I built 2 bench power supplies a couple of years ago. One of the sections of the circuit uses a conventional constant current source with a zener and npn transistor. Textbook stuff.
They used a 3,3V zener which isn't great and varies with temperature along with the Vbe junction of the transistor.
Is there s better way to reference the 3,3V?
Maybe a TL431?
I used a zener as a reference once... I don't think the zener is the problem as much as the current limiting resistor... I had to use very low PPM/°C ±5... The same as I use in the 4~20mA convertor... The normal resistors are the temperature sensors...
I like the 5.1 and 4.7 Zener. +/- 0.03%/C
I must not understand what Mr. Rogers said. I can not see how a small change in the resistor will effect the voltage (much). The left hand two curves are for the 4.7 and 5.1 Zener. A 1% change in resistor will cause a 1% change in current and maybe a 0.1% in voltage.
It is hard to find a data sheet on a old 5% resistor (through home). The first one I found has 500ppm/C.
I went back and found a data sheet from 1950 showing a CC resistor with 0.2%/C temp. A Carbon Comp resistor should never be used where you care about temperature. LOL So a resistor change of 0.2% could cause a voltage change of about 0.02%.
I'm not the best when messing with analogue... But when using a zener diode place your fingers on the limiting resistor and the voltage does move enough to annoy!! when I swapped it out for the low PPM type, it was far more stable... It must have been a carbon type, maybe I wasn't driving the zener with enough current!!
Zeners are used as noise sources so dont forget to put a cap across it, I seem to recall a 5v1 gives a good stability.
A good current source just feeding a resistor would give a reasonable voltage ref, it depends how accurate you want to be.
I agree with the Tl431, its a very popular device.