• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 78 subscribers
  • Views 1176 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.

Which msp430 for my project : Best ADC/Vref perf

Yazou
Prodigy 40 points

Hello,

I'm in the design phase of my project and I'm studing the possibility to use a MSP430 uC. Other candidats uC for trhe moment are silicon labs C8051F931 (8 bits but single cell power supply ) and EMF32 (ARM M3).

The uC i'm looking for must have :

1- a footprint < 4 x 4 mm = 16 mm^2

2- Very low power internal Oscillator. I have no place for a crystal

3- Power consumption < about 8 uW on battery

My application use mainly the Sleep mode (with Clock) anf the ADC to read the voltage output of a sensor (0 < Vout < 1V).

 I need to measure the sensor voltage with a sampling frequency of 200 sps. The uC uses its sleep mode w/RTC to Wake-up every 5 ms.

After a Wake-up the uC will Enable the internal voltage reference and make one ADC measurement . If the read voltage is higher than Vthreshold (0.5V for examaple) then the uC must set a  Digital Port IO to high during 1 ms.

Several msp430 in DSBGA package meet the footprint requirement and the Crytal-less requirements.

The MSP430F5528 is a good candidat but its LPM3 current is high (> 1uA) compared to MSP430F2274.

The main issue with the msp430 is the settling time of the internal Vref. It's 30 us for the F2xx and 75 us for F5xx.

During the settling time the internal Vref alone consumes 100 uA. The average current for Vref is > 75 us x 200 sps x 100 uA = 1.5 uA --> 4.5 uW (@3V) . If I add the LPM3 curent of 1.5 uA then I have 9 uW without even making ADC measurements. 

Any idea about the measurement stability (precision is less important) if I don't respect the 75 us settling time before starting ADC conversion ?  Can I suppose the Vref will still warm during the LPM3 5 ms sleep and avoid the 75 us wait time for every conversion ?

Is there a msp430 with a better ADC/VRef performance ?

Kind regards,

Yaz

  • 0
    Guru 46710 points

    Yazou said:
    Is there a msp430 with a better ADC/VRef performance ?

    Does C8051F931 meets your specs?

  • 0 in reply to Ilmars
    Prodigy 40 points

    Yes,

    C8051F931 could be powered by 1.5V (It requires an inductor 1x 0.6 mm).

    Its static current in sleep mode w/RTC is 0.6uA on VBAT = 1.5V -->  0.9uW

    Its internal DC-DC converter generates 1.9V when active for the CPU and the ADC with an efficiency >85%.

    Its ADC consumes 1.45 mA @ 1.9V but needs 2.85 us for settling & conversion --> ADC consumption with 200 sps is  < 2uW.

    The ADC of the msp430F5xx is better but the Vref settling time is the limiting factor in my case (200 sps).

  • 0 in reply to Yazou
    Guru 46710 points

    Yazou said:
    Its internal DC-DC converter generates 1.9V when active for the CPU and the ADC with an efficiency >85%.

    Seems you don't account regulator start-up time. Using 1.0 uF output cap it could greatly exceed msp430 reference setling time. Actually you shall measure msp430 vref settling time which most probably can be considerably shortened due to fact that time between off/on state is not that long.

    BTW what you do with sampled data? Supposedly you don't send it into void? So CPU core power efficiency comes into equasion.

  • 0 in reply to Ilmars
    Prodigy 40 points

    The Wake-up time of the C8051F9xx when using single cell is 10 us instead of 5us when using 3V battery. So it does not exceed the 75 us of the mps430 Vref. During this Wake-up time power consumption is < 70 uA.

    As I said i'm in the design phase and i cann't test & measure myself the performance of each solution. The ideal solution, will be to buy 3 eval cards for the 3 candidat MCUs and make my benchmark myself. But I'm afraid I don't have enough time.

    Regarding data processing, I compare the measured data with a voltage threshold. That's all.

    Every second i'm supposed to read at least one value which is > Vthreshold, then i must close a switch during  1 ms.

    But even if I need CPU processing, the C8051F9xx is not bad at all (at least on paper) because it consumes 170uA/MHz (@1.8V, 24 MHz) when CPU active.

    I made an Excel sheet to compute the perf of each solution for my project :

    For MSP430F5xx I have :

    • LPM3 static/w RTC = 1.4uA x 3V = 4.2 uW (very optimistic I think)
    • ADC alone = (220uA *3V *200 sps *3 us) = 0.4 uW
    • Vref settling = (100uA for Vref + 80 uA for LPM0 waiting Vref) x 200 sps * 75us x 3 V = 8.1 uW
    • CPU Active (150 uA/MHz) for Wake-up (5 us) & data processing = 1.4 uW

    For C8051F9xx, the computation is a little more complex because of DC-DC voltage conversion but the results are :

    • Sleep static w/RTC = 0.6uA x 1.5V = 0.9 uW
    • ADC + Vref = 1.45 mA x 1.9V x 200 sps x 2.85 us / 0.85 (DC-DC yield) = 1.8 uW
    • CPU Active (170 uA/MHz)  for Wake-up (10 us) & data processing  = 1.9 uW

  • 0 in reply to Yazou
    Guru 46710 points

    Seems like for particular application C8xx wins msp430F5xx. Especially if you have no time to do some tests like vref settling for msp430. I think vref buffer does not leak much during 5ms and so it's settling timeout can be greatly reduced. You shall check msp430 figures at 2.2V supply.

    You made small mistake in c8xx sleep/wRTC calculation by using 1.5V instead of 1.8V or 1.9V

  • 0 in reply to Ilmars
    Prodigy 40 points

    No, It is not a mistake. I plan using a 1.5V single Cell silver oxide battery to power the C8xx.

    The C8051F9xx familly is able to work with voltage down to 0.9V in sleep w/RTC without activating the DC/DC converter. The Wake-up time to power the DC-DC and the CPU is about 2 to 5 us (max is 10 us).

    I cann't count on msp430 figures at 2.2V because my power source is a 3V battery (2 silver oxide in series) with msp430. So currents sinked at 2.2V are also sinked from the battery because i don't use a DC-DC with msp430.

  • 0 in reply to Yazou
    Guru 46710 points

    Good for you. It seems that you already made your mind anyway and we are wasting time here :) Good luck! 

**Attention** This is a public forum