I am having troubles using the LT3092 current source with a BJT as current amplification device.
It behaves very strange.
My goal is this: I want to controll current from V2 from 0 to about 2Amps by setting V1 (in reality it is an output of an opamp). To extend the range of the LT3092 a BJT is used.
When simulating everything works as it should: with v1 increasing at the SET Pin, the same voltage is seen at the OUT Pin, thus controlling the current.
Reality: the current rises uncontrolled until V_IN-V_OUT at the pins of the LT3092 is at about 1.18V. More than 10uA come out of the SET Pin.
Why is there this discrepancy between Spice Sim and reality and why is the LT not keeping SET and OUT voltages exactly in sync but rather uncorrelated?
Any advice is welcome, how to debug it.
Thanks for the post. I don't have time to test your circuit using hardware at the moment, but I will mention that the LT3092 and other similar LDOs will act strangely if SET is driven too far away from OUT. See the link below to a post that discusses OUT and SET.
Keep OUT and SET the same as I mention, and then make sure the circuit is stable. To check stability, try to find a condition where your hardware will regulate and then just do a fast line transient of a few volts and make sure the current does not ring. I will update this post later if I get a chance to try your circuit with hardware. Usually LTspice is accurate in terms of stability predictions for these LDOs. Keep in mind the minimum load for the LT3092 is 500uA.
Ok, thanks, you don't have to build the circuit anymore, I found the problem (My work always progresses like this: Error search -> desperation -> Go to forum to get help -> find the error one minute after the post...). The beta of the bjt is much larger than described in the datasheet, a very simple measurement showed me it is about 280. Assuming the idle current of the LT3092 is 250uA (roughly measured with shunt resistors) we obtain: 0.25mA*280 = 70mA quiescent collector current without any control of the LT. I edited the SPICE Sim by using a higher-beta bjt and got the same results. Choosing a 1k instead of 10k resistor solved the problem completely because now all current is coming from the resistor (250uA*1k = 0.25V instead of 2.5V with the 10k resistor) and the emitter-base diode is not in forward mode. Thank you anyway!