Best Way to Make a Current Source

The GPC-1850D has two adjustable current sources available. Although the adjustment will likely be very coarse as it's just a single turn for the full range.

Download datasheet for LM317 and refer the applications.

Gajanan Phadte

If the OP was referring to the Instek PS that he linked to, I think he made a typo; that supply has a fixed +5V output rated for 3 Amps, not 3 milliamps.

Tom

I looked at a current divider but not adjustable and also it dissipates too much power at the resistors.

I'm not sure what you intend to do with this current source, but consider this. A current source must be able to produce any voltage it needs to maintain the current. So, if your test resistors exceed a given value, your DC power supply will no longer be able to provide that voltage and you will not achieve the current you want. That, will dissipate too much power in your test resistors.

The other side of that is that you will have to use such a low current that the voltage measurement may be difficult. You will need a highly accurate DC volt meter, but you will also have to consider lead-wire resistance and contact resistance. All of this becomes especially difficult when your test value of resistance is less than 5 ohms. Not to mention, that too much current will raise the temperature of your resistor, thereby changing its value. It can be done, but there is more than meets the eye to consider. Remember, test the test to make sure it performs as you expect!

With your 5V/3mA power supply loaded at 100mA the voltage available can only be 5/33 = ~ .15V which may not be high enough to perform the testing you want to do.

Power In = Power Out so 3 each 100mA current sources will require the power supply to produce 300mA at whatever voltage level needed to perform the testing.

Why do you need to add any circuitry at all. This triple output bench supply can be simply adjusted to be a constant source power supply. Presumably you RTFM and you know how to achieve this. This implies that you need to have this supply still working as a constant voltage device for part of your circuitry while another part requires three current sources. Far too often the imp lies but I will assume this requirement for the rest of my reply.

The few times I must still use discrete transistor circuitry to build a current source I like to use the Widlar current source topology. This older circuit topology has many benefits for a hobbyist or somebody learning analog circuit theory that are not obvious when you look at most of the paperwork.

First, you only need to match the transistor Beta gains if you must have known, specific current values driving into the probably varying impedance load. Most of the time the similar but not necessarily identical current from this fairly high output impedance current source works quite well in a circuit. If you do need identical currents then along with matching the Beta by cherry picking transistors or purchasing matched pairs, it is good to keep both transistors at the same temperature by allowing thermal (only) conduction between the two transistors.

Second, if you need multiple (n) current sources then just repeat the Q2, R2 portion (figure 1 in link) for each additional current source. Varying the value of R1 in that circuit will change the collector current for all output transistors (Qn+1).

If you find that the output impedance with this topology is still too low you can improve this by going with a Wilson current mirror topology. The few times I needed this typically involved a need for very high frequency and/or output impedance reasons.