Minimalistic multichip DIY LED build

[mr.wilson]

DIY kits, hobby-grade equipment and online purchases are always hit and miss. The main reason why we all do it is for the sport, not because it is a wise decision. I don't see a problem in sharing our good and bad experiences as long as it is honest and fair.

 

[megadeth72]

So we're raising a flag here saying we're being cheated after using a diy hobby spectrometer?

Really? Have you used one? Mine is extrrmely acurate

Sent from my SCH-I535 using Tapatalk 2

 

[Mike31154]

I agree, by all means share your experience. But the title of the post & the responses that follow are indicative of how readily we get drawn into the anti Asian product syndrome. As for the sport, I personally do not consider online purchases as a sport. I still prefer the physical types of that activity.

 

[Mike31154]

Really? Have you used one? Mine is extrrmely acurate

Sent from my SCH-I535 using Tapatalk 2

No I have not used one & don't intend to. I'm not that concerned over a few nms difference in the LEDs I use. I did have a quick look at the construction though and while I have no experience with the accuracy of the instrument, the FAQs on the site & the fact that it appears to be constructed out of a DVD & some cardboard indicates that results may vary.

 

[mr.wilson]

When people post about their experiences, other members are often able to fairly accurately judge the quality, accuracy and value without actually using it themselves. Just as the user can ascertain the accuracy, quality and value without using any other similar equipment to compare.

 

[Whitebeam]

So we're raising a flag here saying we're being cheated after using a diy hobby spectrometer?

Yes. Here's why.

1) I can quite clearly see just from the photographs that the emission spectra from my three supposedly different LEDs overlay each other exactly.

2) I calibrated the spectrometer using the mercury lines from a CFL tube before taking the LED spectra and took another shot of the same CFL tube immediately after taking the LED spectra to ensure the calibration was still accurate. The confirmation shot read the wavelengths of the mercury 435nm and 546nm lines to within a couple of nm.

Peter

 

[Mike31154]

Yes. Here's why.

1) I can quite clearly see just from the photographs that the emission spectra from my three supposedly different LEDs overlay each other exactly.

2) I calibrated the spectrometer using the mercury lines from a CFL tube before taking the LED spectra and took another shot of the same CFL tube immediately after taking the LED spectra to ensure the calibration was still accurate. The confirmation shot read the wavelengths of the mercury 435nm and 546nm lines to within a couple of nm.

Peter

Cool, thank you for the explanation.

 

[megadeth72]

No I have not used one & don't intend to. I'm not that concerned over a few nms difference in the LEDs I use. I did have a quick look at the construction though and while I have no experience with the accuracy of the instrument, the FAQs on the site & the fact that it appears to be constructed out of a DVD & some cardboard indicates that results may vary.

I am, and these are very accurate meters, I was shocked how accurate they are, I suppose this is because they are calibrated

I do not think the results vary because of the the calibration works, mine have not varied at all and I built a couple of them

 

[TropTrea]

:furious: I've just been using Spectralworkbench to compare 4 supposedly different blue multichip modules bought from different eBay sellers. One module advertised as 430nm, one as 445nm, one as 455nm and one as 460-470nm. Guess what? All except the 430nm have exactly the same 445nm peak spectrum :furious:


Peter

Can you give us information of which chips gave which results? Did the 430 actualy peak at 430nm? The difference between a 455 and 465 should be easily noticable from the naked eye.

For a comparison this setup seems to work however I still question its accuracy at defining the exact frequency of the peaks.

 

[Whitebeam]

Can you give us information of which chips gave which results? Did the 430 actualy peak at 430nm? The difference between a 455 and 465 should be easily noticable from the naked eye.

For a comparison this setup seems to work however I still question its accuracy at defining the exact frequency of the peaks.

The 430 peaked nearer 420nm.

The 445 at 445. The 455 and "460-470" (as it was sold to me) at 445nm.

The 445, 455 & 465 all appeared visually identical to me, hence my move to actually try my best to measure them.

I agree that I wouldn't bet the house on being able to determine absolute wavelength to within a few nm for lines significantly away from the two mercury vapour calibration lines (note that one of these is at 435nm - perfect for our needs here), but I do believe that it can be used for the relative comparison of two spectra with a fair degree of reliability. You will notice from my post above that I took a second shot of the CFL 'calibration' after I measured all of the LED spectra to ensure that the calibration had not been affected by anything whilst I was measuring - I was pretty impressed by the consistency shown between these two shots.

I believe there are things that can easily be done to improve the spectrometer, proper slits, proper gratings, better positioning of cameras relative to grating etc. These are actively being studied on the website associated with the DIY spectrometer and I hope at some point to have some spare money and time to follow some of these up myself. In the mean time I'm happy that I can use this setup for relative comparison of spectra and for indicative absolute measurements. Perhaps we should move such discussions to a separate thread now, as we're straying rather OT?

Peter

 

[Whitebeam]

FWIW, here is the 'calibration' spectrum used for my LED measurements

<iframe width='500px' height='400px' border='0' src='http://spectralworkbench.org/spectra/embed/3453'></iframe>

As you'll see, I can easily resolve the two green peaks, the lower from Terbium and the upper from Mercury, that are 4nm apart.

Unfortunately the large red peak slightly saturated my camera's red sensor, somewhat smearing the line, but none the less this line that should be at 611.6nm reads as between 612nm and 614nm. Close enough, considering it is 'beyond' the calibrated spectrum.

Compare this to the spectra shown here http://en.wikipedia.org/wiki/File:Fluorescent_lighting_spectrum_peaks_labelled.svg taken with a 'proper' spectrometer.

Edit: Having looked at this further, I can make out all of the peaks in the proper spectrum up to peak number 14 in my spectrum. The limit of the measurement error in my spectrum vs the 'proper' line wavelengths is about 2nm across this range.

Peter

 

[TropTrea]

The 430 peaked nearer 420nm.

The 445 at 445. The 455 and "460-470" (as it was sold to me) at 445nm.

The 445, 455 & 465 all appeared visually identical to me, hence my move to actually try my best to measure them.

If you r realy look at the number 430 to 420 that is a 10 nm shift or only 2.3%, the 465 to 445 is a 20 nm shift or 4.3%. You will not see thius level of accrarcy on the Florescent bulbs by any means. Normal standard is the strongest Blue, Green, and Red peak are within 5% of target for wavelenght and 10 % for intensity.

How ever LED production from companies like Cree are completly another thing. They are clasify there led's is brackets and bin them to be within 1 or 2 nm for there peak. This is only possible do to the much narrower bandwidth that LED produce light.

I agree that I wouldn't bet the house on being able to determine absolute wavelength to within a few nm for lines significantly away from the two mercury vapour calibration lines (note that one of these is at 435nm - perfect for our needs here), but I do believe that it can be used for the relative comparison of two spectra with a fair degree of reliability. You will notice from my post above that I took a second shot of the CFL 'calibration' after I measured all of the LED spectra to ensure that the calibration had not been affected by anything whilst I was measuring - I was pretty impressed by the consistency shown between these two shots.

I believe there are things that can easily be done to improve the spectrometer, proper slits, proper gratings, better positioning of cameras relative to grating etc. These are actively being studied on the website associated with the DIY spectrometer and I hope at some point to have some spare money and time to follow some of these up myself. In the mean time I'm happy that I can use this setup for relative comparison of spectra and for indicative absolute measurements. Perhaps we should move such discussions to a separate thread now, as we're straying rather OT?

Peter

One of the thing is I think we are starting to be much pickier on lighting than what the markets have been used to. If a lighting source was off by 20 or 30 nm no one gave a care in the past because light sources were emitting light in band widths of 40 to 100 nm wide. With the newer LED's these band widths are very narrow sometimes as small as only 10 nm wide. What we are looking for is probably not even out there yet as to date we as reefers are the only ones looking for lighting in some of the wavelenghts especialy between 420 and 450 nm.

 

[Whitebeam]

If you r realy look at the number 430 to 420 that is a 10 nm shift or only 2.3%, the 465 to 445 is a 20 nm shift or 4.3%. You will not see thius level of accrarcy on the Florescent bulbs by any means. Normal standard is the strongest Blue, Green, and Red peak are within 5% of target for wavelenght and 10 % for intensity.

Are you talking about the accuracy of fluorescents here for the bulbs we use for tank lighting or for the bulbs used to calibrate the spectrometer?

The clever bit of the spectrometer calibration is that they've chosen the two very precise emission lines from the mercury vapour in the CFL bulbs. These will always be on exactly the specified wavelengths, unless the laws of physics change.

Peter

 

[TropTrea]

I am refering to the industry standards for Florescent tubes of all types. Yes on some bulbs the peaks are very consistant. But on other bulbs with compex compound mixtures a slight change in the compounds can cause a considerable shift in a peak. This is one of the reasons that they do not advertise the ppeaks for some bulbs but istead only give the K temp equivelent. However even the true K temp can vary considerably between batches of the same bulb.

 

[Phishguy]

I'd love to see some pictures or what your tanks look like with this multi-chip system.. I've read a lot of his thread but it's so long that it becomes very confusing.. Please post pics and observations... Also, are these "dream" chips I've read about available? Thanks

 

[miatawnt2b]

You can take a look at my completed setup in my gallery. It's a 50w multichip, 25000k no optics 7" off the surface, 18" cube tank.

 

[acabgd]

How do you like it? How is it growing coral for you?

 

[acabgd]

I'd love to see some pictures or what your tanks look like with this multi-chip system..

Here some photos, good and bad... Single 100 DreamChip.











Softies seem to love this light much more than SPS, but the algae do as well. I will try to lower the white light as I think that might be the problem.

 

[miatawnt2b]

How do you like it? How is it growing coral for you?

I like it a lot. Really bright and I love the color. Much like a 14-20k halide (probably closer to the 14k end) The corals are coloring nicely and growing like crazy. It's been on the tank about 2 months now.

Again, i'm not using the dreamchip. It's just a 50w multichip that I requested the seller send a 25000k version. I got a dimable balast with the kit, but since I don't use optics (didn't like the look) It's running almost full power.

 

[nettraptor]

this is my efford > two cables out 1 for pwm signals , 1 for powering the light
i drive them with an arduino mega , with sunset-sunrise
No lens!