2

I need to drive a ratiometric current sensor ACS758xCB with a nominal supply current of 10 mA. Due to the ratiometric nature of the sensor I would prefer to supply it from the reference voltage of my ADC. Obviously I need to buffer the reference.

However, most op amps, notably those with low offset voltage, require the capacitive loading of the op amp be limited. Since the sensor itself contains active circuitry it requires supply bypassing in the range of 100 nF.

schematic

simulate this circuit – Schematic created using CircuitLab

How does one usually cope with this situation? I've looked into tracking regulators such as TPS7B4250 but the tracking performance of +/-5 mV does not sound too impressive.

Null
  • 7,448
  • 17
  • 36
  • 48
Arne
  • 1,805
  • 3
  • 27
  • 43
  • 1
    The datasheet of the Hall sensor says it can operate on a supply voltage between 3 and 5.5 V. So why use 4.5 V ? Most opamps are not so suited to drive 10 mA with only 0.5 V drop (Vcc - Vout) and into a large capacitor. The tracking performance of the TPS7B4250 might not **sound** impressive but I say that it is **enough**, now **prove that I'm wrong** (by calculations etc.). That is how engineers decide if something is OK or not, not by deciding that *it does not sound impressive*. – Bimpelrekkie May 23 '16 at 08:14
  • 1
    You don't have to supply the sensor with ref voltage, you supply the sensor with available voltage, like Vcc CPU voltage. Then you have to offset the Vout from sensor with Vref voltage by means of Op amp. The offset should be placed at Vadc/2. – Marko Buršič May 23 '16 at 08:19
  • *Due to the ratiometric nature of the sensor I would prefer to supply it from the reference voltage of my ADC* Why ? Provide a proper reason why this is needed. I'd say you cannot provide a good reason because the power supply rejection is not mentioned at all in the datasheet. In my opinion you can feed the sensor directly from +5 V DC. – Bimpelrekkie May 23 '16 at 08:22
  • @FakeMoustache: I fully understand that the dropout of 0.5V between the opamp supply of 5V and the sensor supply of 4.5V is marginal for a 10mA load with significant capacitance. That why I ask "[...] How does one **usually** cope with this situation [...]" – Arne May 23 '16 at 08:26
  • @MarkoBuršič My understanding is that for ratiometric sensors the output is a relative to the supply. Now, when my 5V supply has +/-5% over temperature, but my ADC reference has only +/-1.5% over temperature this induces an error. The error would not be there if both sensor and ADC would run on the same voltage. – Arne May 23 '16 at 08:28
  • There is no **usually** as each design is different. A possible solution is just using the 5V. You assume that you need extra measures to make it more stable etc. but you provide no proof that why you need this. – Bimpelrekkie May 23 '16 at 08:30
  • @FakeMoustache if there were not **usually** all application notes and this this site, too where of absolutely no value. – Arne May 23 '16 at 08:32
  • @Arne The specs suggests that ratiometry work well with 5V supply, other scenarios contact Allegro. OK, if you want ratiometric you can have 5VDC LDO with low pass fiters for noise, and then you forward the Vfref to the MCU with Opamp buffer/divider – Marko Buršič May 23 '16 at 08:35
  • Indeed, the sensor's output is proportional to its supply voltage. Then the question is what kind of accuracy do you need, what measurement bandwidth do you need ? If the 5 V is stable enough you could filter it and use that. Or use a 3.3 V LDO. – Bimpelrekkie May 23 '16 at 08:36
  • @MarkoBuršič I can't feed the 5V back as reference to the ADC. Instead the ADC provides its internal reference on a output pin. – Arne May 23 '16 at 08:38
  • @Ame which shows how this is usually done: by using an ADC that can accept an external reference. – Wouter van Ooijen May 23 '16 at 09:08
  • @Arne Opamp buffer/divider, not 5V direct. You should add more details on your app, exact sensor type, you want measure AC or DC? What MCU? – Marko Buršič May 23 '16 at 09:27

1 Answers1

1

This is a common problem having a common solution:

schematic

simulate this circuit – Schematic created using CircuitLab

The circuit provides direct High Frequency feedback from OPA output to its input via C1. DC feedback goes through R2 R3 and is not affected by C1. R1 prevents shortening of HF feedback by the capacitive load. This design can handle 100 uF of load capacitance or even more.

Some comments on components:

  1. You have to consider input offset / bias current of OPA when choosing R2 R3. The voltage drop created by offset current on parallel connection R2 R3 must be smaller than your allowed error.
  2. C1 can be selected experimentally. Too small values cause the circuit to oscillate. Place at least 2 times larger C2 than the largest one that causes oscillations.
  3. R1 can be 10 to 47 Ohms in your case (10 mA load, right?)
Master
  • 1,289
  • 6
  • 4