Opamps for Howland current source

[jonnybgood]

Can you suggest some opamps to be used in a Howland current source to output 50mA comfortably?

This is the circuit; https://www.indiabix.com/electronics-circuits/howland-current-source/



thanks

 

[Analog Ground]

You have started an interesting thread. Because of the relatively large current you require, 50mA, using the "Improved Howland Current Source" is best. Otherwise, the value of the four resistors is quite low and leads to a very low input impedance for the circuit. Check the link for a very detailed description of both circuits (regular and improved) or search "Improved Howland Current Source" for more information. Back to your question. To get this current, you will need a special op amp. A couple of questions:

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

1. What is the "compliance voltage". What voltage is needed to keep 50 mA into the load?
2. What bandwidth? How fast do you need to vary the current?

 

[jonnybgood]

Hi,
thanks for your interest. I will pass the 50mA through a range of resistance from 10m to 100mOhms. (The I have a precision non inverting amplifier to produce a voltage range from 0.1V to 1V.)

Back to the current source; compliance voltage I think is 0.05A x 0.01 = 0.5mV to 0.05 x 0.1 = 5mV.
Actually there is no hurry for the current source to settle to 50mA. So the bandwidth can be even 1Hz

My main concern is the tolerance of at least 0.1%. so 50mA+/-50uA. It would be very interesting (helpful) if we could
explore other current sources preferibely not the 'voltage regulator' ones like the LM317. I already tried the LM317,
L200 and TL783 and they were successful within these specs. I would like to try with opamps and maybe transistors.

 

[Analog Ground]

I understand, you are making a milliohmmeter. I love this kind of circuit. One easy solution could be a dedicated current source IC such as the Linear Technology LT3092. A Howland Current Source seems more than you need since it is bipolar out. If the LT3092 or one like it doesn't meet your need than probably a combination of voltage reference, op amp and output drive transistor.

https://www.linear.com/product/LT3092

Can I assume you are familiar with the requirement to carefully configure a four terminal measurement? Measuring milli-ohms takes a lot of care with the connections and self heating of the unit under test. If you want 0.1% tolerance than you plan to measure micro-ohms? This is very, very hard. So, I have to question the accuracy requirement. Is this a special laboratory measurement?

 

[FvM]

If you intend an unipolar current, an OP current source with output transistor (preferably MOSFET to get rid of base current errors) seems best suited.

Accuracy is determined by resistors, particularly shunt resistor and reference voltage.

 

[jonnybgood]

0.5% accuracy would be fine for me. LT3092 is not available in my simulator. I would like to try and simulate the "combination of voltage reference, op amp and output drive transistor". Can you guide me to a good circuit? thanks

I will use proteus. Do you know of a better simulator for this application?

 

[FvM]

Here's a principle circuit. I add the LTSpice simulation file and a paper about current source dimensioning, which also applies to this circuit

 

[Analog Ground]

Here is another one which is close to your original post. It uses an improved Howland Current Source and a BJT on the output to increase the current drive to 50mA. For easy simulation, the reference could be a zener diode with a resistor in series from the power supply on on end and the other end of the zener connected to ground. Take the reference voltage from the node of the resistor and zener.

https://www.edn.com/design/analog/4...nimize-errors-in-grounded-load-current-source