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.

Sugestions on low cost/small size strain gauge amplifier

Other Parts Discussed in Thread: MSP430F2013, ADS1231, INA333, ADS1131

Hello, 

I need to develop a "low cost" and small size strain gauge amplifier. I will need something close to 13 bits effective bits on a 1.5mV range and 10Hz frequency BW. The board will be used with a full bridge and will be placed close to the bridge. I have a few ideas in mind and need sugestions. 

From the amp circuit to the processing board there will be a maximum of 2m. In order to prevent values mismatches and noise i am planing to convert the sampled data into PWM to transmit the information and the data would be coded at the duty cycle of the pwm. The advantages of this transmission is that it is imune to frequency mismatches/drift and general noise, however it is susceptible to fast transient glitches. Any comment on that idea?

Regarding the amplication circuit itself. I will be using sigma-delta ADC and i have a few options. To use an MSP430F2013 with internal 16 bits SIgma Delta and an external cheap INA like the microchip MCP6N11 with a gain of 100. The other would be to use a MSP430 with internal 24Bits Sigma Delta and finally the last to use any cheap MSP430 with external Sigma Delta ADC. 

Any sugestions and comments?

Regards,

Luis Filipe Rossi

  • Luis,

    It appears that you want a high end solution at low cost.  1.5mV range is pretty small and noise may be difficult to overcome.  A cheap INA can make the noise problem worse.  One possible solution is to use an inexpensive MSP430 and an external bridge ADC like the ADS1231.  Another is to use the MSP430 with the 24 bit converter that includes the gain stage.  Using the the ADS1231 eliminates the setup for the digital filter and oversampling ratio, but it more expensive.

    You haven't stated your budget so it is hard to give much guidance.  It seems to me that converting the data to PWM is a lot of effort. I think I would just use the UART port on the MSP430.  You could use a number of interface driving methods, including RS232.

    Best regards,

    Bob B

  • Bom thank you for you answer,

    What would you sugest me for such range (actually it is 0.5mV/V). I agree about the cheap INA. Actually it usually got a huge input offset voltage. External ADC is an nice idea, but it got a very limited sample rate. It works for my requirements but i would prefer to have more space for increasing the BW response of the system. The problem i see using the MSP430 with SD24 is the low input impedance (300k Ohm) compared to a INA input. Why do you think converting data to PWM is a lot of effort? Most of MSP430 got a 16 bit timers with 3 CCPs, so changing the duty cycle with 16 bits resolution would be straight forward. I would like to have a 10 USD solution (not incluiding PCB) for small batches (20 pieces). It could be a bit more (like 12 USD) if really necessary.

    Regards,

  • Luis,

    My mistake, I thought the delta-sigma MSP430s had the input buffer.  You could put the INA333 in front.  As far as the ADC, you didn't specify the range of data rates you preferred, or if you wanted to make ratiometric measurements.  With increasing data rates, the level of noise increases dramatically.  Bottom line you say you want 13 ENOB, but that doesn't really tell much of the story.  I'm assuming you are exciting your load cell with 3V based on what you said earlier.  What is the range of the load cell and what is the smallest increment you want to measure noise free based on that range?  Do you prefer to make a ratiometric measurement based on the excitation, or do you have a specific voltage reference that you prefer to use?

    As far as outputing the data, you can do it in whatever way you want.  I personally would not use PWM because there are built in hardware peripherals for communication (and thus the small effort) and the data can be represented with as much resolution as you want.  Its just a personal preference.

    Best regards,

    Bob B

  • Thank you again Bob, 

    In summary my requirements would be something close to SLAA220. In general i would like to have 13 bits enob with 1mV/V. Reducing to 12 bits for a 0.5mv/V is acceptable. 0,6 - 1uV is the step i am expecting but i cant see how to get low noise as that. I am not sure if i can meet such requirements with my price budget. For sure ratiometric measurement is a must . The load cell range is not always the same. It will be placed on a robot arm, so the requiremnts might be different from project to project. Currently a board designed by a few students uses an AD620 with a 16 bits SAR ADC and the cable from the bridge to the electronis is big, so we got a lot of noise. I belive maximum sample rate i would require would be 100SPs and 20Hz BW response. But 10Hz is ok for now.

    Thank you! Best regards,

    Luis Filipe

  • Luis,

    Sounds like you have true engineering situation.  Noise is the issue, as I think you already know.  I see now why you want to do things as you've previously stated.  There is another controller option and that is from our MSC product line.  It is an 8051 based micro, with either 24 bit or 16 bit delta-sigma.  I've attached an app note using this family as weigh scale.  This family has a buffered input, PGA of 128 and can supply the higher data rates.  The app note uses the MSC1210, which is higher cost, but I think you could use the MSC1200 or MSC1202.  The PWM may only be available with the MSC1210 however.

    4073.sbaa092a_MSC1210_in_Weight_Scale_Systems.pdf

    Best regards,

    Bob B

  • Thank you for your advice Bob, 

    Such a nice chip, but VERY expensive. I think i will take a try on the ADS1131. Resonable price with nice performance and very low noise. If i happen to need a bigger sample rate i will take a try with INA333 (i have couple on hands right now that i got for an EMG amp) and MSP with internal ADC. Any advice regardin ADS1131 for this specific application?

    Best regards,

    Luis Filipe Rossi

  • Luis,

    Sounds like a good plan.  A starting point is to take a look at the ADS1131REF design to see one possible connection scheme for the analog inputs.  High quality caps improve noise and drift, but they cost more.  The REF board doesn't really show any filtering (outside of some EMI/RFI filter caps,) although the provision is on the board.  You might want to consider EMI/RFI filtering with a ferrite with maybe 1k series resistance along with the caps.  To limit drift you need to make sure that the differential cap is at least 10 times larger than the common mode caps.  Adding filters will disturb the ratiometric measurement.  You might try various cap values at the reference input to see if there is a value that might keep the noise more in phase between reference and analog inputs. 

    Another thing to notice on the REF board is the layout.  You should have a good solid ground plane.  Make sure that digital and analog portions of the board do not cross paths.

    http://www.ti.com/tool/ads1131ref

    Best regards,

    Bob B