The LM195 was basically a NPN common-emitter voltage amplifier with a threshold of 0.6V and an NPN Darlington Emitter Follower. It's performance was like a FET with a threshold of Vth=0.6V witha n RdsON of ~1Ω but with an extra diode drop on the output. The advantage of the LM195 was that it was rugged with current limit protection but not nearly as low RdsON as HEXFETs or modern power MOSFETs.
Also compared to smart high side automotive switches, it had poor cost*Vdrop performance, so when they changed LM195 to obsolete for new design then phased out.
This is not the 1st time an old product had an error in the application schematic, but then since it was just an example without specs, there are no numbers to benchmark.
TO make it behave better would require the Rb to change from 5k to a much lower value dependent on specs so it might be 10 Ω or 100Ω with a Darlington driver but depends on other factors like ,
how much much heat you can stand in this coffee cup warmer linear regulator.
Then again one could use a MOSFET bypass or even smarter...., add an inductor and make it into a SMPS.
And stilll smarter again.. use a common buck regulator IC.
There are just so many better choices these days.
Next time I insist you add specs with a question, like output current< load regulation, supply power and regulator heat loss.
But to answer your original simple question .. the simple answer is a constant current sink.
Since the design was a variable voiltage, the resistor must be variable with a need to turn on Vbe with 0.7V, so a 35mA CC sinkis better for load regulation at low In-Out voltage drops. This requires 2 transistors and 2 resistors... sorry no simple answer. :laugh:
Tony S