voltage to bidirectional current converter

[jonn]

Hello everyone

How to build a constant current generator from (-1mA to 1mA) controled by voltage. Google gives me links only to one direction current converters mainly 0-20mA. Is there a way to decaple DC from a current? I need my current source to measure impedance using 4 leads, so the current dont have to flow to GND but from lead1 to lead2.
Anyone got any idea how to build something like that?

thx
http://sonysmarteyeglassforums.com/

[Dr. Frank]

Easy, that's a basic circuit which can be found in every EE educational book, I think
I used it to generate LF currents (<1kHz) between 100µA and 100mA, from an 1 VAC input.
Design works well and stable. Phase of current output is also very predictable (nearly zero).
I used a low THD power OpAmp LM1875 most successfully.
+/15V, 200mA PSU required.

The circuit had to be adjusted once, for zero DC current offset output, and for exact 1:1 gain (by 50 Ohm trimmer) of the overall circuit, so that it works as a real constant current source, ie so that the output impedance is as high as possible. Works also fine on inductive loads.

Relays were used, to steer that circuit by PC, but also to not switch AC voltages over long wires.

As the output is referenced symmetrically to GND, no decoupling capacitor is necessary.
The two inputs allow to superpose AC constant current plus DC constant current.

The AC voltage sources I used ( Krohn Hite 4025, 4400A) had a very stable (constant) 1V output, and by this design always had constant 50k input impedance, so that no change occurred, when switching the vernier ranges (Re0-Re3).

Frank

[David Hess]

Search for "Howland current pump".

An alternative is to use a constant current source of I current with a variable current sink that adjusts to from 0 to 2*I.  That allows an output current of +/- I.

There are also integrated operational transconductance amplifiers like the LT1228, LM13700, and  OPA861 but those are probably not what you are looking for.

[Sigmoid]

The main problem I see with Howland current pumps is that one has to trim the resistors to make sure that exactly R1/R2 = R3/R4.

I wonder, since these days there's an IC for everything, has anyone ever seen a Howland Pump, or other voltage-controlled current source/sink IC?

[David Hess]

Matched resistors are not difficult to come by but if one wants to avoid them, then an instrumentation amplifier and operational amplifier can be combined to make a bidirectional current source/sink with a ground referred input.

The operational transconductance amplifiers that I mentioned operate as a voltage-controlled current source/sink but have a limited differential input range and are not intended for this sort of application.

[Dr. Frank]

Hi,

the original Howland current source has additional drawbacks.

Beneath the output impedance trimming , the trimming also depends on the output impedance of the voltage source,  which feeds its signal into to input of the Howland generator. Therefore, the trimming cannot operate with changing voltage sources.

The third disadvantage is the common mode offset of the OpAmp, due to the differential amplifier circuit.


The schematic I have shown, is an advancement of the Howland source:
The trimming is very easily done with that one trim pot, but it is independant of the voltage source output impedance. Furthermore, the input signal is related to GND, not differential.

Both OpAmps have no common mode deflection. This increases stability / precision of the source.

Frank.

[David Hess]

The poster did not say what kind of accuracy is needed so the errors of the simple Howland current source may be acceptable.  As you point out, the simple version has all of the common mode rejection ratio and input impedance limitations of a 4 resistor differential amplifier.  National, now scrubbed by Texas Instruments, published a pretty good discussion of the errors of the various Howland versions:

http://www.ti.com/lit/an/snoa474a/snoa474a.pdf

The two common alternatives I have used include using an instrumentation amplifier to generate a ground referred voltage proportional to current within the feedback loop of an operational amplifier:

http://www.linear.com/solutions/1185

PDF Page 65 of application note 47 shows a fast implementation of this:

http://www.linear.com/docs/4138

And using a pair of current sources of with a fixed I and a variable 2*I to generate +/- I.  This is common in high performance signal generation.

An interesting alternative which can be used as an integrated Howland current pump is the Linear Technology LT1990.  It is just a 4 resistor gain settable instrumentation amplifier but provides the external connections necessary to configure it as a Howland current pump:

http://www.linear.com/solutions/1399
http://www.linear.com/product/LT1990