Voltage divider networks should be linear, right?

Discussion in 'The Projects Forum' started by Camz, Jul 22, 2010.

  1. Camz

    Thread Starter New Member

    Feb 15, 2010
    15
    0
    Little background:

    I'm working on a project with the TI MSP430 microcontroller. Ultimately I'll have to monitor my battery voltage with the chip. There are 2 ways to do this, one using the a/d converter and another using the chip's built-in SVS feature that will raise a flag if the voltage falls below a user set threshold.

    My problem is this:

    I'll be using a 12v battery and the chip can take a max of 3.6v to it's input pins. No problem, I'll use a voltage divider to knock that down to that ballpark. So, I'm going to represent 0-12v with 0-3.6v. If I want to alarm on 10.4v; i would just use this equation to get my resulting threshold right:

    12 : 3.6 , 10.4:x >>> which ultimately ends up being 12x = 10.4(3.6) >> x = 3.12v.

    So in order to alarm on 10.4v; I'd just set my chip to alarm when the a/d converter "sees" 3.12v coming in on the input.

    I set up the circuit with the chip but the voltage values coming off the resistor don't really coincide with my calculated values. Obviously they won't be exact; but i'm getting discrepancies of .2-.3v which is too much when I'm working on this small scale. Any tips, advice? If it doesn't make sense please ask questions.

    Thanks!
     
  2. Markd77

    Senior Member

    Sep 7, 2009
    2,803
    594
    Any current coming out of the centre of the divider changes the voltage. Smaller value resistors would reduce this change.
    Using an op amp buffer would reduce the current coming out and thus the error.
     
  3. Camz

    Thread Starter New Member

    Feb 15, 2010
    15
    0
    thanks Mark; been out of the circuit design game for a while now; things are kinda fuzzy
     
  4. Camz

    Thread Starter New Member

    Feb 15, 2010
    15
    0
    So, using a general purpose 741 opamp coming off the divider would work?
     
  5. Papabravo

    Expert

    Feb 24, 2006
    10,142
    1,790
    Except there are modern alternatives to the "should be obsolete by now" 741.
     
  6. someonesdad

    Senior Member

    Jul 7, 2009
    1,585
    141
    You probably don't need lots of resolution to detect a low voltage condition. I recommend not using the full 3.6 V if that's the upper spec. Use e.g. 3 volts. The reason is that systems can be more reliable if you design more margin in. If that 3.6 V limit is valid, then a rare voltage glitch on the power supply might result on the voltage exceeding the maximum, potentially damaging the chip. If you have other information you didn't give that makes it so you don't worry about this, fine. But it's a good general design tip for both software and hardware.
     
  7. Camz

    Thread Starter New Member

    Feb 15, 2010
    15
    0
    heh; something like the LF353n?
     
  8. Camz

    Thread Starter New Member

    Feb 15, 2010
    15
    0
    good point someone'sdad. The chip has a max voltage of 4.2v probably for situations in which you describe, but it's recommended that the input be in between 1.8-3.6v.
     
  9. nomurphy

    AAC Fanatic!

    Aug 8, 2005
    567
    12
    Change one of the resistors to a trim pot, and then you can adjust the trip for where you want it.
     
  10. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    The LF353, TL07x, TL08x families of opamps will not be suitable because they cannot "see" within 3v of the negative rail.

    Don't forget that when being charged, the battery voltage may exceed 14.5v.

    If an automotive environment, you may get rather high voltage transients. You'll need to make certain that the input can't exceed your power rails.
     
  11. Camz

    Thread Starter New Member

    Feb 15, 2010
    15
    0
    Sgt,

    I'm looking on the datasheet of the LF353 and don't see anything mentioned about the the limits of what it "sees" within the negative rail. What exactly am I looking for that would tell me this?
     
  12. windoze killa

    AAC Fanatic!

    Feb 23, 2006
    605
    24
    Having looked at the datasheet for that micro the current on the ADC I/P pin is less than 1mA so I don't think a voltage drop would be too much of a problem. I would uses a 10K and a 2.2K resistor as the voltage divider. You could also use a trim pot if you want to be able to adjust it.
     
  13. Camz

    Thread Starter New Member

    Feb 15, 2010
    15
    0
    Would this be something more suitable?

    http://www.national.com/ds/LP/LPV521.pdf
     
Loading...