FS: Sharps "505nm" diodes from various batches. (All wl measured)

[Singlemode Laser]

How and at what current did you measure the wavelength? I'm measuring 509nm at 108mA with no lens. :cryyy:

Why to cry right away. I am always willing to help everybody here.

As posted in this thread I did all the measurements at the recommended current regarding the data sheet, which I also attached in this thread. The parameters were:

I = 80mA and temperature stabilized to ~ 25°C.

You seem not to have a propper heatsink if you measuring the wl without a lens (might be wrong though). Are you using a calibrated spectrometer?


Singlemode

 

[Zom-B]

See my edits. I did not see your reply as I kept refreshing page 1 and I also didn't see it pop up in my subscriptions (probably because the system thought I read the reply when I refreshed page 1)

I think heat sinking is adequate. When I keep it running in the closed Aixiz for half an hour I don't feel it getting warm at all. (though it might have just reached skin temperature)

 

[Singlemode Laser]

See my edits. I did not see your reply as I kept refreshing page 1 and I also didn't see it pop up in my subscriptions (probably because the system thought I read the reply when I refreshed page 1)

I think heat sinking is adequate. When I keep it running in the closed Aixiz for half an hour I don't feel it getting warm at all. (though it might have just reached skin temperature)

I will answer your questions:

Edit: I did a reference test with a DPSS 532 (those don't drift much, right?) at 30°C it measures 533.85nm. The calibration sheet of the spectrometer is from 2014 and all wavelengths from 200 to 1100nm are/were within 0.2nm. For one, we should probably recalibrate, but that doesn't explain the 505→509 discrepacy where just a bit higher there's half the error: 532→534.

The usual DPSS operates in the 531nm-533nm range. See this spec sheet f.e. :https://www.thorlabs.com/drawings/b34958d91069df18-42FC1350-CEAF-82DA-E55535DC41CE9034/DJ532-10-SpecSheet.pdf

0.2nm is the resolution but not the accuracy of your spectrometer. The meas. wl will drift with time depending on a ton reasons. The diffraction pattern is not linear, so a propper calibration needs multiple references. 532nm is not so far away though.

You said, that you are measuring without a lens. Since a spe. measures the diffrated light, the high beam divergence of a free diode could create a serious offset in the wl.

Edit2: It looks greener than my 505nm LEDs. My LEDs measure 505.0nm peak, with the median of it's power distribution a tad higher in wavelength.

typical LEDs have a bandwidth of 20nm-30nm with an unsymetric spectrum like you said. The can be used for a calibration at level of some nm at the best. The value you measured for this specific diode batch is the highest ever reported, what could be an indication that the measurument was not optimal.

Since I am always responsible for my statements and actions I can offer you the following:
You can ship the diode back to me, I will measure the wl and post the data here in the forum for everybody. If the wl is not 505nm +/- the half width of the diodes spectrum, I will transfer you the money you paid + the money for the shipping (same shipping type like the one you choose first). If the wl is correct, I will ship it back to you on your own expenses.

Singlemode

 

[paul1598419]

Welcome to the happy world of selling wavelength specific diodes. No way to know what the discrepancy is. I'm glad you and I have not had these kinds of problems, Singlemode.

 

[Zom-B]

0.2nm is the resolution but not the accuracy of your spectrometer. The meas. wl will drift with time depending on a ton reasons. The diffraction pattern is not linear, so a propper calibration needs multiple references. 532nm is not so far away though.

No, the spectrometer is calibrated on 25 wavelengths, all within 0.2nm (0.1nm typically). There are 4 calibration coefficients (cubic curve) with an R value of 0.999995. The resolution you speak of is 0.36 at 188nm and 0.45 at 1036nm (non-linear as you said). The spectrometer is going back for a grating upgrade and re-calibration soon anyway.

You said, that you are measuring without a lens. Since a spe. measures the diffrated light, the high beam divergence of a free diode could create a serious offset in the wl.

I measured the diode inside a 20cm integrating sphere, calibrated with a NIST-traceable standard light source, so it's not the 'bare' diode. I'm not shining the laser into the slit as that would indeed cause a geometric offset on the grating and sensor.

 

[Singlemode Laser]

No, the spectrometer is calibrated on 25 wavelengths, all within 0.2nm (0.1nm typically). There are 4 calibration coefficients (cubic curve) with an R value of 0.999995. The resolution you speak of is 0.36 at 188nm and 0.45 at 1036nm (non-linear as you said). The spectrometer is going back for a grating upgrade and re-calibration soon anyway.

I measured the diode inside a 20cm integrating sphere, calibrated with a NIST-traceable standard light source, so it's not the 'bare' diode. I'm not shining the laser into the slit as that would indeed cause a geometric offset on the grating and sensor.

Hi Zom-B,
which solution do you suggest?

Singlemode

 

[paul1598419]

There are these $70.00 "spectrometers" that have been sold to members here that are made with a camera CCD and a homemade slit that is very wide by professional standards. It was calibrated with only two points, but my understanding is they have new software that can use multiple lines. Still, it is far from a professional spectrometer.

 

[Singlemode Laser]

There are these $70.00 "spectrometers" that have been sold to members here that are made with a camera CCD and a homemade slit that is very wide by professional standards. It was calibrated with only two points, but my understanding is they have new software that can use multiple lines. Still, it is far from a professional spectrometer.

When you replace the slit by a single mode fiber (a "slit" that is some micro meters wide), you can achieve sub pm accuracy and sensitivity!

http://sci-hub.cc/10.1063/1.4765744

Singlemode

 

[paul1598419]

That was an interesting paper on using "off the shelf" CCD and CMOS cameras for measuring transitions to high accuracy in a very narrow bandwidth. The one I was referencing is used to measure the entire VIS spectrum, however, and I think you lose the accuracy by not confining the spectrum to a few nm.

 

[Zom-B]

Indeed. Depending on the grating you could have for example a 500lines/mm grating with a bandwidth of 1000nm but only 1.5nm effective resolution, or a 2400lines/mm with 100nm bandwidth that has 0.15nm resolution. The bandwidth stems from where the first order rainbow starts overlapping the second order rainbow. A smaller slit can further narrow the <s>bandwidth</s> resolution* but I suspect there are even limits to that (diffraction limit maybe?)

[edit] oops

 

[Singlemode Laser]

Indeed. Depending on the grating you could have for example a 500lines/mm grating with a bandwidth of 1000nm but only 1.5nm effective resolution, or a 2400lines/mm with 100nm bandwidth that has 0.15nm resolution. The bandwidth stems from where the first order rainbow starts overlapping the second order rainbow.

Multiple orders overlap only in the case of a very broadband light source. For 2400lines/mm there is even no second order in the visible. The resolution is then determined how many pixels of your sensor you illuminate (more illuminated pixels means higher resolution but smaller bandwidth of coarse), and how small they are, and how good your fit is. The single mode fiber creates a perfect Gaussian beam profile and it's center can be fitted much better, than a slit. Using the CCD instead of just a 1D array increases the amount of measurements of the same signal at the same time (it like 1000 arrays at the same time).

At the end you always trade bandwidth with resolution. A stepper motor could rotate the grating around the point where the beam hits it, so every wl can be accessed. With some calibration lines, this could be a bummer device!!

A smaller slit can further narrow the bandwidth but I suspect there are even limits to that (diffraction limit maybe?)

Diffraction is the limit, when two pixels are illuminated and their size is equal the spectral width of the signal (or half wl at the end). You can increase the distance between grating and sensor to fight this. Some labs have ridiculous huge spectros!

Singlemode

 

[Zom-B]

Using the CCD instead of just a 1D array increases the amount of measurements of the same signal at the same time (it like 1000 arrays at the same time)

Our spectrometer is fitted with a cylindrical lens on the linear CCD to mitigate exactly this problem.

 

[Singlemode Laser]

The 505nm diode is sold to 317718 and already shipped.

Singlemode

 

[paul1598419]

That's great, Singlemode. I'm sure "he" will love it. Even those closer to 505nm are a pretty great color. Hope whatever diodes you have left will sell smoothly.

 

[Zom-B]

Noone else notice 317718 is "billie" up-side-down?

 

[paul1598419]

In fact, he/she pointed that out in the welcome thread.