F28027 ADC 0v offset

Hi Toby,

The ADC result value does not seem to report correctly the values between 0 & 0.3v.

Are the values varying a lot by the mean value or are you getting totally erroneous values?

Regards,

Gautam

Hi Gautam,

We are seeing an ADC result of 0 for values between  0 - 0.03v, above which the ADCs appear to operate as expected, i.e. 0.07v produces the expected equivalent ADC result value as if the ADC's were working correctly.

Thanks

Toby, are you using full range digital values? or are you using ">>x" expression while fetching AdcResults?

My mistake, the result values from the ADC appear to start at 0 at around > 0.03v

so the 0.07v ADC result will be out by -0.03v. e.g. at about 0.04v we see ~4 - 7 in the ADC result register.

I was using the Digital Power Lib ADC single channel macro to read the values from the ADCRESULT register, but now I am observing the register directly in the expressions window and still seeing the same issue.

Thanks

I've noticed that this issue is not present with all the ADC's... ADCIN A0, A1 & A2 appear to function as expected, but the rest of the ADC's follow the issue as described above :\

Yup, there seems to be an offset as you've mentioned earlier. Do you feel ADCREFTRIM register is playing a part in this? Did you try commenting the same and just calling ADC calibration?

Regards,

Gautam

Gautam, 

I will check this out - what is the way to call ADC_Cal()? to date I have used DeviceCal() as presented in the TI-made DevInit_F2802c.x file, as follows:

#pragma CODE_SECTION(InitFlash, "ramfuncs");
#define Device_cal (void   (*)(void))0x3D7C80

void DeviceInit(void)
{
	//...
	EALLOW;
	SysCtrlRegs.PCLKCR0.bit.ADCENCLK = 1; // Enable ADC peripheral clock
	(*Device_cal)(); // Auto-calibrate from TI OTP
	// ...
}

void main (void) {
	DeviceInit();
	//...
}

A little more info on my setup: I have 13 of the ADCs in use, all SOCs triggered from EPWM1 SOCA with an acquisition window size of 7 samples, and each result read via the digital power library single channel ADC macro onto a 32-bit unsigned int "net" variable. I have tried this with two different LanchPad XL PCBs now

Gautam

It is very odd that some of the ADCs (A0,A1,A2) work correctly and the rest dont... this would indicate that the calibration values are not to fault, would it not?

As to change the values to suit the non-working ADCs would surely mean that the currently working ADCs would then no longer work correctly...

EDIT: I just loaded up the Example_2802xAdcSoc example project and changed the channel selection to the one that is not A0,A1 or A2 and ran it, to find that the ADC (B4 in this case) works fine! Obviously this example project is quite a bit different from what I have running in my actual project in that it is exceedingly basic... however some things remain the same, such as the ADC basic initialisation, calibration. Th major difference being that I use stop-start mode rather than continuous mode, and that I use a lot more of the ADCs at the same time.

Another difference is the trigger frequency. In the example project it is very slow in comparison, just 915 Hz. In my project I use a trigger at 43.3 kHz... though changing the example project to use the same period does not seem to impact its performance.

Hi Toby,

It is very odd that some of the ADCs (A0,A1,A2) work correctly and the rest dont... this would indicate that the calibration values are not to fault, would it not?

As to change the values to suit the non-working ADCs would surely mean that the currently working ADCs would then no longer work correctly.

Did you try commenting the ADCREFTRIM register? If yes, what were your observations?

Regards,

Gautam

Gautam,

I ran the example project again, which calls the InitSysCtrl() function from the file DSP2802x_SysCtrl.c. Within this file I commented out the call to DeviceCal(), replacing it with a call to AdcOffsetSelfCal() from the file DSP2802x_Adc.c, which sets OFFTRIM only, not REFTRIM. This didnt seem to result in any change.

I proceeded by commenting out the the ADC cal altogether, which only resulted in the observed offset worsening to ~0.062v.

Rgds

Toby, try these steps:

1. Call InitAdc();

2. Then call AdcOffsetSelfCal();

These routines you would be able to find in "F2802x_Adc.c"

Also, check these links out:

http://e2e.ti.com/support/microcontrollers/c2000/f/171/t/232056.aspx

http://e2e.ti.com/support/microcontrollers/c2000/f/171/t/72784.aspx

FYI, here's ADC accuracy data:

Regards,

Gautam

That's great, Toby!

Good luck and Regards,

Gautam