections from the top surface. If the LED has a flat top facet,
    p-polarized light escapes better, so there's a tendency for the light to
    be somewhat radially-polarized. Textured surfaces and lensed packages
    smear that out pretty well, though, so to leading order your LED should
    be unpolarized.

    Thus, it's a good guess to assume the LED light has equal amounts of s-
    and p-polarized light. These don't interfere, so the total photocurrent
    is just the sum of the s and p photocurrents.

    OK, thanks, makes sense now.

    The LED is an Osram SFH4050, the top surface is slightly frosted so as
    you say, hopefully I can just treat it as 50:50 split between s and p polarization.

    One piece of odd behaviour I did see with this LED - I assumed the output power would be roughly linear with current, and lose efficiency and tail
    off as the die heated up. But going up in 50 uA steps to about 5 mA (max
    is 100) there's a noticeable upward curve. At first I thought I'd
    somehow screwed up the photodiode amp, but I tested it on an Ophir Nova
    II and got the same results. I don't remember seeing this before with
    other LEDs.

    So long as the output power is long-term stable to within a few dB it
    won't matter (there isn't space for a monitor photodiode in the design).
    I will put a few of them on continuously for a few months, just to check.


    Depends on the device. There's normally a bit of a toe at low
    currents--in the low tens of microamps for normal display LEDs--but then
    it's pretty linear.

    If your LED is something unusual it might behave differently.

    Cheers

    Phil Hobbs


    --
    Dr Philip C D Hobbs
    Principal Consultant
    ElectroOptical Innovations LLC / Hobbs ElectroOptics
    Optics, Electro-optics, Photonics, Analog Electronics
    Briarcliff Manor NY 10510

    http://electrooptical.net
    http://hobbs-eo.com

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Rhydian@21:1/5 to Tauno Voipio on Thu Feb 10 23:15:57 2022
    On Thu, 10 Feb 2022 18:52:38 +0200, Tauno Voipio wrote:

    <snip>

    Building a refractometer?

    Sort of, it's a non-contact fluid sensor.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jan Panteltje@21:1/5 to news@rblack01.plus.com on Fri Feb 11 06:54:29 2022
    On a sunny day (Thu, 10 Feb 2022 23:15:57 -0000 (UTC)) it happened Rhydian <news@rblack01.plus.com> wrote in <su46bd$q1o$2@dont-email.me>:

    On Thu, 10 Feb 2022 18:52:38 +0200, Tauno Voipio wrote:

    <snip>

    Building a refractometer?

    Sort of, it's a non-contact fluid sensor.

    I have done that with utrasound (water height in a tank, time delay.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jan Panteltje@21:1/5 to All on Fri Feb 11 07:19:13 2022
    PS
    these days 3 for 9 USD on ebay (I have some)
    https://www.ebay.com/itm/255235085629

    Great fun to play with:
    http://panteltje.com/pub/wind_speed_by_differential_2_ebay_distance_meters_IMG_4891.JPG

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)