knna

Ive received a PM asking for some explanations about the spreadsheet. I prefer to answer it here and try to use a plain english for it. This is the best way people can understand how the sheet works and what each concept mean.

So if anybody dont understand something, please ask for it. Although i believe i already explained all the concepts in the sheet, i dont have any problem in trying it again without technical words, as many times as necessary to it being clear for most people.

BTW, please send me by PM the bulbs you are interested in analyze, if possible with the info required or a link to where to find it (lm, watts, SPD and orientative price).

Yes. The data usually avalaible are watts of the lamp, it emision in lm and SPD (spectral power distribution). We need to convert this into photons. The way to do it is calculate how much the bulb emits on the visible range.

The way the sheet do it is by calculating how many lm emits 1 watt emited as the SPD provided, and multipling it as many times is required to match the total lm emission of the lamp. As the sheet computes how many watts are the SPD, its easy to calculate how many watts emits the bulb.

Once we know how many energy emits the bulb, we can calculate the remaining data.

FC means Photopic Curve. Sorry, i forgot to change it from the original sheet in spanish (Photopic="Fotópica").

The photopic curve gives the relative human perception to each wavelenght. But the max of the PC (photopic curve), at 555nm is 1. 1watt of energy at 555nm correspond to 683 lm (by definition).

So, as we want to calculate how many lm produces 1watt emited in the way of the SPD analyzed, we need to use the figure wich convert miliwatts (mW) directly to lm.This is done by multiplying Watt*PC*683, or the same in the sheet: mW*PC*0,683.

Technically, the curve shown dont correspond to Quantum Yield, so QY curve express the photosynthetic effect (producing oxigen, O2) of a mol (or micromol) of absorbed photons of each wavelenght.

As it isnt avalaible (not any scientific study about Cannabis Quantum Yield), i built a aproximation, using the photosynthetic action curve from Inada, wich express the O2 produced for each mol of photons reaching the canopy at each wavelenght. To do it related to absorbed photons, i multiply it by cannabis absorbance at each wavelenght.

But this curve, as i explained in my last post, is refered to sunlight plants, and would be different for indoor plants. So the PUR value, calculated from this curve, would be valid only for plants in sunlight. I let it because it weights together the relative importance of spectrum and bulb's efficiency, giving a simple figure wich express the capacity of the bulb to produce phothosynthesis. But its necessary to understand its just a orientative figure, because it dont take into account capacity of adaptation of plants.

I use the SPD given by manufacturers directly. Units in the Y axis are irrelevant, because the sheet takes it as relative units. Always the ratio between each wavelenght value are coherent, the sheet works the same. Mark the higher value in the SPD as 100, and each wl value relating to it.

Ive found recently an accurate way to obtain each wl value: edit the SPD's pic (MS Paint works fine for this). Put the cursor in the bottom line and note the pixel value (it appears in the second box in the bottom, near the right square:first number shows the X axis pixel (wavelenght) and the second (decimals) the Y axis pixel). Do the same with the higher value in the SPD. Substract both values and you get the height in pixels (of the SPD). After that, go pointing with the mouse each wl value and calculate the corresponding value (pixels of each wavelenght*100/pixels height gives the 100-each wl emission, due the SPD goes from bottom to up, and pixels count from up to bottom).

If you resize the graph to be 100 pixels in height, this is very simple (any round number works fine).

Info from any graph is limited to graph resolution: if the graph shows wavebands of 10nm (as many Osram SPDs), you need to use the 10nm wavebands model. For SPDs made with 1nm resolution, its necessary to calculate the average of each 5nm waveband, or use the 1nm model (wich i didnt uploaded, but ill do if anybody wants it).

See answers 1 and 2. Lm just serves to calculate energy emited by the bulb. This translated to micromols of photons is the PAR value. The PUR value is PAR weighted by PC.

Because it calculates how many lm emits 1 watt of energy in the SPD way. The 0,683 factor correspond to miliwatts, and there is 1000 mW in 1 watt.

There is an accurate conversion between the energy a wavelenght carries and the number of photons wich carry it. Or said in other way, a photon of each wavelenght carry a amount of energy directly determined by its wavelenght.

As the sheet uses wavebands of 5 or 10nm, in order to minimize error, the wavelenght used is the center of the waveband (wl-2,5 in the 5nm model, and wl-5 in the 10nm model).

The 123105.6748 figure is derived from Planck's equation, wich quantify how many energy carries a photon of any wavelenght.

PPF is the number of photons emited. Weighting it by QY gives an aproximation to the photosyntetic capacity of these photons (as weighting it by PC gives the effect over humans, lm).

PPF is a objetive and true measurement of the light emited by the bulb.

While PUR would be the lm for plants. The problem is humans dont adapt their retina to each lighting, while plants adapt their photosynthetic systems in order to use the light the best they can.

Watts Nominal= The watts of the bulb. Mainly its refered to the watts provided by ballast (the wattage of the ballast that the lamp must use)

Watts Real= Watts the bulb uses actually (sometimes its different from Nominal, specially in blue enhanced HPSs). For example, some "400w" bulbs actually uses 430w. They run in 400w ballast, but consume 430w.

Watts system=Whole system consuption=Bulb Real (true) watts + Ballast Losses.

Klm= 1000lm

The only data not easily avalaible is the ballast losses. Its necessary to know the ballast model and check it in their website. The data used in the sheet for ballast loses are for good magnetic ballast for HIDs, and electronic ballast for floro tubes.

Those using HID's electronic ballast may take into account that they have smaller loses, but actually they make work the bulb with less watts.

EFF is for Efficiency, respect to total consuption. It express how many watts are emited as visible (PAR) light (as percentage of consuption, either Nominal, Real (true bulb's consuption) or system (bulb+ballast)).

Efficiency of Sistem of 30% means that for each 100 watt consumed, 30watts of light are emited.

Knna

__________________


Bulb Analyzer Tool

joyban

Hi

Knna,

Here are three SPD graph from the manufacturers and the bulb properties, if you could please give us a tutorial as how do we now convert the Graph into meaningful data to get the SPD values ?

OSRAM 865 LUMILUX



Lamp Properties:-

FQ 24W/865 HO UNV1
ILCOS FDH-24/60/1B-L/P-G5-16/549


Technical - Electrical Data
Luminous Efficacy in lm/W 67 lm/W
Rated wattage in Watts 24 W

Technical - Light Technical Data
Luminous output in lumen 1600 lm at 25°C
Max. lumen output at 35°C in lumen 1900 lm

Technical - Colors
Colour appearance LUMILUX Daylight
Colour rendering group 1B
Colour rendering index (Ra) Min. 80 Max. 89
Colour temperature in Kelvin 6500 K

Technical - Geometries
Length in mm 549 mm
Tube diameter in mm 16 mm

Technical - Life
Average lamp life in hours 24000 h with preheat ECG
Economic life in hours 18000 h with preheat ECG

The Next One Is Philips 865



Full product name MASTER TL5 HO 24W/865 SLV
ILCOS code FDH-24/65/1B-L/P-G5-16/550
Rated Lamp Wattage[W ] 24W
Colour Code 865 [CCT of 6500K]
System Description High Output
Cap-Base G5
Cap-Base Information Green Plate
Bulb T5[16mm]
Energy Efficiency Label (EEL) B
Mercury (Hg) Content[mg ] 3.0
Colour Rendering Index[Ra8 ] 85
Colour Designation (text) Cool Daylight
Colour Temperature[K ] 6500
Lamp Luminous Flux 25°C EL[Lm ] 1650



The Last One Is from GE:-




Watt 24
Length 549
Product Code F24W/T5/865
CCT 6500
CRI 85
Life 12hr Cycle 36000
Initial Lumen - 35 1900
Initial Lumen - 25 1600
Product Code 90266


So 3 Different types of Graph, and how do we now calculate the SDP Values from each of these ?

A small short tutorial will help all of us a lot ?

Regards

Sujoy

knna

Ok, im going to try to explain how i do it.

Im going to start with the first SPD (Osram Lumilux 860), because its a graph in 10nm wavebands, so its possible to do all by hand.

I right clicked the pic and selected "Save Image as". I go to the destination folder, and again right click in the pic file, and select "Edit". It open the pic with MS Paint. Then i click in "See" (sorry, i dont know how the English version call it, please correct me if i translate it wrong) and select "Zoom" and "Large Size" (its 400% of initial pic).

Yes, i know, the image dont fit in the window, but that isnt a problem.

I put the pointer at the bottom of the page, and note the pixel number wich shows in the bottom right corner:301 (the number after the comma says the Y axis pixel number)

Then i go to the higher value in the graph (531-540nm waveband) and note the pixel number: 69

So the height of the graph is 301-69=232 pixels

Now im going to calculate the value for the first column, corresponding to the 370-380nm waveband. I put the pointer in the top of the blue column and note the pixel number: 297.

So the height of the column is 301-297=4 pixels.

Im going to asign a value of 100 to the max (531-540nm), so the corresponding value to the 370-380nm is (4/232)*100=1,7

Same with the 380-390 column: 301-298=3
(3/232)*100=1,3. This would be the first value to enter in the sheet (the 10nm waveband model), in the E2 cell, corresponding to the 381-390nm waveband (notice the sheet call it just "390nm", i choosed to call each waveband by the last nm in it).

So i open the 10nm model sheet and write 1 in E2. (if you type "1" and press "Intro", the sheet is ready to enter the E3 value). Notice i round closest "full" number. It seems too much for a low value as 1,3, but probably the graph dont have more resolution than that, due 1 pixel has a height of 0,43 (1/232). This is the reason because i zoomed the graph 4x its original size, thats a reasonable error margin).

Im going to do the same with the remaining wavebands, and typing it in the sheet.

Waveband Pixel Value Sheet's cell

391-400nm 276 ((301-276)/232)*100=11 E3
-410nm 264 ((301-264)/232)*100=16 E4
-420nm 267 15 E5
-430nm 191 47 E6

And so on until complete the 38 wavebands in the graph. After that, fill up the required field in the yellow part of the sheet, corresponding to bulb's lm emision and wattage and get the final results.

Ive done it and upload the sheet of this SPD. Note you can add as many differents bulbs wattages using the best SPD as you want. The sheet comes by default with 3 differents wattage options (or same wattage with different ballast), but you just need to select a yellow column, copy and paste it as many times as required.

This looks like a hard work, but its just required once. Anyway, im going to build another spreadsheet wich allows to enter just the pixel number of X axis and Y axis and the wavelengh size of the pic, and just going copying the pixel numbers of the SPD curve result in normalized values for the bulb comparision sheet.

Ive just done this by hand this time, to show how it works. Tomorrow, if i have enough time, ill continue with the "semiautomatic" digitalized SPD sheet and the 2 remaining SPDs.

knna

Attached Files

Osram865.zip (8.8 KB, 26 views)

__________________


Bulb Analyzer Tool

Quantrill

I'll offer an explanation of my technique for digitizing and giving value for the spectral output of each wavelength range.

First I save the image of the SPD graph in question, open it in "paint", go to the "view" menu, go to "zoom", go to "custom" select "800%'"(I'm blind)


then go to "view", go to "zoom", go to "show grid"



then I put the pointer on the horizontal axis at the first wavelength range I want to get data on. In this case I went with 400nm, take note of the horizontal pixel number =105. then go to the last wavelength range you want to measure. I went with 750 nm, take note of the horizontal pixel number = 509.

So 509-105=404 pixels across the total wavelength range you want to get data on.

750nm-400nm=350nm


so we have 404 pixels across the horizontal axis that is supposed to have 350 increments.

350/404=.87

Go to the "Image" menu, then go to "stretch/skew" adjust the horizontal axis to 87%.

This will adjust the image so there are the proper number of pixels across the horizontal axiz for 1 pixel/grid box to each nanometer.

Since I am lazy, and only want to get data once every 5 nanometers, I take 350/5=70

70/350=.20

"stretch/skew" the image again so there are only 70 pixels/grid boxes across the horizontal axis.

now you are ready to get the data to input into the spreadsheet.

Take the pointer and put it on the horizontal axis, take note of the vertical pixel number.

300

put the pointer at the top of the blue colored SPD graph at 400nm, take note of the vertical pixel number. 275

300-275=25

25 is what i enter into the spreadsheet in the 400 nm box.

repeat this vertical box counting for each 5nm range across the horizontal axis until you get to the 750nm box.


then enter the wattage and lumen data in the yellow corner and hit enter.

voila, pertinent info regarding lamps for growth.

Attached Thumbnails

   

__________________
Each man must for himself alone decide what is right and what is wrong, which course is patriotic and which isn't.

~Mark Twain

joyban

Hi Knna & Quantrill

any updates on calculating the SPD for the Philips & the GE Lights, Philips seems easy as it in in 5 nm band but GE seems complex because it is a lenier continous graph ?

Sujoy

knna

Im sorry i delay the second part tutorial so long, i was very busy last week. I expect to have more free time in May.

Anyway, what i wanted to explain has been already explained by Quantrill (thanks, man ). Its a faster way of digitalizing SPDs, by resizing the graph in the way each pixel correspond to the waveband size the graph used (1, 5 or 10nm mostly). This way, filling the SPD column consist only in introducing the height in pixel of each pixel (using the squared mode).

Quantrill explained it perfectly. i just want to note sometimes its impossible to resize the graph in just 1 step, due the "stretch/skew" tool dont accept decimals. So after resizing, check if the graph have so many pixels as wavebands. Very often, there is a 1-3 pixels difference. A second resizing often to 99 or 101% achive the goal of having same pixels as waveband in the graph.

But be aware this way may induce more error, because many times SPD graphs arnt in accurate scale, or because the graph becomes so compressed that is difficult to determine the height. Sometimes its better to resize to have 2 pixels for each waveband. This way the graph is more clear. Just remember to only enter one value for each waveband (of 2 pixels). Other way is resizing to 1pixel per waveband, but checking the original size pic when in doubt.

To show this problems, and the accuracy limit that importing data from graphs implies, i entered the Phillips 865 SPD data in the two ways (resizing to 1 pixel/5nm waveband and to 2 pixels/ 5nm waveband). I upload results, with near 4% differences. In my experience, margin error using the sheet are around 5% mostly, but it may go up to 8%.

In order to process the last SPD, wich has the higher resolution (smaller wavebands, in this case 1nm or maybe less (spectroradiometers often gives 0,5nm and higher resolutions)), there is two ways of doing it:

The first is the faster one. It consist in resizing to 5nm wavebands and continue the same way than in the last SPD. But this way will induce large errors, apart of wasting a graph with high resolution.

The second way consist in resizing to 1nm per pixel and use the 1nm model.

Im redoing the 1nm model. When i finish, ill change the 5 and 10nm models too. Im doing some improvements:

-Include a cell including price, so perfomance/cost can be analyzed (with uE/$ ratios). Maybe some data about useful life too.

-Splitting final results in SPD data, valid for any wattage (PUR/PAR, R/B, Total Blue, PPF/Klm, PUR/Klm and probably other new, as Red/FarRed in narrow and wide wavebands, in order to calculate phytochromes equilibrium) and Result data for each bulb wattage.

-Im recalculating data for Cannabis absorbance at each wavelenght. The data used until now correspond to Colombian Sativa at the end of veg stage. I corrected it to correspond to absorbance along the flowering period, and averaging some strains (12 in total, but a Afghan strain seems to be the average). Anyway, changes are relatively small, and relates mostly to the green region. These are the graphs i used to do it:








-The last change relates to the Quantum Yield curve used to calculate PUR. Until now, it was obtained from multiplying the absorbance curve and the action curve. But lately ive found that the action curve sued correspond to sunlight grown plants, and indoor curve would be very dependent on lighting used. As quantum yield curves are mostly flat with just a small reduced efficacy between 500-600nm when the absorbed photons are considered, im considering two options:

1-Correct PAR just by Cannabis absorbance. This would result in the sheet overstating green photons efficacy.

2-Correct absorbed photons (PAR*absorbance) by a fully invented QY curve. Basically, with a triangle between 520 and 590nm, similar to other known QY curves (im unable to find Cannabis QY) and flat in the blue and red regions. This conversion wouldnt accurate at all, but probably represent better bulb efficacy, always as a orientative data (while PAR keep beeing the accurate and objetive bulb emission measure, but without taking into account photosyntetic efficacy).

I wait opinions about how do you prefer it.

knna

Attached Images

	MJ_Reflectance along Season.bmp (1.19 MB, 778 views)
	MJ_Transmitance along Season.bmp (1.32 MB, 776 views)
	MJ_Reflectance_strains.bmp (1.17 MB, 1448 views)
	MJ_Reflectance_strains2.bmp (1.14 MB, 759 views)

Attached Files

Phillips865.zip (26.1 KB, 15 views)

__________________


Bulb Analyzer Tool

penguin

Hi knna

Like you I have been extremely busy and haven't had a chance to finish the changes I mentioned in my previous post.

I think the changes you are making are great!

Can I suggest that in addition to (reference) pricing, manufacturer rated life is the other important factor contributing to relamping cost. For example, Hortilux 600w has a rated life of 16,000 hours, but Hortilux 1000w and 400w are rated for 24,000. However the retail price for the 600w is very close to that for the 1000w. So even though the 600w produces higher PUR/w, it ends up being more expensive to operate long-term and is thus less efficient. I agree with your previous post that both retail price and rated life numbers have limitations. Obviously pricing and availability vary widely. And my current version calculates replacement frequency based on replacing the lamps at a % of rated life, using the same % for all lamps. This is an oversimplification because some lamps have better radiance maintenance than others, so it should really be calculated at the lamp level. But that would require radiance maintenance data from the manufacturer, which I think is probably impossible to obtain.

I also added CRI and CCT ratings where I could find them, just to have a complete database.

Regarding the derivation of PUR, think it's great that your correcting for the flowering spectrum. I have some reservations about using reflectance spectrum as a proxy for QY but again that seems to the best currently available data, and is easy to change if direct QY data becomes available.

For additional calculations, it would be good to calculate total radiant W as well as PAR W, the latter data is available for a few of the lamps that have been analyzed thus far, and would be good to include for comparison with the manufacturer's rating, and also because of the adaptability of the light-harvesting system that you mentioned in a previous post.

I will try to clean up the version that I have been working on and get it uploaded soon, in case there's anything that you can use.


penguin

__________________

Quantrill

I'll choose option number two


either way you are the man.

__________________
Each man must for himself alone decide what is right and what is wrong, which course is patriotic and which isn't.

~Mark Twain

Quantrill

the error margin is probably pretty high on this one.

Attached Thumbnails

   

__________________
Each man must for himself alone decide what is right and what is wrong, which course is patriotic and which isn't.

~Mark Twain

sleepyrz

someone posted this somewhere else but i thought it might be used here



Full Spectrum Plant Grow Lights from Horizen Hydroponics
http://www.nam.lighting.philips.com/...df/p-5497c.pdf

Made to run on HPS ballasts but with a full spectrum of light. More complete than any other HID bulb produced today.

Check out Full Spectrum HID bulbs for more information.

Full Sun Lamps
Ceramic Full Spectrum HID or HPMH (High Pressure Metal Halide) are the most effective lamps for all stages of plant growth. One lamp for all, with no need for switching from Metal Halide (MH) to Sodium(HPS) during vegetative and/or flowering stages. These lamps have more reds than any type of HPS and also produce the perfect amount of orange, yellow, green, blue, violet and UVA and UVB spectrums needed to simulate the sun. Just like the sun light that hits the Earth, these lamps produce in phase energy, in wavelengths from 380-740nm. These lamps are the most efficient lighting source available, produced in sizes of 100w, 150w, 250w, and 400watts. Special Features: More reds than HPS lamps The right amount of blues Open fixture rated One bulb for veg. & flowering Runs on S51 400w / S50 250w Standard HPS core & coil Ballast

Not to be used with Digital Ballasts.