• State
  • Locked Locked
  • Replies 11 replies
  • Subscribers 31 subscribers
  • Views 2428 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LME49740 voltage follower; DC offset with low source impedance

Bjorn Zetterlund
Prodigy 60 points
Other Parts Discussed in Thread: NE5532, LM4562, LME49860, LME49720

Hi All,

I'm using two LME49740 IC's as panpot buffers in the bus assignment section of a mixing desk.  This is the part where you decide if you send a mono signal to either the left or the right main summing bus (or odd or even subgroups) or to both to have the mono signal sent in equal amounts to left and right.  It is a dual 4K7 linear pot that has 4K7 'pull up' resistors to the wipers (to make the centre attenuation 4.5 dB rather than 6dB).  The wiper contacts go straight to the 4 non-inverting inputs of each LME49740 (one for left, and one for right).  Each opamp channel's output is shorted to the inverting input, and the four output terminals are connected together via 10 Ohm resistors.  The paralleling of four opamp sections is to get the output capacity to drive low impedance summing buses, and to also take advantage of the lower noise of parallel amplifier sections.

AC performance is fine and as expected, but something strange happens to the DC performance as the panpot nears the ends of its range.  When panned centre, the output impedance of the panpot is just short of 1K and the outputs of each quad buffer is practically at 0V.  When the panpot is all the way to one side, the outputs are at about -20mV (even though the input is at 0V, or at the -0.1mV offset of the opamp driving the panpot).  With the panpot about 100 Ohms away from its extremes of travel, output offsets are fine again.  Just as a sanity check, I swapped an LME49740 for the trusty old TLO74 and that behaved fine.  Could this be anything to do with interaction of bias cancellation circuitry?  Any ideas?

Just in case this was something to do with the paralleling of 4 opamp sections, on one of the LME49740's I isolated the input and the output so that it was only a single opamp channel being driven.  The output offset reduced to -7mV but again disappeared when the panpot was not at the extremes of travel.

Has anyone else had this issue with the LME49740 as voltage follower, or am I doing something stupid somehow?  My 100M scope doesn't detect any AC oscillation and each opamp is decoupled at the supply pins.

Many Thanks for any replies,

Bjorn

  • 0
    TI__Guru 52115 points
    Hi Bjorn,

    I will take a look at this issue and try to get back to you soon. In addition, I think it might be better for you to upload your schematic, which will be very helpful for us to understand your issue easily.

    Andy
  • 0 in reply to Andy Liu (BMS)
    Prodigy 60 points

    Thanks Andy,

    Will try to get a schematic up in the next few days. Am just modding an existing Soundcraft 8000 design and only have pencil doodles at this point!  As far as the problem with the LME49740, it really is just a voltage follower after a 4K7 pot driven by another opamp (with negligible offset).  Add the 4K7 pull up resistor from the wiper to the top of the pot and that's it!

    Best Regards,

    Bjorn

  • 0 in reply to Bjorn Zetterlund
    Prodigy 60 points
    That's the PDF, sorry not sure how to post as a picture. Am new to Eagle, don't worry about opamp names, the four on the right are the 49740's. Resistor values are correct.

    Thanks for any help!

    Bjorn
  • 0 in reply to Bjorn Zetterlund
    Prodigy 225 points
    In my opinion, you need once again to test your circuit for high frequency oscillations. Sometimes oscillations can be cancelled with small capacitances, as example - oscilloscope probes. Maybe capacitor between wiper and GND or top of the pot would work...
  • 0
    Prodigy 60 points
    Initially I thought it might be because of the input bias, this being a maximum of 72nA on each device. I have calculated this and even with an input impedance of the full 4K7 and the four lots of input bias you don't get anywhere near -20mV. You could try matching the feedback and input impedances though with some extra resistors. Also is there any current flowing in the zero volts path this can often cause offsets.
  • 0 in reply to Wilson Paul24
    Prodigy 60 points

    Thanks for your input guys! I decided to do another PCB implementing the same circuit with dual opamps rather than quads.  I tried LM4562's first, which I believe are nearly identical to the LME49740.  They showed nearly the same offset behaviour as the LME49740.  Then I tried NE5532's and they did not show any offset problems, even though they have a much higher input bias current.  They also have a very simple input stage.

    Regarding HF stability, the new PCB has 100nF caps immediately below the PCB between the supply pins.  The circuit with the LME49740's had different decoupling for Left and Right to see if that made any difference.  Left was a 200nF cap between the supply pins, Right was two 100nF caps to ground.  In each case the caps were right on the PCB socket.

    Best Regards,

    Bjorn

  • 0 in reply to Bjorn Zetterlund
    Prodigy 225 points
    Hi,
    NE5532 has relatively narrow bandwidth. According all the symptoms, only HF oscillations can get effect, as you observed. Supply bypass capacitors are good thing, but did you try capacitor between wiper of potentiometer and GND, or wiper and top of the pot? Maybe 100pF, or around this...

    Regards.
  • 0
    Prodigy 60 points

    Had a few more  thought on this, I tend to agree with Victor instability is the most likely cause, paralleling devices can cause instability.  Also because you have a 4k7 pot with pull up resistor at mid position this will be generating about 4nv per root hertz of noise.   So the pot will be generating more noise than the op amp i.e. paralleling wont deliver much advantage.  You might be better of devising a scheme to buffer the out put or selecting another device with more output current and using a single device.

    Regards

    Paul

  • 0 in reply to Wilson Paul24
    Prodigy 60 points

    Victor, I have yet to try the 100pF cap as you suggest.  Surely this will be shorted out of circuit when the wiper is at either end (which end depending on if it is connected from the wiper to GND or to the opamp ouput/top of pot).  Still, I do believe like you that it is some HF oscillation that causes the problem, only with voltage followers there isn't much you can do to prevent it unless you add more reistance (and noise) in the circuit.  To be fair to the LME49740, the datasheet doesn't mention paralleling opamps, although it doesn't mention that it is not recommended either.  

    As far as the noise of the pot and pull up resistor, in reality the panpot of a mixer channel will quite often be hard-panned, as in at the extremes of travel, at which point the source resistance is nearly zero, and parallel opamp sections would bring a noise benefit.  I think it is as simple as the LME49740 is not he right tool for the job, and I'm likely to go with the 5532. I have LME49860's on the way as well...interesting to see how they behave.

    Thanks for your input!

  • 0 in reply to Bjorn Zetterlund
    Prodigy 225 points
    LM4562 is the same chip as LME49720 (LM4562 is old name for this chip), and seems also as LME49860. Only one difference for LME49860 - higher limit for the supply voltage (maybe selected in manufacturing process). LME49740 is also the same, but four channels in one package.

    Regards.