Help with using instrumentation amplifier and ADC circuit

Discussion in 'General Electronics Chat' started by naveenz50, Apr 9, 2015.

  1. naveenz50

    Thread Starter New Member

    Apr 9, 2015
    4
    0
    Hello all:

    I have put together a circuit using INA125 and ADS1115 to measure the output of a load cell. I am using INA125 to provide a 10V excitation voltage to the load cell. I am measuring both the excitation votage and the output of the load cell using ADS1115. I have powered the circuit from a 12V 5 amp-hrs lead acid battery. I am attaching my circuit diagram. However, I am encountering some unexpected results. My excitation voltage drops as time progresses. I would expect INA125 to provide a regulated 10V reference voltage. Below are the results. Note that the digital reading is the excitation voltage measured using a trim pot whose output has been buffered using an op-amp to provide a high-impedance input to ADC:

    date digital reading imputed votage ref voltage
    Tue Apr 7 12:20:26 2015 9793 1.22 10.00
    Tue Apr 7 12:26:33 2015 9686 1.21 9.89
    Tue Apr 7 12:32:30 2015 9790 1.22 10.00
    Tue Apr 7 12:38:46 2015 9784 1.22 9.99
    Tue Apr 7 14:09:16 2015 9606 1.20 9.81
    Tue Apr 7 14:18:13 2015 9624 1.20 9.83
    Tue Apr 7 14:23:53 2015 9633 1.20 9.84
    Tue Apr 7 14:29:26 2015 9629 1.20 9.83
    Tue Apr 7 14:34:49 2015 9611 1.20 9.81
    Tue Apr 7 14:48:32 2015 9598 1.20 9.80
    Tue Apr 7 14:55:08 2015 9561 1.20 9.76
    Tue Apr 7 15:13:01 2015 9563 1.20 9.77
    Tue Apr 7 15:18:53 2015 9558 1.19 9.76
    Tue Apr 7 15:24:34 2015 9558 1.19 9.76
    Tue Apr 7 15:30:03 2015 9552 1.19 9.75
    Tue Apr 7 15:39:18 2015 9543 1.19 9.74
    Tue Apr 7 15:45:04 2015 9531 1.19 9.73
    Tue Apr 7 15:50:43 2015 9513 1.19 9.71
    Tue Apr 7 15:56:28 2015 9491 1.19 9.69
    Tue Apr 7 16:15:57 2015 9479 1.18 9.68
    Tue Apr 7 16:22:21 2015 9427 1.18 9.63
    Tue Apr 7 16:28:27 2015 9432 1.18 9.63
    Tue Apr 7 16:35:10 2015 9432 1.18 9.63
    Tue Apr 7 16:41:31 2015 9411 1.18 9.61
    Tue Apr 7 16:52:22 2015 9393 1.17 9.59
    Tue Apr 7 17:04:10 2015 9340 1.17 9.54
    Tue Apr 7 17:10:30 2015 9348 1.17 9.55
    Tue Apr 7 17:21:12 2015 9326 1.17 9.52
     
  2. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    Why did you insert the NPN between the voltage ref output and the bridge?
     
  3. naveenz50

    Thread Starter New Member

    Apr 9, 2015
    4
    0
    I used the NPN transistor to boost the current supply. My load cell is drawing a current of 0.1 amps approx at 10V excitation which INA125 is not capable of supplying. (I have four load cells each with resistance of 350 ohms connected in parallel). The circuit of using TIP 29C is also shown in the datasheet for INA125.
     
  4. dl324

    Distinguished Member

    Mar 30, 2015
    3,250
    626
    What is the drop out voltage associated with the reference voltage? My guess is that your battery voltage is dropping enough to affect the reference voltage.
     
  5. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,693
    2,765
    You will never get any kind of precision or stability with this design.

    You cannot depend upon an NPN pass transistor to supply a stable and accurate excitation voltage for a load cell (regardless of the reason for the drop in emitter voltage WRT supply voltage).

    Believe it or not, your main problem is the A/D you have chosen. It has a reference output but not a reference input.

    If it had a reference input, you could drive the loadcell with *anything*, as long as you derived the reference voltage to be proportional to the loadcell current. The term is "ratiometric" -- and it eliminates drift due to changes in the loadcell supply voltage.
     
  6. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,693
    2,765
    Oh, and a word of advice:

    Draw your schematics with symbols, not IC pin diagrams. This way we don't have to work so hard to help you.
     
    dl324 likes this.
  7. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,693
    2,765
    Finally, I don't believe you need to drive your loadcell at 10VDC. It may be rated and characterized at 10V, but you could most likely get away with 5 or 3VDC.
     
  8. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,693
    2,765
    Question: are your loadcells paralleled on both inputs and outputs? If so, I have some additional comments.
     
  9. naveenz50

    Thread Starter New Member

    Apr 9, 2015
    4
    0
    Thanks joeyd999. Can't I perform ratio metric conversion in the software if I know the excitation voltage? And my load cells are connected in parallel in both input and output. How does that influence the outcome? Thanks for your help.
    Best
    Naveen
     
  10. naveenz50

    Thread Starter New Member

    Apr 9, 2015
    4
    0
    Joey. When I used 5v excitation reference voltage form INA125 I was getting a lot of instability in vout.
     
  11. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,693
    2,765
    The whole point of ratiometric is that you don't need to know what the reference voltage is. In fact, when configured properly, you're "reference" voltage need not even be stable.

    I've been working with load cells for the past 25 years. Paralleling outputs was the 'old' way. Unfortunately, for any kind of accuracy, each load cell must be "matched" to the others for both scaling and offset. This is expensive.

    Better would be to convert each load cell independently, and calibrate and compute total output mathematically. I'd use something like a TI ADS1243 which has 4 independent differential inputs, an internal PGA (up to 128 eliminating the need for analog preprocessing), and an reference voltage input range of up to 5V (allowing direct connection of the excitation voltage to the reference voltage inputs). The wide dynamic range (24 bits) will allow you to eliminate manual offset adjustments.

    The total output would be computed as: Output = C1S1 + C2S2 + C3S3 + C4S4 where Cn are calibration coefficients and Sn are the converted signal outputs of each load cell. The coefficients are initially unknown, but can be computed using simple matrix arithmetic with 4 known different inputs (plus an initial zero), and then saved in non-volatile ram.

    I am about 99.9% sure your instability is not due to 5V vs. 10V.
     
Loading...