Your voltages are so low that temperature change affects the amount of current in the LED.

Also, the current changes exponentially when the base voltage is changed which affects linearity.

 

Your voltages are so low that temperature change affects the amount of current in the LED.

Also, the current changes exponentially when the base voltage is changed which affects linearity.
Attached Thumbnails

Okay, I see what you are saying... If a Vbe of a certain transistor is for example 0.61VDC
Then:

1-0.61 = 0.39V a cross 15ohms
Will be: 0.39/15 ohms = 26 ma

However, if a Vbe of a certain transistor is for example 0.69VDC

1-.69 = 0.31V a cross 15ohms
Will be: 0.31/15 ohms = 20 ma

Which makes quite a difference!!

Right now, I only have 5VDC and 3.3VDC available and I don't have any regulators at hand that can give me 10V.

On another note, even if put RE (Led limiting resistor) on the side of the collector of T2, then I won't have 1V at T2's base anymore... I will have Vbe instead! and that's no good because T2 base has to ramp in the linear region only.

I can try to add in a series resistor from Vc of T1 to the base of T2 to limt current
in T2' base but then I end up with a transistor configuration that has no feedback resistor to keep beta stable from various transistors.

I tried all I can do. With an Re resistor of approximately 20 ohms and using the last diagram, I have tried many different transistors of pn2222 and with a heat source as a hair dryer for over 20 seconds as I measured transistors going from 23 Deg C to 50 deg C and in most transistors the led current varied only a 4 ma...

The circuit is to be used indoors!! When does indoor temperature ever get to 50 deg C. And on top of that worst comes to worst, I can always adjust a 4 ma offset by raising Vs a little. Don't you think we are splitting hairs here....

Thanks all
r