0
\$\begingroup\$

First of all I am a newbie in electronics.

I am building some weight scale applications using (as seen everywhere) load cells wired to a HX711 ADC and a ESP8266. I am now giving a try to the Texas Instruments CC1312R MCU (datasheet), still work in progress :)

The CC1312R features include a 8 channels 12-Bit ADC. My needs require at least a 16-Bit ADC.

Since this internal ADC provides 8 channels, I was wondering if I could somehow serialize them to achieve some extra bits resolution, thus not having to use any external ADC, is that possible?

Thanks for your help!

\$\endgroup\$
8
  • \$\begingroup\$ I think you are talking about measuring the same signal with all 8 channels and averaging the measurements. I doubt that will be any better than just averaging the measurements from one channel. You can sometimes achieve improved resolution by averaging, but I don't think it will be good enough for 16 bits. \$\endgroup\$ – Justin May 29 '20 at 14:28
  • 3
    \$\begingroup\$ It's just one ADC being multiplexed to all 8 channel so sampling on multiple channels won't do anything. \$\endgroup\$ – DKNguyen May 29 '20 at 14:28
  • 1
    \$\begingroup\$ In my experience, you might get an extra 2 bits of resolution, but the effective number of bits for the 12 bit channel is going to be less than 12 in the first place. analog.com/media/en/training-seminars/tutorials/MT-003.pdf \$\endgroup\$ – Peter Smith May 29 '20 at 14:40
  • 1
    \$\begingroup\$ Related: Accuracy of RMS of sine measured from N samples of ADC conversion. You can increase the effective resolution by combining multiple samples, but ONLY to the extent that the errors in the measurements are uncorrelated. And the improvement scales with the square root of the number of samples, so for each additional bit of resolution, you need to quadruple the number of samples. \$\endgroup\$ – Dave Tweed May 29 '20 at 14:56
  • \$\begingroup\$ @DKNguyen I understand, and what about piping the ADC channel to the internal DAC and then refeed an ADC channel (suggested here)? \$\endgroup\$ – Cinn May 29 '20 at 15:16
6
\$\begingroup\$

If you dither the signal and average more than 256 samples you can get some apparent resolution (not accuracy) increase, but frankly you’d be far better off ignoring the on-chip ADC and using an external ADC designed for load cell applications.

The one you mention (HX711, made by Avia in Xiamen, China) is only about 40 cents (24-bit delta-sigma with on-board PGA and bandgap reference) so it's not like it's going to break the bank!

\$\endgroup\$
3
  • \$\begingroup\$ Things like this quickly gives diminishing returns.. Though the ultimate example is the sigma-delta converter magically turning 1-bit into a nice 16-bit output. \$\endgroup\$ – pipe May 29 '20 at 14:43
  • \$\begingroup\$ For uncorrelated noise the maximum theoretical improvement is the square root of the number of samples, so 256 potentially gives 4 bits of improvement. Reality is going to be a lot more disappointing, especially with a typical crummy on-chip MCU ADC peripheral. \$\endgroup\$ – Spehro Pefhany May 29 '20 at 14:54
  • 1
    \$\begingroup\$ Delta-sigma is more than just about averaging. Noise shaping and post filtering are important parts of the process. \$\endgroup\$ – Dave Tweed May 29 '20 at 14:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.