SPD simulator for DIY builders

ChiefRunningPhist

Well-Known Member
The question of how much of a certain color is often asked. I'm not professing to know the exact ratios, but attached is a little excel file to help you figure it out.

You only need to enter the number of chips, your desired brightness from 0 - 5 and then a custom efficiency if the LEDs are less efficient than the ones I've modeled the file after. The wattage is populated after you've entered your brightness, and this is the value you'd want to base your design from. ALL inputs are based on a per ft2 basis. The PPFD is assuming that the # of chips you entered is meant for 1ft2, so if you had an 800 chip panel you were using in a 2×4, in order to model average PPFD, you'd take 800 chips and divide by 8ft2, to arrive at a total of 100 chips/ft2. PPFD is the only cell that will be effected like this. μmol/J, system efficiency, ect should all be golden regardless. Its not an exact science but just to get you close to your target. The chips used in the file are actual chips and the most efficient in their respective color. The efficiency (if not customized) is dynamic and scales with added power for all the chips I was able to find multiple power/efficiencies for. The static efficiencies are the values calculated for max brightness so if anything they should increase with less than a value of "5" brightness.

You'll need Microsoft excel to use.

This 100 chip/ft2 example would end up scaling to 442.32W & 1500+ PPFD in a 2×4 area... Walls ect will absorb some of the photons emitted, so this is just a system estimator...
Screenshot_2020-01-20-00-46-26~2.png

55.29/ft2 × 8ft2 = 442.32W
 

Attachments

Last edited:

ChiefRunningPhist

Well-Known Member
Hopefully it works for you guys. I didn't know you could use Google Sheets offline otherwise I would have used that instead. I was building on my buddies computer first, and then transferred it over to mine after awhile so there might be an "Edit Link" pop-up that you'll get initially when first opening, just click continue and you should be fine. Has anyone been able to get it to work?
 

nachooo

Well-Known Member
Hopefully it works for you guys. I didn't know you could use Google Sheets offline otherwise I would have used that instead. I was building on my buddies computer first, and then transferred it over to mine after awhile so there might be an "Edit Link" pop-up that you'll get initially when first opening, just click continue and you should be fine. Has anyone been able to get it to work?
It works for me! Yes... lot of days waiting for the LIbreoffice Software to open it.. Lol! Nice done! Sadly my base leds are 3500K not 3000K
 

ChiefRunningPhist

Well-Known Member
It works for me! Yes... lot of days waiting for the LIbreoffice Software to open it.. Lol! Nice done! Sadly my base leds are 3500K not 3000K
Lol oh nooo! Hopefully its snappy enough when finally loaded. It stalls for a second on opening for me, but after it's loaded the graph is snappy and changes in RT with different inputs entered. I first started building with Google Sheets but as the data set grew it became real laggy trying to run it on a web-based program so I tried using excel and found it to be much better. I resorted to buying my own copy of office 365 to finish (reluctantly lol). I wish I would have remebered about officeLibre!

Ya, I need to add a few more PC LEDs (3500K, 4000K, 6500K ect), I originally had Opti's but after some messing around decided they weren't for me. Amber and orange were pretty lacking with just the monos listed, but the amber and orange monos were dismal efficiency so that's why I added the 3000k PC (modeled on the LM301H SK bin). It was an efficiency improvement but doesn't give quite the precise spectrum control I was looking for. Perhaps I'll make a Google sheet version but maybe I'll just make a website (or not do anything more lol) where one can go to use. Working on transferring it into JS rn so I can get an instaneous SPD graph on the web interface as I mess around with the various channel intensities.

If you look at the Samsung data sheets, there seems to be only a small difference between 3500K & 3000K, obviously enough to change CCT, but the red and blue peaks are centered pretty much exactly the same although the 3500K blue peak is ~80% while the 3000K blue peak is closer to ~70%. This is of course assuming that the charts I'm looking at are actually 3000K & 3500K. The data sheet seems to have a duplicate label?

Zoomed out...
Screenshot_2020-01-21-11-09-00~2.png

Zoomed in...
Screenshot_2020-01-21-11-08-51~2.png

Amplitudes...
USER_SCOPED_TEMP_DATA_orca-image-470378630.jpeg_1579630452277.jpeg

Overlay...
1579630673122~2.png

EDIT:
I'm guessing the graph labeled "3000K" is actually a 2700K, and that the first labeled 3500K graph is actually the 3000K...
 
Last edited:

ChiefRunningPhist

Well-Known Member
I've modeled the 3000K LM301H SK to emit ~2.91μmol/J at 65mA, or 0.177W, how accurate this is(?), idk. As you increase brightness or increase the wattage the efficiency scales to the point that at ~190mA or 0.55W/chip the efficiency drops to 2.583μmol/J.

The 6500k SM bin are touted as being 3.03μmol/J, but I'm guessing that there's a higher CE or phosphor conversion efficiency at CCT's closer to the base pump. That's why I've dropped the μmol/J down to 2.91 for the 3000K SK bin. It may actually need to be dropped more. If looking at HLG's R-Spec LM-79, they've a measured 2.6μmol/J. This is measured from a mixed array of PC white & 660nm red. One might surmise that without the 660nm the 2.6 figure would be lower (depending on 660nm efficiency), which would mean that an efficiency of 2.583μmol/J at 0.55 w/chip is actually too high, and may need to be adjusted lower. I've been told that the LM301H really are able to achieve 3.0+ efficiencies at low power, and these statements seem to jive with 3rd party data, although when using 3rd party data efficiency estimates at higher or more practical wattages per chip, the high efficiencies don't seem to jive with 3rd party testing. So I'm not sure exactly where these chips fit for true efficiency. The xslx file will give a decent approximation but one may want to edit or insert their own custom efficiencies given the ambiguity ect. I've tried to err on the side of underperformance rather than over performance, but perhaps using an even lower efficiency than given may ensure you're designing for the correct PPFD if the datasheets truly are overstating their efficiencies.
 

ChiefRunningPhist

Well-Known Member
I've modeled the 3000K LM301H SK to emit ~2.91μmol/J at 65mA, or 0.177W, how accurate this is(?), idk. As you increase brightness or increase the wattage the efficiency scales to the point that at ~190mA or 0.55W/chip the efficiency drops to 2.583μmol/J.

The 6500k SM bin are touted as being 3.03μmol/J, but I'm guessing that there's a higher CE or phosphor conversion efficiency at CCT's closer to the base pump. That's why I've dropped the μmol/J down to 2.91 for the 3000K SK bin. It may actually need to be dropped more. If looking at HLG's R-Spec LM-79, they've a measured 2.6μmol/J. This is measured from a mixed array of PC white & 660nm red. One might surmise that without the 660nm the 2.6 figure would be lower (depending on 660nm efficiency), which would mean that an efficiency of 2.583μmol/J at 0.55 w/chip is actually too high, and may need to be adjusted lower. I've been told that the LM301H really are able to achieve 3.0+ efficiencies at low power, and these statements seem to jive with 3rd party data, although when using 3rd party data efficiency estimates at higher or more practical wattages per chip, the high efficiencies don't seem to jive with 3rd party testing. So I'm not sure exactly where these chips fit for true efficiency. The xslx file will give a decent approximation but one may want to edit or insert their own custom efficiencies given the ambiguity ect. I've tried to err on the side of underperformance rather than over performance, but perhaps using an even lower efficiency than given may ensure you're designing for the correct PPFD if the datasheets truly are overstating their efficiencies.
Driver inefficiency will further reduce the total system efficency (ie the efficiency figures from the spreadsheet), I forget that from time to time.

Example:
2.5μmol/J system estimate & a 94% efficient driver.

(2.5μmol/J) × (0.94)
=
2.35μmol/J
 

dabby duck

Well-Known Member
Driver inefficiency will further reduce the total system efficency (ie the efficiency figures from the spreadsheet), I forget that from time to time.

Example:
2.5μmol/J system estimate & a 94% efficient driver.

(2.5μmol/J) × (0.94)
=
2.35μmol/J
Is that the best way to represent that? 2.5 as the output with a PF, lil lower than 100 is drawing a couple extra watts on the AC side isnt it? Not ultimately reducing the 2.5 ouput on the DC?
I get your representation though....maybe I am just confusing myself!
 

ChiefRunningPhist

Well-Known Member
Is that the best way to represent that? 2.5 as the output with a PF, lil lower than 100 is drawing a couple extra watts on the AC side isnt it? Not ultimately reducing the 2.5 ouput on the DC?
I get your representation though....maybe I am just confusing myself!
Haha I hear ya and had to think for a second myself.

μmol/J is a figure that takes the total μmol/s output and divides by the total wattage input (J/s). If the driver efficiency were 100% then you could simply add up all the individual chips' μmol/s emmissions, and divide by the sum of all the DC chip wattages because the AC wattage in and the DC wattage out would be equal. Though as soon as your driver efficiency drops from 100% your AC wattage in, is no longer equal to your DC wattage out. Though like you said, the wattage on the DC side, as well as the μmol/s emissions caused by the DC side wattage, remain constant. The total wattage increases, but the DC side spreadsheet estimates would be unaffected.

Example:
100W DC, 250μmol/s - 94% efficient AC/DC PS

(250μmol/s) ÷ (100W DC)
=
2.5μmol/J DC side


(100W DC) ÷ (0.94)
=
106.38297872W AC side


(250μmol/s) ÷ (106.38297872W AC)
=
2.35μmol/J AC


Derivation:
(Total μmol/J) = (μmol/s out) ÷ (AC watts in)
(AC watts in) = (DC watts) ÷ (PS efficiency)
(DC μmol/J) = (μmol/s out) ÷ (DC watts)



(Total μmol/J)
=
(μmol/s out) ÷ (AC watts in)


(Total μmol/J)
=
(μmol/s out) ÷ [(DC Watts) ÷ (PS efficiency)]


(Total μmol/J)
=
(μmol/s out) × [(PS efficiency) ÷ (DC watts)]


(Total μmol/J)
=
[(μmol/s out) ÷ (DC watts)] × (PS efficiency)


(Total μmol/J)
=
(DC μmol/J) × (PS efficiency)


"μmol/s out," divided by "DC watts" in, equals the "DC side μmol/J" efficiency figure. One can now see why multiplying the PS (power supply) efficiency, by the DC side μmol/J figure, will ultimately result in a "Total μmol/J" figure, or an AC side μmol/J figure.
 
Last edited:

ChiefRunningPhist

Well-Known Member
@ChiefRunningPhist , any movement or progress on our new light fixture yet, it must be nearly ready by now?
I think I've a photo of your prototype........ lol
I been slacking on the PFC, but I'm getting back at it. I can pump out a proto now but the power supply, while extremely efficient (99%), had poor power factor and if I were to run several units I'd want to have good power factor. Residential doesn't pay for PF but industrial does. There's a few options I can utilize, one is a dedicated PFC controller IC, but I'm trying to build one with Arduino. I took a break cause it was hurting my head lol but getting back at it. Hopefully have something soon. The spreadsheet needs to be added to the control app as well, and working on transferring into JS so that it can all work as intended. The holidays along with work and travel has made it a bit tedious and I just needed a breather, but getting back at it. If I don't develop an arduino solution soon I'll just buy dedicated ICs, they are really cheap tbh ($1-$10), but I just wanted to make my own as a bit of a challenge to myself.
 

ChiefRunningPhist

Well-Known Member
Ive been stuck on how an induced magnetic field collapses after initial current has been cut and not sure if the rate of collapse is dependent upon the amount of resistance/impedance in the electrical circuit in which the inductor is part of.

So if you're boosting voltage in a conventional boost topology, and lets say there's 2 scenarios, 1 scenario your storage cap is at 385V, and the 2nd scenario your storage cap is 40V, does the induced magnetic field about the boost inductor collapse at the same rate, regardless the opposing voltage in the inductor discharge path (from storage cap), or is the mag field collapse dependent on oppositional voltage? I think all the graphs and charts I've seen on inductor discharge characteristics are representing a discharge across a linear resistance, or linear load, and Ill be dealing with non linear loads so I'm not sure how this effects the discharge rate. I can determine energy stored in a magnetic field but I need to determine the rate at which that enegy is discharged so I can calculate switching Hz and PWM for the desired power needed. I think I can just assume that a 50% duty cycle will result in sufficient OFF time to allow a full inductor discharge. Its hard to get the answers for specific small details from Google sometimes, so I'm leaning towards just using a dedicated PFC IC, but I feel like I've put a bit of time into comprehending what's needed, so I'd like not to give up when I feel like I'm close, but then more time goes by so..
 

hybridway2

Amare Shill
Hopefully it works for you guys. I didn't know you could use Google Sheets offline otherwise I would have used that instead. I was building on my buddies computer first, and then transferred it over to mine after awhile so there might be an "Edit Link" pop-up that you'll get initially when first opening, just click continue and you should be fine. Has anyone been able to get it to work?
I wish i knew how to use the computer. I was and still am very interested in this since you first mentioned it but am dumb as an ox on the computer.
If anybody (especially you) would like to play with me sometime ill ask my mommy.
For real, I'd love to throw some nm's at that n see what comes back.
 
Top