ADC Accuracy

Gautam Iyer

Guru 192875 points

Other Parts Discussed in Thread: TMS320F28069, TMS320F28027, TMS320F28335

I'm working on TMS 320F2808 Control Card. The issue that I'm facing is with ADCs; the digital values that I tend to get contains last 4-5 bit variations. Thus my ADC value is unstable which is in turn effecting my calculations. I've attached the waveform of the digital values from ADCINA2 of the card. Out of 12bits, last 2 bit variations is acceptable but not more. Here I'm getting around 13-20mv of variations. How do I reduce this variation to less than 1mV ???

Regards,

Gautam Iyer

over 10 years ago

0 Igor Gorbachev over 10 years ago

Guru 20525 points

Hello!

First check the operation of the ADC if its input is short-circuit. 10-20 mV could have been by the induced interference.

Regards,

Igor

0 Adam Dunhoft over 10 years ago

TI__Expert 7885 points

Gautam,

Have you scoped the signal you are supplying to the ADC input to confirm there is no noise on the signal?

Have you tried to use a buffered dc voltage as the ADC input? If not, try that and see if the conversions are accurate.

Regards,

Adam

0 Gautam Iyer over 10 years ago in reply to Adam Dunhoft

Guru 192875 points

The above variations in waveform is with constant Agilent's DC supply, have tried with many other supplies too; but all in vain. So now I'm using the average principle to minimize the variations in adc values.

0 Igor Gorbachev over 10 years ago in reply to Gautam Iyer

Guru 20525 points

Hello!

There's something wrong. Check your input circuits and make sure equipment is properly grounded. It is induced interference or it is a problem reading data. 12-bit ADC can not have such a big inaccuracy. Averaging the result is not the best solution.

Regards,

Igor

0 Gautam Iyer over 10 years ago in reply to Igor Gorbachev

Guru 192875 points

Hi Igor,

Exactly, I agree with you; but the heights here are that I checked the ADCs on TMS320F28069 control stick, TMS320F28027 launch pad and now I'm working on TMS320F28335 control card all of them have this variations. For TMS320F28335 its between 0-9mV; but I'm unable to find the root cause for this noise. Checked with 3 dc power supplies and verified them on DSO too. The signals seems too be fine; still ADCs are introducing the noise.

Regards,

Gautam

0 Igor Gorbachev over 10 years ago in reply to Gautam Iyer

Guru 20525 points

Hi Gautam!

And yet are you sure you do not have any noises (interferences) on the pins REFHI/REFLO (for F28069 & F28027) or on the pin ADCLO (for F28335)? Forming "Digital Value" depends on them directly. I can not believe that so bad all!

Regards,

Igor

0 Gautam Iyer over 10 years ago in reply to Igor Gorbachev

Guru 192875 points

Hi Igor,

No improvement, those pins are internally grounded. Anyways thanks for all the help. Finally was able to reduce the variation to 3mV for TMS320F2335 and happy with the results.

Regards,

Gautam

0 Gautam Iyer over 10 years ago in reply to Gautam Iyer

Guru 192875 points

Here's the final result with TMS320F2335, these are the pure ADC values; through different power source though!

0 Igor Gorbachev over 10 years ago in reply to Gautam Iyer

Guru 20525 points

Hi!

Congratulations! Now it looks very nice! What was the solution (averaging or not)?

Regards,

Igor

0 Gautam Iyer over 10 years ago in reply to Igor Gorbachev

Guru 192875 points

Just used another channel of Agilent power supply....! Also what I feel is the ADCLO pin is supposed to be fed to Analog ground (datasheet says: Low Reference (connect to analog ground) ), but on the control card the analog and digital grounds are same. That might also introduce some erroneous values.

Regards,

Gautam

0 Igor Gorbachev over 10 years ago in reply to Gautam Iyer

Guru 20525 points

Thanks.

Good luck,

Igor

C2000™︎ microcontrollers

C2000 microcontrollers forum

ADC Accuracy