If we "begin with the end in mind", what we really want in an engine (stuff prior to the LCD):

1) Highly efficient ... percent of lamp output converted to illumination of the LCD.
2) Highly collimated ... light passing through the LCD should be perpendicular and only perpendicular.
3) Evenly distributed ... we need to overcome the vignetting problem of current DYI designs.
4) Properly colored ... high CRI, white is white
5) Low cost of ownership ... investment, lamp replacement, power usage, parts availability
6) Scaleable ... low to high power, small to large panels, low to high resolutions
7) Simple enough for DIY

Current DIY designs fail in 1, 3, and 4.

We do not use a high percentage of the light thrown off by the bulb, we have vignetting problems at a 4:3 ratio screen, and many experience challenges achieving an even distribution of colors due to bulb choices.

Commercial designs fail in 5, 6, and 7.

Commercial projectors have high initial costs, overwhelming bulb replacement costs, designs that are highly crafted to the particular model and use machining, prisms and the like which go way beyond a reasonable DIY approach.

So, I challenge myself and the "team" ... how do we improve 1, 3 and 4 without losing sight of the things we do well already? How do we illuminate the panel with collimated light evenly distributed with good white light?

well, not to derail this thread before it even starts...but I just want to point out that the lcd can make a huge difference, we've measured them anywhere between 4%(my last one, durachko and others) , 5%*dazzz') ,6%(simul8r) and the holy grail 8.5%(supraguy).... so the same light engine that produces 150 lumens on a 4% lcd could produce 300 lumens on another lcd...just something to always keep in the back of your mind...
as for vignetting, with a precondenser, I personally have seen vignetting in the mid 70% range, commercial pjs are around 90%
while keeping vignetting, and cost in mind. lets be more specific on the ultimate goal....
how about 500 lumens on screen with a 5%(seems to be the average) LCD ? most commercial pjs when properly calibrated and in cinema mode have been measured to be in the 400-500 lumen range....
to "eliminate the effect of an lcd transmisiiveness, and considering that we think that the triplet itself eats about 25% of the lumens(not sure if anyone has ever measured it but its been calculated to be about that) what we come down to that we need a light engine that will deliver highly collimated light on the order of about 13,000 lumens to the lcd.... while keeping cost, vignetting and color in mind...
theres your goal..... hows that sound?

That's too much loss for the triplet. For the LL Pro I calculate only a 3.7% loss.

really? well, Like I said, the "transmissiveness" of the standard triplet was calced at .76% by someone at one time, don't know how they arrived at that though, I like yours better
though I must say it seems pretty low, how did you calc that?

edit: if we use your figure then we need to deliver about 10,000 collimated lumens to the lcd....

Using this page and the index of refraction values for the types of glass used in the triplet.
http://hyperphysics.phy-astr.gsu.edu/hbase...pt/antiref.html

The HCI produces 14000 so we need an efficiency of about 70%. Sounds reasonable maybe

you would think so wouldn't you? paladin ahs pointed me to a couple of fiber optic illumminators using the cdm 150 sa/r that only produce 3700 lumen at the port though.... although that may be due to some peculiar property of fiber optics useage.... so there you have a reflector specifically designed for a 12,000 lumen lamp and the best they can do is 35% or so...seems really odd to me.... I would think 70% is doable, or at least close to it....
when I get luckys sa/r reflector I p[lan on testing it pretty thouroghly, though its designed for a 6mm arc not a 10 like I have...

400 out of 14,000 is an overall efficiency of about 3%.

I haven't seen anyone get over 1% yet...but I would like to

daz's new light engine design with the parabolic reflector and two lenses seems to hold promise.

and elken's satellite dish/wok reflector might do wonders...

Its not a WOK,,,, big holy in da centre,, tried it all the rice fell out hehehehhe