This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS1235: Using only AIN2 and AIN3 for AC-Excitation - Scaling the Ref Voltage

Part Number: ADS1235
Other Parts Discussed in Thread: DRV8837, OPA2210, UCC27524, , ADS1261

Hello,

We are thinking about using the ADS1235 as new standard IC for load cell measurements.

I have some questions regarding AC-excitation and Ref-voltages. There is one dedicated reference input pair as well as the possibility to use AIn0 and AIn1 for reference input. The dedicated reference input pins will be used for the excitation voltage (5V). For low signal load sensors (~1mV/V) and low SPS this reference might be too high to get the full (noise free) resolution. We therefore would like to use a reduced bridge voltage as reference voltage (5V --> 1V via resistor divider) at AIn0/AIn1 and multiplex it when neccessary. 

  1. In figure 84 from the datasheet (attached), a typical connection is shown. I would copy that filter for AIn0/AIn1, but I am not sure where to place the resistor divider. Is it better to use position 1 or position 2? I am also not quite sure about the 100kOhm resistor which connects both reference inputs. I would guess it is for the input currents if the bridge is not connected. I am also not quite sure about the values. Would I need 400-200-400 Ohms or is it possible to use higher values?
  2. Is it possible to use the "4-WIRE MODE (CHOP[1:0] = 11)" and connect only the ACX1 and ACX2 signals on AIn2 and AIn3 via GPIO_DIR and GPIO_CON registers? I then would use a motor driver like DRV8837 to reverse the voltage at the bridge.
  3. I think I got from figure 82 the information that with ACX1 high (and ACX2 low) the reference voltage is expected non-reversed (higher voltage at REFP than REFN) and vice versa (in AC-excitation mode). Is that right or does it pick automatically the the correct polarity?
  4. To save current, I would like to disable current consumption from the bridge. For this, I would disable the AC-mode and drive both pins low, which leaves the output of the motor driver floating. Since the reference inputs are then connected to floating inputs, is there any problem with that (besides the fact that I cannot measure)?

The favored connection would then be:

REFP0 -> positive output of bridge driver

REFN0 -> negative output of bridge driver

AIn0 -> as REFP1 -> divided positive output of bridge driver

AIn1 -> as REF3n1 -> divided negative output of bridge driver

AIn2 -> as GPIO -> ACX1 

AIn3 -> as GPIO -> ACX2

AIn4 -> as analog input -> InP

AIn5 -> as analog input -> InN

Thanks in advance, Gerald

  • Instead of using a reduced ref voltage at AIn0 and AIn1 I could also imagine using the channels as input for an amplified signal. I would configure the two amplifiers (in e.g. OPA2210 ) as the gain stage in a three amp instrumentation amplifier with a gain of 8. Then I could amplify the small signals from the bridge by 4 (setting the PGA of the ADS1235 to x64) or by 8.

    Would you recommend to follow this path? It is quite hard to estimate the "real world" effect on the noise when I change the input stage since the ADS1235 has a lot of features to reduce noise.

  • Hi Gerald,

    Welcome to the E2E forum!  You have asked a lot of questions, but I have some overall concerns as to what you are trying to do.  You are attempting to lower the noise and offset error, but you are also adding in error at the same time.  For example, if you add an external amplifier, the measurement is not longer truly ratiometric and you have added potential gain error of the amplifier and have also gained any external noise prior to the ADC.  Although there is some potential benefit, there may be some other factors that could be a detriment.

    Also, if you use a resistor divider circuit you will be subject to error and drift of the resistor components.  If you make the values large, then thermal noise of the resistors will be added to the measurement.  There is also a potential settling time issue when charging and discharging the reference input filter while operating in AC excitation mode.

    What you appear to be describing is 2-wire AC excitation mode.  This method is used with the ADS1235EVM using the driver UCC27524.  This driver has much less propagation delay when switching and has been tested to work.  The DRV8837 should also work, but propagation delays may affect the results.

    If your bridge sensitivity is 1mV/V what kind of noise-free scale counts (scale resolution relative to full-scale load cell output) are you hoping to achieve?  What are you hoping to achieve using AC excitation?

    Best regards,

    Bob B

  • Hi Bob, thanks for your answer.

    Bob Benjamin said:

    Although there is some potential benefit, there may be some other factors that could be a detriment.

    This is the reason why I am asking here. I can calculate most effects from discrete components, but mixing an advanced analog IC with discrete components is not my daily business.

    Bob Benjamin said:

    What you appear to be describing is 2-wire AC excitation mode.  This method is used with the ADS1235EVM using the driver UCC27524.

    I found a side note on the documentation for the ADS1235EVM which describes exactly the requested mode "ALTERNATIVE 2-WIRE AC EXCITATION MODE: Control the UCC27524DR (Dual non-inverting) driver using ACX1/ACX2 connected to INA/INB, respectively. Set the ADS1261 to 4-wire ACX mode, but ONLY enable the ACX1/ACX2 GPIOs."

    Bob Benjamin said:

    If your bridge sensitivity is 1mV/V what kind of noise-free scale counts (scale resolution relative to full-scale load cell output) are you hoping to achieve?  What are you hoping to achieve using AC excitation?

    Until now we were using a discrete amplifier and "normal" ADC without PGA (as attached) and got around 15-16 Bit resolution @10Hz. Since some components are obsolete now, I am looking for a way to improve this. I am sure that the ADS1235 alone is a better solution than this, but I am still interested in improvement. If I could improve the noise by 1 bit for the cost of an amplifier it is worth think about it. But I read between the lines that you would simply stick to the ADS1235 without any modification. Is this correct?

    AC excitation is also one of this improvements. It costs a few mosfets/ a driver IC (<1€) but erases different (temperature dependent) offsets.

  • Hi Gerald,

    There is no doubt that you can fine tune a circuit to maximize the dynamic range using an external amplifier.  It is difficult to say if the benefits of the external amplifier will overcome the total noise.  The benefit of using the ADS1235 alone is that the excitation and the reference will result in the output signal noise/drift of the source to cancel in the measurement.  There will be source drift cancellation in the circuit you have shown, but the noise through the amplifier stage will be added to the signal as well as any external noise that may also be amplified.

    The second benefit of using only the PGA stage is input referred noise actually reduces with increase gain.  There is some limit to this as at the highest gain (128) as the gain is a digital gain, so the low noise effects will even out.  Using the internal PGA as opposed to using external gain does have benefit.  Might it make sense to use a combination?  Certainly, but you must also consider the timing of the output data and any settling with respect to AC excitation.  As you change your excitation signal, the entire signal chain will also be changing as well.  The settling time may be significant and needs to be a consideration.

    As to your original thoughts on a voltage divider for the reference, this too will have analog settling behavior and there will be additional error due to drift of the reference circuit.

    In the end it is difficult for me to say which approach might be best.  You will be attempting to resolve very low signal levels and noise can greatly impact the result. From using the converter alone and gain of 128 with sinc4 filter, 10sps the noise-free bits is 20.8.  With the reduced range from 156mV to 5mV the noise-free scale counts is better than 200k (17.8 bits).  Using just this method will also simplify AC excitation.

    But if you attempt additional external amplification or a change in reference you may increase resolution but you will have other factors to deal with and AC excitation will complicate things as I already stated.

    I would suggest getting the ADS1235EVM and trying your load cell and compare the results with just using the ADS1235 alone.  You could then configure for AC excitation.  If you would like to compare to your original circuit, you could take the output of your current discrete approach and connect to the ADS1235 inputs.

    Best regards,

    Bob B