4.5V from OPamp LMV321

[mathieusan]

Designing a power supply section of an audio circuit (not shown) which needs VCC of 9V and 4.5V for some bias voltage of op-amps. I want to supply the 4.5V bias with as low impedance as possible so I am using another opamp as a voltage follower. In the power supply circuit (attached) the 9V supplies go through a voltage divider (2 x 22k) to obtain 4.5V in which I feed into the Vin+ of the voltage follower op-amp. This op-amp is LMV321 and has a 5.5V max operating so I am also supplying power to this op-amp using the 4.5V from the voltage divider. Though I am now afraid there may be not enough current left to supply the LMV321 (?).

I expected to have 4.5V at the Vin+ and supply pin 5 of the LMV321 but I measure 3.22V; and at the output I was expecting 4.5V and got 2.65V. I don't quite understand what's happening here.

[mathieusan]

now that I read some more about op-amps, they have a "voltage swing" rated limit, for which they cannot output a voltage higher than the (Vcc used to supply the chip minus the "swing"). So I will have to switch to a different opamp..

[mikerj]

This is doomed to failure.  1) You can't expect to supply op-amps through high impedance supplies (yours is 11K) and have it to work, 2) You can not use an op-amp to buffer a signal at the same DC level as it's supply, there needs to be some headroom.  The solution is to use a higher voltage op-amp powered from the 9v rail (directly, not via 11k worth of resistance) which will have plenty of headroom to buffer your 4.5v level.

[mathieusan]

Understand. Thank you for your reply! Learning every day! I redesigned it using TL072 (attached). Now this package comes with 2 separate opamps, and I am not using the second side. Would I want to ground the Vin- and Vin+ of the second side to minimize any chances of noise (Vout1 supplies an audio circuit)? Or would it not matter at all if left floating?

[mathieusan]

I cleaned-up up the schematic (attached) by tying the unused opamp Vin+ to GND, and Vin- to Vout and thus minimizing current consumption of the chip at least. Not sure it will actually reduce any risk of noise but I suppose is not going to induce extra noise.

[magic]

TL071 is the simplest way of dealing with an unused half of TL072.

[dobsonr741]

Why do you need low impedance for biasing?
The 22uF capacitive load on the opamp will get you in trouble, need a series resistor.
What current difference the split rail need to provide? The opamp is limited at +/- 25mA.
The noise/PSRR will not be better than the RC splitter provides to feed the opamp - why bother with the opamp at all, if it’s indeed for biasing?

[mathieusan]

the audio circuit has a major buzz (noise) and I suspect it's from the bias voltage that has too high of an impedance. Why would the 22uF capacitor create trouble? (the one before the voltage follower or after?). Are you suggesting I remove the capacitors after the Vout? (and only keep a 100nF near where the Vbias is needed in the actual audio circuit?)

[ledtester]

Understand. Thank you for your reply! Learning every day! I redesigned it using TL072 (attached). Now this package comes with 2 separate opamps, and I am not using the second side. Would I want to ground the Vin- and Vin+ of the second side to minimize any chances of noise (Vout1 supplies an audio circuit)? Or would it not matter at all if left floating?

Guidance from TI:

https://www.ti.com/lit/an/sboa204a/sboa204a.pdf

[dobsonr741]

Noise is not b/c of the bias high impedance, unless you have a very special corner case, which I doubt. More probable explanation is your VB_pre is noisy. Where’s the 9V coming from?

Show the full schematics to give better advice. How’s the noise looking on an oscilloscope? Is it buzz or hum? Under what conditions do you observe the noise?

As far as capacitive loading of opamps - the opamp can go to oscillation. Google phase margin, and look for opamp rail splitters. Your 4.5V bias scenario happens in guitar amps, typical 2x10K || 100uF is fine for biasing. Phase margin issues aside, if you have noise on VB_pre the opamp will try to force in on the capacitor. It will fail at it, and the total outcome is worse than when you’d replace the unity gain opamp with a wire.

[mathieusan]

when playing with the device ("normal operation"), the 9VDC power supply come from a ground-isolated brick: "Fender Engine Room LVL12 Power Supply". I power several devices with this power brick, some being my design, some being commercials, and I found no buzz with any other devices except that one design in particular.

I put some slides together to explain the circuit (to share with someone at work but he didn't had time to review yet).
https://www.dropbox.com/scl/fi/6e1iap52g4ousu5oahgr6/Charlie-Brown-pedal-noise.pptx?dl=0&rlkey=smcvic55g0n5b3dnvuur76o0s

Slide 8 and also Slide 10 you can see my current power section (picture on the right) with the Vbias from a simple voltage divider. Slide 11 is a schematic of the device (not exactly my implementation but representative for this discussion) where you can see where the Vbias is used (called BIAS on this)

[mathieusan]

One possible additional culprit is the opamp LM833 used on the audio circuit calls for a power supply of at least 10V and I am using 9V. But this LM833 op-amp is used for this design in this application (from other builders) without the buzz issue. I noticed their design uses a voltage follower to supply the Vbias.