• State TI Thinks Resolved
  • Locked Locked
  • Replies 1 reply
  • Answers 1 answer
  • Subscribers 48 subscribers
  • Views 1132 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

PGA112: Measuring single-ended values close to GND

Adam Berger
Prodigy 70 points
Part Number: PGA112
Other Parts Discussed in Thread: REF3230, OPA2364, TINA-TI, OPA364, OPA363

I am intending to use the PGA112 amplifier to measure voltages with a wide dynamic range. Peak voltage to be measured is 60 V, scaled to < 3V with a 100k / 4.7k resistor divider, and low-passed as appropriate to the application. The output of the PGA112 will then go through an output filter to the 12-bit ADC of a Piccolo or Delfino uC.

Application details: This is for sensing phase voltage in a BLDC motor controller where the controller can both be used with a wide variety of bus voltages (Vbus = 6V - 60V), and where the phase voltage will vary at 0 - 500 Hz between 0 V and duty_cycle*Vbus for 0 < duty_cycle < 1.

The application will use the gain of the PGA to adjust for changes in Vbus; for example, when running at 6V a gain of 8 will be used. However, I am not intending to change gain dynamically (between sequential readings of voltage) if at all possible. Therefore, at a given gain, the question of usable dynamic range of the PGA/ADC combination arises. Essentially, the ADC has a nominal resolution of 3.3V / 2^12 = 0.8 mV/count, but the bottom 50 mV = 62 counts are indistinguishable. If my input varies from 0V to 100 mV (very reasonable for scaled phase voltages at low duty cycle), a full half of the output is clipped and indistinguishable.

QUESTION 1: Is there a different approach I should be taking here, either with the PGA112 or a different part, to solve this problem? I considered using no amplifier gain and instead a single-ended-to-differential amplifier and the 16-bit mode of the Piccolo/Delfino ADC, but for various reasons this is not ideal.

QUESTION 2: If I continue to use the PGA112, can I set Vref to 100 mV to offet the output? I understand that this would restrict me to gains of 2+ (as Vref has no impact with gain=1), and slightly restrict my output range (from 100 mV to DVdd - 50 mV). Are there other impacts I'm missing? The datasheet discusses Vref=GND and Vref=AVdd/2 cases, but not the general case.

QUESTION 3: What's the most appropriate way to get a Vref in this range, with the necessary current source/sink requirements? I expect to have a REF3230 available for Vcal; should I buffer this, resistor-divide by 30, and re-buffer? Is an OPA2364 appropriate for both buffering stages here? Is it acceptable to have the output of the second buffer (the 100 mV Vref) shared by 4x PGA112s? (Output current of OPA2364 = 45 mA >> 4x PGA112 Vref current = 4*2 mA.)

QUESTION 4: Is there a PSPICE / TINA-TI model of the PGA112 available so I can answer some of these questions on my own?

  • 0
    TI__Expert 6250 points

    Hi Adam,

    I'm actually not sure that the issue you're seeing is being caused by the ADC. I suspect it is being caused by the PGA112's output swing limitation (50mV from either rail). While boosting Vref up would fix this, it causes an added issue that your output offset will vary with configured gain. It's possible you could compensate for this in software, but another solution would be to substitute the following circuit instead of your existing resistor divider:

    This will scale a 0-60V input to a 0.1 to 1.6V output (1.6V to account for the 2x gain of the PGA when using Vref). Simulating this over the 0-60V range, the Vref input only consumes ~140uA, so you could probably even get away with directly connecting this to the REF3230.

    If you then provide a 0.1V Vref to the PGA, you should get a ~100mV to ~3.3V output from the PGA112. Since the input is now offset by .1V, the gain setting of the PGA112 no longer changes the offset - both cancel out. 

    For the PGA Vref, I'd recommend using a simple buffer. It looks like the Vref input also consumes ~120uA, so I don't think individual buffered Vrefs for each PGA are required. As a quick calculation, a 10 inch long, 10-mil wide trace on 1-oz copper is 500mΩ. I'd expect a worst-case voltage drop of ~62uV at the Vref pin, or ~.1LSB. If this is an issue, you could add dedicated buffers for each PGA.

    I used the OPA363 (OPA364 with an enable pin) for this test, but I think any low-supply op-amp should work here.

    I've attached the TINA file I used for this below. We don't have a model for the PGA112 currently. I mocked up the Vref input using an additional OPA363. This isn't an accurate model of the PGA112, but it should give a ballpark idea of current into the Vref pin.

    TINA-TI File