[SOLVED] my simple current regulator build, maybe too simple?

I've been working for weeks/months on a simple bench supply. I've learned quite a few things that don't work, and better ideas, constantly starting over, bread-boarding, etc. What I want to accomplish is a simple linear bench supply with current and voltage regulaton, <30v, <7A. Once I have accomplished this, I want to explore LM723, L200, LT1083, dual tracking, and switching supplies. I know that 7A is a tall order, but I have already built, and fried, an LM317 based voltage/current supply(back EMF), so I want something more robust. It doesn't have to be efficient. I don't care about efficient at this stage of my learning. I'll include output protection diode this time around, promise. This is my sketch, plan:

First stage is the transformer: 10A/26v, diode bridge, caps.
Second stage: mosfet voltage pre regulator.
Third stage: CURRENT ADJUST circuit.
Fourth stage: LM338.

The third stage is what I am working on. I copied a portion of a circuit I found online, and it
seems to work well, with very fine control. I can light a couple of 12v/35w halogens at full intensity,
or, I can gently creep up on an LED. I mounted the twin TIP102 to a cpu heatsink.

Here is the original circuit, and the portion I removed, and modified.



I'm just working with the current portion. Here is my schematic, with mods:



If there is a better way, or if there seems to be problems, I sure would appreciate any feedback.
Maybe I need to place the current regulation before or after the pre regulation, or LM338?

Anytime you limit current the drop in voltage results in dissipated power, so the smart solution is to adjust your pre-regulated voltage to prevent heat build-up. I don't know if you use a Triac for pre-regulation, but there are much more sensitive and efficient current regulators using smaller shunts and Op Amp to drive MOSFETs with LC low pass filters to smoothen any oscillations if you have any. Ultimately any linear 150W supply with current limiting will have excess power wasted in regulation, so a SMPS is superior for efficient regulation.

Something like this for pre regulation:



My thought was it would help spread the heat, and MOSFET might run cooler.
Anyway, the LM338 will not give rated amps if the input is too far outside the
final target voltage. I can bypass the pre regulator with a switch for the higher voltages,
say, anything above 15v.

This is just to get me up an running. SMP will come later. Can't make a power supply
without, er, a power supply!