Unable to realize summing amplifier/voltage follower using lm358 opamp!

Discussion in 'The Projects Forum' started by Abhimanyu G, Mar 23, 2015.

  1. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    Hello guys,
    I have built a Summing amplifier using lm358P such that the output follows the sum of input signals (atleast that is what i wanted it to do..!) , i've built it to connect 6 guitars and 1 audio input (mostly from the head phone jack of PC). This Amplifier is followed by a pre amp with controllable gain whose output is fed to a speaker.

    I can only use a single power supply of 12v, hence i've offset-ed the summed signal by vcc/2.

    The problem i am facing is that the summing amplifier stage is not linear, meaning , the output does not follow the input within the voltage swing range and it is getting distorted and i'm unable to understand why. Plz help!
     
  2. ericgibbs

    AAC Fanatic!

    Jan 29, 2010
    2,504
    380
    hi AG,
    You circuit will not work as you expect.
    The leftside OPA is configured as an Inverting amp, so if you have Vcc/2 on the Input, what do you think the OPA Vout will be.?
    Also the overall gain of the OPA will be determined by input resistance of all of your inputs
    E
     
  3. DickCappels

    Moderator

    Aug 21, 2008
    2,664
    634
    You should connect the noninvering (+) inputs to your 6 volt reference and put a large capacitor from your 6 volt reference to ground.

    You also need some other capacitors in your circuit.
    Put a capacitor in series with each audio input resistor with the negative end connected to the pot wiper.

    Put a decoupling capacitor (100 uf should be fine for an LM358) between pins 8 and 4, the positive end connected to pin 8.

    Put a capacitor in series with the audio output to remove the 6 volt reference from the audio signal, with the positive end connected to the LM358 output.

    It looks like you might be driving a speaker directly with the LM358 output. That would not get good results; better to use something like an LM380.

    Incidental things:
    The LM358 is not very good for audio, but it is ok for experimenting. It lacks bandwidth and slew rate and its noise performance is not very good eithere.

    The 1k resistors are on the low resistance side and might cause problems with dynamic range. 10k to 47k is more typical of "casual" audio circuits.
     
  4. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    Overall gain should be 1, because V0=(-Rf/Ri)(V1+V2+V3+.....)... since Rf=Ri=1k gain should be 1. Plz correct me if i am wrong
     
  5. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    I purposefully connected it to Inverting terminal because , when the dc ref signal shifts the summed signal to higher voltage (vcc/2 in this case), the AC signals tend to go towards GND (Virtual GND actually) and this will avoid crosstalk, is this assumption wrong?
     
  6. DickCappels

    Moderator

    Aug 21, 2008
    2,664
    634
    Your assumption is correct with respect to the first inverting stage. Part of the problem is that the output of the second inverting amplifier will try to go below ground.

    If you connect the non-inverting input to your 6 volt reference, the output of both amplifiers will swing above and below that voltage if you AC couple the input signals.

    A side effect of direct-coupling the 5k pots into the summing node (via 1k summing resistors) is that as you adjust the pots the DC offset will change.
     
  7. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    I'm not that worried about 2nd stage as of now, my concern is if i connect 6v to Non-Inverting terminal will it still avoid cross talk? if so plz explain.
     
  8. ericgibbs

    AAC Fanatic!

    Jan 29, 2010
    2,504
    380
    hi,
    The three 1K are in parallel to thats ~330R

    If the two, 5K pots are on the centre setting they are effectively two 2.5K's in parallel.
    That is 1.25K per pot equivalent to 625R for the pair of pots

    The two 1k's on the divider are in parallel, so thats 500R

    So overall thats 330R+625R+500R = 1458R say 1500R, the OPA Gain would be 1500/1000 = 1.5.

    BUT as you change the settings of the 5K pots the Gain will change.

    Do as Dick has suggested AC coupling and Non Inverting Offset.

    E
     
  9. DickCappels

    Moderator

    Aug 21, 2008
    2,664
    634
    I don't understand you mean by "crosstalk" -please explain and I will try to answer your question.
     
  10. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    I mean, Signal comming from S1 should enter opamp and should not go into any other channel. Let me give an example, say S1 and S2 have sinusoidal signals connected to them, at some instant of time say S1 is 3v and S2 is -3v. Now since S2 potential is lower than S1, S1 tends to go to S2 rather than going to virtual ground terminal (since virtual ground node is now at Vcc/2). I "Don't" want this to happen !
     
  11. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    I do not understand which 3 1k resistors are in parallel ?!
     
  12. DickCappels

    Moderator

    Aug 21, 2008
    2,664
    634
    The summing node will still be a virtual ground, except that it will be at +6 volts instead of ground, so there will be no crosstalk, in theory. No signal current enters the +6 volt reference because the only thing connected to it are the inputs of the opamps and one hopes a suitable bypass capacitor to remove any signal that might appear on the power supply.

    You may be skeptical but this is a very common way to sum signals in a single supply circuit.

    [​IMG]
    http://www.next.gr/circuits/Single-Supply-Summing-Amplifier-for-Audio-Mixer-l52754.html
     
  13. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    Thank you Dick, i applied your suggestion and simulated the circuit in proteus. It seem like i am getting an output :) let me try to implement it on the circuit board. Thank you once again :)

    In the mean time could you suggest me some good and easily available dual opamps for audio applications.?
     
  14. ericgibbs

    AAC Fanatic!

    Jan 29, 2010
    2,504
    380
    I have shown the effective Rinput resistance on your diagram, this does not include the Source resistances the individual input are driven from.

    Dick's circuit is the conventional way of summing inputs, I would use it.:)
     
  15. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    Each 1K resistors are dedicated to single channel, so if 5k pots are at 50% i.e 2.5k the signal flow will be something like this... Signal->2.5k->1K->input of the opamp. Conventionally parallel means two ends are connected to same node, i fail to see how this can be termed as "Parellel" to each other... plz explain..
     
  16. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,546
    1,252
    The LM358 is an old but still excellent general purpose opamp. However, audio is not a general purpose application, often requiring both gain and bandwidth for ow distortion. One of the first opamps designed specifically for audio was the NE5534. The dual version is NE5532. RC4560, RC4580, OPA2604, OPA2134 are other excellent audio devices. Because these have much more gain and bandwidth than the LM358, pc board layout is more important to prevent unwanted oscillations.

    ak
     
  17. Abhimanyu G

    Thread Starter New Member

    Mar 23, 2015
    18
    1
    [QUOTE=" Because these have much more gain and bandwidth than the LM358, pc board layout is more important to prevent unwanted oscillations.
    [/QUOTE]

    AnalogKid- What do you mean by it? could you elaborate?
     
  18. DickCappels

    Moderator

    Aug 21, 2008
    2,664
    634
  19. ericgibbs

    AAC Fanatic!

    Jan 29, 2010
    2,504
    380
    Hi AG.
    Many apologies, I mis-read the original question, sorry if I confused you.:(
    I will go outside and beat myself with a wet Birch tree branch!
    E
     
  20. crutschow

    Expert

    Mar 14, 2008
    13,056
    3,245
    Don't quite understand that statement about "overall gain".
    The gain of each input is independent of the gain of the other inputs and is equal to the feedback resistance divided by the input resistance of that input.
    The gain of all the inputs paralleled together is sometimes called the noise gain since the op amp noise, bandwidth, and DC offset is determined by that value.
     
Loading...