In Reply to: current source posted by Francis L'Heureux on 03/28/00 at 12:44 PM:
: I want to know how design a current souce, maximum 20mA with op-amp or transistors. I'd appreciate if you could provide to me a paper or the formula.
: Francis L'Heureux
I got your e-mail.
There really is no one particular current source configuration which is better than all others. The two circuits I sent you are probably the best examples of that.
First of all, with the dual op-amp current source, I like to let both the two R1 resistors and the R2 resistors be 10K. That gives a gain of unity between the two input voltages and that voltage which gets impressed across R3.
The nice thing about this circuit is its versatility. Op-amp choice is simply in keeping with the current requirements. You can use tiny little surface mount devices for very low currents or you can use a really big, high power device in a modified TO-3 package for a current source in the multiple ampere range.
Of course, the voltage follower can be a low power device in either case since it only has to drive one end of an R2. The reason for the follower is that the output impedance of the current source is ideally infinity and the voltage follower prevents R2 from corrupting that characteristic. Although the follower can be eliminated with some additional analysis, I personally prefer to use it because it makes circuit analysis a very simple matter.
This circuit, by the way, is simply one variation on what is called a Howland current pump. A Professor Howland at M.I.T. first published it back in 1959. A few years ago, a series of articles on all kinds of current sources was published in either EDN or Electronic Design (I don't recall which) by National Semiconductor. Their applications people can probably help you find it.
One disadvantage of the Howland circuit however, is the need for matched components. Assuming your op-amps have lots and lots of open loop gain, the precision of resistor matching is important, perhaps even critical for some applications.
The JFET based circuit has some advantage in that regard. No matter what the tolerances on the three resistors which set the value of current flow happen to be, the circuit itself will remain a true current source at SOME current value, whatever that value turns out to be. However, this circuit requires a feedback capacitor for loop stability and is therefore not as fast as the Howland circuit is capable of being if high speed op-amps are used.
As before, I like to use equal value resistors, say 10K each, to drive the op-amps input voltage divider and then choose the third resistor in accordance with the desired current. Bear in mind though, there are no hard and fast rules for resistor value selection. You don't REALLY have to have unity gain via the Howland circuit and you don't REALLY have to have a 2 to 1 voltage divider for the JFET circuit.
Temperature stability of either of these circuits will be very good if very high gain and temperature stable op-amps are used.
John Dunn - President