• State
  • Locked Locked
  • Replies 9 replies
  • Subscribers 79 subscribers
  • Views 2088 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.

SD24_B in the msp430F6746

jim wang2
Intellectual 270 points
Other Parts Discussed in Thread: MSP430F6746, MSP430F4619, MSP430F2013, MSP430FG4619, MSP430F6779

I have been used ADC12 of MSP430F4619.now I'm trying to learn msp430F6746 for my next project, I don't know how to use SD24_B, is it similar to ADC12 ? for example, for ADC, it need a REF,like 1.5V, I have to adjust input signal to 1.5V when it is full range. How about SD24_B ? I don not want to know the detail of princple, but I want to know how to use it, could you give me some link of sample code and application? in my appliation, I will measure Pressure and Temperature, the input is 0-10mv signal. before I use OP to adjust signal to 0-1.5V, then add it onto ADC12, how about now if I use SD24_B in the msp430F6746 ?

  • 0
    Guru 227245 points

    jim wang2 said:
     I don't know how to use SD24_B, is it similar to ADC12 ? for example, for ADC, it need a REF,like 1.5V, I have to adjust input signal to 1.5V when it is full range. How about SD24_B ?

    The SD converters are Delta-Sigma converters. They are designed for low-cost, low speed, high-precision and require continuous operation on an input channel. Also, the ones in the MSP do have a fixed 1.2V (+-0.6V on some devices) reference.

    ADC10/12 converters are SAR type and each converison is  separate operation. Connect to signal source, charge the sampling capacitor, disconnect, convert using charge distribution. Hten connect to next signal source etc.

    If you work with an SD16/SD24, forget your ADC10/12 experience and start from scratch. Both are totally different, in target application, configuration, and data handling.

  • 0 in reply to Jens-Michael Gross
    Intellectual 270 points

    Thank you very much. I need measure pressure which is from14.00PSI to 1500.00PSI, so need 16bits AD converter, I'm interested in using SD24, but not continous signal, is it OK? and how slow it is, hong long it need to complete one conversion? Is the function of 1.2V REF same as ADC10's 1.5V/2.5V REF? so I have to adjust input signal to close it, right?

  • 0 in reply to jim wang2
    Guru 23000 points

    SD converters are good for +/- measurements and small values (internal gain).

    If you only need a higher resolution you can use Over-Sampling, 256 times give you 4-bit resolution extra.

    But in your case I would consider to use a cheap MSP430F2013 SD16 converter without extra Op-Amps.

    Conversion time is explained in the Data Sheet. With the MSP430F2013 at 1Mhz (conversion clock) & OSR=256 you can get about 3,900 samples per second.

    -Leo

  • 0 in reply to Leo Bosch
    Intellectual 270 points

    Thank you, for my application, the high resolution AD converter is only one point, I need big flash 256KB, 3 timers+ RTC, 2 UART+IrDA, I2C, 160LCD drives, ports interruption,internal temperature, voltage measurment, very low power, AES encryption, becasue I had been used msp430FG4619 in our current products, so only SD24 and AES encryption are new for me, I want to make sure I can use them before I select it, of course it is better to try it but there is no sample available.

     

  • 0 in reply to jim wang2
    Guru 23000 points

    The MSP430F6779 is available.

  • 0 in reply to jim wang2
    Guru 227245 points

    Leo Bosch said:
    If you only need a higher resolution you can use Over-Sampling, 256 times give you 4-bit resolution extra.

    If you mean to get a higher resolution from an SAR by oversampling, you're wrong. Oversampling an SAR like the ADC12 will increase SNR by averaging teh noise, but won't increase the resolution of the result.
    To increase the resolution, you need to superimpose the signal with a high-frequency sine (or triangle) signal and average the samples during one of these superimposed waves. (n bits per 2^2n samples)

    On Delta-Sigma converters, however, a higher oversampling will indeed increase resolution. But won't increase SNR. So your added resolution will soon vanish below the noise floor.

    jim wang2 said:
    hong long it need to complete one conversion?

    The SD16 converters usually run with a conversion (or rather: recursion or bit) clock of 1MHz. Minimum number of samples for a conversion is 32 (1 on SD24B, but then the result is 2 bit resolution only), maximum is 1024, which gives a conversion rate of 1 to 31 kHz. Also, after starting a continuous measurement (or switching input channels), the first three consecutive conversions are void.

    jim wang2 said:
    Is the function of 1.2V REF same as ADC10's 1.5V/2.5V REF?

    Its overall funciton is the same, yes. But its use differs in details. So on SD modules with differential input, the differential input range is from -0.6 to 0.6V (while both inputs must be positive to GND), on MSPs with SD modules for power metering, even negative voltages can be measured (-0.6 to 0.6V relative to  GND).

    Also, the usable input range is limited if you want to have best performance. You should stay withing +-0.5V or precision degrades. This is caused by the Delta-sigma working method in conjunction with supply voltage. The ADC10/12 don't have this effect.

  • 0 in reply to Jens-Michael Gross
    Intellectual 270 points

    Thank you very much for your answer, I look the document slaa323, and also run it's sample code. for my understand, it means use 256 times sampling to add 4 bit more resolution, of course it improve the SNR too. base on sample code, the AD 16bits result is the sum of 256 times 12bits conversion and then divided by 16, it is 16bits AD result, I do not konw it is correct or not. now I'm trying to prove it ?! 

  • 0 in reply to jim wang2
    Guru 23000 points

    I didn’t explain the oversampling method.

    As long as there is enough noise (in lot of cases) oversampling will work, in other case you can use your method by adding noise.

    Oversampling by 256 times and right-shift the value by 4 gives 4-bits more resolution.

  • 0 in reply to Leo Bosch
    Guru 227245 points

    Leo Bosch said:
    As long as there is enough noise (in lot of cases) oversampling will work

    Depends on the noise - and doesn't apply to internal noise of the converter at all.

    So in most (but - rather coincidentally - not all cases), simple attempts to increase resolution by oversampling only are just averaging and do not increase resolution.

    Leo Bosch said:
    in other case you can use your method by adding noise.

    Best is not noise, but a clean signal (sawtooth, triangle, sinus) of 1LSB amplitude that is aligned with the sampling frequency (e.g. signal period spans 256 samples for 4 more bits). With simple noise, you might add resolution but also decrease(!) SNR. So if SNR is already less than the converter's dynamic range, then you won't gain anything, you will actually lose precision and only gain fake resolution.

**Attention** This is a public forum