F28335 ADC shows incorrect values

Certainly you do need to add the power up delay to your code as spec'd in the DS after the ADC is turned on; but I do agree that if you are single stepping/breakpointing after this power up happens then you are most likely observing the required time. 

I do not see where you have set the HSPCLK divider to get the ADC clock to 25MHz, this should be in your main as well but I want to make sure.  If ADC Clock >25MHz results will be unpredictable.  I'm assuming you are running the core @150MHz, so you will need a value of 3(which gives divider of 6) to get 25MHz.

From code you are sampling ADCINB0, have you confirmed voltage at the pin?

Best,

Matthew

Hi Matthew,
thank you for a quick reply. I added a delay loop:  for(i=0; i<1E+4; i++) for(i1=0; i1<5; i1++);
checked the clock: SysCtrlRegs.HISPCP.all = 0x0003; and   SysCtrlRegs.PCLKCR0.bit.ADCENCLK = 1; // ADC

and connected the ADCINB0 to the ground but it still consistently shows sth around half the max value. The ADC hardware is reacting to my setup as it shows just two conversions, as I specified. Hope you might have some more hints ...

Can you post/attach the contents of all ADC registers?  I just want to look at everything that is setup in case there is something I am missing.

Along these lines; this device requires ADC calibration SW routine be ran before conversions, to properly trim the internal circuits to meet the DS. 

1)Can you confirm this is done

2)Can you confirm there is NOT a SW reset issued to the ADC regs(ADCTRL1 I think) after this is done?  Reason is that ADC Reset will clear out the trim registers and cause ADC to give bad results.  In this case ADC calibration would need to be re-called by the SW.  Data dump will tell me if these are in correct state or not.

Best,

Matthew

Thank you for this solid and attentive support. I have been silent for a while since in the meantime I have assembled another prototype PCB and it seems the hardware is at fault. The same software works OK for the new PCB. I also measured the resistance of the ADC inputs to the ground and on the new, working PCB all of them showed some hudrends Ohm but on the not-working old one out of four ADC inputs that are available on my PCB  (A0, A1, B0, B1)  three showed also 600-700 Ohm but one - B1 showed low resistance ca. 15 Ohm. However, it is not only the B1 input on the old one that is not working, I got the same 0x7F0 values when I applied ground and followed the other inputs. The uC on the old PCB does not show any other misbehaviour, processor and PWM-s work flawlessly.

I would simply switch to the new PCB for testing and forget the old one but the problem is I have managed to burn the new one soon after the test and now I'm left with the old one with little possibility to replace either uC as f28335 is not available at the moment at any distributors in the world...
Is it possible that the ADC section would be somehow broken while the rest OK ? Or there is a chance there is sth wrong with my PCB that makes the ADC section mulfunction - any way to check it somehow ?

Best regards,

Pawel

Pawel,

I think that if one ADC channel is showing too low resistance then it also has the potential to disturb the other channels since they share a  mux with the ADC.   It sounds like there has been some damage to the analog portion of the device.  Is the current draw on the analog power domain within the DS limit?  If not it would be a clear indication some damage has occurred.

Can you comment if the ADC channels are seeing a volage before the power is applied to the device?  Any voltage on the ADC inputs should be kept at or below the 3.3V power rail as it ramps up during power up.  In the same vein the 3.3V analog power should be ramped at the same time as the 3.3V digital supply.

Additionally even while running the analog inputs must stay at or below VDDA level; which is 3.3V nominally.

Please let me know your observations.

Best regards,
Matthew

I thinks I also encounterd some problems with ADC conversion, to much noise, for example, when I connected A0 channel to GND, of course, most of the time I saw the ADCRESULT0 show around 0x000, but sometimes it jump like a crazy horse, even to 0x1xxx! 

Here's my ADC code:

InitAdc();

// Specific ADC setup for this example:

  AdcRegs.ADCTRL1.bit.ACQ_PS   = ADC_SHCLK;

  AdcRegs.ADCTRL3.bit.ADCCLKPS = ADC_CKPS;

  AdcRegs.ADCTRL1.bit.SEQ_CASC = 1;        // 1  Cascaded mode

  AdcRegs.ADCTRL1.bit.CONT_RUN = 0;       // Setup non-continuous run

  AdcRegs.ADCTRL1.bit.SEQ_OVRD = 0;       // Disable Sequencer override feature

  AdcRegs.ADCMAXCONV.bit.MAX_CONV1 = 0x0001; // Setup 2 conv's on SEQ1

  AdcRegs.ADCCHSELSEQ1.bit.CONV00 = 0x0; // Setup ADCINA0 as 1st SEQ1

  AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x1; // Setup ADCINA1 as 2nd SEQ1

while(1)

{

delay_ms(1);    //delay 1ms

AdcRegs.ADCTRL2.bit.RST_SEQ1 = 1;       // Reset SEQ1

AdcRegs.ADCTRL2.bit.SOC_SEQ1 = 1; // Start ADC conversion

AdcRegs.ADCTRL2.bit.INT_ENA_SEQ1 = 1;  // Enable SEQ1 interrupt (every EOS)

}

interrupt void  adc_isr(void)

{

Vd1_new= AdcRegs.ADCRESULT0 >>4;

Vd2_new= AdcRegs.ADCRESULT1 >>4;

   AdcRegs.ADCST.bit.INT_SEQ1_CLR = 1;       // Clear INT SEQ1 bit

AdcRegs.ADCTRL2.bit.INT_ENA_SEQ1 = 0;  // Disable SEQ1 interrupt 

   PieCtrlRegs.PIEACK.all = PIEACK_GROUP1;   // Acknowledge interrupt to PIE

  return;

}

Please give me some advices!

Duy, please start a new thread for your issue.  Thanks.