I've been wondering just how much of the lamp's output we are actually using. I plugged some formulas into a spreadsheet and I was unpleasantly surprised.

The basic concept of this calculator is that the solid angle created by a rectangular plane (LCD) at a certain distance from a point (BULB) can be approximated by a square or circle of the same area at the same distance. You can then calculate within a relatively small margin of error how much light actually hits your LCD. Unfortunately, you have to make some assumptions about how efficient some of our components are. Please let me know if you have a more reliable reflectance/transmittance factor, and I'll update the spreadsheet.

I sure hope we can make a better engine than this!

Update:
I've added calculations for a precondenser. I can't upload an excel file directly for some reason, so let me know if there's another format that will work for you and I'll try to post it.
Don't forget to adjust the angular magnification and precondenser transmittance to match your particular setup. The example values reflect a 25% increase in the (2D) angle of light that hits the rear fresnel because of the precondenser lens. (about .262 radians or 15°)

for lcd transmittance it looks like you have 20%(.2) it should be more like 6% on average (from research in the extreme mods thread) and I think stainless steel is 65% not 80.... plugged those in and get 86 lumens which I think is low, but who knows... building a new haas_man case now and planning on measuring most of these factors just out of curiosity, have lux meter as do some others, maybe we can get a better idea... good thread.... standard assumption is around 200-250 lumens

edited: just realized I was looking at the wrong column, now says 142 with a reflector...probably pretty close actually.....

Thanks mik, I think you're right about the LCD, I've seen that figure quoted in a Lumiled PDF file before. I found a spec for stainless steel mirrors that listed .86, but I assumed we wouldn't have perfectly spherical or polished surfaces. How about .75?

Ack! Looks like you were right about the stainless as well, mik. Updating...
I found a whole bunch of optics info: http://www.anchoroptics.com/documents/

this will come in handy, thanks for doing it, with any luck I can finish up my new pj(haas style, the wife saw it last night and was very impressed...of course if you'd seen the previous attempt(duct taped dog coffin ;-) you would be too....anyway I'm using a 250 watt setup so it'll be interesting to see how close it is for mine..was just looking at another thread,,,,had a spot 106 lumen measurement.....think it might have been ozstang in the stripping antiglare extreme mod thread.... you should start a "measurements" thread so we can all see what its capable of..... there are more than a few people with light meters ...

and I'm kind of suspicious of the fresnel and lexan percentages...I think elken measured his before and after taking the lexan out and it was like 12%.... I wish the search here took 3 letter (like "lux" ;-) or even the % sign..hard to search for stuff...

Research: I've found several specs for fresnels at ~92%. I've seen Lexan specs from 85%-92%, but rarely with thickness listed. One plastic distributor had 1/8" lexan is listed at 86%. Of course, these figures are for new, clean panels. We should all consider how many surfaces we leave open for dust to collect on and perhaps micro-filtration when planning our PJs. (That deserves a thread of it's own) Updating...

heres some measurements from another thread:
http://www.lumenlab.com/forums/index.php?s...opic=8534&st=0#

sigh, speaking of serching I always forget about google being able to do this(not great but its better sometimes than the search on LL)

"lux site:lumenlab.com"

hey, could you upload that as a regular spreadsheet? I can't get it to run on my home pc without having office2003, the web components I downloaded let me view it but not change any of the figures

Just something for you to consider. I believe the stated transmittance of these materials is at zero degrees incidence. So any light that is at an angle will suffer from reflection and reduce the transmittance

DJ

Agreed, Dazz. I've deliberately attemped to find transmittance factors that were tested with diffuse light sources to minimize this.

I've updated the calculator to include a precondenser.

Here is my post from that thread which attempts to explain the transmittance of panels (The Slightly off topic part):

http://www.lumenlab.com/forums/index.php?s...indpost&p=98562

Mark.

hey IG, starting with a 14000 mean lumen setup(250 W venture bulb) I measured 13000 lux (average,9 spots) hitting the first fresnel, standard LL setup, ikea reflector, ED28 formfactor bulb...that works out to about 1850 lumens( figuring a 9x12 fresnel) right? boy, I must need some serious tweaking I'm not even in the same ballpark as I should be eh? I realize your calcs are "perfect world" so to speak but..sigh.... and my first rough measurement of the screen lumens was avg of 20 lumens(ranging from 58 lux (center) to 11(corner) 9 spots on a 1.1 square meter screen .... yeah, thats what I said, I must be doing something wrong in my math.... or I've got a bad light meter!

What were your individual 9 spot readings at the first fresnel?
And would it possible for you to take 9 readings on the opposite side of the fresnel?

mmmm..I have it written down at home, I am planning on putting it into a spreadsheet the next day or so so I can keep track , I also started a lux/lumen methods and results thread... mostly so I can see if(ie "how badly" ;-) I am in alignment ... anyway, right offhand(I took the worst of the measurements together) it was like :
top row: 12 ,22,13
middle: 23 ,58, 11
bottom: 11,22,12

oops...you said the fresnel...no, I just remember it went from 15000 at the center to 11000 at the corner , added them up in my head and averaged them very roughly, I just wanted a ball park at that moment since I had solidifying the pj to do still...

however I will do them again, as I have everything screwed down now, I will do before fresnel,after fresnel...then (I am using unsplit optics) I will put in the 2nd fresnel and measure that , then after the lcd, then the screen.... prob in the next day or so...

edit: this was at the standard 8.75(roughly) inches from the arc center

you measured 22 lumens output on your projector, why do you think your maths are wrong?

We all would like to be 2000 lumens output my frend, but it is not posible with your setup. I always said 200 lumens is the averaged lumenlab. Some will have 250 maybe, you are contributing to drop this value to 200.

try playing with the positions of lamp.. fresnells...

well, if I moved up to the ushio or venture 400 watt I would still be getting only about 50-60 lumens right?(about 2- 2.5 x more lumens) ... I guess I'm wondering where the "200" lumen figure came from.... and want to compare to see if I do have a severe alignment problem(its not perfect but I don't think its that bad ;-), just finished my new case so yes, I will be playing with all the distances .... but actually, its a very watchable picture in the dark, I can't imagine how bright 200 lumens would be if this is 20!

why is it precondensor angula rmagnification 1.25?


the 200 lumes came from real readings done just like you did. some were higher, some lesser... 200 average was concluded.

well, I figured that, do you have any links? I'd liek to see their setup (and how they measured it to get some tips)... I started a thread for lux measurements and methods, it would be nice to have it in one area so someone can look at it and if their measurements are waaaay out of line they'll know womething is wrong and maybe where to look.. ...

one of the issues here may be a little confusion between lux and lumens
edit oops sorry spoke too quickly. didn't see the conversion

I don't have MS Office, and open office doesn't seem to help opening your html, so I can't tell exactly what you've done. However, I was just working on something like this last week - here is what I came up with:

[(IL/2)(GPF)+(IL/2)(GPF)(RF)(RR)](IRT)(UVT)(CFT)(PGT)(LCDT)(FFT)(MT)(TT)(AN)=FL
Slightly simplified:
[(IL/2)(GPF)(1+(RF)(RR))](IRT)(UVT)(CFT)(PGT)(LCDT)(FFT)(MT)(TT)(AN)=FL

IL=Initial lumens from lamp
GPF= Geometrical percentage from front of bulb - what percentage of light from the front of lamp intersects the LCD? One way to figure this out could be to take the area of the LCD and divide it by the area of the sphere with radius equal to the average distance from the lamp to the LCD. For a 15" this is about 255mm.
RF=Reflector factor - if you have a spherical reflector you theoretically double the GP. However, the light has to pass back through the lamp envelope. So a spherical would be 100%, but two glass passes make it closer to 77%.
RR=Reflector reflectance
IRT=IR glass transmission %
UVT=UV glass transmission %
CFT=Condenser fresnel transmission %
PGT=Plain glass transmission %
LCDT= LCD transmission %
FFT=Field fresnel transmission %
MT=Mirror reflectance %, for those with mirrors
TT=Triplet transmission %
AN=Apeture number - A "fudge factor" percentage representing the effect triplet diameter has on light transmission
FL=Final lumens

Sample Calculation

IL= 400W MH - 33,000 mean lumens
GPF= 15" LCD, 220mm fresnel - (LCD Area)/(Sphere Area)=(697/8171)= 9 %
RF= Spherical - 100%*88%*88%=77 %
RR= polished steel - 55 %
IRT= IR glass - 88 % *guess
UVT= UV glass - 88 % *guess
CFT= 80 % *guess
PGT= Plain glass - 88 %
LCDT= 9 % *guestimate, loosely based on data
FFT= 80 % *guess
MT= 100 %, no mirror
TT= 80 % *wild guess
AN= 100 % *no idea what this should be

[(IL/2)(GPF)+(IL/2)(GPF)(RF)(RR)](IRT)(UVT)(CFT)(PGT)(LCDT)(FFT)(MT)(TT)(AN)=FL
Slightly simplified:
[(IL/2)(GPF)(1+(RF)(RR))](IRT)(UVT)(CFT)(PGT)(LCDT)(FFT)(MT)(TT)(AN)=FL
Plug'n'chug:
[(33000 /2 )(.09 )(1 +(.77 )(.55 ))](.88 )(.88 )(.80 )(.88 )(.09 )(.80 )(1.00 )(.80 )(1.00 )=FL
Theoretical answer:
66 lumens !?

The 1.25 angular magnification was derived from a quick and dirty ray trace using a 30mm focal point lens and measuring the difference in the (2D) angle of additional light that hit the fresnel. It added about 15°. Ex: 75°/60° = ~1.25

I've backdated the HTML to Excel 97 and put the formulas next to all the values. Give it the download another try.

thanks , got it in excel now.... well I inputted my data, based on 14000 mean lumens...of course it could be as high as 18000( how would I measure actual bulb output? any ideas?) but I'd guess as far as the first fresnel I'm within 20% of the calcs.... will have to find some time for better measurements this week and see where I (or it) goes wrong ;-)
and this is where the value of this comes in right now, in that at least I have some theoretical numbers to shoot for... well see if rox or anyone else has any suggestions for mods to the calcs, now if we can get some measurements from others we can have some kind of base to work from ..... but yeah, it comes down to the light engine... wonder what the pro reflector will do....think its 90% reflective but smaller than the ikea...
well, plugging in just the 90% figure increase end lumens about 15% ...not bad for what I think is about 20-25$

No Problem. If this exercise has proven anything it's that small changes in efficiency can have a big impact on the final output.

Mikyd,
Measuring your bulb output: You have a round lux meter right? You can use the spreadsheet to help, It's already got the formula for the solid angle of a cone in it. Input the surface area of your LUX sensor (in mm², ex:2500) where "Area" is, then hold the meter at the same distance from your bulb as the fresnel. (ex: 220mm) In this position, the spreadsheet will calculate the solid angle from the bulb to your sensor surface and put it in "Solid angle of the cone"(ex: 0.051). It will also calculate what fraction of the bulb's output you are measuring at that location and put it in "Spherical output of the bulb directed toward the fresnel"(ex: 0.0041). Take an average reading from several locations. Multiply your sensor area (in m­², i.e. mm²/1,000,000) by the average LUX reading to get the average number of lumens striking your sensor (ex: 11,000 X .002500 = 27.5). Then divide by the fraction of the bulb's output (ex:27.5 / 0.0041 = ~6776 Lumens).

I appreciate you trying to make it accesible, but it seems that although Open Office lets you look at Excel files, it cannot deal with them as html. I looked at the html source but couldn't make heads or tails of it. If it isn't too much trouble, could you post it as a straight Excel file (non html)?

yeah I was looking at it last night and it occurred to me I could measure the diameter of the sensor(yes, round, maybe 3" dia) and do the math...will do that later today...I did some tweaking with the bulb location and managed to even out my brightness better(tweaked by the old fashioned method...looked at white screen and randomly moved the bulb;-) ... before the first fresnel(at its location I got avg 16000 lux(really evenly distributed all 9 points were within 30 lux of each other, much better than last measurements).. I'll see how your method does tonight....the tweaking also upped my average for onscreen, maybe closer to 30 now..
so I can't add very well in my head..my average is 22.2 onscreen though it seems much brighter, possibly because it is much more evenly distributed, the center level was down, corneres up...

Grrr...I'll try it again. The server wouldn't let me upload an excel file directly last time. Failing that, I'll try to find other hosting and post a link.

hmm,. I tried to yesterday after I changed it to excel also and it wouldn't let me...lets see if it will take a zip of it

edit, well it will take a zip file but not an excel, must be some virus blocker or something? I removed it, it had my lumens in it, try it as a zip file

Okay, it let me upload a zip file, give that a shot.

yep, worked like a charm, thanks!

ok, just got done with some measurements upshot is I get approx 46 lumens out of my 14000 lumen bulb(unsplit config,ikea reflector,no lexan) , though the bulb has less than 100 hours on it so I'm thinking its closer to its init lumens of 18000 ..anyway, plugging the 14000 figure in gets me 62 lumens .... about 35%% of theoretical..plugging in 18000 gets me 80, about 75% more than I'm getting...(I put in 1.0 for the lexan to compensate for not having any)
I also did a quick 7 point reading AFTER the lcd(unsplit), I wouldn't bet on these but from 3200 lux at the center to 1060 on one of the corners, average was 1680 lux which on a 9x12 window(~.07 sq meter) is approx 117 lumens ..and my average on screen was 46 lumens.... which means a lot of my light isn't getting into the triplet? hmmm.... because of the small size of my projection I wonder if the light cone is diverging again when it hits the triplet, it sure doesn't look like I'm losing that much light though...hmmm(again;-) looking at the excel, with the 18000 figure as the initial lumens it says 106 lumens exits the front fresnel ......(since mine is unsplit, this would be the equivalent of mine after the lcd) actually considering how rough my measurements were thats pretty darn close to what I got after the lcd(117).... I may have to try a bigger projection and see what happens.... I do know from just putzing around with things that the smallest arc image formed is happening before my triplet, by at least an inch or two, but to project a 3x4 foot screen I need to have the triplet further out than standard.... b ut could it be that much of a lumen diff? see, this is why I like measuring..makes you stop and think..thanks for this tool IG....given me lots to think about

hmmm....sounds like your lamp placement is no longer optimal for such a small projection?
ie the arc image is focussing too close and mucho light is not going through the triplet
you could try moving the light closer to compensate...

yeah, that occurred to me, I'm going to add another inch to the bulbs adjustability and see...

hey IronGecko.....would it be like super hard to figure out an ellipsoidal reflector as an option on this calculator?

What do you think of this Mikyd?

My first impression is that you'd get 3/4-7/8 of the availible light passing through the aperature (which is a huge improvement). However, the bright center/dark corner problem looks to be worse than with a standard setup. It may be possible to correct for this by placing the lamp slightly away from the focal point of the ellipse or mounting the bulb end-on through the center of the ellipse.

I've been thinking. MH bulbs don't emit light equally in all directions (something the calculator completely ignores) but I assume that most bulbs output ratings are based on measurements done with real bulbs. I'm wondering if I should base the percentage of the spherical output available on less than a whole sphere. I'm not sure how much of an angle is blocked by the the ends of the inner glass tube. Some measurements and observations of light dropoff would be helpful. ...Anyone?...Anyone?...Bueller?...Bueller?...

Taking this into account would make the standard setup have a slightly higher possible total output. It would also make bulb mounting postion more important for elliptical and parabolic reflector light engines.

how would you measure something like that? ...have bulb will measure! as for the corner problem....well... depending on how much light you have I'm sure we could think of something... if you can suggest any readings I will do them... I think this basic calculator is probably pretty close.... so the elliptical...the second focus goes at the focal point of the first fresnel, right?

Correct. As for measurements, we need to know the dimensions of the arc tube, and dimensions and distance from the arc of any surfaces that don't pass light (ends?, base). Do some bulbs have opaque arc tube ends while others are clear? Does the arc completely fill the arc tube? Has anyone taken a picture of their arc while it was running or had a look at it through welding glass? Is the arc in the same position with a horizontal or vertical burning bulb? Lumen/LUX measurements at a constant distance and incremental angles from base to tip would be helpful in designing the ideal mounting position for any light engine. I wonder if this information is available from the manufacturers...google time!

well for the internal dimensions, maybe someone ahs a dead bulb the could break open? I know my arc tube is opaque at both ends...hmmmm maybe I can look at it with crossed polars? not sure what you mean by this: "Lumen/LUX measurements at a constant distance and incremental angles from base to tip " ... I am imaginatively challenged..got a drawing?

You know me. I'm always ready with a drawing. (Gotta love Visio!)

got it.. I can tie a string to the meter and mark off the angles on paper, I'll try to get it this weekend....

Heres a few pics, taken through welding glass, of my HQI 250w (double ended)

The arc is roughly 24mm in length by these pics:




The inner bulb is fairly diffusive so its hard to say what proportion of the light comes from the arc itself, and how much is emitted by the bulb surface:




hehe... might just try that

Rockin! Thanks MichaelJ! It's apparent that the entire tube produces light due to the diffusion. It also looks like the tube ends block a considerable amout of light. If mikyd's measurements bear this out, it could make a big difference in how the lumens should be calculated and how to design more efficient light engines. It also looks like the central arc is a bit brighter near the ends. (mostly the left) Do you see that IRL?

Click to view attachment

this is basically what you are discussing. This was posted somewhere in the forums some months back.

The picture shows how is the angle dependent brightness distribution for a G12 lamp and a double ended lamp in the 3 planes (X Y Z).

Amazing! Thanks Rox! That's better than I had hoped. I'm going to attempt to extract a useful solid angle from those graphs and update the calculator. Thanks everyone!

I would say that most of the light comes from the actual arc, first two pics. The third pic is probably suffering from over exposure. So most light from the arc, some light from the vessel.

DJ

nope, because in fact the arc is flickering - this is just a moment in its cycle when it's nearly extinguised
the 'overexposure' in the third pic is intentional - it shows what most people would see if they looked at the lamp -hopefully with very dark shades

Rox, I find that diagram quite hard to decipher - could you explain a little how to read it? eg which colours are relevant in this app?
is the dip in the orange curve (double ended) what you would expect looking down on this lamp or am I misreading?

the colors are the lamp cut planes (X Y Z). The graphic is a polar graphic. For instance at the lower side there is like 120 degrees of 4000 candelas brightness in all directions. That side is the one I would use on the rear fresnell side.

the circles are the candela values, (brightness). there you can see how the base (FC2) does decrease the brightness as well as the small glass imperfection on the tube.

Nothing more to coment, candelas Vs angle graphic represntation.