Constant current source with supply voltage swing

[haecifer]

Hello everyone,

I am looking for a simple current source that is able to provide a constant(ish) current even with a small supply voltage swing.
My ultimate goal is to use this current source to provide a stable reference current to a current mirror I have.

As an example of the problem that I am facing, please consider the image attached. The current mirror is being polarized by the resistor R4. However, any variation on -Vcc will change the reference current considerably and that is an unwanted effect to me. Thus, I need a current source with a certain resistance to supply voltage variation to polarize transistor Q6.
One limitation that I have is that I can only work with transistors, resistors or capacitors; so no zener diodes allowed.

Is anyone aware of a solution to this problem?

Thank you all in advance!

[Ian.M]

Disconnect the top end of R4 from Gnd and stabilise it with a 'Vbe multiplier' between there and the negative supply, with a pullup to Gnd or the positive supply depending on the headroom required. 

[Jay_Diddy_B]

Hi,

Welcome to the Forum !!

Here is another idea that might work for you. It depends on how accurate you need the current source and over what temperature range.


I have added R1 and D1 to stabilize the voltage across R4 and therefore the current.





D1 was chosen at 6.2V to be relatively stable with temperature.

Regards,

Jay_Diddy_B

[Ian.M]

@Jay: O.P. said "One limitation that I have is that I can only work with transistors, resistors or capacitors; so no zener diodes allowed. "

[Zero999]

To be picky what you've posted is a current sink, not a source. Reverse the polarity, diodes and change the transistors to PNP and you have a current source.

Here are two ways to make a constant current sink which doesn't vary much, as the power supply voltage changes.

[haecifer]

Thank you all for the input, it was very helpful to me.
The Vbe multiplier that was suggested by @Ian.M (and was also on the image @Hero999 posted) will do the trick for me so thanks once again.

[Zero999]

No a V_BE multiplier is something completely different.

[Ian.M]

+1
Its basically a 'roll your own' Zener substitute but with much worse slope resistance and temperature stability