on chip spectrometer?...

[whit3rd]

On Sunday, October 23, 2022 at 4:37:32 PM UTC-7, John Larkin wrote:

On Sun, 23 Oct 2022 15:44:10 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:
On Sunday, October 23, 2022 at 2:19:55 PM UTC-7, John Larkin wrote:
On Sun, 23 Oct 2022 15:42:32 -0400, Joe Gwinn <joeg...@comcast.net
wrote:

One can cobble something together with a replica grating and a silicon
photo detector array of some kind.

.<https://www.sargentwelch.com/store/product/8885837/replica-diffraction-gratings

How wide a spectral range can a grating cover before things get
ambiguous?

Huh? If you just want to distinguish 850 from 1350 nm, that\'s no problem.
The $1 replica that looks like a 35mm slide will do it fine. You can worry about
blaze angles and UV transmission (the slide is a transparency, but not to UV) and
line spacing fineries, but why?

I was thinking that a grating could fire into several detectors, each
with a different spectral range, and ...

Or, you could swing a detector over a range of angles and register a \'hit\'.
It\'ll take only seconds, why bother with a computer analysis?
I don\'t think a grating will cast a wavelenght-linear unambiguous
image over a wide spectral range, like 5:1 or so.

I think the 700 nm lines will wind up on top of the 1400\'s. And
probably much worse.

https://tinyurl.com/ext8r82d

\"The diffracted beams of different colors and corresponding to
consecutive orders can overlap, this phenomenon becomes more likely to
grow in the order of diffraction.\"

So what? If you\'re talking about a monochromatic (laser) as a source,
there aren\'t any different colors. Is it room lighting that you would find
disturbing? Scanning the angles, starting at zero degrees (zero order), your first peak is
it at order #1, and isn\'t subject to confusion.

 

[Martin Brown]

On 24/10/2022 00:37, John Larkin wrote:

On Sun, 23 Oct 2022 15:44:10 -0700 (PDT), whit3rd <whit3rd@gmail.com
wrote:

On Sunday, October 23, 2022 at 2:19:55 PM UTC-7, John Larkin wrote:
On Sun, 23 Oct 2022 15:42:32 -0400, Joe Gwinn <joeg...@comcast.net
wrote:

One can cobble something together with a replica grating and a silicon
photo detector array of some kind.

.<https://www.sargentwelch.com/store/product/8885837/replica-diffraction-gratings

How wide a spectral range can a grating cover before things get
ambiguous?

Huh? If you just want to distinguish 850 from 1350 nm, that\'s no problem.
The $1 replica that looks like a 35mm slide will do it fine. You can worry about
blaze angles and UV transmission (the slide is a transparency, but not to UV) and
line spacing fineries, but why?

I was thinking that a grating could fire into several detectors, each
with a different spectral range, and ...

Or, you could swing a detector over a range of angles and register a \'hit\'.
It\'ll take only seconds, why bother with a computer analysis?

I don\'t think a grating will cast a wavelenght-linear unambiguous
image over a wide spectral range, like 5:1 or so.

But if you have a monochromatic emitter you will know the wavelength to
within a factor of 2 or 4 so it should be fairy obvious except possibly
with the odd frequency doubler (like 1066, 533, 266 or 800, 400).

OTOH any of the visible ones you can pretty much identify by eye.

I think the 700 nm lines will wind up on top of the 1400\'s. And
probably much worse.

https://tinyurl.com/ext8r82d

\"The diffracted beams of different colors and corresponding to
consecutive orders can overlap, this phenomenon becomes more likely to
grow in the order of diffraction.\"

Stick to first order and interpose a long pass or high pass filter to
eliminate the ambiguity then. The trick is described here:

https://www.edinst.com/blog/second-order-diffraction/

They call it an order sorting wheel in fluorescence spectroscopy.
(which sounds a bit Harry Potter to me).

--
Regards,
Martin Brown

 

[Martin Brown]

On 23/10/2022 16:15, John Larkin wrote:

On Sun, 23 Oct 2022 14:27:56 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

They are seriously nice pieces of kit. PE did an atomic
absorption/emission spectroscope using a similar configuration and early
CCDs back in the 1990\'s. Must have been ~95 because I saw it in Japan at
one of the big analytical trade fairs where we were also exhibiting.

The spectrometer business seems to be a race for resolution in narrow
bands. There\'s no wide-range low-resolution stuff that we can find.

What are you roughly trying to do? You can\'t go too far either side of
the visible band without running into problems with glass transparency.

Something like a grating and a bunch of detectors could work. It would
have a lot of wavelength overlap confusion which could be mostly
computed out.

That\'s why they use one high dispersion and one low dispersion at 90
degrees it maps what would be a line spectrum onto a rectangle.


--
Regards,
Martin Brown

 

[Martin Brown]

On 23/10/2022 22:15, John Larkin wrote:

On Sun, 23 Oct 2022 12:34:11 -0700 (PDT), Rich S
richsulinengineer@gmail.com> wrote:

On Sunday, October 23, 2022 at 7:04:21 PM UTC, whit3rd wrote:
On Sunday, October 23, 2022 at 10:49:29 AM UTC-7, John Larkin wrote:

We buy all sorts of lasers and LEDs and we can\'t be sure they are the
right wavelength. Even 1% wavelength resolution would be plenty.

Nobody makes it.
[snip}

would this work? Handheld, ca. $1500,
https://www.intl-lighttech.com/products/ilt350-chroma-meter

That covers the visible, which would check the colors of LEDs, which
is not really much of a problem.

The more serious problem is that we buy a bunch of lasers in the 800
to 1550 sort of range and we\'d like to make sure they are right.

I\'m assuming for the moment that you are serious.

Semiconductor laser lines tend to be on *very( specific wavelengths.
(as do all the common laser and excimer gas mixes)

We tended to be on 1066, 533, 266 ie NdYAG + doublers.

I always found the amount of visible green in the beam a bit worrying
and the promise that the perspex enclosure was entirely opaque to 266nm.
I never spent much time in close proximity to the systems when running.

Low pass filters plus a few detectors should be good enough to
discriminate between most of them. I\'d have thought that for a known
part number whether or not it emits laser light or not would be good
enough. I vaguely recall an amateur astronomers tunable visible
wavelength dichroic filter (I know someone who bought a prototype) but
it seems to have sunk without trace. I\'ll ask next time I see him.

Otherwise try Young\'s slits to figure out the wavelength.

--
Regards,
Martin Brown

 

[Martin Brown]

On 23/10/2022 22:10, John Larkin wrote:

On Sun, 23 Oct 2022 12:04:17 -0700 (PDT), whit3rd <whit3rd@gmail.com
wrote:

On Sunday, October 23, 2022 at 10:49:29 AM UTC-7, John Larkin wrote:

We buy all sorts of lasers and LEDs and we can\'t be sure they are the
right wavelength. Even 1% wavelength resolution would be plenty.

Nobody makes it.

I don\'t know what \'all sorts\' means, but for red visible and most IR, a silicon
photodiode is a good detector. A grating (those start at a dollar or so)
and a protractor will complete the ensemble. You already have a milliammeter, I trust.

\"Don\'t be a jerk. Nobody likes jerks.\"

He isn\'t that far off the mark though.

Your choices for discriminating between various LED laser emitters might
be met by using one of each device in a 5x5 grid and illuminating them
with the DUT. Sum the current that they produce with a opamp and with a
bit of cunning you should be able to approximately get the wavelength
from the actual current it produces. That\'s one device per 50nm.

Only the diodes emitting the same or lower energy photons will generate
photo electrons when so illuminated.

You might get away with just one example of each of the common laser
diodes. I assume the problem is to check that anonymous black plastic
blobs are in fact the emitting the right wavelength ones.

The other option would be one or more photo detectors and various Schott
of Hoya low pass filter glasses available from various dealers.

I recall a tunable visual band dichroic interference filter intended for
amateur astronomers from a few years back that was quite impressive but
I think only covered 400-700nm. I know someone who has one next time I
see him I\'ll ask for the spec. I don\'t think they ever took off.

Not sure you really need to do the visible ones since there are not all
that many semiconductor laser lines possible in practice.

--
Regards,
Martin Brown

 

[Phil Hobbs]

Martin Brown wrote:

On 23/10/2022 09:38, Jan Panteltje wrote:
On a sunny day (Sun, 23 Oct 2022 09:13:53 +0100) it happened Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote in <tj2t42$1jlg$1@gioia.aioe.org>:

On 23/10/2022 06:39, Jan Panteltje wrote:
on chip spectrometer?
   https://www.sciencedaily.com/releases/2022/10/221020140615.htm

Possibly. I\'d like to see a bit more of the specifications and light
intensity it requires before I take that press release at face value.

There is a bit more here but the main article is behind a paywall

https://www.science.org/doi/10.1126/science.add8544

I see.
Well, CCD sensor with prism in front of it should work too?

The best super high resolution systems use an echelle method modest
dispersion prism one way and a very high dispersion grating at almost 90
degrees to it so as to map a linear spectrum onto a 2D rectangular CCD.

https://solarsystem.nasa.gov/resources/390/the-solar-spectrum/

That example was actually observed with a Fourier transform method and
then displayed in the fashion of a traditional echelle spectrum. It is a
very impressive piece of kit even it it only works on bright stars:

https://www.jstor.org/stable/26660057#metadata_info_tab_contents

This is a real physical highres echelle spectroscope

https://www.shelyak.com/le-woppshel-un-spectro-echelle-a-grande-resolution/?lang=en


They are seriously nice pieces of kit. PE did an atomic
absorption/emission spectroscope using a similar configuration and early
CCDs back in the 1990\'s. Must have been ~95 because I saw it in Japan at
one of the big analytical trade fairs where we were also exhibiting.

Yeah, a cross-dispersed echelle system can get resolving powers
(lambda/FWHM) up near 1E6, which is pretty impressive.

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

 

[John Larkin]

On Mon, 24 Oct 2022 10:36:41 +0100, Martin Brown
<\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

On 23/10/2022 16:15, John Larkin wrote:
On Sun, 23 Oct 2022 14:27:56 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

They are seriously nice pieces of kit. PE did an atomic
absorption/emission spectroscope using a similar configuration and early
CCDs back in the 1990\'s. Must have been ~95 because I saw it in Japan at
one of the big analytical trade fairs where we were also exhibiting.

The spectrometer business seems to be a race for resolution in narrow
bands. There\'s no wide-range low-resolution stuff that we can find.

What are you roughly trying to do? You can\'t go too far either side of
the visible band without running into problems with glass transparency.

Test various lasers to see if they are the right wavelength, as in
1550 vs 1310 vs 850 vs 800.

We can gross separation with a fiber WDM splitter and several
detectors, but that\'s klunky.

 

[John Larkin]

On Mon, 24 Oct 2022 10:24:58 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Martin Brown wrote:
On 23/10/2022 09:38, Jan Panteltje wrote:
On a sunny day (Sun, 23 Oct 2022 09:13:53 +0100) it happened Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote in <tj2t42$1jlg$1@gioia.aioe.org>:

On 23/10/2022 06:39, Jan Panteltje wrote:
on chip spectrometer?
   https://www.sciencedaily.com/releases/2022/10/221020140615.htm

Possibly. I\'d like to see a bit more of the specifications and light
intensity it requires before I take that press release at face value.

There is a bit more here but the main article is behind a paywall

https://www.science.org/doi/10.1126/science.add8544

I see.
Well, CCD sensor with prism in front of it should work too?

The best super high resolution systems use an echelle method modest
dispersion prism one way and a very high dispersion grating at almost 90
degrees to it so as to map a linear spectrum onto a 2D rectangular CCD.

https://solarsystem.nasa.gov/resources/390/the-solar-spectrum/

That example was actually observed with a Fourier transform method and
then displayed in the fashion of a traditional echelle spectrum. It is a
very impressive piece of kit even it it only works on bright stars:

https://www.jstor.org/stable/26660057#metadata_info_tab_contents

This is a real physical highres echelle spectroscope

https://www.shelyak.com/le-woppshel-un-spectro-echelle-a-grande-resolution/?lang=en


They are seriously nice pieces of kit. PE did an atomic
absorption/emission spectroscope using a similar configuration and early
CCDs back in the 1990\'s. Must have been ~95 because I saw it in Japan at
one of the big analytical trade fairs where we were also exhibiting.


Yeah, a cross-dispersed echelle system can get resolving powers
(lambda/FWHM) up near 1E6, which is pretty impressive.

Cheers

Phil Hobbs

How would you do a small spectrometer that works from 750 to, say,
1900 nm? Better yet 350 to 1900.

(Mo just gave me a bybye hug. She says hi.)

 

[John Larkin]

On Mon, 24 Oct 2022 10:37:39 +0100, Martin Brown
<\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

On 23/10/2022 22:15, John Larkin wrote:
On Sun, 23 Oct 2022 12:34:11 -0700 (PDT), Rich S
richsulinengineer@gmail.com> wrote:

On Sunday, October 23, 2022 at 7:04:21 PM UTC, whit3rd wrote:
On Sunday, October 23, 2022 at 10:49:29 AM UTC-7, John Larkin wrote:

We buy all sorts of lasers and LEDs and we can\'t be sure they are the
right wavelength. Even 1% wavelength resolution would be plenty.

Nobody makes it.
[snip}

would this work? Handheld, ca. $1500,
https://www.intl-lighttech.com/products/ilt350-chroma-meter

That covers the visible, which would check the colors of LEDs, which
is not really much of a problem.

The more serious problem is that we buy a bunch of lasers in the 800
to 1550 sort of range and we\'d like to make sure they are right.

I\'m assuming for the moment that you are serious.

Would anyone joke about that? We buy thousands of pcb mount fiber
lasers per year, most of them at four wavelengths and the occasional
oddball, and they all look alike. I\'d like to final test our products
and be sure we are shipping the correct wavelength.

Semiconductor laser lines tend to be on *very( specific wavelengths.
(as do all the common laser and excimer gas mixes)

We tended to be on 1066, 533, 266 ie NdYAG + doublers.

I always found the amount of visible green in the beam a bit worrying
and the promise that the perspex enclosure was entirely opaque to 266nm.
I never spent much time in close proximity to the systems when running.

Low pass filters plus a few detectors should be good enough to
discriminate between most of them.

Yes, a WDM splitter or two and several detectors and a bunch of
opamps. We might build that in a box, with an LED to indicate each
band.


Recently a customer asked for a some units with 800 nm lasers, which
we hadn\'t done before. We offer 850 as one of our standards. At least
you can see the 800, to make sure it\'s not one of the IRs.

I\'d have thought that for a known
part number whether or not it emits laser light or not would be good
enough.

Our policy is to test everything that has a spec. Mixups are possible.

I vaguely recall an amateur astronomers tunable visible

wavelength dichroic filter (I know someone who bought a prototype) but
it seems to have sunk without trace. I\'ll ask next time I see him.

Otherwise try Young\'s slits to figure out the wavelength.

I want something that looks like a DVM... not an optical bench.

Maybe there is a linear or circular graduated bandpass filter. Rotate
a marked knob for max output. Or a metal disc with multiple,
selectable bandpass windows. I\'d rather buy something.

We could build N boxes, each a bandpass o/e converter. Run a
production source into the right one and it should light up.

 

[whit3rd]

On Monday, October 24, 2022 at 8:37:46 AM UTC-7, John Larkin wrote:

How would you do a small spectrometer that works from 750 to, say,
1900 nm? Better yet 350 to 1900.

Just as your eyes don\'t cover that whole range, so most detectors don\'t.
This one <https://www.thorlabs.com/thorproduct.cfm?partnumber=DSD2>
is actually a pair, and only covers 400 to 1700 nm, and not terribly sensitive
at the endpoints, nor inexpensive.

If you can modulate the light source, you can use a bolometer or photoacoustic sensor
with wide range. The scheme will be, effectively, a lock-in amplifier.

 

[John Larkin]

On Mon, 24 Oct 2022 10:39:25 +0100, Martin Brown
<\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

On 23/10/2022 22:10, John Larkin wrote:
On Sun, 23 Oct 2022 12:04:17 -0700 (PDT), whit3rd <whit3rd@gmail.com
wrote:

On Sunday, October 23, 2022 at 10:49:29 AM UTC-7, John Larkin wrote:

We buy all sorts of lasers and LEDs and we can\'t be sure they are the
right wavelength. Even 1% wavelength resolution would be plenty.

Nobody makes it.

I don\'t know what \'all sorts\' means, but for red visible and most IR, a silicon
photodiode is a good detector. A grating (those start at a dollar or so)
and a protractor will complete the ensemble. You already have a milliammeter, I trust.

\"Don\'t be a jerk. Nobody likes jerks.\"

He isn\'t that far off the mark though.

Your choices for discriminating between various LED laser emitters might
be met by using one of each device in a 5x5 grid and illuminating them
with the DUT. Sum the current that they produce with a opamp and with a
bit of cunning you should be able to approximately get the wavelength
from the actual current it produces. That\'s one device per 50nm.

Only the diodes emitting the same or lower energy photons will generate
photo electrons when so illuminated.

You might get away with just one example of each of the common laser
diodes. I assume the problem is to check that anonymous black plastic
blobs are in fact the emitting the right wavelength ones.

The other option would be one or more photo detectors and various Schott
of Hoya low pass filter glasses available from various dealers.

I recall a tunable visual band dichroic interference filter intended for
amateur astronomers from a few years back that was quite impressive but
I think only covered 400-700nm. I know someone who has one next time I
see him I\'ll ask for the spec. I don\'t think they ever took off.

Not sure you really need to do the visible ones since there are not all
that many semiconductor laser lines possible in practice.

We really need to check IR fiber lasers. LEDs are pretty obvious.

I\'d need a dispersal mechanism and a bunch of detectors. If there is
any wavelength ambiguity, a bunch of software might look at all the
detectors and untangle things and report a single wavelength in plain
sight; this would be a production test, not a research project.

Software might work from an imager, in the case of 2d dispersion, but
it might be tough to find a wideband imager chip. Lotta work.

 

[John Larkin]

On Mon, 24 Oct 2022 08:59:04 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Monday, October 24, 2022 at 8:37:46 AM UTC-7, John Larkin wrote:

How would you do a small spectrometer that works from 750 to, say,
1900 nm? Better yet 350 to 1900.

Just as your eyes don\'t cover that whole range, so most detectors don\'t.
This one <https://www.thorlabs.com/thorproduct.cfm?partnumber=DSD2
is actually a pair, and only covers 400 to 1700 nm, and not terribly sensitive
at the endpoints, nor inexpensive.

And doesn\'t report wavelength.

If you can modulate the light source, you can use a bolometer or photoacoustic sensor
with wide range. The scheme will be, effectively, a lock-in amplifier.

Our sources are always pulsed, so we could use an ac-coupled detector,
like a few wavelength-specific SFP modules maybe.

No, bad idea, they have wide range AGC.

 

[Martin Brown]

On 24/10/2022 16:52, John Larkin wrote:

On Mon, 24 Oct 2022 10:37:39 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

On 23/10/2022 22:15, John Larkin wrote:

That covers the visible, which would check the colors of LEDs, which
is not really much of a problem.

The more serious problem is that we buy a bunch of lasers in the 800
to 1550 sort of range and we\'d like to make sure they are right.

I\'m assuming for the moment that you are serious.

Would anyone joke about that? We buy thousands of pcb mount fiber
lasers per year, most of them at four wavelengths and the occasional
oddball, and they all look alike. I\'d like to final test our products
and be sure we are shipping the correct wavelength.

Fair enough.

Low pass filters plus a few detectors should be good enough to
discriminate between most of them.

Yes, a WDM splitter or two and several detectors and a bunch of
opamps. We might build that in a box, with an LED to indicate each
band.

That might be the simplest solution.

Recently a customer asked for a some units with 800 nm lasers, which
we hadn\'t done before. We offer 850 as one of our standards. At least
you can see the 800, to make sure it\'s not one of the IRs.

I found 800nm dim red and 400nm disappointing dim in the purple.

I\'d have thought that for a known
part number whether or not it emits laser light or not would be good
enough.

Our policy is to test everything that has a spec. Mixups are possible.

Indeed.

I vaguely recall an amateur astronomers tunable visible
wavelength dichroic filter (I know someone who bought a prototype) but
it seems to have sunk without trace. I\'ll ask next time I see him.

Otherwise try Young\'s slits to figure out the wavelength.

I want something that looks like a DVM... not an optical bench.

Maybe there is a linear or circular graduated bandpass filter. Rotate
a marked knob for max output. Or a metal disc with multiple,
selectable bandpass windows. I\'d rather buy something.

Yes that was basically it a rotating disk with a graduated shift in
bandpass as you moved it. This isn\'t like the one I remember but a
related idea based on crossed polars and an electrical tuneable cavity.

https://www.nature.com/articles/s41598-018-29544-x

Selectivity isn\'t great but it might be good enough for your application.

and references therein especially (1)

Xiang, J. et al. Electrically Tunable Selective Reflection of Light from
Ultraviolet to Visible and Infrared by Heliconical Cholesterics. Adv.
Mat. 27, 3014–3018 (2015).

If they have any working prototypes they might be worth talking to...

We could build N boxes, each a bandpass o/e converter. Run a
production source into the right one and it should light up.

I was thinking more in terms of a splitter and a grid of sensors each
one covered with a different bandpass (or cheaper low pass filter). You
can get annealed selelenium glass low pass filters over quite a wide
range of wavelengths from any of the glass companies.


--
Regards,
Martin Brown

 

[Phil Hobbs]

John Larkin wrote:

On Mon, 24 Oct 2022 10:24:58 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Martin Brown wrote:
On 23/10/2022 09:38, Jan Panteltje wrote:
On a sunny day (Sun, 23 Oct 2022 09:13:53 +0100) it happened Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote in <tj2t42$1jlg$1@gioia.aioe.org>:

On 23/10/2022 06:39, Jan Panteltje wrote:
on chip spectrometer?
   https://www.sciencedaily.com/releases/2022/10/221020140615.htm

Possibly. I\'d like to see a bit more of the specifications and light
intensity it requires before I take that press release at face value.

There is a bit more here but the main article is behind a paywall

https://www.science.org/doi/10.1126/science.add8544

I see.
Well, CCD sensor with prism in front of it should work too?

The best super high resolution systems use an echelle method modest
dispersion prism one way and a very high dispersion grating at almost 90
degrees to it so as to map a linear spectrum onto a 2D rectangular CCD.

https://solarsystem.nasa.gov/resources/390/the-solar-spectrum/

That example was actually observed with a Fourier transform method and
then displayed in the fashion of a traditional echelle spectrum. It is a
very impressive piece of kit even it it only works on bright stars:

https://www.jstor.org/stable/26660057#metadata_info_tab_contents

This is a real physical highres echelle spectroscope

https://www.shelyak.com/le-woppshel-un-spectro-echelle-a-grande-resolution/?lang=en


They are seriously nice pieces of kit. PE did an atomic
absorption/emission spectroscope using a similar configuration and early
CCDs back in the 1990\'s. Must have been ~95 because I saw it in Japan at
one of the big analytical trade fairs where we were also exhibiting.


Yeah, a cross-dispersed echelle system can get resolving powers
(lambda/FWHM) up near 1E6, which is pretty impressive.


How would you do a small spectrometer that works from 750 to, say,
1900 nm? Better yet 350 to 1900.

For lab use with collimated beams, probably an 800 l/mm grating, a white
card with a scale, and a lead salt vidicon camera. (I have a couple
that I need to repair one of these times--both tubes work but there\'s
something wrong in the black level circuitry that makes the picture
disappear after a few tenths of a second.)

For barefoot diode lasers, maybe a shear plate instead of the
grating--you know the radius of curvature of the wavefront, because it\'s
just the perpendicular distance between the plate and the laser, so the
fringe spacing gives you the wavelength. I have a couple of nice shear
plate devices that would probably work at some level, although the
coatings would be badly mistuned.

> (Mo just gave me a bybye hug. She says hi.)

Say hi back. She\'s a jewel.

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

 

[Phil Hobbs]

John Larkin wrote:

On Sun, 23 Oct 2022 15:44:10 -0700 (PDT), whit3rd <whit3rd@gmail.com
wrote:

On Sunday, October 23, 2022 at 2:19:55 PM UTC-7, John Larkin wrote:
On Sun, 23 Oct 2022 15:42:32 -0400, Joe Gwinn <joeg...@comcast.net
wrote:

One can cobble something together with a replica grating and a silicon
photo detector array of some kind.

.<https://www.sargentwelch.com/store/product/8885837/replica-diffraction-gratings

How wide a spectral range can a grating cover before things get
ambiguous?

Huh? If you just want to distinguish 850 from 1350 nm, that\'s no problem.
The $1 replica that looks like a 35mm slide will do it fine. You can worry about
blaze angles and UV transmission (the slide is a transparency, but not to UV) and
line spacing fineries, but why?

I was thinking that a grating could fire into several detectors, each
with a different spectral range, and ...

Or, you could swing a detector over a range of angles and register a \'hit\'.
It\'ll take only seconds, why bother with a computer analysis?

I don\'t think a grating will cast a wavelenght-linear unambiguous
image over a wide spectral range, like 5:1 or so.

I think the 700 nm lines will wind up on top of the 1400\'s. And
probably much worse.

https://tinyurl.com/ext8r82d

\"The diffracted beams of different colors and corresponding to
consecutive orders can overlap, this phenomenon becomes more likely to
grow in the order of diffraction.\"

If you use a low-res grating, you\'ll get multiple orders from everybody.
Since you know a priori that it\'s only one wavelength, there\'s no
ambiguity.

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

 

[Martin Brown]

On 24/10/2022 17:00, John Larkin wrote:

I\'d need a dispersal mechanism and a bunch of detectors. If there is
any wavelength ambiguity, a bunch of software might look at all the
detectors and untangle things and report a single wavelength in plain
sight; this would be a production test, not a research project.

Software might work from an imager, in the case of 2d dispersion, but
it might be tough to find a wideband imager chip. Lotta work.

I think your problem will be with the longer wavelengths even thinned
back illuminated CCDs fall off in sensitivity steeply at around 1000nm.


--
Regards,
Martin Brown

 

[Joe Gwinn]

On Sun, 23 Oct 2022 14:19:47 -0700, John Larkin
<jlarkin@highlandSNIPMEtechnology.com> wrote:

On Sun, 23 Oct 2022 15:42:32 -0400, Joe Gwinn <joegwinn@comcast.net
wrote:

On Sun, 23 Oct 2022 10:49:22 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Sun, 23 Oct 2022 17:23:12 GMT, Jan Panteltje
pNaonStpealmtje@yahoo.com> wrote:

On a sunny day (Sun, 23 Oct 2022 09:58:18 -0700) it happened John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote in
m0salh9pp7c3gqortagaj6ualh9g23ar4n@4ax.com>:

On Sun, 23 Oct 2022 15:26:39 GMT, Jan Panteltje
pNaonStpealmtje@yahoo.com> wrote:

On a sunny day (Sun, 23 Oct 2022 08:10:42 -0700) it happened John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote in
96malhldkco3pnjog22sqb7dq8jm4c7hdb@4ax.com>:

On Sun, 23 Oct 2022 05:39:23 GMT, Jan Panteltje
pNaOnStPeAlMtje@yahoo.com> wrote:

on chip spectrometer?
https://www.sciencedaily.com/releases/2022/10/221020140615.htm

I\'ve always wanted a handheld DVM-like spectrometer that covers a wide
range, specifically 1600 to 300 nm to verify LED and laser
wavelengths.

That should not be difficult to make

In the UNI we had small spectrometers that consisted of a rotating prism
and a photocell looking at it through a slot (to look at a spectral line)
A \'white\' light source and a tube with the stuff that had to
be investigated in the light beam.
Big knob on top to rotate the prism, the knob had a scale with numbers on it,
the wavelength.
Small box, 3x3 inch or so I think.
Had to repair one once.
You could perhaps use a rotating prism, a slot and a photocell, tune for maximum and
read the rotation from the knob?

The challenge is to make a spectrometer with a wide wavelength range.
It would at least need several detectors, and a prism or grating that
would work over about a 5:1 wavelength range.

Nobody seems to make one.

We\'re lucky in electronics. We can easily measure resistance and
capacitance and frequency over million or billion or sometimes
trillion-to-one spans.

Our Keysight counter can measure picoseconds to kiloseconds, microHz
to gigahertz.

A Fluke DVM can measure microvolts and kilovolts.

I can measure femtofarads to kilofarads with the gear on my little
workbench.

If you a need wide range IR detector:
https://www.irlabs.com/products/bolometers/bolometer-systems/#:~:text=Bolometers%20are%20detectors%20used%20to,5000%C2%B5m%20(30THz%20to%2060GHz).
20 THz to 150 GHz 15 to 2000 um
I have a cryocooler also workbench size.
You have a very strong signal using laser output,
should make things easier.

I want a spectrometer. We don\'t need quantified power measurement but
it would be nice.

We buy all sorts of lasers and LEDs and we can\'t be sure they are the
right wavelength. Even 1% wavelength resolution would be plenty.

Nobody makes it.

One can cobble something together with a replica grating and a silicon
photo detector array of some kind.

.<https://www.sargentwelch.com/store/product/8885837/replica-diffraction-gratings

This is one example. There are many others.

You will need a calibration source of some kind. A neon tube or the
like, to provide some known lines for reference.

Joe Gwinn

How wide a spectral range can a grating cover before things get
ambiguous?

If it\'s one source with a dominant wavelength, one can disentangle the
overlapping diffraction orders with software. If the source contains
a frequency doubler, there will also be some of the original un-
doubled drive also present, but this approach may still work.

I was thinking that a grating could fire into several detectors, each
with a different spectral range, and the resulting confusion might be
sorted out in software.

I would use a linear photodetector array, but silicon won\'t work for
1550 nm at all. There may be a detector material that will span 800
nm to 2000 nm, but it may not be suitable for a camera.


More generally, if the task is simply to detect mixups in production,
and we are testing a bright laser source, I\'d go for simple and
rugged. This approach others have mostly suggested:

Acquire a four-way passive optical power divider made of Ge-doped
fused silica optical fiber. Step index is OK. This divider will
provide four outputs of roughly equal power (if so designed). Attach
each output to a detector for one of the possible (nominal)
wavelengths, using optical filters as needed. Estimate the dominant
wavelength from the combined output currents in such a way that the
source brightness mostly cancels out.

If the response patterns are complex, four dedicated
pattern-recognizers in parallel may be used between outputs and the
decision process.

Model the decision algorithm on Receptive Fields in Biology:

..<https://en.wikipedia.org/wiki/Receptive_field>

This will give an unambiguous answer representing the best guess of
the mixup-detecting box.

One can also use the four outputs to drive four green LED lights is a
square, and let the human to the receptive-field processing visually.

Or both, at least initially.

This can be done in analog hardware, or in code.


Joe Gwinn

 

[Martin Brown]

On 24/10/2022 21:30, Joe Gwinn wrote:

On Sun, 23 Oct 2022 14:19:47 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Sun, 23 Oct 2022 15:42:32 -0400, Joe Gwinn <joegwinn@comcast.net
wrote:

On Sun, 23 Oct 2022 10:49:22 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Sun, 23 Oct 2022 17:23:12 GMT, Jan Panteltje
pNaonStpealmtje@yahoo.com> wrote:

On a sunny day (Sun, 23 Oct 2022 09:58:18 -0700) it happened John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote in
m0salh9pp7c3gqortagaj6ualh9g23ar4n@4ax.com>:

On Sun, 23 Oct 2022 15:26:39 GMT, Jan Panteltje
pNaonStpealmtje@yahoo.com> wrote:

On a sunny day (Sun, 23 Oct 2022 08:10:42 -0700) it happened John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote in
96malhldkco3pnjog22sqb7dq8jm4c7hdb@4ax.com>:

On Sun, 23 Oct 2022 05:39:23 GMT, Jan Panteltje
pNaOnStPeAlMtje@yahoo.com> wrote:

on chip spectrometer?
https://www.sciencedaily.com/releases/2022/10/221020140615.htm

I\'ve always wanted a handheld DVM-like spectrometer that covers a wide
range, specifically 1600 to 300 nm to verify LED and laser
wavelengths.

That should not be difficult to make

In the UNI we had small spectrometers that consisted of a rotating prism
and a photocell looking at it through a slot (to look at a spectral line)
A \'white\' light source and a tube with the stuff that had to
be investigated in the light beam.
Big knob on top to rotate the prism, the knob had a scale with numbers on it,
the wavelength.
Small box, 3x3 inch or so I think.
Had to repair one once.
You could perhaps use a rotating prism, a slot and a photocell, tune for maximum and
read the rotation from the knob?

The challenge is to make a spectrometer with a wide wavelength range.
It would at least need several detectors, and a prism or grating that
would work over about a 5:1 wavelength range.

Nobody seems to make one.

We\'re lucky in electronics. We can easily measure resistance and
capacitance and frequency over million or billion or sometimes
trillion-to-one spans.

Our Keysight counter can measure picoseconds to kiloseconds, microHz
to gigahertz.

A Fluke DVM can measure microvolts and kilovolts.

I can measure femtofarads to kilofarads with the gear on my little
workbench.

If you a need wide range IR detector:
https://www.irlabs.com/products/bolometers/bolometer-systems/#:~:text=Bolometers%20are%20detectors%20used%20to,5000%C2%B5m%20(30THz%20to%2060GHz).
20 THz to 150 GHz 15 to 2000 um
I have a cryocooler also workbench size.
You have a very strong signal using laser output,
should make things easier.

I want a spectrometer. We don\'t need quantified power measurement but
it would be nice.

We buy all sorts of lasers and LEDs and we can\'t be sure they are the
right wavelength. Even 1% wavelength resolution would be plenty.

Nobody makes it.

One can cobble something together with a replica grating and a silicon
photo detector array of some kind.

.<https://www.sargentwelch.com/store/product/8885837/replica-diffraction-gratings

This is one example. There are many others.

You will need a calibration source of some kind. A neon tube or the
like, to provide some known lines for reference.

Joe Gwinn

How wide a spectral range can a grating cover before things get
ambiguous?

If it\'s one source with a dominant wavelength, one can disentangle the
overlapping diffraction orders with software. If the source contains
a frequency doubler, there will also be some of the original un-
doubled drive also present, but this approach may still work.


I was thinking that a grating could fire into several detectors, each
with a different spectral range, and the resulting confusion might be
sorted out in software.

I would use a linear photodetector array, but silicon won\'t work for
1550 nm at all. There may be a detector material that will span 800
nm to 2000 nm, but it may not be suitable for a camera.

Detectors are available typically InGaAs for the NIR as are whole module
solutions but I don\'t think he is going to like the price! eg.

https://www.stellarnet.us/spectrometers/portable-nir-spectrometer-line-up/

or from Edmund

https://www.edmundoptics.com/p/900---1700nm-bw-tek-ingaas-nir-spectrometer/43100/

--
Regards,
Martin Brown

 

[piglet]

On 23/10/2022 16:15, John Larkin wrote:

On Sun, 23 Oct 2022 14:27:56 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

On 23/10/2022 09:38, Jan Panteltje wrote:
On a sunny day (Sun, 23 Oct 2022 09:13:53 +0100) it happened Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote in <tj2t42$1jlg$1@gioia.aioe.org>:

On 23/10/2022 06:39, Jan Panteltje wrote:
on chip spectrometer?
https://www.sciencedaily.com/releases/2022/10/221020140615.htm

Possibly. I\'d like to see a bit more of the specifications and light
intensity it requires before I take that press release at face value.

There is a bit more here but the main article is behind a paywall

https://www.science.org/doi/10.1126/science.add8544

I see.
Well, CCD sensor with prism in front of it should work too?

The best super high resolution systems use an echelle method modest
dispersion prism one way and a very high dispersion grating at almost 90
degrees to it so as to map a linear spectrum onto a 2D rectangular CCD.

https://solarsystem.nasa.gov/resources/390/the-solar-spectrum/

That example was actually observed with a Fourier transform method and
then displayed in the fashion of a traditional echelle spectrum. It is a
very impressive piece of kit even it it only works on bright stars:

https://www.jstor.org/stable/26660057#metadata_info_tab_contents

This is a real physical highres echelle spectroscope

https://www.shelyak.com/le-woppshel-un-spectro-echelle-a-grande-resolution/?lang=en

They are seriously nice pieces of kit. PE did an atomic
absorption/emission spectroscope using a similar configuration and early
CCDs back in the 1990\'s. Must have been ~95 because I saw it in Japan at
one of the big analytical trade fairs where we were also exhibiting.

The spectrometer business seems to be a race for resolution in narrow
bands. There\'s no wide-range low-resolution stuff that we can find.

Something like a grating and a bunch of detectors could work. It would
have a lot of wavelength overlap confusion which could be mostly
computed out.

Regular silicon detectors cover 900-300nm (or thereabouts) at low cost
so I wonder if a non-linear crystal can be switched in ahead of the
grating to double or triple 1600nm into the Si-detectable range. JL
wants to verify single bright monochromatic sources so losses needn\'t be
a worry?

piglet

 

[John Larkin]

On Tue, 25 Oct 2022 16:03:40 +0100, piglet <erichpwagner@hotmail.com>
wrote:

On 23/10/2022 16:15, John Larkin wrote:
On Sun, 23 Oct 2022 14:27:56 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

On 23/10/2022 09:38, Jan Panteltje wrote:
On a sunny day (Sun, 23 Oct 2022 09:13:53 +0100) it happened Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote in <tj2t42$1jlg$1@gioia.aioe.org>:

On 23/10/2022 06:39, Jan Panteltje wrote:
on chip spectrometer?
https://www.sciencedaily.com/releases/2022/10/221020140615.htm

Possibly. I\'d like to see a bit more of the specifications and light
intensity it requires before I take that press release at face value.

There is a bit more here but the main article is behind a paywall

https://www.science.org/doi/10.1126/science.add8544

I see.
Well, CCD sensor with prism in front of it should work too?

The best super high resolution systems use an echelle method modest
dispersion prism one way and a very high dispersion grating at almost 90
degrees to it so as to map a linear spectrum onto a 2D rectangular CCD.

https://solarsystem.nasa.gov/resources/390/the-solar-spectrum/

That example was actually observed with a Fourier transform method and
then displayed in the fashion of a traditional echelle spectrum. It is a
very impressive piece of kit even it it only works on bright stars:

https://www.jstor.org/stable/26660057#metadata_info_tab_contents

This is a real physical highres echelle spectroscope

https://www.shelyak.com/le-woppshel-un-spectro-echelle-a-grande-resolution/?lang=en

They are seriously nice pieces of kit. PE did an atomic
absorption/emission spectroscope using a similar configuration and early
CCDs back in the 1990\'s. Must have been ~95 because I saw it in Japan at
one of the big analytical trade fairs where we were also exhibiting.

The spectrometer business seems to be a race for resolution in narrow
bands. There\'s no wide-range low-resolution stuff that we can find.

Something like a grating and a bunch of detectors could work. It would
have a lot of wavelength overlap confusion which could be mostly
computed out.


Regular silicon detectors cover 900-300nm (or thereabouts) at low cost
so I wonder if a non-linear crystal can be switched in ahead of the
grating to double or triple 1600nm into the Si-detectable range. JL
wants to verify single bright monochromatic sources so losses needn\'t be
a worry?

piglet

A typical source is a connectorized fiber-coupled laser of a couple of
milliwatts. Multiplication usually requires very high powers.

The longest wavelength lasers that we now use are 1550, and we have
corresponding photodiodes. The shortest are 800, ditto.

A wavelength splitter and three photodiodes would at least identify
the band of a laser: 1500ish, 1300ish, and 850ish.