INA326 help

Discussion in 'The Projects Forum' started by saiello, Sep 4, 2011.

  1. saiello

    Thread Starter Member

    Jul 1, 2007
    24
    0
    Hi folks,
    Got hold of a few INA326's, have been perusing the datasheet and trying them out in a test circuit to try and get a feel for things. I have tried to follow the example circuit as I saw it on page 9 of the datasheet. Gain is arbitrarily set at 2.4 . I have some ASIC pressure transducers that I am intending to use in conjunction with these IC's. Unfortunately I don't seem to be getting the results I was expecting. When Vin- and Vin+ are relatively close together the gain drops dramatically. After a Vdiff of around 200mV things start to settle down. I have done some other tests with Vin- at 0V and 2.5V and I get very similar results with the gain rapidly settling towards 2.4 after 200mV. It is more than likely something I am doing wrong which I haven't understood from the datasheet. I have attached the test circuit and tabulated the results. If someone can point me in the right direction that would be great...! :D

    Thanks.
     
  2. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    I don't think it has anything to do with your problem but the output shows some non-linearity near the rail voltage, zero in your case.
    To make sure it's not that non-linearity that causes any problems you could use the setup in figure 6 to introduce an offset into the output voltage.

    If it's not this, I have no other ideas at the moment.
     
  3. saiello

    Thread Starter Member

    Jul 1, 2007
    24
    0
    Thanks for the reply. You say output shows non-linearity near the rail voltage? With Vin- at 100mV and take say a Vin+ of 200mV giving a Vdiff of 100mV. The output in this case is 227mV which corresponds to a gain error of around 5%. Surely this is well away from any rail voltages, and the INA326 should not be producing this much error if the datasheet is anything to go by..? I should have mentioned that supply is single +5V.
     
  4. mcasale

    Member

    Jul 18, 2011
    210
    12
    It looks like you are just trying to verify the operation of the amplifier. My advice is to SIMPLIFY. Get rid of the 1Meg pot and put in a good quality 200K, 1% or 0.1% tolerance part. Use the same part for R2. That way, you know you have an overall gain of 2. Your gain is affected by the tolerance of these resistors. Looking at the data sheet, you should not have a problem with CM input signals as long as you stay above ground and below 5 volts. Also, note you will amplify the offset voltage of the amplifier by its gain. Also, verify what your +5V supply is -it too has a tolerance. I hope this helps.
     
  5. saiello

    Thread Starter Member

    Jul 1, 2007
    24
    0
    Hi,
    I included the pots, in particular the Vin- and gain pots because I intend to have them in the final circuit for voltage translation and gain adjust. I use a pot on the Vin+ input to simulate the transducer that will be attached whose output varies from 1V-6V, hence the need for the Vin- pot to initially match the 1V output of the transducer.

    I initially chose the ina326 because I was under the impression that it was a true rail-to-rail input/output device, but this turns out not to be the case. It IS truly rail-to-rail for the inputs, and it differs in this respect to other instrumentation amplifiers/op amps, but the output is not rail-to-rail. According to the datasheet, and at a worst case scenario, it will only go from GND+75mV to Vs-75mV, so it appears I will need to include an output offset 'pedestal' of around 100mV, as per figure 6 in the datasheet. I really wanted the full range output of 0V-5V as this is going to be fed into an ADC but I will just have to live with the reduced range.

    Whether it is because I have not catered for this in my test circuit that I am getting the erroneous test results I am really not sure. I have my doubts as this output offset is not explicitly mentioned in the Applications Information section as among the necessary requirements for linearity. All the other things as given in the basic circuit ( figure 1 ) that would have a major affect I have already implemented in my test circuit. Having said that, there is a footnote (3) in figure 1 that suggests for measurements near zero output you will need an output offset. Given that this mention is virtually hidden and the fact that I would imagine most applications would require measurement near zero ( as I do ) seems to suggest to me that the datasheet is saying that not having an offset would not really affect linearity. But if this really is the root cause of my problem then it should really be given more prominence in the datasheet. As my test results stand, the linearity is way off the suggested specifications so something is seriously amiss and I have yet to understand why....:confused:
     
    Last edited: Sep 7, 2011
  6. mcasale

    Member

    Jul 18, 2011
    210
    12
    I was suggesting to eliminate the pot just for testing. If you need an adjustment in the application, so be it. Also, any resistors above 500KOhms can be problematic if they get dirty or wet. Leakage current changes the effective resistance.

    Rail-to-rail on either ins or outs is a problem with instrumentation amps, especially with a single supply. In such a case, I use the AD8571 op amp. The inputs are really rail-to-rail, and the outputs are within a few millivolts of each rail. Maybe you can build your own differential amplifier using 3 good op amps. Use the AD8574 version.
     
  7. saiello

    Thread Starter Member

    Jul 1, 2007
    24
    0
    Hi, Managed to get things seemingly working ok. A pedestal/offset output voltage is definitely required but is a necessity rather than an option as indicated in the datasheet. Typical output voltage swing is within 5-10mV of the rails, 75mV being the max. I initially applied a 50mV output offset as per figure 6 and immediately got results more accurate than I could measure with my cheapo multimeter. Figure 6 suggest 100mV offset voltage and I will go with this for 'safety' and practical reasons as this requires use of a 10M resistor. I considered making a discrete 3 op-amp circuit but this would have had a bigger footprint on the circuit board and is never going to be as accurate as a dedicated IC... ;)
     
Loading...