Need a simple transistor amp circuit

Discussion in 'The Projects Forum' started by wayneh, Sep 21, 2010.

  1. wayneh

    Thread Starter Expert

    Sep 9, 2010
    12,126
    3,048
    Is there a simple, single transistor voltage amplifier to give me a voltage gain of at least 4X, using an input voltage of 0.1-0.3v?

    I'm using the LM35 thermometer IC in a thermostat application; it gives a voltage of 10mV per °C. My LM339 comparator seems to have hysteresis of about ±5 mV. That means my thermostat has about ±0.5°C of hysteresis. Not bad, but if I could amplify the LM35 voltage then I think I could decrease my temperature deadband.

    I'm vaguely familiar with how to amplify an AC audio signal (see http://www.falstad.com/circuit/e-ceamp.html), but in my case it's a pure DC I need to boost. Can I simply omit the decoupling caps?
     
  2. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
  3. wayneh

    Thread Starter Expert

    Sep 9, 2010
    12,126
    3,048
    Op amps I know, but I've got a drawer full of transistors and limited board space. I guess that's not much of an excuse. An op amp and two resistors will probably take less room (and work better) than a bunch of biasing resistors and a transistor.
     
  4. wayneh

    Thread Starter Expert

    Sep 9, 2010
    12,126
    3,048
    Just to follow up: I went ahead and added an op-amp to my thermostat circuit, to amplify the voltage put out by an LM35, eg. 200 mV at 20°C. I used a negative feedback gain of about 16X on the op-amp (an MC33178 I had laying around), which brought my temperature output up to ~3v.

    This was quick and easy - much easier than tuning a transistor circuit - and reduced my temperature hysteresis to about ±0.15°C instead of ±0.5°C. I'm a little surprised it didn't get even touchier than that, but anyway this level is plenty tight for my application. Problem solved, thanks to beenthere for reminding me how much better op-amps are for simple amplification tasks.
     
  5. PRS

    Well-Known Member

    Aug 24, 2008
    989
    35
    Op Amp is your best choice. Your low voltages suggest this. When you get into really low voltages like yours you do best with a dual power supply that is capable of zeroing zero volts right at zero volts. Op Amps allow for this.
     
  6. wayneh

    Thread Starter Expert

    Sep 9, 2010
    12,126
    3,048
    Woot! I just corrected a problem and took my deadband down to the ±0.02°C range. That's crazy precise for such a simple circuit and device. Gotta love those op amps. Oh, and I'm dividing the 5v power supply to compare the amplified temperature to, so I don't need dual supply to give me zero. It ends up making the cooler on-off switch at ~3v. But I get your point.

    The last problem was caused by a grounded output on the unused op-amp on a MC33178 dual op-amp. A datasheet for the 1458 showed grounding both inputs and the output of any unused op-amp. (I can't find that reference!) I assumed (there's that word) this was the right procedure for general purpose op amps, but it sure wasn't good for the MC33178. It was drawing enough juice to heat up the 7805 supplying it, although it was not warm itself (it's a "high power" op amp). The 7805 cooled off immediately when I pulled the ground pin and like magic my hysteresis all but disappeared.

    Does it make sense to leave the output open or should I run a 10K to ground or something? Leave the inputs pinned to ground?
     
  7. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    If the opamp is unity-gain stable, you can wire the inverting (-) input to the output, and the non-inverting (+) input to the other output, or a voltage source between ground and Vcc.

    If the opamp can sense and output down to the negative rail, then you can wire the non-inverting input to ground.

    If the opamp is not unity-gain stable, you will need to provide more information.

    Keep in mind that you can use the unused opamp output to provide an artificial ground reference by feeding the + input from a voltage divider network, and wiring the - input to the output. You may need a Bucherot cell (low-value resistor in series with a 0.1uF cap to ground) in order to suppress oscillations.
     
  8. wayneh

    Thread Starter Expert

    Sep 9, 2010
    12,126
    3,048
    That makes sense, and I've got a ~3v on-board reference.

    Hmmm, I'm not really sure but since I have the ref mentioned above, I can use that instead of ground.

    All I can say is, here's the datasheet.
     
Loading...