100mA (or whatever) the easy way

[Terry Bites]

A simple circuit can make for a very precise current source (no, not a sink).
You need an opamp with a CMV range to V+.
I had an AD820in the drawer but any rail to rail input amp should work fine.
Use a better shunt ref if required.

[David Hess]

Replace Q1 with a JFET or MOSFET to remove the base current error.

[Alex Nikitin]

Replace Q1 with a JFET or MOSFET to remove the base current error.

There is no base current error in this circuit. It is a classic configuration of a current source, used at least in two units on my bench - Keithley 263 and HP3245A.

Cheers

Alex

P.S. - the inputs of the opamp need reversing!

[David Hess]

Replace Q1 with a JFET or MOSFET to remove the base current error.

There is no base current error in this circuit. It is a classic configuration of a current source, used at least in two units on my bench - Keithley 263 and HP3245A.

Ah, you are right of course.  This is the circuit which I commonly see in books describing voltage and current references.

[Terry Bites]

I make no claims to originality, for a fixed and upgradable source it's simple and handy.
It doesn't appear much in current source examples on the interwotzit

A MOSFET could be used but it will reduce the avaiable voltage compliance.
A bit more compliance can be squeezed out using a Sziklai pair instead of a Darlington.
Some of stability components don't go amiss.

[magic]

I think the opamp doesn't need to work to V+, its inputs normally see over 2.5V less than the output is.

OTOH, it needs to be functional down to V- if you want the circuit to work accurately with low resistance loads. And even if you don't, it must have no phase reversal or it could blow up on such loads.

The value of R2 makes the reference run at a fairly high current when load voltage is high, and starves it when load voltage falls below 0.5V. An active sink would fare much better.

[wasedadoc]

Ah, you are right of course.  This is the circuit which I commonly see in books describing voltage and current references.

Both of those figure show a V_REF developed across a short circuit.  Should be across R_S