Differential ADCs and negative signals

Go To Last Post
8 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The new AVR DA family has a 12 bit differential ADC. My application involves reading small AC (sine-wave) signals from a CT with a burden resistor across it with one leg connected to ground. With a single-ended ADC, the signal gives a reading in the positive half of the sine wave and reads as 0 during the negative half. Does having a differential ADC mean that it can read the negative part of the signal also, the part which is actually a negative voltage w.r.t ground? 

-Sam

Last Edited: Wed. Aug 5, 2020 - 02:40 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No.

 

Instead of reading a single ended ADC input with respect to Ground, it will read the difference between two ADC input pins, but both pins must still be a (+) voltage.

 

On some of the chips the differential input also goes to an op-amp, so one can have gain inside the uC without adding an external op-amp.

 

JC 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

You didn't post a schematic, but perhaps:

 

Instead of connecting the one end to Ground, you should connect it to a "virtual ground" at Vcc/2.

 

Now the other end deviates a little bit above Vcc/2 and a little bit below Vcc/2, and you can feed that into the (single end, (+) voltage only) ADC.

 

The VG is most easily constructed with an op-amp voltage follower.

 

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:

You didn't post a schematic, but perhaps:

 

Instead of connecting the one end to Ground, you should connect it to a "virtual ground" at Vcc/2.

 

Now the other end deviates a little bit above Vcc/2 and a little bit below Vcc/2, and you can feed that into the (single end, (+) voltage only) ADC.

 

The VG is most easily constructed with an op-amp voltage follower.

 

JC

This will only work if the source and the microcontroller are galvanic isolated.

 

You can use two resistors on each of the two differential inputs. Like a voltage divider, but with one side connected to your controllers supply so that the input voltage to the ADC is level shifted and always >0V.

The supply voltage doesn't have to be very stable, it does not affect the ADC result because of the differential inputs.

 

Unfortunately, the DA's do not have a gain stage. If the input voltages are very low this won't work, in that case you can use an external differential amplifier with an op-amp.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:
The VG is most easily constructed with an op-amp voltage follower.

The AVR-DA MCUs have a DAC that can be used for this, so no external parts are needed.

 

eauth wrote:
This will only work if the source and the microcontroller are galvanic isolated.

Indeed. We don't want stuff exploding.

 

eauth wrote:
Unfortunately, the DA's do not have a gain stage. If the input voltages are very low this won't work, in that case you can use an external differential amplifier with an op-amp.

Well, at least it's possible to use the 1.024V internal reference as full scale, so compared to 5V VDD as reference it's already equivalent to 5x amplification, better than nothing I guess...

The DAC virtual gnd would be 0.512V.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

This will only work if the source and the microcontroller are galvanic isolated.

 

Isn't this the case with a current transformer?

 

JC 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hopefully yes, unless the MCU is somehow being powered from the transformer, we never know...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Now the other end deviates a little bit above Vcc/2 and a little bit below Vcc/2, and you can feed that into the (single end, (+) voltage only) ADC.

 Or perhaps, with your transformer output leg raised to Vcc/2, measure both that and the other lead (differential)....now even if you are at Vcc/2.08, your reading will be right on the money.... this is true only if the ADC has the ability to provide negative (signed) results (I didn't look)...you can get negative results from two positive voltages, like 2.3-2.7 = -0.4.  The cheapest differential setups only allow a positive result from two positive voltages.

 

A 100% true diff setup allows positive or negative results from voltages that are each positive or negative.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!