18-02-2018, 11:27 AM
Dave, my point about the cascode is that the high voltage transistor might well be good enough on its own. The tempco of a 5V zener is close to zero but that of the base emitter junction of the transistor is not. Hence the tempco of the zener should be approximately matched to that of the transistor. Though this calculation also needs to account for the emitter resistor. If the drop across the emitter resistor is large compared to the change in Vbe that will improve the stability.
One easy way of getting good stability is a current mirror which ultimately relies on the matching between 2 transistors. Again emitter resistors improve the stability.
Of course all this may be academic. The actual current from the CCS may be fairly uncritical. Another factor is the output impedance at the frequencies of interest. Again we don't know how much this matters. Any CCS will be a lot better than a simple resistor.
Al, the tempco of the LEDs or zener needs to have the same polarity as the BE junction. Otherwise it will makes things worse, not better.
LEDs are a handy way to get a not very precise voltage drop. For example I've used them to get roughly 5V from an 8V supply. You get a free pilot light too
One easy way of getting good stability is a current mirror which ultimately relies on the matching between 2 transistors. Again emitter resistors improve the stability.
Of course all this may be academic. The actual current from the CCS may be fairly uncritical. Another factor is the output impedance at the frequencies of interest. Again we don't know how much this matters. Any CCS will be a lot better than a simple resistor.
Al, the tempco of the LEDs or zener needs to have the same polarity as the BE junction. Otherwise it will makes things worse, not better.
LEDs are a handy way to get a not very precise voltage drop. For example I've used them to get roughly 5V from an 8V supply. You get a free pilot light too
www.borinsky.co.uk Jeffrey Borinsky www.becg.tv
