P3D Re: White LED efficiency.

[Tom Hubin <thubin@xxxxxxxxx>]

Greg Erker wrote:

> >I don't think so. Think about a cube around the bulb. The condenser can
> >usually collect light from much of one face. A good afocal reflector
> >could reverse the light from the opposite face. But nobody uses these.
> >They use a simple spherical mirror which is too close to the filament.
> >As a result, much of the reflected light misses the collection optics.
> >Some of the light hits the filament itself and never makes it through to
> >the collection optics. The mirror is usually less than 85% reflective.
> >All together I figure you are lucky if you can capture and control 20%.
> >I have no references for this but simple geometric optics is where I
> >come up with this.
> 
>   I think projector bulbs are the exception. They have
> the filiament wound in flat coils so that the reflector
> images the coil back to fill in the gaps between the
> coils. Thus you have a flat emitting surface which will
> put most of the light towards the condensor or back to
> the reflector.
> 
>   So with projector bulbs I believe you would get
> more than 20% of the light to the film. THough I
> don't have numbers to prove this. Perhaps BobH could
> help there.
> 
>   With the bulbs that are usually used in viewers
> I would guess that you are correct about the 20%.
> 
> Greg E.

Hello Greg,

First problem is that the filament spacing is ill defined. For example,
the CZX 500 watt lamp used in the TDC116 has a filament coil arranged
like an 8mm square. There is little free space between the coils and no
spec on how that should be done. So any reflected light that gets
between the coils is by luck and not by design.

The second problem is the concave mirror that is too close to the
filament. The mirror is spherical and centered on the center of the
filament about 35mm away as I recall. Light from the center of the
filament will reflect from the mirror and pass right back through the
center. It will then follow the same path as the forward traveling light
to the collection optics.

Light from a corner of the filament will reflect and pass through the
opposite corner of the filament. Thus we have a mirror image of the
filament for the reflected light. But, the light is traveling at an
angle which is NOT the same as the forward traveling light. Much of the
reflected light will miss the collection optics.

The 85% reflection of the cheap aluminum mirror is also a loss.

A better arrangement would be to use a high reflectance flat mirror, say
97%, with an AR coated condenser lens to make an afocal reflector. The
condenser lens would be in place of the spherical mirror. Focal length
37mm with the filament at the focal point. Then the flat mirror at the
other focal point of the lens. Light from a filament corner would pass
through the lens, hit the mirror, come back through the lens, and form
an image at the opposite corner. The light would be traveling just as if
it was emmitted from this corner in the first place. Headed for the
collection optics. This arrangement would double the light collected
except for mirror losses and lens reflection losses.

But nobody does this.

To get around the coil spacing problem I offset the filament. I have a
1.7mm x 0.5mm filament coil in a 20w, 12v Halogen lamp that I use for
microscope illumination. I offset the filament by 0.6mm. The reflected
filament and the forward filament look like a pair of side by side
filaments that now occupy 1.7mm x 1.2mm. I use a simple 50mm radius
mirror. This is nearly as good as afocal because the mirror radius is
much longer that the filament size. 

Tom Hubin
thubin@xxxxxxxxx