A Sun Simulator for Solar Thermal Collector Testing

I've been doing some solar thermal collector testing (here and here) and plan to do more.   The method I've used is to test a baseline collector side by side with the new design at the same time under the same sun.  This works well, but it has the disadvantages that you have to wait for good steady sun, and you have to run the test and collect all the data for both collectors, which is time consuming.

I've decided to have a go at building an indoor sun simulator that will allow me to do indoor tests of moderate size collectors.  If it works, this has the advantages of being able to test under the same sun conditions anytime without waiting for the right conditions, and without the need to always do side by side tests with a reference collector. 

I've run into some difficulties with the reflector design on the sun simulator and would appreciate any help or ideas you might have.

The stuff below goes through what I've done in getting to a sun simulator design, and the initial test of the design -- then into the problem I'm having with the reflector design.

NEW -- the "Production" version...


What Kind of Lamps to Use?

I should say first that indoor sun simulators are nothing new, and there are are many commercial and collector research sun simulators out there.  The problem with the commercial ones is that my budget is hundreds of dollars, and the commercial sun sim prices are more like tens of thousands of dollars and up.  So, the question is, can a person put together a useful sun simulator on a low budget.

My goals are to:

My first thought was to use halogen lamps.  I thought about the inexpensive halogen shop lamps and the halogen PAR spot lamps and did a little experimenting with each.  The problem with the halogen lamps is that their color temperature is low compared to the sun (about 2900K vs 5500K) and they put out LOTS of their energy in the IR.  So, while it looked like it was possible to use halogen lights (and some commercial sun sims do), it seems like the halogen spectrum being so heavily weighted to the IR would make testing results doubtful.

 Looking around at the various existing sun simulators, and the data on various lamp types, it looked like metal halide lamps offer the best overall properties --

Metal halide lamp spectrum for the lamps I'm using.

The spectrum of the metal halide lamps can be varied by using different mixes of metals in the arc tube.  The most common ones have a color temperature of about 4200K. 

Initial Setup

After a bit of thought I settled initially on four of the 400 watt metal halide lamps mounted on a backing board.  The lamp backing board supports the lamp sockets and the reflectors.  The lamp backing board is mounted to a base box that sits on the floor.  The ballasts are mounted on the base plate, and the weight of the ballasts on the base plate provides a pretty stable support for the lamp board.

The initial mounting of four 400 watt
lamps on the lamp board.
Its hard to see the lamps because
of the reflector material behind them,
but they are there.

Ballasts mounted at the base of the
light support board.  The grey junction
boxes are the connections to the
mogul socket mh lamps, which
are on the opposite side of the board.
The weight of the ballasts keeps the
whole thing steady and stable.

This was the initial arrangement -- see below
for the new arrangement.

Makes a lot of light!
Pulls about 15.5 amps at 120VAC

The pictures just above show the initial arrangement of four mh lamps on the lamp support board.  The idea is that each lamp gets a reflector, and the whole thing is positioned a couple feet away for the collector under test with each lamp covering about 1/4 of the collector area.  I later decided to go to 6 mh lamps instead of 4 -- these are on order.

So, this initial setup gave me the chance to experiment with the mh lamps some -- I have zero experience with them before.  Some initial impressions:

- They put out a lot of light :)

- The temperatures around the lamps were cooler than I expected -- the mounting board runs very cool.  Reflectors placed quite close to the lamps run relatively cool.  Having the lamps mounted on a wood board does not appear to  be a problem.

- The ballasts also run fairly cool -- about 170F is the highest temperature I've measured.

- As expected the lamps take a while to startup, and then a while more to reach full brightness.

I realize that this mounting arrangement is unconventional, and that there may be some safety and fire issues.  I keep a careful eye on the whole setup and do not leave it unattended.

I've since decided to go up from four lamps to six of the 400 watt lamps, and have ordered these.  The new arrangement has six of the mh lamps which are setup with the lamps oriented horizontally.  The new lamps are the T15 style, which are only 2 inches in diameter.  The combination of horizontal orientation and smaller diameter makes it easier (I hope) to do more effective reflectors.

This is the spec for the lamp...  From 1000bulbs ...   (I don't usually recommend companies, but I have to mention that 1000bulbs.com has been really helpful through two orders of lamps and ballasts and odds and ends -- real humans who know what their products plus good prices and fast shipping).  The lamp is 1.8 inches in diameter by 9.75 inches long -- 10 bucks a lamp.  The lamps produce light by maintaining an arc through the arc tube that you can see inside the out glass envelope.  The arc length is about 1.5 inches.  Each lamp has a ballast that provides the right arc starting voltages and maintains the arc current at the right level once the arc is established.  The arc operates at about 135 volts once established.  To save money, I bought ballast "kits" which have the ballast transformer and capacitor, but no case.

The rough logic for using six 400 watt mh lamps providing the needed light levels is:

- Each 36K lumen lamp will light up an about 18 by 18 inch (2.2 sqft) area -- six of these lights would then light up about 12 sqft -- a bit larger than the planned 2 by 4 ft test collector.

- I would like to be able to produce an illuminance level equivalent to full sun on the target -- full sun is about 10K foot-candles (fc)
(A foot-candle is 1 lumen per sqft)

- If the full 36K lumen were uniformly spread over the 2.2 sqft target, the illuminance on the target would be about (36K/2.2 sf) = 16K fc
This is about 1.6 suns, but assumes a perfect reflector, which is not realistic.

- Reflectors are not perfect, and one way to account for this is to apply a coeficient of utilization (CU)-- a value of about 0.6 appears to be typical.  The CU is basically the fraction of the light output from the lamp that actually illuminates the target area -- the remaining part being lost to other areas.
If a 0.6 CU was achieved, the the illuminance would be (16K fc)(0.6) = 9.6K fc, which would meet the goal of full sun.

So, I ordered the 6 lamps assuming that I could work out a reflector that achieved the CU of 0.6 and get the 1 full sun on the test collector.  That is proving to be harder than I expected. 

I've done these estimates in terms of lumens, which is a unit of luminous flux that is weighted to the response of the human eye.  The collector does not react to radiation in the same way as the human eye -- it is able to turn a wider range of light frequencies into heat.  But, it is difficult to find the data on lamps that would allow working over the full spectrum that collectors respond to, and it seems like if the numbers work out in lumens they will probably be reasonably close for the collectors thermal response? 


A good reflector that puts the lamps light on the target is critical.  Just to illustrate this, if you have a 400 watt metal halide lamp that is radiating 36000 lumens, that light energy is going in all directions.  If you have (say) a 2 ft by 2 ft target at (say) 5 ft from the lamp, then the percentage of the 36000 lumens that end up on the target is the ratio of the surface area of the target to the surface area of the 5 ft radius sphere -- which is (2)(2) / (4)(pi)(5^2)  = 1.2%   -- so, without a good reflector, very little light gets to the target where you want it.   Basically the CU without a reflector is about 0.012.

To make the six 400 watt lamp configuration work, I need a reflector that delivers at least 60% of the each lamps 36K lumens to the 18 by 18 inch target area for each lamp, and (hopefully) spreads it fairly uniformly.

1 st Try

At first look, a parabolic reflector seems like a good choice.  If the lamp arc is placed at the parabola's focal point, light reflected off the reflector surfaces will leave parallel to the parabola's center line.  So, if the big end of the reflector is about the same size as the target, it seems like a perfect match.

Here is a parabola that is 18 inches wide (to match the target), and 8 inches deep.

This is done on Parabola Calculator 2.0...

The problem with this is that if the target is about 2 ft away from the reflector, and the target is 18 inches wide, than a lot of light will escape out the sides as shown just below.  All of the light between the two marked light rays does not hit the reflector or the target and is just lost.

If you make the parabola deeper to reduce this area of lost light like this 24 inch deep parabola:

As shown just below, the deeper parabola greatly reduces the fan of light lost between the two light ray directions shown.  But, the making the parabola deeper moves the focal point (and lamp arc) such that quite a bit of the parabola would be inside the lamp glass envelope, which would be tough to do.  The focal point for the deep parabola is only 0.8 inches from the apex of the parabola, and the part of the parabola near the apex all lies within the lamp envelope.


This seems to make the parabola not so good a candidate for the reflector?  Am I missing something here?

2nd Try

The way I ended up doing the first cut reflector was to layout the lamp on the board in the picture below.  I then drew radials from the center of the lamp arc location at 30 degree intervals all the way around the circle.  I then set up a laser pointer at the arc location, and positioned a mirror within each 30 degree segment, and turned it until the laser pointer from the focal point would reflect off the mirror and onto the target.  The 30 degrees is a fairly close match to the desired target width in that the reflected ray off one end of the mirror hits the left edge of the target and the reflected ray off the other edge of the mirror hits the right end of the target.

This is, of course, not an exact method in that the lamp arc is not a point, but is about 1.5 inches long -- but the idea (hope) was that it was close enough. 

lamp arc is located where all the radials meet.
Laser pointer is pinned at the arc location.
The target is the piece of insulation board to the right.
The mirror segment has the small block of wood
attached to it below the laser pointer.

The mirror angle as adjusted for each segment so that
the mirror reflected the laser onto the target.
The laser dot can be seen on the target toward the right side.
This picture has the mirror lines drawn in for the first
few mirror locations.

The mirror layout I actually used is closer in to the lamp then the one shown in the picture above (see pic below).  The first segment lies within the lamp envelope, and can't be used.  I combined the next three 30 degree segments into one because the angles were close and it would have been difficult to fabricate these small single segments accurately.  By moving the first segments in closer to the lamp, the big end segments also moved in and got down to about the same width as the target. 

Building the Reflector Prototype

So, I ended up building a reflector with 3 segments that is 24 inches deep, and 18 inches in diameter at the big end.  The hole in the small end is about 2.5 inches in diameter, and just fits over the lamp with a quarter inch clearance.  See pictures below.

So, each segment of the reflector is a portion of a cone, which can be made from a flat piece of sheet metal bent around and joined.  I laid these out using the slant lengths of the cone from the diagram on the board.  The piece for the middle segment of the collector looks like this in flat:

I made a cardboard template for each segment, and then used the template to cut out the shapes in aluminum flashing sheet.

After the sheet was cut to shape, I glued aluminized mylar to the inside surface with spray adhesive.

Then bent the flat sheet around into the cone segment and fastened the edges together with short self tapping screws.   I cut the sheet an extra inch long to allow for the overlap needed for fastening to the other two cone segments.

Here is the finished middle segment:

I made the other two segments the same way, and then attached all three segments together to make the whole reflector.

The gluing job is a bit lumpy, but all in all this way of making the reflector seems pretty workable and produces a reasonably precise product.

The 2d outline for the built reflector is in the dark green lines -- six 30 degree segments in all.  Segment 1 lies inside the bulb, so it can't be built.  The next 3 segments (2, 3, and 4) are combined into a single segment of a cone on the built reflector as they are close in angle and it would be tedious to build them as separate segments -- the red line shows the cone frustum representing segments 2, 3, and 4.  
Segments 5 is a separate cone frustum in the built reflector, as is segment 6.

The built reflector.  It is made from 3 segments, each is a frustum of a cone.
The brown cone is segments 2, 3, and 4 from the sketch above.
The middle section of a cone is segment 5.
The right section of a cone is segment 6.

Light Pattern

To get an idea how this was going to work out without building all 6 reflectors, I set up the one reflector on one mh lamp and aimed it at a simulated collector to see how well the one lamp lights up its 18 by 18 inch share of the collector.  The lamp I used was not the one shown above as I've not received them yet, but its a 400 watt mh and should be pretty close.

The reflector encloses the lamp, and nearly touches it at the back end -- I was concerned about what kind of temperatures the reflector might see.  After an hour of operation, the back of the middle segment (hottest part of the reflector) was at 160F.  The smaller segment while closer to the lamp actually ran cooler at 120F -- so, no really high temperatures.  You can touch any part of the reflector with your hand at least for a few seconds -- most of the reflector is quite cool.  The aluminized mylar seems to be surviving OK.  I think that in the final design, I'll have a bit of forced air circulation around the reflectors and lamps.


This shows the light pattern on a white board representing the collector.  The squares are 6 inches.  The 18 inch target area goal is the blue dotted line that is a bit hard to see. 

These are the light levels measured with a light meter at the grid intersection points.   It should be mentioned that the light meter is only certified accurate up to 50K lux, and some the readings are twice that, so there is some question about the accuracy.

The table gives the light levels in foot-candles with the big end of the reflector 18 inches from the target.  So, the light level right at the center of the grid is 10,480 fc (a bit over 1 full sun). 

  12 inches
6 inches
Vertical Center 6 inches
12 inches
12 inches
6 inches
  1407 3717 1352    
1208 2462 10480 3643 880 352
6 inches
  1574 4182 2777    
12 inches

The big lighted circle has a radius of about 20 inches.  Outside the circle, light levels drop to about 60 fc.

So, looking at the picture and the table -- I'd tentatively conclude:

- While it reaches full sun levels at the center, the average of over the 18 by 18 target area is substantially less that full sun.  Maybe only about 35% of full sun over the full 18 by 18 area.

- The pattern has a distinct hot area in the middle, but drops off to much lower values well inside the 18 inch square. 

- Spill over from adjacent lights when all 6 are operating  might increase the light levels up to about 50% of full sun.

- Assuming the light is actually operating at 36K lumen output (an unknown), the CU is low -- maybe only about 25% compared to the desired 60%. 

If the "sun" is moved back so that the big end of the reflector is 33 inches from the target (pic below), the light levels drop substantially.  For example, the center of grid reading drops to 3800 fc.

light pattern with light moved back to 33 inches from the target.

As a way to compare "real" sun to the simulator, I did this little box below.  Its made from insulation board with and absorber of thin (0.01 inch) black aluminum and a plexiglass glazing.  A surface mount thermocouple is mounted to the back of the absorber.  The stagnation temperature that the thermocouple registers is an indication of the sun intensity.  This is not a perfect system for comparing simulator to outdoor in that the temperature it registers also depends on the ambient temperature.

For the simulator, the temperature reads 176F with the reflector 17 inches from the box and the shop temperature at 48F.  The other day outside with full sun (probably about 1100 watts/sm), the temperature read 220F.  So, this agrees with the light meter readings to the extent that it indicates that the simulator is not quite up to full sun levels.

The stagnation box that compares solar intensity.


3rd Try

The 3rd try uses a smaller diameter light with a parabolic reflector behind the lamp to reflect light from the lamp toward the target, and a long light tube in front of the light to channel light from the lamp toward the target. 

The parabolic reflector is about 11.5 by 11.5 inches.  It is shallow enough that the arc tube in the light can be right at the focal point of the lamp without any cutouts in the reflector to clear the lamp envelope. 

The light tube is 11.5 by 11.5 at the lamp end tapering to about 14 by 14 at the far end. 

The back parabolic reflector mounted behind the now
horizontal lamp with arc tube parallel to long axis
of reflector.  The two end plywood forms keep the
parabolic shape fairly accurately.

The light tube that extends forward from the parabolic
reflector to channel light from the front of the lamp
to the target.  The cutout is for the lamp socket.


This shows the whole works in place and shining on the target.  I realize this is not pretty, but I'll clean it up when I get to something that works.

This is the parabola used for the back reflector.  It is 12 inches at the big end, and 6 inches deep.  The focal point is 1.5 inches from the apex of the parabola.  The new smaller diameter lamps allow the arc tube to sit at the focal point without any cutouts in the reflector for lamp clearance.  The long axis of the arc tube is parallel to the long axis of the reflector.

The parabola is relatively shallow because if it is made deeper the apex moves closer to the focal point and the lamp envelope interferes with the reflector.  

So, the parabola takes light from the back half of the bulb (and a bit more), and directs it at the target.  The side walls that extend toward the target from the big end of the parabola take light from the front half of the lamp and direct it toward the target.  Without the side walls, a large fraction of the light from the front half of the lamp would miss the target.  Some of the light off the front half of the lamp requires two reflections to get to the target. 

Because the back parabola is a segmented parabola, the reflected rays don't come out perfectly parallel to the axis of the parabola, but a little work with the laser pointer at the focal point indicates that while they may not be parallel, they generally hit the target, and the few areas that are wide of the target mostly get picked up by the long light tube and then onto the target.  The arc tube is, of course, not  even close to a point light source, so that further spreads the reflections.  The long axis of the arc tube is not aligned with the parabolic cylinder, which should help.

This setup is working better.

This is the light pattern as measured by the Extech light meter -- in foot candles -- full sun is about 10K fc:

  12 inches
6 inches
Vertical Center 6 inches
12 inches
12 inches
6 inches
    11430 2080    
  13580 14560 800 1860  
6 inches
2510 7990 7160 3070    
12 inches

The stagnation box got up past 220F and was still going up a bit when I shut it down -- similar to the values in full "real" sun.

The 18 by 18 inch target area for one lamp is lit up pretty well -- its at 1.5 suns in the middle, and much of it is just above or just below 1 sun.  There is still a hot spot in the middle, but not quite as bad as it was, and when the other lamps are in, there will be some fill in of the areas toward the outside of each 18 by 18 area by the other lights.   It the hot spot is still too bright with all lamps in place, I'll try the reflective button in front of the arc tube that reduces direct radiation from the arc tube to the center of the target. 

This is with the end of the light tube 14 inches from the target -- I can get to lower light levels by increasing the distance.

The main thing that I like is that the light levels for the first time look like full sun is achievable -- I think that the uneven distribution is something I can fix at least well enough to do the job.

So, in the end there would be six of these 400 watt, small diameter lamps mounted in a 2 wide by 3 high pattern  -- each mounted just as this test one is mounted with a back parabola.   Before building the individual light tubes for each lamp, I'm going to try just a big light tube around the whole thing as has been suggested.  If that does not work, I'll build light tubes for each lamp.

The reflective material I'm using now is a thin  aluminized mylar from the local Planet Natural.  I will probably change to shiny thin aluminum (maybe just aluminum foil) for the parabola as its pretty close to the lamp.  The temperatures I measured on the back reflector were about 135F max after half an hour -- not so bad given that its now pretty closed in.  Thank goodness for the metal halides with the lower IR output. 

This may take a while as some of the ballasts and sockets are back ordered.  When 1000bulbs.com said they were out of the $3.58 mogul sockets for the lamps,  I tried our local Platt Electric -- they could get them in 2 days for $60 each!  -- decided I could wait.

Want  to thank everyone for the suggestions -- I've used some of them and plan to try more.

If you have any further ideas, I'd very much like to hear them


So, here are the questions:

- What can be done to the reflector design to get a higher percentage of the lamp light output on the target?

- What can be done to the reflector design to get a more even pattern of light over the 18 inch square?

- Is there a fundamentally different approach that would work better?

- Is there a better way to compare real sun thermal content to simulator thermal content than the stagnation box?

- One question I've not found an answer for on the mh lamps is that they have to run for a few hours before they come up to full output -- question is, how many hours, and how much lower is the initial output?  That is, are my lamps with only a couple hours on them not yet up to full output?

If you know of any successful, low budget sun simulator designs, please let me know.  If you know of any details on how the lamp/reflector designs are done on any of the commercial or research sun simulators, I'd love to see them.

Any ideas or suggestions would be appreciated.  There is a comments section below, or if you would rather email me directly, I'm at  gary@BuildItSolar.com

blog comments powered by Disqus