Why ADC results have big difference each time?

Hi Xiaowei,

It looks like the average of these conversions is roughly what you would expect, so mostly the concern is the noise?  Is the battery connected directly to the ADC input, or is there an op-amp and/or anti-aliasing filter?  If not, you should at least add an RC filter, or possibly an op-amp and RC filter.  You can also take many samples and average to reduce the noise.  

Thanks for your help. I added an op-amp and filter. The measurement results look better now. But still too much noise for our application. The following is the results. Normally, what is the acceptable error percentage of this ADC?

Xiaowei,

There are quite a few ways to specify the ADC error.  In the datasheet right now we have gain, offset, and linearity.  These will tell you how accurate the ADC is from a DC perspective.  

With respect to noise, the SNR of the ADC will be around 68dB.  It is certainly be possible to get a DC codespread of <= 4LSBs.  

Your results still seem pretty noisy.  Have you tried supplying the ADC directly from the output of a signal source like a function generator (you should still use the op-amp and filter circuit)? 

I just made a measurement from the output of the function generator with op-amp. Input voltage is 1.623, still a lot of noise!

Your values are gradually increasing now, but actually not very noisy.  I think you may need to increase ACQPS or decrease source impedance.  

The ADC result is getting better and better with your help.  The difference between input voltage(1.622V) and The average result of 16 measurements is 0.0002549V.  Do you think I need to cal  Device_cal() function? Because I saw the following information.  If needed, how to cal Device_cal() function?  I use the following way to call the function. It didn't work.

#define Device_cal (void   (*)(void))0x000000

...........

(*Device_cal)();

// *IMPORTANT*
    // The Device_cal function, which copies the ADC & oscillator calibration values
    // from TI reserved OTP into the appropriate trim registers, occurs automatically
    // in the Boot ROM. If the boot ROM code is bypassed during the debug process, the
    // following function MUST be called for the ADC and oscillators to function according
    // to specification. The clocks to the ADC MUST be enabled before calling this
    // function.
    // See the device data manual and/or the ADC Reference
    // Manual for more information.

Hi Tommy,

Thanks for your help. May I ask a simple question? what are TMX and TMS shorted for?

TMS and TMX are prefix codes that describe the maturity level of the device.  There should be definitions in the datasheet similar to this snapshot.

If I am fully understood, we can only get samples for TMX320F28377D  at this moment.  TMS version is still on PREVIEW status.

Do you an idea when will TI release TMS320F28377D?

The current estimate is early 2015.

Hello!

I have seen in the errata that we need to connect the ADC's Vreflo to the ground in order to avoid the ADC-to-ADC crosstalk and to reduce the noise. But the problem is that I can't found anywhere in the datasheet what are the pins number on the EVM F28377DZWT? Could you give me a link or anything else where they are referenced?

Hi Baptiste,

I am not sure if your are looking for this or not. 

TMDSCNCD28377D_180cCARD_pinout_R1_1.pdf

\ti\controlSUITE\development_kits\~controlCARDs\TMDSCNCD28377D_v1_1\R1_1

Hi Xiaowei,

Thank you. Obviously I had not seen this PDF!