Bipolar ADC possibility

[drobison]

Posted on December 31, 2012 at 16:49

I will use the STM32 to drive a power supply.  The input signals to the ADC do not need any gain, so I'd like to avoid using as many OPAmps as possible.  The problem is that many of the potentials are negative or transition from positive to negative.  It's annoying to have to incorporate OPAmps onto the board for things like making the negative potential positive or shifting the zero up or down.  I would have to do these things if the ADC can only be unipolar.  If I can have a bipolar ADC I can eliminate many OPAMPS.

The problem is the document for the ADC is terrible.  I have no idea what potential is allowable on the VSSA, VDDA, Vref-, Vref+ pins.  The manual even contradicts itself and makes no sense.  IS there a better resourse of the ADC wiring and the possibility of operating it in bipolar instead of unipolar mode? #adc-unipolar-biploar

[jj2]

Posted on January 02, 2013 at 04:50

''there is a vref- pin...'' suspect this does not mean, ''negative with respect to ground'' - thus your confusion.  This pin - instead - connects to the most negative potential of your introduced signal or source - but is constrained to remain w/in the MCU's power supply limits.  (i.e. not negative)

Deep w/in the Analog ''Spec'' section of the datasheet - should be voltage limits for this/similar MCU pins - and this sets the signal swing limits.   Agree with earlier fellows ideas/suggestions - yet we note that you ''attack'' both the use of Op-Amps (blameless) and the MCU (again blameless) - while leaving the ''real source'' of your misery - your ''below ground'' signal source or generator - free from disdain!  And why is this? Perhaps you may not require ''true'' positive and negative supplies to correctly operate this source/sensor/generator.  While the output will likely shrink (perhaps by 1/2) at least this may eliminate the troublesome, negative excursions.  Seems inconsistent that the offending ''source'' has escaped blame...  May prove fruitful to look there... Should the source be unable to be ''tamed'' - quad op-amp should easily ''pull'' 4 of your most negative signals to 0V - allowing safe/proper introduction to MCU ADC - so long as the more positive portion of such signals do not exceed Vadc(max).

[trevor23]

Posted on December 31, 2012 at 16:55

The ADC is not bipolar (I've never seen a microcontroller ADC that was).

[drobison]

Posted on December 31, 2012 at 17:06

So, even though there is a vref- pin and the manual says(possibly) it can be attached to negative potential, it may not be connected to negative potential?  Maybe I am not using the terms bipolar and unipolar correctly. 

You have to understand my confusion.  There are a vref+ and a vref- pin.  I'd like to connect them to 3.3 and -3.3 then I can just subtract half of the ADC range off in the software.  Is this not possible?  What are the allowable potentials for the vref+ and vref - pins.  Also where do you get any info on this?  I've spent hours searching.

[frankmeyer9]

Posted on December 31, 2012 at 17:30

There are external ADC's, which support true differential inputs. But id you look closer, the have a positive and negative supply voltage.

Albeit theoretically possible, vendors avoid the trouble of implementing such ADC's, which are rarely needed. The easiest option for you would be either such an external ADC, or an OpAmp to shift/amplify the input voltage to the input range of the internal ADC. And this ext. OpAmp would need a negative supply voltage, too. There is no way around, otherwise you only feed the clamping diodes...

[trevor23]

Posted on December 31, 2012 at 18:05

One reason for vref- is that you can have an analog ground that might be slightly differient to GND (due to ground currents) but only by a small amount. As FM has said either use an external bipolar ADC (e.g. with SPI interface) or use some OP-AMPs to bias the signal.