ADS1231: Multiple channels, accuracy analysis, and improving device dynamic range

Hi Maxwell,

See my responses below.

Best regards,

Bob B

Maxwell A. Smith said:

Part Number: ADS1231

I am working on a measurement system and I have three primary questions:

1. What is the recommended implementation method for multiple channels with this device? Since it is I2C without configurable addresses I assume that a mux is the only way to implement? Are there any recommendations or drawbacks to doing this? The reason I wish to use multiple channels of this single channel device is so that I can populate only the specific channels requested and keep my design versatile. [Bob] Actually the communication interface is SPI even though it is 2 wire (SCLK and MISO).  You could implement an input mux, or you could use the 2-channel ADS1232 or the 4-channel ADS1234.  Each of those devices have an internal mux.  Also those devices have lower noise and adjustable gain.
2. Is there an application note, or recommended reference material (book, app note, etc) for calculating a performance or accuracy analysis on a load cell-amplifier system? Specifically, in reference to this response by Bob Benjamin I realized I might have missed some key parameters in my analysis. [Bob]  This is something we are looking into doing, but have no specific material at this time.  If you use an external amplifier you would need to consider the additional error of introducing another amplifier into the system.  Primarily this will be noise, offset, gain error and any drift component this will add into the measurement result.  I created this document some time ago that you might find helpful. 6281.Weigh Scale_Design.doc
3. Based upon the suggested parameters given by Bob in that prior post, I calculate that my load cell (1mV/V, 5kg) will get about 0.85g steps because it is not using anywhere close to the maximum dynamic range of the system. This is reflected in my prototype setup and with environmental conditions I see +/- 1g reliably. Ideally I would like to see better numbers say 0.1g. Is this is as simple as increasing excitation voltage say with a 5V or 10V voltage reference to increase the available signal? [Bob] The ADS123x devices use a ratiometric approach to making the conversion.  Ratiometric refers to the output result being a ratio of the value at the input by using the same excitation voltage as the ADC reference voltage.  This method greatly reduces any noise in the conversion result that appears in the excitation and also removes any drift of the excitation source.  So even though you may benefit by increasing the dynamic range there is no guarantee that the response will be better scale resolution.  This is true of both the external amplifier and increasing the excitation voltage.  If you increase the excitation voltage you will need to use bipolar supplies so that input range of the ADS1231 is maintained to the required value of AVDD - 1.5V to GND +1.5V.  One example would be 7.5V and -2.5V for the excitation voltage.  You must also provide a reference using some other method other than directly connecting the excitation voltage to the reference inputs.  Basically you are looking for 50000 noise free counts with a measurable resolution of 100nV while using 5V reference.  Doubling the excitation helps some as now the measurable resolution doubles to 200nV, but the noise-free resolution for a 5V system is typically best case for the ADS1231 at 231.9nV.  So some type of averaging my help you achieve the goal, but external noise and any drift of voltage divider components for the reference may nullify the response.

Thanks kindly.