this Led spectrum has a negative R9 value, if you want to fit a cluster of data, you surely want the variance of the data is small. for R9 the variance is quite large, so R9 is indeed not easy.
Specifically:
Would you get the LM4? Or are the issues significant enough to hold off or get the LM3?
Are there any alternatives under $100?
LM3 has no R9 and duv and other R values
2 Thanks
That seems a bit surprising to me considering the existence of hobbies. For instance, in all the years I spent in manufacturing, mostly as a CNC machinist, my knowledge of the characteristics of various battery chemistries was never relevant to my job, yet I knew (all too well) the memory effect inherent in NiCad’s, as well as how much heat was generated by pushing 30 amps through FETs could generate due to my hobby of RC cars. Knowing about light is important for theatre, photography, and other artistic endeavors that some do on the hobbyist level. And those who paint, even if only gaming miniatures or or plastic models, would likely know a little about color even if it works differently with pigments than it does with lighting.
So it strikes me as a little unusual that not a single one of the CS grads you’ve encountered have hobbies that have some crossover involving colorimetry, even if only tangentially. Granted, it’s not common, but I would thin that you would’ve run across at least one.
Just as fluorescent lights are generally UV-excited. In NNPTC, you had to recite the entire operation of an automatic-starting fluorescent light in detail and would lose points on tests if you did not mention the heating of the bimetallic strip in the starter, the inductive kick from the ballast, the vaporization of the mercury, or the phosphors converting the UV light of the mercury vapors to visible light.
I knew far more than I cared to about phosphors and excitation than I cared to before I got into this hobby. Forgotten a lot too, but not enough to not be unimpressed by how far LEDs have come, or the beauty of (and science behind) a good 9080 emitter.
Knowing how to use light, and knowing the physics and mathematics to calculate and reconstruct spectra off some pretty wonky datapoints from a cheap sensor is a whole different beast though…
It’s a bit like expecting flashlight enthusiasts to be able to come up with their own competitive LiIon cell chemistries and have a company make top notch batteries by their design because they have experience with batteries already.
2 Thanks
In my opinion, it’s like wondering why a proctologist can’t treat teeth. They are both doctors… I apologize if the comparison is harsh. I graduated from university with a degree in information systems (programmer). Programming is almost everywhere now: medicine, space, fintech, … It is impossible to study all subject areas. I have no programmer friends who understand anything about lighting engineering.
2 Thanks
can’t agree with you any more
My proctologist treats my teeth.
He takes the scenic route. 
1 Thank
I think the reason why its hard finding engineers in this area is because colorimetry is in the domain of color science, not computer science. If you are CS you may have exposure to image processing but its not the same thing.
So an app like this, is by its very nature, interdisciplinary … a mix of color science + software engineering + bluetooth + app development. A MS or PhD in color science might be able to implement the tricky algos. This is probably the best approach, and then have a software engineer take the code and make sure its robust, testable, and maintainable.
A CS person with background in computer graphics and image processing may also be able to do it, but would still need BLE and mobile app development.
originally I thought that optical engineers would know how to calculate colorimetry metrics, such as CRI/R9/circadian stimulus, but astonishingly not so. optical engineers mainly work on photometric aspect thing, such as design optical lens(a plastic thing as far as I know). so colorimetry and photometry are different. I am wondering if there is a major in college named colorimetry?
I don’t think CS students ever learn colorimetry in college, I did learn digital image processing in college but never heard of colorimetry.
2 Thanks
That’s exactly what I was thinking, but apparently I stated it in a way that invited dogpiling. I think some folks simply read too much into, “… some crossover involving colorimetry, even if only tangentially.”, and don’t believe that many folks know things that they did not take as a college course.
Optical engineers are a bit more concerned with refraction and transmissivity, which is a bit more relevant to opticians (and, to a lesser extent, optometrists) than those dealing with lighting. To an optical engineer, color is merely a wavelength.
Colorimetry is far more relevant to photographers and cinematographers than to optical engineers. Someone who is particular about the camera settings and scene lighting to get the perfect shot will know color better than someone who studied refractive indices and/or ABBE values. And there’s plenty of folks who use cameras, still-image or motion, as a hobby.
you are right. so when the supervisor/leader of the optical group said that they don’t know colorimetry caused me to think they are just passing the buck, and I had no background in this field, this really made me suffered a lot last year, sometimes I can’t get asleep in the night.
LM3 has 4 team members, LM4 got 3. for LM3, there is a specific engineer for coding/spectral thing/machine learning thing, for LM4 I am actually doing all the things except tristimulus values.
3 Thanks
I think I am still not very good at communication and how to deal with colleagues(office politics?), there is a guy in optical group who knows colorimetry a lot, however, …
I think there is shortcomings in my character/temper, I am afraid of conflicts with others, I want harmonious relationships with others.
I can’t elaborate more. what I can say is that I am already giving out 100% of my effort on this project.
4 Thanks
I think we should put correct person in a correct position, nobody knows all fields, that’s why there are so many majors in colleges, right? I believe in /quote an old Chinese saying from 《The Analects of Confucius》here， “The real fact is that one might have learned the doctrine earlier than the other, or might be a master in his own special field”.
have a nice day, guys:-)
2 Thanks
My friend, interfacing with a hobby community and learning new skills just to make such a product better is 110%, maybe 120%.
We have our critiques but it is amazing to be able to interface with you on this so thank you!
2 Thanks
thank you for your understanding.
I am actually not saying this forum is not friendly, from a user’s perspective when you bought a product that is buggy or defective, your anger is totally acceptable for me. what I am saying is a harmonious working environment is crucial for getting things done. here we have somebody who knows colorimetry well, why not let him do this part? I’d like to be responsible for the whole stuff, but originally I just didn’t have this capability to do colorimetry thing. but I think now I have mastered the calculation of all metrics/variables for LM4.
in a word, I think this team deserve better management/planning, this is just team problem. Opple still is a great company.
6 Thanks
Precisely so. And that’s why teams are so important; to add their knowledge together. In time, everyone will learn at least a little about what their colleagues do, even if there is one who will always know more about a particular thing.
I just received my LM4 - did not have time to test it yet, apart from a quick check with daylight (obviously 100 CRI and no flicker).
I have a few questions, mainly for other LM users. How do you measure with yours? Do you point the flashlight directly at the LM4? Do you ceilingbounce? Did you make some cheapo “integrating sphere” in the shape of a white-coated plastic tube?
Can the LM4 log brightness over time (e.g. check how long a light can hold turbo, how the regulation works etc)?
Will test it with a few lights over the next couple of days. I have Nichia 519A, LH351D, SST-20, SST-40 and XHP70.3 lights in various color temperatures and CRIs here.
What’s the current status regarding reliability of measurements? Which are okay, which are not? I guess CCT and Tint are accurate? Flicker as well? CRI and R9 still guesstimates?
I put the LM3 on a table and I have a flashlight mounted on a tripod half a metre, a metre or two metres away. The beam is pointing at the white diffuser.
I try to get the flashlight as close as possible to the LM3 before it overpowers the sensor.
Sometimes each measument can be taken at one metre. Sometimes Moonlight isn’t bright enough or Turbo overpowers the sensor.
I start with the lowest mode and make my way up to Turbo. I use the Android app and take screenshots. Select the mode, wait a few seconds, take a screenshot, select the next mode etc. It’s pretty fast.
Once all the measurements are saved, I enter them into a table on my PC (i.e. CCT, CRI, x, y).
I use a UNI-T UT383BT to take lux measuments at 30 seconds with the flashlight five or one metre away. I charge the battery between taking measurements. I go from Turbo down to Moonlight.
1 Thank