LT1013: Question regarding the voltage controlled current source

Part Number: LT1013
Other Parts Discussed in Thread: OPA2180

Tool/software:

Hi eyeryone, I am designing a howland currents source which suppose to output -4mA to 4mA constant current through R9 while the Vin is 0v to 5.2v. On LTspice simulation it works fine, but in reality it doesn't produce enough current through R9 and the current isn't symmety as well (2.7mA to -4.7mA) ...  I tried to change the resistance of R10 but did not work well, should I use another op-amp or is there something went wrong in the circuit?
Thank you very much in advance and best regards.

 

  • Hi Liu,

    How are you measuring the current in your circuit? Can you measure the input and supply voltages going to your circuit as well as the voltages at the output of each amp?

    It may help to probe the output voltages with an oscilloscope to see if there are unexpected oscillations that don't show up on a DC DMM measurement.

    I also notice the center opamp in your circuit is labelled LT1012...  I assume this is an error? Can you confirm all three amplifiers in your circuit are LT1013?

    Thanks,

    Zach

  • Hi Zach, thank you for your reply.
    The LT1012 is indeed a typo, so acturally its 3 LT1013.
    I tested the supply voltage of opamps and they are 24.6v DC correctly.
    But when I measure the output of each opamp, it's not perform as the simulation result, and here is the specifics.
    Is there something wrong with one of these opamp or should I choose another opamp for better performance?


    Thank you!

  • Hi Liu,

    I agree these results do not match with the simulation. How are you applying the input voltages labelled Vin and V5 on the schematic? Are you able to measure these nodes as well to confirm V5 is 2.6V and Vin is 5.2V or 0V?

    Were you able to probe the outputs with an oscilloscope to check for oscillations or other unexpected AC behavior?

    How are you implementing the circuit? Do you have a PCB layout I can check?

    Regards,

    Zach

  • Hello Zach, sorry for the late reply.
    For the input voltage, I used a DCDC power source (adafruit PowerBoost 500), and for V5 I used a DAC module (MCP4725).
    I measured the V5 as 2.6V. Vin is set to be changed from 5.2v to 2.6v to 0v for testing purposes (each last 1000ms then switch to the next value).
    I haven't used an oscilloscope to test the ripple and noise of the power supply part, but I tried to provide 24.6v supply voltage with a DC power supply and got similar results...

    (Ideally, result should be 4, 0 and -4mA)

    Here is how the circuit was implemented.
    I was wondering what factor(s) could lead to this much error? Have you encountered similar situation before?
    Thank you very much for your help!

  • Howland current sources are very sensitive to resistor tolerances. You get similar results for the OPA2180 and LT1013, so it looks as if the amplifier itself is not a problem.

    See AN-1515 A Comprehensive Study of the Howland Current Pump for how to adjust the resistors.
    Analysis of Improved Howland Current Pump Configurations suggests to use an integrated difference amplifier with trimmed resistors.

  • Thank you for your reply! While designing the PCB I wasnt aware of the resistor tolerance problem, guessing now it's perhaps better to try with resistors with very low tolerance (0.01%) and see what happens.

  •  Here is my current PCB design, the unconnected part is GND.

  • Hi Liu,

    I suggest testing with high precision resistors and see if this resolves the issue.

    Regards,
    Mike

  • Thank you Mike, in the current prototype it uses 1% precision resistors for most part, but I was wondering if the precision of resistors can lead to so much error in howland current source. I have checked references but they didn't provide a quantitative result for the impact. Did you encountered similar situation before? 

    Beat Regards,

    Liu

  • Hi Liu,

    It looks like you are using rotary potentiometers for some of your resistors, labelled VR1 and VR2 in the schematic. It is extremely difficult to tune a rotary pot within 0.1% accuracy. This could be contributing to your error.

    I also notice you are relying on a 10k/10k resistor divider (R13,R10) to bring your 5.2V down to the desired 2.6V at the input of U1. In practice, the resistance of the U1 feedback path (R3 and R4) will interact with the 10k divider resistance and produce a large error. These 10k input resistors will also affect your gain in a way that was not simulated in your spice circuit. I recommend going back to your simulation including these resistors and verify the V5 voltage in your circuit.

    Regards,

    Zach