I am pretty sure that when he posted this, the light had either a 1 or 3 watt Luxeon in it. A lot of the LEDs in that era struggled to have perfectly uniform color output across the entire emitting area, especially right at the edges. Early Cree parts were especially notorious for this - emitting blue at the edge, and more yellow in the middle.
On average and with diffuse reflectors, the effect didn't matter, but with highly focusing optics, there was color separation in the beam. You can see the yellow and blue color separation in the beam in various fringe areas. That used to be a huge challenge, even with higher end LED, when trying to match the Lux performance of incan filaments.
I did a lot of work with 12mm diameter reflector + fresnel lens setups - around $4K worth of parts, plus my time. It can work, but in order to really work, the assembly precision becomes really critical. I figured out a way to do laser aligned, epoxy bonded assemblies in my garage to make it all work. A few years later I was visiting a precision optical assembly place in Germany, and learned that for many years, virtually all precision optics are laser aligned, in multiple dimensions, and epoxy bonded, in temperature controlled environments.
If only I had hired an optical engineer in the first place, I could have saved so much time and money.
I used a Cree XML or XPG at that time. Given the 8-9cm at 2m I'm guessing it was the XPG. The blue/yellow distribution of LEDs is caused by how the phosphor is applied and not the issue I was describing. This color change is usually a problem when using extreme refraction angles. In my case you get a sharp image of the LED die with some white glow around it. I also have this with the Ahorton lens, so it is not caused by the rings of the fresnel, although it is possible that they contribute a tiny bit to this.
Regards,
Johan
I used Cree XM