• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 82 subscribers
  • Views 2528 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

MSP430FR5969: Strain Gauge Interface through ADC12_B with external amplifier

Paul Shepherd17
Intellectual 595 points
Part Number: MSP430FR5969
Other Parts Discussed in Thread: THS4513, MSP430F6736A, OPA4277-SP, INA333-HT, INA826S, INA826

Hi all,

I am working on a design which includes a full-bridge strain gauge to measure tank pressure. It does not need to be rad-hard at this time, but I would like to be able deliver a rad-hard version in the future with minimal redesign. For this reason, the MSP430FR5969 (and its -SP version) are my best choice for the microcontroller. I am trying to decide if I can get away with using the internal 12 bit SAR ADC.

Unlike some MSP430s (such as MSP430F6736A), the ADC on the FR5969 does not have a programmable gain amplifier. Because of that, I am looking at using the THS4513 (fully differential) to provide the gain necessary for my strain gauge into the MSP430FR5969 ADC12 input. Have I made any obvious mistakes in setup here?

My intended supply voltage is 3V (single ended). According to the THS4513 datasheet, the input and out ranges are +/-400mV and +/-450mV around mid-rail.

Maximum bridge output voltage (with 3V supply): 486uV differential
Minimum bridge output voltage to be resolved: 1.6uV differential
ADC Vref+: 1.2V internal reference voltage (ADC12VRSEL = 0001b)
ADC Vref-: AVSS
ADC ENOB with differential input: 10.7 bits typical (Table 5.26 in datasheet - 2.5V Vref, not sure how much this drops with a 1.2V reference)
ADC settings: 12 bit unsigned, differential inputs (ADC12DIF =1b, ADC12DF = 0b, ADC12RES = 10b)
ADC Voltage resolution based on ENOB: 721uV

My calculations tell me that I need about 55dB of gain to match the 1.6uV differential output of the strain gauge to the 721uV sensitivity of the ADC. Part of my concern is that the open loop gain of the THS4513 is only 68dB, and shooting for 55dB of gain will require a large resistor (Rf of 267k with Rg of 47 ohms).

This design seems borderline to me. I could go ahead and order the THS4513 EVM and try it all out, but if my chances of success are low, I should go ahead and start looking for an external ADC which works well with the MPS430. As a side note, I don’t have any other external chips which will require an I2C/SPI/etc interface, which is one reason I am hesitant to add an external ADC.

 Thanks,
Paul

  • 0
    Genius 3255 points
    Remember that large resistors mean more noise and temperature drift (a recipe for disaster). This of course gets amplified by the gain, which could add significant noise to your measurement. Also, thermal EMF of copper wire to tin can easily go beyond 1.6uV/ºC.
    You should probably look into an ADC that has sufficient resolution without/or with very moderate gain.
  • 0 in reply to Mike Tooma
    Intellectual 595 points
    Thanks for the feedback! Some of those issues I was already considering (such as the noise issue of large resistor values). I am concerned that the small sensitivity of my strain gauge can only be fixed by a larger supply voltage (say 12V), but then I have the issue of level-shifting the strain gauge output down to the voltage range allowed by the MPS430 (3.3V or less). I don't believe an external ADC would really fix the level shifting issue, as they seem to have analog supply ranges of 5V or less.
  • 0 in reply to Paul Shepherd17
    Genius 3255 points
    An external ADC will, in fact, resolve your issue. Most ADCs have allowed input of GND to VRef (+/- Vref/2 for differential). Their Vref is usually independent of their Vcc.
  • 0 in reply to Mike Tooma
    Intellectual 595 points

    I am new to spec'ing ADCs, so I appreciate you working through this with me. Unfortunately, I don't think a using a different Vref will solve my problem. The issue is that I have a very small differential voltage compared to my common mode voltage. I still need a very low Vref or a lot of gain on my strain gauge. The final alternative would be a lot more bits of precision, but the options are limited, particularly for components with a rad-hard option.

    The good news is that I just pointed out to the mechanical engineers how difficult this was going to be, and they are backing off the precision requirement. I am now trying to get 25uV of precision from the strain gauge bridge at 3V supply (up from 1.6uV), so I can reduce my differential amplifier gain to around 30dB. I am a lot more optimistic that we can make this work.

    I am still interested in the reality check of whether or not I have the uC settings correct for sensing a differential signal with a 1.5V CM voltage, and the differential voltage will only be positive (won't ever be sensing a negative pressure/differential voltage)

    Thanks,
    Paul

  • 0 in reply to Paul Shepherd17
    TI__Genius 12315 points

    Hi Paul,

    I think ADS1282-SP which is radiation tolerant high-resolution delta sigma ADC with PGA may meet your requirements. And the following resources should be helpful for your design of strain gauge.

    Best Regards,

    Winter Yu

  • 0 in reply to Winter Yu
    Intellectual 595 points

    Hi Winter,

    As I mentioned in my original post, I'd really prefer to avoid having to write the code for an external peripheral, when there is already a SAR ADC on board.

    I was talking over my intended design with a friend, and he pointed out a very big oversight I made... The THS4513 is an RF part, so using it in a very low frequency application creates a lot of layout and implementation issues which I am not prepared to handle. He suggested looking at the INA333-HT instrumentation amplifier as a hi-rel (but not rad-hard) choice. I am also looking at building a discrete INA using the OPA4277-SP or similar, along with a resistor array for reasonable matching.

    Thanks,
    Paul

  • 0 in reply to Paul Shepherd17
    Genius 3330 points
    I am currently using an INA826S to amplify a load cell with good results.
  • 0 in reply to eltury
    Intellectual 595 points
    Hi eltury,

    I think you have hit the nail on the head. An instrumentation amplifier is really the correct solution. I originally had a concern about CM level shifting, but with the right device, the CM input can be shifted down to ground, and the signal to the MSP430FR5969 can be single-ended.

    There are a few rad-hard INAs out there, but they are not rated for low supply voltages as far as I can find. I am going to buy a dedicated non-rad-hard part like the INA826, as well as layout a discrete INA from matched op-amps. I will use the integrated INA to baseline performance, and then see how well I can do with opamps that are available in a rad-hard version.

    Thanks!
    Paul

**Attention** This is a public forum