Developing specialized voltmeter

Discussion in 'The Projects Forum' started by onesandzeros, Aug 6, 2010.

  1. onesandzeros

    Thread Starter New Member

    Aug 6, 2010
    Hi everybody, new member & first post.

    I'm having a problem similar to the original posters'. A little background: I'm developing a specialized voltmeter. It needs high impedance, bipolar operation, and auto-ranging from 0 to 200V AC/DC. It must be small and battery powered. I did a prototype using through-hole pcb and things seemed to work fine. I have now migrated to surface mount and I'm having a problem with a voltage follower.

    The input signal enters thru a 20 meg series resistor followed by clamp diodes to limit the input to +/- 3V. There follows some range selecting via analog switches and DC/AC selection relay (I have disconnected these and the clamp diodes to eliminate them from the problem). There is also a 20 meg resistor to ground, effectively halving the input voltage. Then comes a LMC6042 op amp connected as a voltage follower, input to the non-inverting pin. V+ is 3.0 volts, V- is -3.0V, well bypassed, from 3-terminal linear regulators. Following the voltage follower is a precision rectifier and polarity detection circuit and finally an A/D converter. The problem is, as the DC input voltage is gradually increased from 0 up, as soon as the input voltage at the voltage follower exceeds 1.8 volts, the op amp output immediately goes to the + rail & stays there til the voltage is reduced. Because of the high impedance of the 20 meg resistors on the input side, I can only measure the voltages at the input point and at the output of the voltage follower- on the input side a scope or voltmeter probe loads the signal too much. Given the input spec of the LMC6042 (Differential Input Voltage = ±Supply Voltage) I would expect the input could go up to almost 3 volts. This problem did not happen with the through hole prototype board, where the amp was an 8-pin DIP. In the SM board it is a SO-8 package. Also, this does not happen for negative inputs, I can go from 0 to -2.5 volts no problem. This is a 2-channel meter with two independent analog boards and the same thing happens on both boards. I am a little lost on analog op amps specs, but I think the LMC6042 meets most of SgtWookie's and kkazems's caveats (except micropower, battery device!).

    Given the circumstances, I am inclined to think that the problem may be board surface leakage paths, since the geometry is a lot closer on the SM board. In retrospect I probably should have put 'guard bands' around the op amp inputs because of the high input impedance at the input circuit. This would of course require re-doing the boards- very bad news for a project that is already over budget and behind schedule and with anticipated product introduction in early September. Any suggestions or direction to relevant source material would be greatly appreciated. Very frustrating to do the all the digital stuff (logging to SD card using FAT16, user interface, etc, and be humbled by a simple voltage follower.

    Sorry for the over-long post.....
  2. beenthere

    Retired Moderator

    Apr 20, 2004
  3. eblc1388

    Senior Member

    Nov 28, 2008
    With a 20MΩ input, the guard band around the opamp inputs are a must in your design to avoid leakage current from affecting the operation.

    You have input voltage exceeding the common mode voltage range of the opamp, as show below.

  4. sage.radachowsky


    May 11, 2010
    Yes, I agree. You're outside the limits of the device. Can you provide higher supply rails?

    I have wondered if this device is the input stage to the RadioShack 22-820 compact multimeter. I have a Fluke and other multimeters, but this little thing is *amazing* for its ultra-low input impedance. I can charge a 10uF poly cap to 153 mV and read the voltage, and it stays steady at the same millivolt for minutes. If I test the voltage with the Fluke, it drops like 2 mV per second. It on another planet in terms of input impedance.

    What is this project for, if you don't mind at least giving general clues? I am interested to know what's going on in ultra-low current voltage sensing.

    Sometimes the worst thing is for a circuit to work beyond spec, and then you build it and find that it doesn't work with other parts of the same model... it's happened to me. But I am pretty sure you need to give a higher (+) rail at least, if you want to follow up to 3V. 5V might work but to really follow spec you need 5.5V or 6V.
  5. eblc1388

    Senior Member

    Nov 28, 2008
    Yes many are caught up assuming the input impedance of the Fluke is still at 10MΩ even in the lowest voltage range. I don't know about others model but my Fluke 75 has an input imedance of only 2~3MΩ for the 300mV range.

    It is definitely too low as even my very old LED Schlumberger DVM can mange over 50MΩ in the mV range using bootstrapping.