topic Re: K60 ADC problem - Results are zero or random in Kinetis Microcontrollers

K60 ADC problem - Results are random

michalhrouda — Tue, 12 Mar 2013 23:58:06 GMT

Hi,

I have chosen MK60DN256ZVLL10 for my design and I have a problem with ADC (with both ADCs)

When I set input channel between 0 and 0x1C (including internal temp sensor) the result of the ADC is zero or random

(On channels 0 ~ 23 i get 0 or random values depending on MUXSEL setup and temperature sensor channel result is always random :smileyshocked: )

I get non zero (and correct) results only when i choose REFLH or Bandgap channels.

I tried Processor expert ADC component, Init_ADC component and code written by myself only (same behavior)

I tried all possible configurations of conversion modes, sampling timing, sampling time and calibrating feature.

Voltage on VREFH pin is correct (3.5 V) and voltage on pins that I would like to measure is correct and without any noise too.

List of pins that I use for measurement:

HUMIDITY	=> PTC2/SPI0_PCS2/UART1_CTS_b/FTM0_CH1/FB_AD12/ADC0_SE4b/CMP1_IN0/TSI0_CH15 [72]
BAT_MEAS	=> PTE0/SPI1_PCS1/UART1_TX/SDHC0_D1/I2C1_SDA/ADC1_SE4a [1]
PROX_MEAS	=> PTB1/I2C0_SDA/FTM1_CH1/RMII0_MDC/MII0_MDC/FTM1_QD_PHB/ADC0_SE9/ADC1_SE9/TSI0_CH6 [54]
PWRSRC_MEAS	=> PTC1/SPI0_PCS3/UART1_RTS_b/FTM0_CH0/FB_AD13/ADC0_SE15/TSI0_CH14 [71]

 //enable bandgap
    PMC_REGSC |= PMC_REGSC_BGBE_MASK;
  
    //Processor Expert Init
    AD0_Init();
    AD1_Init();

    //Calibration
    ADC0_SC3 |= 0x80;
    while ((ADC0_SC1A && 0x80) == 0);

    ADC1_SC3 |= 0x80;
    while ((ADC1_SC1A && 0x80) == 0);

    for(;;)
    {
  
        ADC1_SC1A = 4;
        while ((ADC1_SC1A && 0x80) == 0);
        bat = ADC1_RA;
  
  
        ADC0_SC1A = 15;
        while ((ADC0_SC1A && 0x80) == 0);
        pwr = ADC0_RA;
  
        ADC0_SC1A = 9;
        while ((ADC0_SC1A && 0x80) == 0);
        prox = ADC0_RA;
  
        ADC0_SC1A = 4;
        while ((ADC0_SC1A && 0x80) == 0);
        hum = ADC0_RA;
  

        FRTOS1_vTaskDelay(500/portTICK_RATE_MS);

    }

Re: K60 ADC problem - Results are zero

adriansc — Wed, 13 Mar 2013 16:54:50 GMT

Hi,

I am not really sure what is going wrong there. In the attachments you can find an ADC driver for any Kinetis Family MCU, compare this with your initialization and read codes.

Hope this helps.

Re: K60 ADC problem - Results are zero

michalhrouda — Mon, 02 Nov 2020 13:07:38 GMT

Thank you for your answer and example code I tried your code but i get same (wrong) results:

Code:


static void vAdcTestTask(void *pvParameters)
{
    //enable bandgap
    PMC_REGSC |= PMC_REGSC_BGBE_MASK;
       
    //Init
    ADC0_Init16b();
    ADC1_Init16b();

    //Calibration
    ADC0_SC3 |= 0x80;
    while ((ADC0_SC1A && 0x80) == 0);
   
    ADC1_SC3 |= 0x80;
    while ((ADC1_SC1A && 0x80) == 0);
   
    for(;;)
    {
        ADC0_Values[channel] = ADC0_Read16b(channel);
        ADC1_Values[channel] = ADC1_Read16b(channel);
                   
        channel++;
        if (channel >= 31) channel = 0;

        //10 ms delay
        FRTOS1_vTaskDelay(10/portTICK_RATE_MS);
   
    }
           
}

Here is layout of my pcb:

Re: K60 ADC problem - Results are zero

adriansc — Thu, 14 Mar 2013 16:21:31 GMT

Hi,

Did you try just one channel? I mean, just read one of the channels with problems just to dismiss that the problem can be because the time that other channels demand for conversion.

Is there a possibility to test your code in a TWR-K60?

Regards.

Re: K60 ADC problem - Results are zero

JimDon — Thu, 14 Mar 2013 16:25:20 GMT

Adrian,

I think what you are really questioning is if your sample code correctly waits for a conversion to complete.

Only the channel number you write to the register will be converted.

Re: K60 ADC problem - Results are zero

michalhrouda — Thu, 14 Mar 2013 16:41:11 GMT

Yes, I did. I get zero all the time on all channels that I would like to measure (I was measuring on one channel and putting values into buffer)

Most confusing to me is the fact that the value measured on internal bandgap channel is always correct (19101/65535 * 3,5V = 1V) and value measured on internal temperature sensor is always zero

I'm going to try my code on TWR-K60 today

Re: K60 ADC problem - Results are zero

michalhrouda — Thu, 14 Mar 2013 18:19:32 GMT

Same code is working on TWR-K60N512 without any problems. That means, that there is a problem with hardware design :smileysad:

I suspect that problem could be in 3.5V power supply (typical is 3.3V).

Update: Power supply voltage is not the issue = same behaviour @ 3.3V

Re: K60 ADC problem - Results are zero

adriansc — Thu, 14 Mar 2013 19:02:58 GMT

As we say: Al mal paso darle prisa. This is like "Let's get it over and done with" keep us posted. I recommend you to have a look at Freescale's layout design, preferably TWR-K53 ,it has more ADC functions.

Regards.

Re: K60 ADC problem - Results are zero

michalhrouda — Fri, 15 Mar 2013 00:42:41 GMT

I'm just tried to resolder mcu and there was probably cold junction. Internal temperature sensor is working now. But channels that i'd like to measure (1~24) are still wrong but different. Values are not zero but random... :smileyconfused:

Re: K60 ADC problem - Results are zero

JimDon — Fri, 15 Mar 2013 02:33:25 GMT

Since you have confirmed that the code is correct, then I suggest there is still a hardware problem. It ain't fixed until it's fixed.

Could be that the VREF connections are bad or any number of interconnect issues. You are in for a tough day, because you need to crawl through the hardware and figure out what's ups. Perhaps the chip has gone bad due to a short??. If there was a bad solder joint then most likely there are more. Only you can figure this out, as asking about symptoms is really tough for anyone else to answer when it is a hardware problem.

Re: K60 ADC problem - Results are zero

Florijan — Fri, 15 Mar 2013 09:34:50 GMT

Dear michalhrouda,

the first thing I have noticed is that in sentence

while ((ADC0_SC1A && 0x80) == 0);

you are using double '&&' instead of single '&', so you should write

while ((ADC0_SC1A & 0x80) == 0);

try this and report.

Best regards,

Florijan

Re: K60 ADC problem - Results are zero

michalhrouda — Fri, 15 Mar 2013 10:06:31 GMT

Thank you for your remark. This mistake was in my code all the time. But my application still behaves in the same way even after fixing that error. The problem is in hardware

Best regards,

Michal

Re: K60 ADC problem - Results are zero

michalhrouda — Fri, 15 Mar 2013 10:08:51 GMT

Do you think that problem is in layout/hardware design or in soldering quality?

Re: K60 ADC problem - Results are zero

Florijan — Fri, 15 Mar 2013 10:30:38 GMT

Hi,

I have a lot of experience with 8-bit mcus and their ADC modules and if you overclock the ADC it will broke down. In such case it will behave exactly as you described. I think you should replace MCU and be very careful with new one not to overclock or underclock the ADC.

Please report.

Best regards,

Florijan

Re: K60 ADC problem - Results are zero

michalhrouda — Fri, 15 Mar 2013 17:38:39 GMT

I'm sure that I did not overclock or underclock ADC. Pay attention to the fact that measurements on internal bandgap, internal temperature sensor, VREFH and VREFL are correct. Could overclock or underclock affect/damage ADC channel multiplexor only?

Re: K60 ADC problem - Results are zero or random

michalhrouda — Sat, 16 Mar 2013 18:16:54 GMT

I've tried ADC measurements on my other K60 board and it behaves in the same way... :smileysad: Internal bandgap and internal temperature sensor measurements are correct but measurements on channels 1 ~ 24 are wrong (not zero, but some random value)

Here is part pcb layout and scheme of my other design

Re: K60 ADC problem - Results are zero or random

Florijan — Sat, 16 Mar 2013 20:00:13 GMT

Today I have put my hands on Kinetis M0+ MCU the first time and guess what? I have the same problem as you do. This mcu is part of demo board purchased directly from Freescale, so I think it is most probably something wrong with the initialization of ADC rather than pcb layout issue. I have made initialization using Device Initialization GUI and maybe there is some bug in it. I will definitely open a new service request and I suggest you do the same.

best regards

Re: K60 ADC problem - Results are zero or random

JimDon — Sun, 17 Mar 2013 02:43:48 GMT

Ok this is a PE project generated for a FREEDOM board, which is an M0+

I absolutely know beyond any shadow of a doubt that it 100% works.

I used the default cpu setup, added the ADC component, set the clock for it, added 5 inputs and set the pins for the 5 inputs.

I also noticed that if your inputs are floating, guess what - you get random data.

So, import this project and study it.

Hope this helps.

Re: K60 ADC problem - Results are zero or random

JimDon — Sun, 17 Mar 2013 03:01:50 GMT

florijan,

It is hard to help you if you do not post your code. If you used PE to generate the code, it does work, it may just not be clear how to use it.

Re: K60 ADC problem - Results are zero or random

Florijan — Sun, 17 Mar 2013 10:42:06 GMT

Hi,

thanks for reply. First I have to say that measured ADC (11th)channel pin is not floating and is defined using 4k7 potentiometer at about a half Vcc (1,6V) so it is well defined and steady. My code is:

int main(void)

{

MCU_init(); /* call device initialization */

for(;;) {

ADC0_SC1A=11;

while ((ADC0_SC2 & 0x80));

while (!(ADC0_SC1A & 0x80));

result=ADC0_RA;

}

return 0;

}

Init code for ADC generated by Device Initialization GUI:

/* ### Init_ADC init code */

/* SIM_SCGC6: ADC0=1 */

SIM_SCGC6 |= (uint32_t)0x08000000UL;

/* ADC0_CFG1: ADLPC=0,ADIV=3,ADLSMP=1,MODE=1,ADICLK=0 */

ADC0_CFG1 = (uint32_t)0x74UL;

/* ADC0_CFG2: ADACKEN=0,ADHSC=0,ADLSTS=0 */

ADC0_CFG2 &= (uint32_t)~0xFFFFFFEFUL;

/* ADC0_CV1: CV=0 */

ADC0_CV1 = (uint32_t)0x00UL;

/* ADC0_CV2: CV=0 */

ADC0_CV2 = (uint32_t)0x00UL;

/* ADC0_OFS: OFS=4 */

ADC0_OFS = (uint32_t)0x04UL;

/* ADC0_SC2: ADACT=0,ADTRG=0,ACFE=0,ACFGT=0,ACREN=0,DMAEN=0,REFSEL=0 */

ADC0_SC2 = (uint32_t)0x00UL;

/* ADC0_SC3: CAL=0,CALF=0,ADCO=0,AVGE=0,AVGS=0 */

ADC0_SC3 = (uint32_t)0x00UL;

/* ADC0_SC1A: COCO=0,AIEN=0,DIFF=0,ADCH=0x1F */

ADC0_SC1A = (uint32_t)0x1FUL;

/* ADC0_SC1B: COCO=0,AIEN=0,DIFF=0,ADCH=0x1F */

ADC0_SC1B = (uint32_t)0x1FUL;

1. If I am stepping cpu the first pass of infinite loop the result is 0 at next pass the result is about 1,4V at next pass the result is 1,9V,... The result is wrong by as much as 30%.

2. But if I run the cpu without stepping it, it never reaches the line"result=ADC0_RA;" if I pause the execution it stops at line "while (!(ADC0_SC1A & 0x80));" like the conversion is never finished.

It is true that I am not using ADC calibration yet but this shouldn't be the cause of described behavior.

Best regards