I went ahead and tested some cascode BJT current sinks as a companion test to my EZ Sink tests from last week. These sinks are very similar to the C4S CCS. These were all biased around 9mA. The reference voltage is meant to be as part of the power supply bleeder network. I wanted to use a 2N3904 instead of a PN2222A for more gain and better rejection, but I didn't have any on hand nor any 1k resistors. Can we say blasphemy? That won't happen again.
Note that I used LTSpice just to draw the basic circuit and these are real bench tests. I started with a 1.3K resistor at R2 and swapped in a red LED to see if there was any difference - there was zero difference. These tests basically were to see how the quality of the reference voltage and its voltage drop varied the CCS. Walt Jung notes in his previously mention articles that the higher quality voltage reference helps rejection. We also know that a higher voltage at the base means that our current set resistor at R3 can be increased, thus increasing the effectiveness of our CCS.
References compared: LM336 2.5V, LM336 5V, 5.1V zener and 10V zener. I ran out of time to test a dual red LED ref - I need to do that, too. 1.3K and red LED for MJE340 base showed no difference.
Shannon
#1
20Hz 61M
1kHz 48M
20kHz 2.8M
100kHz 582K
#2
20Hz 48M
1kHz 29M
20kHz 1.7M
100kHz 365K
#3
20Hz 68M
1kHz 43M
20kHz 2.4M
100kHz 511K
#4
20Hz 54M
1kHz 41M
20kHz 2.4M
100kHz 499K